Base_JSON/prefixO/json/O17/O17-2001.json

{
    "paper_id": "O17-2001",
    "header": {
        "generated_with": "S2ORC 1.0.0",
        "date_generated": "2023-01-19T07:59:29.389793Z"
    },
    "title": "An Empirical Comparison of Contemporary Unsupervised Approaches for Extractive Speech Summarization",
    "authors": [
        {
            "first": "\u5289\u58eb\u5f18",
            "middle": [
                "\uf02a"
            ],
            "last": "\u3001\u9673\u51a0\u5b87",
            "suffix": "",
            "affiliation": {},
            "email": ""
        },
        {
            "first": "\uf02a",
            "middle": [],
            "last": "\u3001\u65bd\u51f1\u6587",
            "suffix": "",
            "affiliation": {},
            "email": ""
        },
        {
            "first": "\uf02b",
            "middle": [],
            "last": "\u3001\u9673\u67cf\u7433",
            "suffix": "",
            "affiliation": {},
            "email": ""
        },
        {
            "first": "\uf02b",
            "middle": [],
            "last": "\u3001\u738b\u65b0\u6c11",
            "suffix": "",
            "affiliation": {},
            "email": ""
        },
        {
            "first": "\uf02a",
            "middle": [],
            "last": "\u3001\u8a31\u805e\u5ec9",
            "suffix": "",
            "affiliation": {},
            "email": ""
        },
        {
            "first": "Shih-Hung",
            "middle": [],
            "last": "Liu",
            "suffix": "",
            "affiliation": {},
            "email": ""
        },
        {
            "first": "Kuan-Yu",
            "middle": [],
            "last": "Chen",
            "suffix": "",
            "affiliation": {},
            "email": "[email protected]"
        },
        {
            "first": "Kai-Wun",
            "middle": [],
            "last": "Shih",
            "suffix": "",
            "affiliation": {},
            "email": ""
        },
        {
            "first": "Berlin",
            "middle": [],
            "last": "Chen",
            "suffix": "",
            "affiliation": {},
            "email": "[email protected]"
        },
        {
            "first": "Hsin-Min",
            "middle": [],
            "last": "Wang",
            "suffix": "",
            "affiliation": {},
            "email": ""
        },
        {
            "first": "Wen-Lian",
            "middle": [],
            "last": "Hsu",
            "suffix": "",
            "affiliation": {},
            "email": "[email protected]"
        },
        {
            "first": "\uf02a",
            "middle": [],
            "last": "\u4e2d\u592e\u7814\u7a76\u9662\u8cc7\u8a0a\u6240",
            "suffix": "",
            "affiliation": {},
            "email": ""
        }
    ],
    "year": "",
    "venue": null,
    "identifiers": {},
    "abstract": "Due to the rapid-developed Internet and with the big data era coming, the automatic summarization research has been emerged a popular research topic. The aim of automatic summarization is in attempt to select important text or spoken sentence to represent the topic (theme) of original text or spoken document according to a predefined summarization ratio. In this study we frame automatic summarizaiton task as an ad-hoc information retrieval (IR) problem and employ the mathematical sound language modeling (LM) framework for extractive speech summarization, which can perform important sentence selection in an unsupervised manner and has shown its preliminary success. The main contribution of this paper is threefold. First, by the virtue of relevance modeling, we explore several effective sentence modeling formulations to enhance the sentence models involved in the LM-based summarization framework and the first use of tri-mixture model to improve the performance of extractive speech summarization. Second, since the language modeling will suffer from data sparseness problem and the common solution is to adopt smoothing techniques, in this research we investigate three different smoothing approaches to evaluate how they influence the summarization performance. Third, we further apply the well-studied ranking model (BM25) and also its variants in IR community for ranking important sentence in extractive speech summarization. Experiments conducted on public avaiable dataset (MATBN) and the results show that our applied methods have effective summarization performance when compared to the other well-practiced and state-of-the-art unsupervised methods.",
    "pdf_parse": {
        "paper_id": "O17-2001",
        "_pdf_hash": "",
        "abstract": [
            {
                "text": "Due to the rapid-developed Internet and with the big data era coming, the automatic summarization research has been emerged a popular research topic. The aim of automatic summarization is in attempt to select important text or spoken sentence to represent the topic (theme) of original text or spoken document according to a predefined summarization ratio. In this study we frame automatic summarizaiton task as an ad-hoc information retrieval (IR) problem and employ the mathematical sound language modeling (LM) framework for extractive speech summarization, which can perform important sentence selection in an unsupervised manner and has shown its preliminary success. The main contribution of this paper is threefold. First, by the virtue of relevance modeling, we explore several effective sentence modeling formulations to enhance the sentence models involved in the LM-based summarization framework and the first use of tri-mixture model to improve the performance of extractive speech summarization. Second, since the language modeling will suffer from data sparseness problem and the common solution is to adopt smoothing techniques, in this research we investigate three different smoothing approaches to evaluate how they influence the summarization performance. Third, we further apply the well-studied ranking model (BM25) and also its variants in IR community for ranking important sentence in extractive speech summarization. Experiments conducted on public avaiable dataset (MATBN) and the results show that our applied methods have effective summarization performance when compared to the other well-practiced and state-of-the-art unsupervised methods.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Abstract",
                "sec_num": null
            }
        ],
        "body_text": [
            {
                "text": "\u96a8\u8457\u5927\u8cc7\u6599\u6642\u4ee3\u7684\u4f86\u81e8\uff0c\u773e\u591a\u7684\u6587\u5b57\u53ca\u591a\u5a92\u9ad4\u5f71\u97f3\u8cc7\u8a0a\u88ab\u5feb\u901f\u5730\u50b3\u905e\u4e26\u5206\u4eab\u65bc\u5168\u7403\u5404\u5730\uff0c \u8cc7\u8a0a\u8d85\u8f09(Information Overload)\u7684\u554f\u984c\u4e5f\u96a8\u4e4b\u7522\u751f\u3002\u5982\u4f55\u80fd\u8b93\u4eba\u5011\u5feb\u901f\u4e14\u6709\u6548\uf961\u5730\u700f\u89bd\u8207 \u65e5\u4ff1\u589e\u7684\u6587\u5b57\u8cc7\u8a0a\u6216\u591a\u5a92\u9ad4\u5f71\u97f3\u8cc7\u8a0a\uff0c\u5df2\u6210\u70ba\u4e00\u500b\u523b\u4e0d\u5bb9\u7de9\u7684\u7814\u7a76\u8ab2\u984c\u3002\u5728\u773e\u591a\u7684\u7814\u7a76 \u7576\u4ee3\u975e\u76e3\u7763\u5f0f\u65b9\u6cd5\u4e4b\u6bd4\u8f03\u65bc\u7bc0\u9304\u5f0f\u8a9e\u97f3\u6458\u8981 3 \u65b9\u6cd5\u4e2d\uff0c\u81ea\u52d5\u6458\u8981(Automatic Summarization)\u88ab\u8996\u70ba\u662f\u4e00\u9805\u4e0d\u53ef\u6216\u7f3a\u7684\u95dc\u9375\u6280\u8853 (Lin & Chen, 2010; Liu & Hakkani-Tur, 2011 ) \u3002 \u81ea \u52d5 \u6458 \u8981 \u4e4b \u76ee \u7684 \u5728 \u65bc \u64f7 \u53d6 \u55ae \u4e00 \u6587 \u4ef6 (Single-Document)\u6216\u591a\u91cd\u6587\u4ef6(Multi-Document)\u4e2d\u7684\u91cd\u8981\u8a9e\u610f\u8207\u4e3b\u984c\u8cc7\u8a0a\uff0c\u85c9\u6b64\u8b93\u4f7f\u7528\u8005 \u80fd\u66f4\u6709\u6548\u7387\u5730\u700f\u89bd\u8207\u7406\u89e3\u6587\u4ef6\u7684\u4e3b\u65e8\uff0c\u4ee5\uf965\u5feb\u901f\u5730\u7372\u5f97\u5176\u6240\u9700\u7684\u8cc7\u8a0a\uff0c\u907f\u514d\u82b1\u8cbb\u5927\u91cf\u6642 \u9593\u5728\u5be9\u8996\u6587\u4ef6\u5167\u5bb9\u3002\u53e6\u4e00\u65b9\u9762\uff0c\u8a9e\u97f3\u662f\u591a\u5a92\u9ad4\u6587\u4ef6\u4e2d\u6700\u5177\u8cc7\u8a0a\u7684\u6210\u5206\u4e4b\u4e00\uff1b\u5982\u4f55\u900f\u904e\u8a9e \u97f3(\u6587\u4ef6)\u6458\u8981\u6280\u8853\u4f86\u81ea\u52d5\u5730\u3001\u6709\u6548\uf961\u5730\u8655\u7406\u5177\u6642\u5e8f\u6027\u7684\u591a\u5a92\u9ad4\u5f71\u97f3\u5167\u5bb9\uff0c\u4f8b\u5982\uff1a\u96fb\u8996\u65b0 \u805e\u3001\u5ee3\u64ad\u65b0\u805e\u3001\u90f5\u4ef6\u3001\u96fb\u5b50\u90f5\u4ef6\u3001\u6703\u8b70\u53ca\u6f14\u8b1b\uf93f\u97f3\u7b49 (Ostendorf, 2008; Nenkova & McKeown, 2011 (Mani & Maybury, 1999 )\uff1a",
                "cite_spans": [
                    {
                        "start": 202,
                        "end": 220,
                        "text": "(Lin & Chen, 2010;",
                        "ref_id": "BIBREF20"
                    },
                    {
                        "start": 221,
                        "end": 244,
                        "text": "Liu & Hakkani-Tur, 2011",
                        "ref_id": "BIBREF24"
                    },
                    {
                        "start": 478,
                        "end": 495,
                        "text": "(Ostendorf, 2008;",
                        "ref_id": "BIBREF37"
                    },
                    {
                        "start": 496,
                        "end": 519,
                        "text": "Nenkova & McKeown, 2011",
                        "ref_id": "BIBREF36"
                    },
                    {
                        "start": 520,
                        "end": 541,
                        "text": "(Mani & Maybury, 1999",
                        "ref_id": "BIBREF32"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "\u7dd2\u8ad6 (Introduction)",
                "sec_num": "1."
            },
            {
                "text": "(1) \u4f86\u6e90\uff1a\u6839\u64da\u6587\u4ef6\u4f86\u6e90\uff0c\u53ef\u4ee5\u5206\u70ba\u55ae\u4e00\u6587\u4ef6\u6458\u8981\u8207\u591a\u91cd\u6587\u4ef6\u6458\u8981 (Cai & Li, 2013) \uff1b \u55ae\u4e00\u6587\u4ef6\u6458\u8981\u662f\u4f9d\u64da\u4e8b\u5148\u5b9a\u7fa9\u597d\u7684\u6458\u8981\u6bd4\u4f8b\uff0c\u9078\u53d6\u80fd\u5920\u4ee3\u8868\u6587\u4ef6\u7684\u53e5\u5b50\u7576\u4f5c\u6458\u8981\uff1b\u800c\u591a\u91cd \u6587\u4ef6\u6458\u8981\u662f\u6536\u96c6\u591a\u7bc7\u76f8\u4f3c\u7684\u6587\u4ef6\uff0c\u9700\u8981\u79fb\u9664\u6587\u4ef6\u9593\u5f7c\u6b64\u5197\u9918\u6027(Redundancy) \u7684\u8cc7\u8a0a (Carbonell & Goldstein, 1998 )\uff0c\u8003\u616e\u6587\u4ef6\u63cf\u8ff0\u4e8b\u4ef6\u767c\u751f\u7684\u5148\u5f8c\u9806\u5e8f(Causality) (Kuo & Chen, 2006 )\uff0c\u4e26\u4e14\u78ba\u8a8d\u6587\u4ef6\u4e4b\u9593\u7684\u56e0\u679c\u95dc\u4fc2\uff0c\u7d93\u7531\u9019\u4e9b\u8cc7\u8a0a\u5e0c\u671b\u80fd\u7522\u751f\u6709\u9023\u8cab\u6027\u7684\u6587\u4ef6\u6458\u8981\u3002",
                "cite_spans": [
                    {
                        "start": 32,
                        "end": 48,
                        "text": "(Cai & Li, 2013)",
                        "ref_id": "BIBREF2"
                    },
                    {
                        "start": 135,
                        "end": 163,
                        "text": "(Carbonell & Goldstein, 1998",
                        "ref_id": "BIBREF3"
                    },
                    {
                        "start": 193,
                        "end": 210,
                        "text": "(Kuo & Chen, 2006",
                        "ref_id": "BIBREF16"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "\u7dd2\u8ad6 (Introduction)",
                "sec_num": "1."
            },
            {
                "text": "(2) \u9700\u6c42\uff1a\u4f9d\u64da\u4f7f\u7528\u8005\u9700\u6c42\u4e0d\u540c\uff0c\u6458\u8981\u5167\u5bb9\u53ef\u5340\u5206\u70ba\u5177\u6709\u8cc7\u8a0a\u6027(Informative)\u3001\u6307\u793a\u6027 (Indicative)\u3001\u4ee5\u53ca\u8a55\uf941\u6027(Critical)\u3002\u5177\u6709\u8cc7\u8a0a\u6027\u7684\u6458\u8981\u662f\u7528\u4f86\u8868\u9054\u6587\u4ef6\u63cf\u8ff0\u7684\u4e3b\u65e8\u5167\u5bb9\u8207\u6838 \u5fc3\u8cc7\u8a0a\uff1b\u5177\u6307\u793a\u6027\u7684\u6458\u8981\u662f\u5e0c\u671b\u5c07\u6587\u4ef6\u4e2d\u7684\u4e3b\u984c\u5167\u5bb9\u505a\u7c21\u55ae\u7684\u63cf\u8ff0\uff0c\u4e26\u5c07\u6587\u4ef6\u5206\u6210\u4e0d\u540c\u7684 \u4e3b\u984c\uff0c\u4f8b\u5982\uff1a\u653f\u6cbb\u6027\u3001\u5b78\u8853\u6027\u3001\u9ad4\u80b2\u6027\u548c\u5a1b\u6a02\u6027\u6587\u4ef6\uff0c\u56e0\u6b64\u6240\u7522\u751f\u7684\u6458\u8981\u4e0d\u8981\u6c42\u50b3\u9054\u8a73\u7d30 \u7684\u539f\u59cb\u6587\u4ef6\u5167\u5bb9\uff1b\u5177\u8a55\uf941\u6027\u7684\u6458\u8981\u63d0\u4f9b\u6587\u4ef6\u6b63\u9762\u8207\u53cd\u9762\u7684\u89c0\u9ede(Positive and Negative Sentiments) (Galley, McKeown, Hirschberg & Shriberg, 2004 )\u3002",
                "cite_spans": [
                    {
                        "start": 244,
                        "end": 289,
                        "text": "(Galley, McKeown, Hirschberg & Shriberg, 2004",
                        "ref_id": "BIBREF9"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "\u7dd2\u8ad6 (Introduction)",
                "sec_num": "1."
            },
            {
                "text": "(3) \u65b9\u5f0f\uff1a\u53ef\u6982\u5206\u70ba\u4e8c\u5927\u985e\uff0c\u7bc0\uf93f\u5f0f(Extractive)\u6458\u8981\u8207\u62bd\u8c61\u5f0f(Abstractive)\u6458\u8981(\u6216\u91cd\u5beb \u5f0f\u6458\u8981) \u3002\u524d\u8005\u4e3b\u8981\u662f\u4f9d\u64da\u7279\u5b9a\u7684\u6458\u8981\u6bd4\u4f8b\uff0c\u5f9e\u6700\u539f\u59cb\u7684\u6587\u4ef6\u4e2d\u9078\u53d6\u91cd\u8981\u7684\u8a9e\u53e5\u4f86\u7d44\u6210\u6458\u8981\uff1b \u800c\u5f8c\u8005\u662f\u5728\u5b8c\u5168\u7406\u89e3\u6587\u4ef6\u5167\u5bb9\u4e4b\u5f8c\uff0c\u91cd\u65b0\u64b0\u5beb\u7522\u751f\u6458\u8981\u4f86\u4ee3\u8868\u539f\u59cb\u6587\u4ef6\u7684\u5167\u5bb9\uff0c\u5176\u6240\u4f7f\u7528 \u4e4b\u8a9e\u5f59\u6216\u6163\u7528\u8a9e\u4e0d\u4e00\u5b9a\u662f\u5168\u7136\u5730\u4f86\u81ea\u65bc\u539f\u59cb\u6587\u4ef6\uff0c\u6b64\u7a2e\u6458\u8981\u65b9\u5f0f\u662f\u6700\u70ba\u8cbc\u8fd1\u4eba\u5011\u65e5\u5e38\u64b0\u5beb \u6458\u8981\u7684\u5f62\u5f0f\u3002\u7136\u800c\u62bd\u8c61\u5f0f\u6458\u8981\u9700\u8981\u8907\u96dc\u7684\u81ea\u7136\u8a9e\u8a00\u8655\u7406(Natural Language Processing, NLP) \u6280\u8853\uff0c\u5982\u8cc7\u8a0a\u64f7\u53d6(Information Extraction)\u3001\u5c0d\u8a71\u7406\u89e3(Discourse Understanding)\u53ca\u81ea\u7136\u8a9e\u8a00 \u5289\u58eb\u5f18 \u7b49 \u751f\u6210(Natural Language Generation)\u7b49 (Paice, 1990; Witbrock & Mittal, 1999) \uff0c\u56e0\u6b64\uff0c\u8fd1\uf98e\u4f86 \u7bc0\uf93f\u5f0f\u6458\u8981\u4e4b\u7814\u7a76\u4ecd\u70ba\u4e3b\u6d41\u3002 (4) \u7528\u9014\uff1a\u4f9d\u6458\u8981\u7528\u9014\u53ef\u5206\u70ba\u4e00\u822c\u6027(Generic)\u6458\u8981\u8207\u4ee5\u67e5\u8a62\u70ba\u57fa\u790e(Query-focused)\u7684\u6458 \u8981\u3002\u524d\u8005\u662f\u5f9e\u6574\u7bc7\u6587\u4ef6\u4e2d\u8403\u53d6\u51fa\u80fd\u5920\u7a81\u986f\u6574\u7bc7\u6587\u4ef6\u5168\u9762\u6027\u4e3b\u984c\u8cc7\u8a0a\u7684\u8a9e\u53e5\uff0c\u671f\u671b\u6458\u8981\u7522\u751f \u7684\u5167\u5bb9\u53ef\u4ee5\u6db5\u84cb\u6574\u7bc7\u6587\u4ef6\u6240\u6709\u91cd\u8981\u7684\u4e3b\u984c\uff1b\u5f8c\u8005\u900f\u904e\u4f7f\u7528\u8005\u6216\u7279\u5b9a\u7684\u67e5\u8a62\u4f86\u7522\u751f\u8207\u67e5\u8a62\u76f8 \u95dc\u7684\u6458\u8981\u3002 (5) \u6a21\u578b\u6280\u8853\uff1a\u7c21\u55ae\u5206\u6210\u4e09\u5927\u985e\uff0c(i)\u4ee5\u7c21\u55ae\u7684\u8a9e\u5f59(Lexical)\u8207\u7d50\u69cb(Structural)\u7279\u5fb5\u505a\u70ba \u5224\u65b7\u6458\u8981\u8a9e\u53e5\u7684\u6a21\u578b\u6280\u8853 (Zhang, Chan & Fung, 2010 )\uff0c(ii)\u76e3\u7763\u5f0f\u6a5f\u5668\u5b78\u7fd2(Supervised Machine Learning) \u4ee5 \u53ca (iii) \u975e \u76e3 \u7763 \u5f0f \u6a5f \u5668 \u5b78 \u7fd2 (Unsupervised Machine Learning) (Liu & Hakkani-Tur, 2011 )\u4e4b\u6a21\u578b\u6280\u8853\u3002\u96d6\u7136\u975e\u76e3\u7763\u5f0f\u6a5f\u5668\u5b78\u7fd2\u7684\u65b9\u6cd5\u5728\u4e00\u822c\u7684\u60c5\u6cc1\u4e0b\u5176\u6548\u80fd\u6c92\u6709 \u76e3\u7763\u5f0f\u6a5f\u5668\u5b78\u7fd2\u65b9\u6cd5\u4f86\u7684\u597d\uff0c\u4f46\u975e\u76e3\u7763\u5f0f\u6a5f\u5668\u5b78\u7fd2\u65b9\u6cd5\u4e0d\u9700\u8981\u4e8b\u5148\u6e96\u5099\u5927\u91cf\u4eba\u5de5\u6a19\u8a18\u7684\u8a13 \u7df4\u8cc7\u6599\uff0c\u4ee5\u53ca\u5177\u6709\u5bb9\u6613\u5be6\u4f5c(Easy-to-Implement)\u7684\u7279\u6027\uff0c\u4ecd\u5438\u5f15\u8a31\u591a\u5b78\u8005\u9032\u884c\u7814\u7a76\u8207\u767c\u5c55\uff0c \u672c\u8ad6\u6587\u4e3b\u8981\u4e5f\u662f\u63a2\u8a0e\u4e14\u6bd4\u8f03\u975e\u76e3\u7763\u5f0f\u6a5f\u5668\u5b78\u7fd2\u7684\u65b9\u5f0f\u65bc\u81ea\u52d5\u6458\u8981\u4e4b\u4efb\u52d9\u3002 \u7d9c\u89c0\u4e0a\u8ff0\u5404\u500b\u9762\u5411\uff0c\u672c\u8ad6\u6587\u4e3b\u8981\u63a2\u7a76\u4e00\u822c\u6027\u3001\u55ae\u4e00\u6587\u4ef6\u7bc0\uf93f\u5f0f\u8a9e\u97f3\u6458\u8981\u554f\u984c\uff0c\u4e26\u6bd4 \u8f03\u5404\u5f0f\u975e\u76e3\u7763\u5f0f\u6a5f\u5668\u5b78\u7fd2\u6a21\u578b\u6280\u8853\u3002\u8fd1\uf98e\u4f86\uff0c\u5404\u5f0f\u57fa\u65bc\u8a9e\u8a00\u6a21\u578b\u4e4b\u975e\u76e3\u7763\u5f0f\u6a21\u578b\u6280\u8853\u904b \u7528\u5728\u8cc7\u8a0a\u6aa2\u7d22\u9818\u57df\u4e2d\u5df2\u5448\u73fe\u5353\u8d8a\u7684\u7814\u7a76\u6210\u679c (Zhai, 2008) \uff0c\u9019\u4e9b\u6280\u8853\u4e5f\u521d\u6b65\u5730\u88ab\u61c9\u7528\u65bc\u8a9e \u97f3\u6587\u4ef6\u6458\u8981\u4e4b\u7814\u7a76\u4e0a (Lin, Yeh & Chen, 2011) \uff0c\u4ea6\u7372\u5f97\u4e00\u5b9a\u7684\u6458\u8981\u6210\u6548\u3002\u672c\u8ad6\u6587\u5c07\u5ef6\u7e8c\u6b64 \u4e00\u7814\u7a76\u4e3b\u8ef8\uff0c\u63d0\u51fa\u4e09\u500b\u4e3b\u8981\u7684\u7814\u7a76\u8ca2\u737b\u3002\u9996\u5148\uff0c\u6709\u9451\u65bc\u95dc\u806f\u6027(Relevance)\u8cc7\u8a0a\u7684\u6982\u5ff5\u5df2 \u88ab\u61c9\u7528\u65bc\u8cc7\u8a0a\u6aa2\u7d22\u9818\u57df\u4e4b\u4e2d (Zhai & Lafferty, 2001a; Lavrenko & Croft, 2001 ",
                "cite_spans": [
                    {
                        "start": 345,
                        "end": 358,
                        "text": "(Paice, 1990;",
                        "ref_id": "BIBREF38"
                    },
                    {
                        "start": 359,
                        "end": 383,
                        "text": "Witbrock & Mittal, 1999)",
                        "ref_id": "BIBREF46"
                    },
                    {
                        "start": 613,
                        "end": 638,
                        "text": "(Zhang, Chan & Fung, 2010",
                        "ref_id": "BIBREF51"
                    },
                    {
                        "start": 740,
                        "end": 764,
                        "text": "(Liu & Hakkani-Tur, 2011",
                        "ref_id": "BIBREF24"
                    },
                    {
                        "start": 1018,
                        "end": 1030,
                        "text": "(Zhai, 2008)",
                        "ref_id": "BIBREF49"
                    },
                    {
                        "start": 1056,
                        "end": 1079,
                        "text": "(Lin, Yeh & Chen, 2011)",
                        "ref_id": "BIBREF22"
                    },
                    {
                        "start": 1158,
                        "end": 1182,
                        "text": "(Zhai & Lafferty, 2001a;",
                        "ref_id": "BIBREF47"
                    },
                    {
                        "start": 1183,
                        "end": 1205,
                        "text": "Lavrenko & Croft, 2001",
                        "ref_id": "BIBREF18"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "\u7dd2\u8ad6 (Introduction)",
                "sec_num": "1."
            },
            {
                "text": "\u672c\u8ad6\u6587\u5c07\u904e\u53bb\u6458\u8981\u7814\u7a76\u6240\u9678\u7e8c\u767c\u5c55\u51fa\u7684\u81ea\u52d5\u6458\u8981\u6a21\u578b\u6280\u8853\u5927\u7565\u5730\u6b78\u7d0d\u6210\u4e09\u5927\u985e (Mani & Maybury, 1999 )\uff1a",
                "cite_spans": [
                    {
                        "start": 35,
                        "end": 56,
                        "text": "(Mani & Maybury, 1999",
                        "ref_id": "BIBREF32"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "\u81ea\u52d5\u6458\u8981\u6a21\u578b\u6280\u8853 (Techniques of Automatic Summarization)",
                "sec_num": "2."
            },
            {
                "text": "(1) \u4ee5\u7c21\u55ae\u8a5e\u5f59\u8207\u7d50\u69cb\u7279\u5fb5\u70ba\u57fa\u790e\u4e4b\u81ea\u52d5\u6458\u8981\u6a21\u578b\u6280\u8853\uff1a\u5728 1950 \u5e74\u4ee3\uff0c\u6709\u5b78\u8005\u63d0\u51fa\u4f7f\u7528\u8a5e \u983b(Frequency)\u4f86\u8a55\u91cf\u6bcf\u4e00\u500b\u8a5e\u7684\u91cd\u8981\u6027\u8207\u8a08\u7b97\u6587\u4ef6\u4e2d\u6bcf\u4e00\u500b\u8a9e\u53e5\u7684\u986f\u8457\u6027(Significance Factor) (Luhn, 1958) (Strzalkowski, Wand & Wise, 1998 )\u3001\u4fee\u8fad\u7d50\u69cb(Rhetorical Structure) (Zhang et al., 2010) \u7b49\u3002\u53e6\u6709\u5b78\u8005\u5728\u5be9\u8996 200 \u7bc7\u79d1\u6280\u6587\u4ef6\u5f8c\uff0c\u767c\u73fe\u6709 85%\u7684\u91cd\u8981\u8a9e \u53e5\u51fa\u73fe\u5728\u6587\u4ef6\u4e2d\u7684\u7b2c\u4e00\u6bb5\uff0c7%\u7684\u91cd\u8981\u8a9e\u53e5\u51fa\u73fe\u5728\u6700\u5f8c\u4e00\u6bb5 (Baxendale, 1958) \u3002\u56e0\u6b64\uff0c\u63d0\u51fa \u4e86\u8a9e\u53e5\u5728\u6587\u4ef6\u4e2d\u7684\u4f4d\u7f6e(Position)\u8cc7\u8a0a\u662f\u9032\u884c\u6458\u8981\u8a9e\u53e5\u9078\u53d6\u6642\u7684\u4e00\u9805\u95dc\u9375\u7dda\u7d22\u3002",
                "cite_spans": [
                    {
                        "start": 106,
                        "end": 118,
                        "text": "(Luhn, 1958)",
                        "ref_id": "BIBREF27"
                    },
                    {
                        "start": 119,
                        "end": 151,
                        "text": "(Strzalkowski, Wand & Wise, 1998",
                        "ref_id": "BIBREF43"
                    },
                    {
                        "start": 181,
                        "end": 201,
                        "text": "(Zhang et al., 2010)",
                        "ref_id": "BIBREF51"
                    },
                    {
                        "start": 262,
                        "end": 279,
                        "text": "(Baxendale, 1958)",
                        "ref_id": "BIBREF1"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "5",
                "sec_num": null
            },
            {
                "text": "(2) \u4ee5\u975e\u76e3\u7763\u5f0f\u6a5f\u5668\u5b78\u7fd2\u70ba\u57fa\u790e\u4e4b\u81ea\u52d5\u6458\u8981\u6a21\u578b\u6280\u8853\uff1a\u975e\u76e3\u7763\u5f0f\u6a5f\u5668\u5b78\u7fd2\u901a\u5e38\u5c07\u81ea\u52d5\u6458\u8981 \u4efb\u52d9\u8996\u70ba\u5982\u4f55\u6392\u5e8f\u4e26\u6311\u9078\u5177\u4ee3\u8868\u6027\u8a9e\u53e5\u4e4b\u554f\u984c\uff0c\u5176\u65b9\u6cd5\u901a\u5e38\u662f\u8a08\u7b97\u51fa\u4e00\u7a2e\u6458\u8981\u7279\u5fb5\u4f9b\u8a9e\u53e5 \u6392\u5e8f\u4f7f\u7528\uff0c\u5e38\u898b\u7684\u7279\u5fb5\u6709\uff1a\u8a9e\u53e5\u8207\u6587\u4ef6\u76f8\u95dc\u6027 (Gong & Liu, 2001 )\u3001\u8a9e\u53e5\u6240\u5f62\u6210\u7684\u8a9e\u8a00\u6a21\u578b \u751f\u6210\u6587\u4ef6\u4e4b\u6a5f\uf961\u7b49 (Chen, Chen & Wang, 2009) \u3001\u8a9e\u53e5\u9593\u4e4b\u76f8\u95dc\u6027 (Erkan & Radev, 2004; Mihalcea & Tarau, 2004; Wan & Yang, 2008) \u3001\u6216\u8a9e\u53e5\u8207\u6587\u4ef6\u5728\u6f5b\u85cf\u4e3b\u984c\u7a7a\u9593\u4e2d\u7684\u8ddd\u96e2\u95dc\u4fc2 (Lin & Chen, 2009 )\u7b49\u3002",
                "cite_spans": [
                    {
                        "start": 103,
                        "end": 120,
                        "text": "(Gong & Liu, 2001",
                        "ref_id": "BIBREF10"
                    },
                    {
                        "start": 143,
                        "end": 168,
                        "text": "(Chen, Chen & Wang, 2009)",
                        "ref_id": "BIBREF4"
                    },
                    {
                        "start": 178,
                        "end": 199,
                        "text": "(Erkan & Radev, 2004;",
                        "ref_id": "BIBREF8"
                    },
                    {
                        "start": 200,
                        "end": 223,
                        "text": "Mihalcea & Tarau, 2004;",
                        "ref_id": "BIBREF34"
                    },
                    {
                        "start": 224,
                        "end": 241,
                        "text": "Wan & Yang, 2008)",
                        "ref_id": "BIBREF44"
                    },
                    {
                        "start": 263,
                        "end": 280,
                        "text": "(Lin & Chen, 2009",
                        "ref_id": "BIBREF19"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "5",
                "sec_num": null
            },
            {
                "text": "(3) \u4ee5\u76e3\u7763\u5f0f\u6a5f\u5668\u5b78\u7fd2\u70ba\u57fa\u790e\u4e4b\u81ea\u52d5\u6458\u8981\u6a21\u578b\u6280\u8853\uff1a\u76e3\u7763\u5f0f\u6a5f\u5668\u5b78\u7fd2\u901a\u5e38\u5c07\u81ea\u52d5\u6458\u8981\u4e4b\u4efb \u52d9\u8996\u70ba\u4e8c\u5143\u5206\u985e\u554f\u984c(Binary Classification)\uff0c\u4ea6\u5373\u5c07\u8a9e\u53e5\u5340\u5206\u70ba\u6458\u8981\u8a9e\u53e5\u6216\u975e\u6458\u8981\u8a9e\u53e5\u3002\u6211 \u5011\u5fc5\u9808\u4e8b\u5148\u6e96\u5099\u597d\u4e00\u4e9b\u8a13\u7df4\u6587\u4ef6\u4ee5\u53ca\u5176\u5c0d\u61c9\u7684\u4eba\u5de5\u6a19\u8a3b\u6458\u8981\u8cc7\u8a0a\uff0c\u7136\u5f8c\u900f\u904e\u5404\u7a2e\u5206\u985e\u5668\u7684 \u5b78\u7fd2\u6a5f\u5236\uff0c\u9032\u884c\u5206\u985e\u6a21\u578b\u7684\u8a13\u7df4\u3002\u5c0d\u65bc\u5c1a\u672a\u88ab\u6458\u8981\u4e4b\u6587\u4ef6\uff0c\u6b64\u985e\u65b9\u6cd5\u5c07\u6587\u4ef6\u88e1\u7684\u6bcf\u500b\u8a9e\u53e5 \u9032\u884c\u4e8c\u5143\u5206\u985e\uff0c\u5373\u53ef\u4f9d\u5176\u7d50\u679c\u7522\u751f\u51fa\u6458\u8981\u3002\u6b64\u985e\u65b9\u6cd5\u8f03\u8457\u540d\u7684\u76f8\u95dc\u7814\u7a76\u5305\u62ec\u7c21\u55ae\u8c9d\u6c0f\u5206\u985e \u5668(Na\u00efve-Bayes Classifier) (Kupiec, Pedersen & Chen, 1995) \u3001\u9ad8\u65af\u6df7\u5408\u6a21\u578b(Gaussian Mixture Model, GMM) (Murray, Renals & Carletta, 2005 )\u3001\u96b1\u85cf\u5f0f\u99ac\u53ef\u592b\u6a21\u578b(Hidden Markov Model, HMM) (Conroy & O'Leary, 2001 ) \u3001\u652f\u6301\u5411\u91cf\u6a5f(Support Vector Machines, SVM) (Kolcz, Prabakarmurthi & Kalita, 2001; Zhang & Fung, 2007) \u8207\u689d\u4ef6\u96a8\u6a5f\u5834\u57df(Conditional Random Fields, CRF) (Shen, Sun, Li, Yang & Chen, 2007) \u7b49\u3002\u76e3\u7763\u5f0f\u6a21\u578b\u53ef\u540c\u6642\u7d50\u5408\u591a\u7a2e\u6458\u8981\u7279\u5fb5\u4f86 \u8868\u793a\u6bcf\u4e00\u8a9e\u53e5(\u901a\u5e38\u662f\u7531\u4e0a\u8ff0\u4ee5\u8a5e\u5f59\u6216\u7d50\u69cb\u70ba\u57fa\u790e\u4e4b\u6458\u8981\u65b9\u6cd5\u3001\u6216\u662f\u5404\u5f0f\u975e\u76e3\u7763\u5f0f\u6458\u8981\u6a21\u578b \u91dd\u5c0d\u8a9e\u53e5\u6240\u8f38\u51fa\u7684\u5206\u6578\u6216\u6a5f\uf961\u503c) \uff0c\u7d9c\u5408\u5404\u7a2e\u6458\u8981\u7279\u5fb5\u6240\u5f62\u6210\u7684\u7279\u5fb5\u5411\u91cf\u5c07\u88ab\u7528\u4f86\u505a\u70ba\u76e3\u7763 \u5f0f\u6458\u8981\u6a21\u578b\u5224\u65b7\u8a9e\u53e5\u662f\u5426\u5c6c\u65bc\u6458\u8981\u8a9e\u53e5\u7684\u4f9d\u64da (Lin & Chen, 2009 (Chen et al., 2009) (Zhai & Lafferty, 2001a; Lavrenko & Croft, 2001; Hiemstra et al., 2004) \uff0c\u4f46\u5728\u6458\u8981\u4efb\u52d9\u4e2d\u537b\u662f\u76f8\u5c0d \u8f03\u5c11\u7814\u7a76\u7684\uff0c\u503c\u5f97\u4e00\u63d0\u7684\u662f\uff0c\u96d6\u7136\u95dc\u806f\u6a21\u578b\u3001\u7c21\u55ae\u6df7\u5408\u6a21\u578b\u5df2\u521d\u6b65\u88ab\u61c9\u7528\u5728\u6458\u8981\u4efb\u52d9\u4e0a (Chen, Chang & Chen, 2013; Liu et al., 2014) ",
                "cite_spans": [
                    {
                        "start": 242,
                        "end": 273,
                        "text": "(Kupiec, Pedersen & Chen, 1995)",
                        "ref_id": "BIBREF17"
                    },
                    {
                        "start": 311,
                        "end": 343,
                        "text": "(Murray, Renals & Carletta, 2005",
                        "ref_id": "BIBREF35"
                    },
                    {
                        "start": 381,
                        "end": 404,
                        "text": "(Conroy & O'Leary, 2001",
                        "ref_id": "BIBREF7"
                    },
                    {
                        "start": 444,
                        "end": 482,
                        "text": "(Kolcz, Prabakarmurthi & Kalita, 2001;",
                        "ref_id": "BIBREF15"
                    },
                    {
                        "start": 483,
                        "end": 502,
                        "text": "Zhang & Fung, 2007)",
                        "ref_id": "BIBREF50"
                    },
                    {
                        "start": 543,
                        "end": 577,
                        "text": "(Shen, Sun, Li, Yang & Chen, 2007)",
                        "ref_id": "BIBREF42"
                    },
                    {
                        "start": 702,
                        "end": 719,
                        "text": "(Lin & Chen, 2009",
                        "ref_id": "BIBREF19"
                    },
                    {
                        "start": 720,
                        "end": 739,
                        "text": "(Chen et al., 2009)",
                        "ref_id": "BIBREF4"
                    },
                    {
                        "start": 740,
                        "end": 764,
                        "text": "(Zhai & Lafferty, 2001a;",
                        "ref_id": "BIBREF47"
                    },
                    {
                        "start": 765,
                        "end": 788,
                        "text": "Lavrenko & Croft, 2001;",
                        "ref_id": "BIBREF18"
                    },
                    {
                        "start": 789,
                        "end": 811,
                        "text": "Hiemstra et al., 2004)",
                        "ref_id": "BIBREF11"
                    },
                    {
                        "start": 864,
                        "end": 890,
                        "text": "(Chen, Chang & Chen, 2013;",
                        "ref_id": "BIBREF5"
                    },
                    {
                        "start": 891,
                        "end": 908,
                        "text": "Liu et al., 2014)",
                        "ref_id": "BIBREF25"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "5",
                "sec_num": null
            },
            {
                "text": "Theorem)\u5c07 P(S|D)\u5c55\u958b\u6210()\uff1a ) ( ) ( ) | ( ) | ( D P S P S D P D S P \uf03d (1) \u7576\u4ee3\u975e\u76e3\u7763\u5f0f\u65b9\u6cd5\u4e4b\u6bd4\u8f03\u65bc\u7bc0\u9304\u5f0f\u8a9e\u97f3\u6458\u8981 7 \u5176\u4e2d ) (D P \u662f\u6587\u4ef6 D \u7684\u4e8b\u524d\u6a5f\uf961\uff0c\u7531\u65bc ) (D P \u4e0d\u5f71\u97ff\u8a9e\u53e5\u7684\u6392\u5e8f\u7d50\u679c\uff0c\u6545\u53ef\u7701\u7565\u4e0d\u8a0e\u8ad6\uff1b \u53e6\u4e00\u65b9\u9762\uff0c ) (S P \u662f\u8a9e\u53e5 S \u7684\u4e8b\u524d\u6a5f\u7387\uff0c\u53ef\u4ee5\u4f7f\u7528\u5404\u5f0f\u975e\u76e3\u7763\u5f0f\u65b9\u6cd5\u6216\u76e3\u7763\u5f0f\u65b9\u6cd5\u4f86\u6c42 \u5f97",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "5",
                "sec_num": null
            },
            {
                "text": "EQUATION",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [
                    {
                        "start": 0,
                        "end": 8,
                        "text": "EQUATION",
                        "ref_id": "EQREF",
                        "raw_str": "\u3002\u672c\u8ad6\u6587\u7684\u7814\u7a76\u5047\u8a2d\u8a9e\u53e5\u7684\u4e8b\u524d\u6a5f\uf961\u70ba\u4e00\u500b\u5747\u52fb\u5206\u5e03(Uniform Distribution)\uff0c\u6240\u4ee5 ) (S P \u4ea6\u53ef\u7701\u7565\u3002\u6700\u5f8c\uff0c ) | ( S D P \u662f\u8a9e\u53e5 S \u6240\u5f62\u6210\u7684\u8a9e\u8a00\u6a21\u578b\u751f\u6210\u6587\u4ef6 D \u4e4b\u6a5f\uf961(\u6216\u7a31\u4f5c\u6587\u4ef6\u76f8\u4f3c\u5ea6)\uff0c\u53ef\u4ee5\u7528\u4f86\u8868\u793a\u6587\u4ef6 D \u8207\u8a9e\u53e5 S \u4e4b\u9593\u7684\u76f8\u4f3c\u95dc\u4fc2\uff0c\u5982\u679c\u8a9e \u53e5 S \u751f\u6210\u6587\u4ef6 D \u7684\u6a5f\u7387\u503c\u6108\u9ad8\uff0c\u4ee3\u8868\u8a9e\u53e5 S \u8207\u6587\u4ef6 D \u6108\u70ba\u76f8\u4f3c(\u8a9e\u53e5\u6108\u80fd\u4ee3\u8868\u6587\u4ef6 D)\uff0c \u5373\u6108\u6709\u53ef\u80fd\u662f\u6458\u8981\u8a9e\u53e5\u3002\u6211\u5011\u53ef\u4ee5\u66f4\u9032\u4e00\u6b65\u5730\u5047\u8a2d\u6587\u4ef6 D \u4e2d\u8a5e\u8207\u8a5e\u4e4b\u9593\u662f\u7368\uf9f7\u7684\uff0c\u4e26\u4e14 \u4e0d\u8003\u616e\u6bcf\u4e00\u500b\u8a5e\u5728\u6587\u4ef6 D \u4e2d\u767c\u751f\u7684\u9806\u5e8f\u95dc\u4fc2(\u5373\u8a5e\u888b\u5047\u8a2d(Bag-of-Word Assumption))\uff0c\u5247 \u8a9e\u53e5 S \u751f\u6210\u6587\u4ef6 D \u7684\u6587\u4ef6\u76f8\u4f3c\u5ea6\u91cf\u503c(Document Likelihood Measure, DLM) ) | ( S D P \u53ef\u62c6 \u89e3\u6210\u6587\u4ef6 D \u4e2d\u6bcf\u4e00\u7684\u8a5e w \u500b\u5225\u767c\u751f\u7684\u689d\u4ef6\u6a5f\uf961\u4e4b\u9023\u4e58\u7a4d\uff1a \uf0d5 \uf0ce \uf03d D w D w C S w P S D P ) , ( ) | ( ) | ( (2) \u6b64\u7a2e\u65b9\u6cd5\u662f\u70ba\u8a9e\u53e5 S \u5efa\uf9f7\u4e00\u500b\u8a9e\u53e5\u6a21\u578b(Sentence Model) ) | ( S w P \uff0c w \u662f\u51fa\u73fe\u5728\u6587\u4ef6 D \u4e2d\u7684\u8a5e\uff0c ) , ( D w C \u662f\u8a5e w \u51fa\u73fe\u5728\u6587\u4ef6 D \u4e2d\u7684\u6b21\u6578\u3002\u5176\u4e2d\uff0c\u6211\u5011\u53ef\u5229\u7528\u6700\u5927\u5316\u76f8\u4f3c\u5ea6\u4f30\u6e2c (Maximum Likelihood Estimation, MLE)\u7684\u65b9\u5f0f\u4f86\u5efa\u7acb\u6bcf\u4e00\u500b\u8a9e\u53e5\u7684\u8a9e\u53e5\u6a21\u578b\uff1a | | ) , ( ) | ( S S w C S w P \uf03d (3) \u5728(3)\u4e2d\uff0c ) , ( S w C \u8868\u793a\u8a5e w \u5728\u8a9e\u53e5 S \u4e2d\u51fa\u73fe\u7684\u6b21\u6578\uff0c S \u5247\u8868\u793a\u8a9e\u53e5 S \u7684\u7e3d\u8a5e\u6578\u3002\u503c\u5f97\u6ce8 \u610f\u7684\u662f\uff0c\u7531\u65bc\u8a9e\u53e5 S \u901a\u5e38\u50c5\u7531\u5c11\u6578\u5b57\u8a5e\u6240\u7d44\u6210\uff0c\u56e0\u6b64\u5bb9\u6613\u906d\u9047\u8cc7\u6599\u7a00\u758f(Data Sparseness) \u7684\u554f\u984c\uff0c\u9019\u6703\u4f7f\u5f97\u8a9e\u53e5\u6a21\u578b\u4f7f\u7528\u6700\u5927\u5316\u76f8\u4f3c\u5ea6\u4f30\u6e2c\u6642\uff0c\u4e0d\u50c5\u53ef\u80fd\u7121\u6cd5\u6e96\u78ba\u5730\u4f30\u6e2c\u6bcf\u4e00\u500b \u8a5e\u5728\u8a9e\u53e5\u4e2d\u771f\u6b63\u7684\u6a5f\uf961\u5206\u4f48\uff0c\u4e5f\u53ef\u80fd\u56e0\u70ba\u67d0\u4e9b\u8a5e\u7684\u689d\u4ef6\u6a5f\u7387\u503c\u70ba\u96f6\uff0c\u5c0e\u81f4\u8a9e\u53e5 S \u7522\u751f\u6587 \u4ef6 D \u7684\u6a5f\uf961\u503c\u70ba\u96f6\u3002\u70ba\u4e86\u6e1b\u8f15\u4e0a\u8ff0\u7684\u73fe\u8c61\uff0c\u53ef\u63a1\u7528\u5e73\u6ed1\u5316(Smoothing)\u6280\u8853\u4f86\u9054\u6210\uff0c\u5e38\u898b \u7684\u5e73\u6ed1\u5316\u6280\u8853\u5305\u542b\u6709 Jelinek-Mercer \u5e73\u6ed1\u5316\u3001Dirichlet \u5e73\u6ed1\u5316\u3001Add-delta \u5e73\u6ed1\u5316 (Zhai & Lafferty, 2001b)\uff0c\u672c\uf941\u6587\u4f7f\u7528 Jelinek-Mercer \u5e73\u6ed1\u5316\u6280\u8853\u85c9\u7531\u4f7f\u7528\u4ee5\u5927\u91cf\u6587\u5b57\u8a9e\u6599\u8a13\u7df4\u800c \u6210\u7684\u80cc\u666f\u55ae\u9023\u8a9e\u8a00\u6a21\u578b(Background Unigram Language Model)\u4f86\u8abf\u9069\u8a9e\u53e5\u6a21\u578b(Zhai & Lafferty, 2001b)\uff0c\u6545 ) | ( S D P \u53ef\u9032\u4e00\u6b65\u5730\u8868\u793a\u6210\uff1a \uf0d5 \uf0ce \uf0d7 \uf02d \uf02b \uf0d7 \uf03d D w D w C B w P S w P S D P ) , ( )] | ( ) 1 ( ) | ( [ ) | ( \uf06c \uf06c (4) \u5176\u4e2d\uff0c ) | ( B w P \u662f\u8a5e w \u5728\u80cc\u666f\u55ae\u9023\u8a9e\u8a00\u6a21\u578b B \u4e2d\u4e4b\u6a5f\uf961\u503c\u3002 3.2 \u865b\u64ec\u76f8\u95dc\u56de\u994b (Pseudo-Relevance Feedback) \u901a\u5e38\uff0c\u6587\u4ef6\u4e2d\u7684\u8a9e\u53e5\u50c5\u7531\u5c11\u8a31\u7684\u8a5e\u5f59\u6240\u7d44\u6210\uff0c\u7576\u8a9e\u53e5\u6a21\u578b\u4f7f\u7528\u6700\u5927\u5316\u76f8\u4f3c\u5ea6\u4f30\u6e2c\u6642\uff0c\u5bb9 \u6613\u906d\u9047\u8cc7\u6599\u7a00\u758f\u7684\u554f\u984c\uff1b\u518d\u8005\uff0c\u7531\u9019\u8a9e\u53e5 S \u4e2d\u4e9b\u8a31\u7684\u8868\u9762\u8a5e\u5f59\u662f\u9060\u4e0d\u5920\u6b63\u78ba\u4f30\u7b97\u8a9e\u53e5 S \u8207\u88ab\u6458\u8981\u6587\u4ef6 D \u4e4b\u9593\u7684\u76f8\u4f3c\u5ea6(\u6216\u4f4e\u4f30\u4e86\u6b64\u76f8\u4f3c\u5ea6)\uff0c\u6240\u4ee5\u85c9\u7531\u80cc\u666f\u8a9e\u8a00\u6a21\u578b\u9032\u884c\u8a9e\u53e5\u6a21 \u578b\u4e4b\u8abf\u9069\u70ba\u6700\u5e38\u898b\u7684\u65b9\u6cd5\u4e4b\u4e00(\u53c3\u7167\u5f0f(4))\u3002 \u5289\u58eb\u5f18 \u7b49 \u70ba\u4e86\u6709\u6548\u89e3\u6c7a\u8a9e\u53e5\u7684\u8cc7\u6599\u7a00\u758f\u53ca\u76f8\u4f3c\u5ea6\u88ab\u4f4e\u4f30\u7684\u554f\u984c\uff0c\u6211\u5011\u53ef\u5229\u7528\u5728\u8cc7\u8a0a\u6aa2\u7d22 (Information Retrieval)\u9818\u57df\u88ab\u5ee3\u6cdb\u61c9\u7528\u7684\u865b\u64ec\u76f8\u95dc\u56de\u994b(Pseudo Relevant Feedback, PRF) \u6280\u8853\u4f86\u5f37\u5316\u8a9e\u53e5\u6a21\u578b(\u91cd\u65b0\u4f30\u6e2c\u6216\u5c0d\u5176\u505a\u8abf\u9069)(Chen, Chen, Chen. Wang & Yu, 2014)\u3002\u70ba \u6b64\u76ee\u7684\uff0c\u7576\u865b\u64ec\u76f8\u95dc\u56de\u994b\u904b\u7528\u65bc\u6587\u4ef6\u6458\u8981\u9818\u57df\u4e2d\u6642\uff0c\u6703\u5c07\u6bcf\u4e00\u8a9e\u53e5 S \u7576\u6210\u662f\u4e00\u500b\u67e5\u8a62 (Query)\uff0c\u7136\u5f8c\u8f38\u5165\u5230\u4e00\u500b\u8cc7\u8a0a\u6aa2\u7d22\u7cfb\u7d71\u4e2d\uff0c\u627e\u51fa\u4e00\u4e9b\u8207\u8a9e\u53e5 \u6700\u53ef\u80fd\u76f8\u95dc\u7684\u6587\u4ef6\uff0c\u800c\u9019 \u4e9b\u6587\u4ef6\u5c31\u7a31\u4e4b\u70ba\u865b\u64ec\u76f8\u95dc\u6587\u4ef6(Pseudo Relevant Documents)\uff1b\u4e00\u500b\u6700\u7c21\u55ae\u7684\u65b9\u5f0f\u5373\u662f\u9078\u53d6 \u6392\u540d\u6700\u524d\u9762(\u6aa2\u7d22\u5206\u6578\u6700\u9ad8)\u7684\u5e7e\u7bc7\u6587\u4ef6(Top-ranked Documents)\u3002\u6709\u4e86\u9019\u4e9b\u865b\u64ec\u76f8\u95dc\u6587\u4ef6 \u5f8c\uff0c\u5c31\u53ef\u4ee5\u5229\u7528\u5b83\u5011\u4f86\u589e\u9032\u8a9e\u53e5\u6a21\u578b\u4ee5\u89e3\u6c7a\u8a9e\u53e5\u8cc7\u6599\u7a00\u758f\u53ca\u5176\u76f8\u4f3c\u5ea6\u4f4e\u4f30\u4e4b\u554f\u984c\uff0c\u5176\u865b \u64ec\u95dc\u806f\u56de\u994b\u793a\u610f\u5716\u5982\u5716 1 \u6240\u793a\u3002\u6240\u4ee5\u672c\u8ad6\u6587\u91dd\u5c0d\u8a9e\u53e5\u6a21\u578b\u8abf\u9069\u9032\u884c\u521d\u6b65\u7814\u7a76\uff0c\u7576\u6211\u5011\u900f \u904e\u8cc7\u8a0a\u6aa2\u7d22\u7cfb\u7d71\u5df2\u53d6\u5f97\u865b\u64ec\u76f8\u95dc\u6587\u4ef6(\u6700\u9ad8\u6392\u5e8f\u6587\u4ef6) \uff0c\u63a5\u8457\u5c31\u8981\u505a\u8a9e\u53e5\u6a21\u578b\u7684\u8abf\u9069\u4f30\u6e2c\uff0c \u5e95\u4e0b\u4ecb\u7d39\u5e38\u898b\u7684\u8abf\u9069\u6a21\u578b\u5305\u542b\u6709\u95dc\u806f\u6a21\u578b (Relevance Model, RM)\u3001\u7c21\u55ae\u6df7\u5408\u6a21\u578b(Simple Mixture Model, SMM)\u4ee5\u53ca\u4e09\u6df7\u5408\u6a21\u578b(Tri-Mixture Model, TriMM)\u3002 \u5716 1\u3001\u865b\u64ec\u95dc\u806f\u56de\u994b\u793a\u610f\u5716 [Figure 1. Illustration of pseudo-relevance feedback.] 3.2.1 \u95dc\u806f\u6a21\u578b (Relevance Model, RM) \u95dc\u806f\u6a21\u578b\u7684\u57fa\u672c\u5047\u8a2d\u662f\u8a8d\u70ba\u6bcf\u4e00\u8a9e\u53e5 S \u7686\u662f\u88ab\u7528\u4f86\u63cf\u8ff0\u4e00\u500b\u6982\u5ff5\u3001\u60f3\u6cd5\u6216\u4e3b\u984c\uff0c\u6211\u5011\u7a31 \u4e4b\u70ba\u8a9e\u53e5\u7684\u95dc\u806f\u985e\u5225(Relevance Class)\u3002\u5728\u672c\u8ad6\u6587\u4e2d\uff0c\u6211\u5011\u7684\u76ee\u6a19\u662f\u60f3\u9032\u4e00\u6b65\u5730\u6a21\u578b\u5316\u95dc \u806f\u985e\u5225\u6240\u4ee3\u8868\u7684\u8cc7\u8a0a\uff0c\u85c9\u6b64\u4f86\u8c50\u5bcc\u8a9e\u53e5\u6a21\u578b\u6240\u80fd\u50b3\u9054\u7684\u8a9e\u610f\u5167\u5bb9\u6216\u4e3b\u984c\u7279\u6027\u3002\u7136\u800c\uff0c\u5be6 \u969b\u4e0a\u6bcf\u4e00\u8a9e\u53e5 \u7684\u95dc\u806f\u985e\u5225 \u662f\u975e\u5e38\u96e3\u4ee5\u6c42\u5f97\u7684\uff1b\u70ba\u6b64\uff0c\u6211\u5011\u900f\u904e\u865b\u64ec\u76f8\u95dc\u56de\u994b(Pseudo Relevant Feedback, PRF)\u4f86\u5c0b\u627e\u8207\u95dc\u806f\u985e\u5225\u53ef\u80fd\u76f8\u95dc\u7684\u4e00\u4e9b\u6587\u4ef6\uff0c\u4e26\u85c9\u7531\u9019\u4e9b\u6587\u4ef6\u4f86\u8fd1\u4f3c \u95dc\u806f\u985e\u5225\u3002\u66f4\u660e\u78ba\u5730\uff0c\u5728\u5be6\u4f5c\u4e0a\u6211\u5011\u5c07\u865b\u64ec\u76f8\u95dc\u6587\u4ef6(\u6700\u9ad8\u6392\u5e8f\u6587\u4ef6)D Top ={D 1 ,D 2 ,\u2026,D M } \u7528\u4ee5\u4ee3\u8868\u95dc\u806f\u985e\u5225 \u3002\u63a5\u8457\uff0c\u900f\u904e\u6aa2\u8996\u8a5e\u5f59 w \u8207\u8a9e\u53e5 S \u5728\u9019\u4e9b\u865b\u64ec\u76f8\u95dc\u6587\u4ef6\u4e2d\u540c\u6642\u51fa\u73fe \u4e4b\u95dc\u4fc2\uff0c\u53ef\u8a08\u7b97\u51fa\u8a5e\u5f59\u8207\u8a9e\u53e5\u7684\u806f\u5408\u6a5f\u7387(Lavrenko & Croft, 2001)\uff1a \u7576\u4ee3\u975e\u76e3\u7763\u5f0f\u65b9\u6cd5\u4e4b\u6bd4\u8f03\u65bc\u7bc0\u9304\u5f0f\u8a9e\u97f3\u6458\u8981 9 , ) ( ) | , ( ) , ( \uf0e5 \uf0ce \uf03d Top D RM m D m m D P D S w P S w P (5) \u7576\u6211\u5011\u9032\u4e00\u6b65\u5730\u5047\u8a2d\u5728\u7d66\u5b9a\u67d0\u4e00\u7bc7\u865b\u64ec\u76f8\u95dc\u6587\u4ef6\u6642\uff0c\u8a5e\u5f59\u8207\u8a9e\u53e5\u662f\u7368\u7acb\u7684\uff0c\u4e26\u4e14\u8a9e\u53e5\u5167 \u7684\u8a5e\u5f59\u4e5f\u662f\u7368\u7acb\u4e14\u4e0d\u8003\u616e\u5176\u5148\u5f8c\u6b21\u5e8f(\u5373\u6240\u8b02\u7684\u8a5e\u888b\u5047\u8a2d)\uff0c\u5247\u900f\u904e\u865b\u64ec\u76f8\u95dc\u56de\u994b\u6240\u4f30\u6e2c \u7684\u8a9e\u53e5\u6a21\u578b\u70ba\uff1a , ) ( ) | ' ( ) ( ) | ( ) | ' ( ) | ( ' ' ' D ' ' D RM Top ' Top ' m D S w m m D m S w m D P D w P D P D w P D w P S w P m m \uf0e5 \uf0d5 \uf0e5 \uf0d5 \uf0ce \uf0ce \uf0ce \uf0ce \uf03d",
                        "eq_num": "(6"
                    }
                ],
                "section": "5",
                "sec_num": null
            },
            {
                "text": "(Zhai & Lafferty, 2001a)\uff1a )], | ( ) | ( ) 1 log[( ) , ( BG w P S w P D w c LL m D V w m \uf0d7 \uf02b \uf0d7 \uf02d \uf0d7 \uf03d \uf0e5 \uf0e5 \uf0ce \uf0ce \uf061 \uf061 SMM D D Top Top (7) \u5176\u4e2d\u03b1\u70ba\u5e73\u8861\u53c3\u6578\uff0c\u7528\u4f86\u63a7\u5236\u6a21\u578b\u4f30\u6e2c\u6642\u662f\u8981\u6bd4\u8f03\u504f\u597d\u7c21\u55ae\u6df7\u5408\u6a21\u578b\u6216\u662f\u80cc\u666f\u8a9e\u8a00\u6a21\u578b\uff0c c(w,D m )\u70ba\u8a5e\u5f59 w \u5728\u865b\u64ec\u76f8\u95dc\u6587\u4ef6 D m \u7684\u6b21\u6578\uff0c\u5f0f(7)\u7684\u6700\u5927\u5316\u53ef\u900f\u904e\u671f\u671b\u503c\u6700\u5927\u5316\u8fed\u4ee3\u66f4 \u65b0\u5f0f\u4f86\u9054\u6210\uff1a \u671f\u671b\u503c\u6b65\u9a5f\uff1a , ) | ( ) 1 ( ) | ( ) | ( ) ( SMM ) ( SMM ) ( BG w P S w P S w P l l l w \uf0d7 \uf02d \uf02b \uf0d7 \uf0d7 \uf03d \uf061 \uf061 \uf061 \uf074 (8) 10 \u5289\u58eb\u5f18 \u7b49 \u6700\u5927\u5316\u6b65\u9a5f\uff1a , ) , ( ) , ( ) | ( ) ( ) ( ) 1 ( \uf0e5 \uf0e5 \uf0e5 \uf0ce \uf0a2 \uf0ce \uf0a2 \uf0ce \uf02b \uf0d7 \uf0a2 \uf0a2 \uf0d7 \uf03d V w D l w m D l w m l m m D w c D w c S w P Top Top D D SMM \uf074 \uf074 (9) \u5176\u4e2d l \u8868\u793a\u671f\u671b\u503c\u6700\u5927\u5316\u7684\u7b2c l \u6b21\u8fed\u4ee3\u3002\u9019\u500b\u7c21\u55ae\u6df7\u5408\u6a21\u578b\u7684\u4f30\u6e2c\u6703\u52a0\u5f37\u5177\u6709\u7368\u7279\u6027 (Specificity)\u7684\u8a5e\u5f59\u4e4b\u6a5f\u7387\uff0c\u4f8b\u5982\u67d0\u8a5e\u5f59\u6c92\u6709\u5728\u80cc\u666f\u8a9e\u8a00\u6a21\u578b\u4e2d\u6709\u597d\u89e3\u91cb(Well-Explained) \u5247\u6703\u88ab\u52a0\u5f37\u5176\u6a5f\u7387\uff0c\u9019\u6a23\u4f7f\u5f97\u6b64\u6a21\u578b\u70ba\u66f4\u5177\u6709\u9451\u5225(Discriminant)\u80fd\u529b\u7684\u8a9e\u53e5\u6a21\u578b\uff1b\u53cd\u4e4b\uff0c \u82e5\u662f\u6c92\u6709\u7368\u7279\u6027\u7684\u8a5e\u5f59\uff0c\u5247\u5176\u6a5f\u7387\u5c31\u6703\u88ab\u80cc\u666f\u8a9e\u8a00\u6a21\u578b\u6240\u5438\u6536\u3002 3.2.3 \u4e09\u6df7\u5408\u6a21\u578b (Tri-Mixture Model) \u53e6\u4e00\u65b9\u9762\uff0c\u672c\u8ad6\u6587\u5617\u8a66\u5c07\u4e09\u6df7\u5408\u6a21\u578b(Tri-Mixture Model)(Hiemstra, Robertson & Zaragoza, 2004)\u7528\u65bc\u8a9e\u97f3\u6458\u8981\u4efb\u52d9\u3002\u4e09\u6df7\u5408\u6a21\u578b\u53ef\u8996\u70ba\u662f\u8907\u96dc\u5316\u5f8c\u7684\u7c21\u55ae\u6df7\u5408\u6a21\u578b\uff1b\u5b83\u66f4\u9032\u4e00\u6b65\u7684 \u5047\u8a2d\u865b\u64ec\u76f8\u95dc\u6587\u4ef6 D Top \u88e1\u7684\u8a5e\u5f59 w \u662f\u6e90\u81ea\u65bc\u4e09\u7a2e\u6210\u5206\u6a21\u578b(Component Models)\uff0c\u5176\u4e00\u70ba \u6587\u4ef6\u6a21\u578b P(w|D m )\uff0c\u5176\u4e8c\u70ba\u4e09\u6df7\u5408\u6a21\u578b P TriMM (w|S)\uff0c\u6700\u5f8c\u70ba\u80cc\u666f\u8a9e\u8a00\u6a21\u578b P(w|BG)\u3002\u4e09\u6df7 \u5408\u6a21\u578b\u7684\u4f30\u6e2c\u4e5f\u662f\u85c9\u7531\u671f\u671b\u503c\u6700\u5927\u5316\u6f14\u7b97\u6cd5\u4f86\u6700\u5927\u5316\u865b\u64ec\u76f8\u95dc\u6587\u4ef6\u7684\u5c0d\u6578\u76f8\u4f3c\u5ea6\u4ee5\u9032\u884c \u6a21\u578b\u7684\u4f30\u6e2c\uff0c\u5176\u865b\u64ec\u76f8\u95dc\u6587\u4ef6\u7684\u5c0d\u6578\u76f8\u4f3c\u5ea6\u7684\u5b9a\u7fa9\u5982\u4e0b(Hiemstra et al., 2004)\uff1a )], | ( ) | ( ) | ( ) 1 log[( ) , ( BG w P D w P S w P D w c LL m D V w m m \uf0d7 \uf02b \uf0d7 \uf02b \uf0d7 \uf02d \uf02d \uf0d7 \uf03d \uf0e5 \uf0e5 \uf0ce \uf0ce \uf06d \uf06c \uf06d \uf06c TriMM D D Top Top (10) \u5176\u4e2d\u03bb\u548c\u03bc\u70ba\u5e73\u8861\u53c3\u6578\uff0c\u7528\u4f86\u63a7\u5236\u6a21\u578b\u4f30\u6e2c\u6642\u662f\u8981\u6bd4\u8f03\u504f\u597d\u4e09\u6df7\u5408\u6a21\u578b\u6216\u6587\u4ef6\u6a21\u578b\u4ea6\u6216 \u662f\u80cc\u666f\u8a9e\u8a00\u6a21\u578b\uff0cc(w,D m )\u70ba\u8a5e\u5f59 w \u5728\u865b\u64ec\u76f8\u95dc\u6587\u4ef6 D m \u7684\u6b21\u6578\uff0c\u5f0f(10)\u7684\u6700\u5927\u5316\u53ef\u900f\u904e \u671f\u671b\u503c\u6700\u5927\u5316\u8fed\u4ee3\u66f4\u65b0\u5f0f\u4f86\u9054\u6210\uff1a \u671f\u671b\u503c\u6b65\u9a5f\uff1a , ) | ( ) | ( ) | ( ) 1 ( ) | ( ) , ( ) | ( ) | ( ) | ( ) 1 ( ) | ( ) 1 ( ) , ( TriMM , TriMM TriMM , \uf0ef \uf0ef \uf0ee \uf0ef \uf0ef \uf0ed \uf0ec \uf0d7 \uf02b \uf0d7 \uf02b \uf0d7 \uf02d \uf02d \uf0d7 \uf0d7 \uf03d \uf0d7 \uf02b \uf0d7 \uf02b \uf0d7 \uf02d \uf02d \uf0d7 \uf02d \uf02d \uf0d7 \uf03d m m r D w m m D w D w P BG w P S w P D w P D w c e D w P BG w P S w P S w P D w c r m m \uf06c \uf06d \uf06d \uf06c \uf06c \uf06c \uf06d \uf06d \uf06c \uf06d \uf06c (11) \u6700\u5927\u5316\u6b65\u9a5f\uff1a , ) | ( ) | ( , , , , \uf0ef \uf0ef \uf0ee \uf0ef \uf0ef \uf0ed \uf0ec \uf03d \uf03d \uf0e5 \uf0e5 \uf0e5 \uf0ce w D w D w m D w D w m D w m m m m e e D w P r r S w P Top D TriMM (12) \u7576\u4ee3\u975e\u76e3\u7763\u5f0f\u65b9\u6cd5\u4e4b\u6bd4\u8f03\u65bc\u7bc0\u9304\u5f0f\u8a9e\u97f3\u6458\u8981 11 \u904b\u7528\u6b64\u4e09\u6df7\u5408\u6a21\u578b\u4f86\u8abf\u9069\u8a9e\u53e5\u6a21\u578b\u6642\uff0c\u53ef\u53d6\u4ee3\u539f\u672c\u7684\u8a9e\u53e5\u6a21\u578b\u6216\u8207\u4e4b\u7dda\u6027\u7d50\u5408(linearly interpolation)\uff1a ), | ( ) 1 ( ) | ( ) | ( S w P S w P S w P TriMM \uf0d7 \uf02d \uf02b \uf0d7 \uf03d \uf067 \uf067 (13) \u5176\u4e2d 1 0 \uf03c \uf0a3 \uf067 \uff0c\u7576 0 \uf03d \uf067 \u4ee3\u8868\u4f7f\u7528\u4e09\u6df7\u5408\u6a21\u578b\u53d6\u4ee3\u539f\u672c\u7684\u8a9e\u53e5\u6a21\u578b\u3002 \u95dc\u806f\u6a21\u578b\u3001\u7c21\u55ae\u6df7\u5408\u6a21\u578b\u53ca\u4e09\u6df7\u5408\u6a21\u578b\u5728\u8cc7\u8a0a\u6aa2\u7d22\u9818\u57df\u4e2d\u5df2\u88ab\u5ee3\u6cdb\u61c9\u7528",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "5",
                "sec_num": null
            },
            {
                "text": "\u5176\u4e2d\uff0cc(w,B)\u70ba w \u5728\u6587\u4ef6 D \u4e2d\u7684\u51fa\u73fe\u6b21\u6578\uff0cB \u70ba\u80cc\u666f\u8cc7\u8a0a\u4e2d\u6240\u6709\u6587\u4ef6\u6578\u76ee\uff0cn(w)\u70ba w \u5728 \u80cc\u666f\u8cc7\u8a0a\u4e2d\u51fa\u73fe\u7684\u6587\u4ef6\u6578\u76ee\uff0c|S|\u70ba\u8a9e\u53e5\u9577\u5ea6\uff0cavgsl \u70ba\u6587\u4ef6 D \u4e2d\u8a9e\u53e5\u7684\u5e73\u5747\u9577\u5ea6\uff0ck 1 \u3001k 2 \u548c b \u70ba\u53ef\u8abf\u7684\u6a21\u578b\u53c3\u6578\u3002 BM25 \u662f\u4e00\u500b\u878d\u5408\u8a9e\u53e5\u7684\u8a5e\u983b\u8cc7\u8a0a\u3001\u6587\u4ef6\u76f8\u4f3c\u5ea6\u4ee5\u53ca\u53cd\u6587\u4ef6\u983b\u51fd\u6578\u4e4b\u6392\u5e8f\u8a08\u7b97\u516c\u5f0f\u3002 \u5728 BM25 \u7684\u8a08\u7b97\u516c\u5f0f\u4e2d\uff0c\u5b57\u8a5e\u51fa\u73fe\u5728\u6587\u4ef6 D \u7684\u983b\u7387\u8cc7\u8a0a\u6703\u7d93\u7531\u6b0a\u91cd\u51fd\u6578 F(w,D)\u9032\u884c\u9069\u7576 \uf0e5 \uf0ce \uf0d7 \uf0d7 \uf03d S w B w IDF S w Sim D w F B D S BM ) , ( ) , ( ) , ( ) , , ( 25 ) , ( ) 1 )( , ( ) , ( 2 2 k D w c k D w c D w F \uf02b \uf02b \uf03d ) | | 1 ( ) , ( ) 1 )( , ( ) , ( 1 1 avgsl S b b k S w c k S w c S w Sim \uf02b \uf02d \uf02b \uf02b \uf03d 5 . 0 ) ( 5 . 0 ) ( log ) , ( \uf02b \uf02b \uf02d \uf03d w n w n B B w IDF 12 \u5289\u58eb\u5f18 \u7b49 \u7684\u8abf\u6574\uff1a\u7576\u53c3\u6578 k 2 \u8a2d\u5b9a\u70ba 0 \u6642\uff0c\u5247\u8868\u793a BM25 \u50c5\u8003\u616e\u5b57\u8a5e\u662f\u5426\u6709\u51fa\u73fe\u65bc\u6587\u4ef6\u7576\u4e2d\uff0c\u800c\u4e0d \u8003\u616e\u5176\u51fa\u73fe\u7684\u983b\u7387\uff0c\u82e5\u53c3\u6578 k 2 \u7684\u8a2d\u5b9a\u4e0d\u70ba 0\uff0cBM25 \u5c07\u4e0d\u50c5\u8003\u616e\u5b57\u8a5e\u7684\u51fa\u73fe\u8207\u5426\uff0c\u4e26\u4e14 \u9032\u4e00\u6b65\u5730\u5c07\u5b57\u8a5e\u65bc\u6587\u4ef6\u4e2d\u51fa\u73fe\u7684\u983b\u7387\u8cc7\u8a0a\u505a\u9069\u7576\u7684\u52a0\u6b0a\uff1b\u6587\u4ef6\u76f8\u4f3c\u5ea6 Sim(w,S)\u5247\u7528\u65bc\u8a08 \u7b97\u5019\u9078\u6587\u4ef6\u4e2d\u8207\u67e5\u8a62\u5171\u540c\u51fa\u73fe\u7684\u8a5e\u5f59\u65bc\u6587\u4ef6\u4e2d\u7684\u91cd\u8981\u6027\uff0c\u67e5\u8a62\u7684\u8a5e\u5f59\u5728\u5019\u9078\u6587\u4ef6\u4e2d\u4ea6\u626e \u6f14\u8209\u8db3\u8f15\u91cd\u7684\u89d2\u8272\uff0c\u82e5\u67e5\u8a62\u7684\u8a5e\u5f59\u5171\u540c\u51fa\u73fe\u8f03\u591a\u6b21\u4e14\u53c3\u6578 k 1 \u7684\u8a2d\u5b9a\u4e0d\u70ba 0\uff0c\u5247\u8868\u793a\u6b64\u7bc7 \u5019\u9078\u6587\u4ef6\u61c9\u88ab\u8ce6\u4e88\u8f03\u9ad8\u7684\u6392\u5e8f\u5206\u6578\uff1b\u53cd\u6587\u4ef6\u983b\u51fd\u6578 IDF(w,B)\u662f\u7528\u65bc\u6c7a\u5b9a\u6bcf\u4e00\u500b\u8a5e\u5f59\u7684\u91cd \u8981\u6027\uff0c\u4e5f\u5c31\u662f\u52a0\u5f37\u5167\u5bb9\u5b57\u8a5e(Content word)\u7684\u6b0a\u91cd\uff0c\u4e26\u524a\u5f31\u529f\u80fd\u5b57\u8a5e(Function word)\u7684\u8ca2 \u737b\u5ea6\u3002 \u8fd1\u5e74\u4f86\uff0c\u6709\u5b78\u8005\u5c07 BM25 \u904b\u7528\u65bc\u610f\u898b\u6458\u8981(Opinion Summarization)\u7814\u7a76\u4e2d\uff0c\u70ba\u4e86\u7b26\u5408 \u610f\u898b\u6458\u8981\u6240\u504f\u597d\u7684\u8a9e\u53e5\u7279\u6027\uff0c\u4ed6\u5011\u9032\u4e00\u6b65\u5730\u5c07 BM25 \u4fee\u6539\u70ba(Kim, Castellanos, Hsu, Zhai, Dayal & Ghosh, 2013)\uff1a \uf0e5 \uf0ce \uf0d7 \uf03d S w E E E B w IDF S w Sim B D S BM ) , ( ) , ( ) , , ( 25 (18) ) | | 1 ( ) , ( ) 1 )( , ( ) , ( 1 1 avgsl S b b k S w c k S w c S w Sim E \uf02b \uf02d \uf02b \uf02b \uf03d (19) 5 . 0 ) , ( 5 . 0 ) , ( | | log ) , ( \uf02b \uf02b \uf02d \uf03d B w c B w c B B w IDF E (20) \u5176\u4e2d\uff0cc(w,B)\u70ba w \u5728\u80cc\u666f\u8cc7\u8a0a B \u4e2d\u7684\u51fa\u73fe\u6b21\u6578\uff0c|B|\u70ba\u80cc\u666f\u8cc7\u8a0a\u6240\u6709\u5b57\u8a5e\u7684\u6b21\u6578\u3002\u6bd4\u8f03\u5f0f (14)\u8207(18)\uff0cBM25 E \u5728\u5c0d\u8a9e\u53e5\u9032\u884c\u6392\u5e8f\u6642\uff0c\u7701\u7565\u4e86\u8003\u616e\u67e5\u8a62\u8a5e\u5f59\u51fa\u73fe\u983b\u7387\u7684\u8cc7\u8a0a\uff0c\u50c5\u8003\u616e \u8a5e\u5f59\u662f\u5426\u6709\u51fa\u73fe\u65bc\u67e5\u8a62\u4e2d\uff1b\u53e6\u5916\uff0c\u5176 IDF E (w,B)\u7684\u7b97\u6cd5\u662f\u4f7f\u7528\u5b57\u8a5e w \u5728\u80cc\u666f\u8cc7\u8a0a B \u4e2d\u51fa \u73fe\u7684\u6b21\u6578\uff0c\u800c\u4e0d\u662f\u4f7f\u7528\u5b57\u8a5e w \u5728\u80cc\u666f\u8cc7\u8a0a B \u4e2d\u51fa\u73fe\u7684\u6587\u4ef6\u6578\u76ee\u3002",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "5",
                "sec_num": null
            },
            {
                "text": "\u7576\u8a9e\u53e5\u5f88\u9577\u7684\u6642\u5019\uff0c\u6587\u4ef6\u76f8\u4f3c\u5ea6 Sim(w,S)\u5728\u50b3\u7d71\u7684 BM25 \u6392\u5e8f\u516c\u5f0f(\u53c3\u7167\u5f0f 16 ",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "BM25L and BM25+",
                "sec_num": "4.2"
            },
            {
                "text": ")\u4e2d\u6703\u8b8a\u5f97 \u5f88\u5c0f\uff0c\u610f\u5373\u50b3\u7d71\u7684 BM25 \u8a08\u7b97\u516c\u5f0f\u5bb9\u6613\u504f\u597d\u77ed\u8a9e\u53e5\u3002\u6709\u9451\u65bc\u6b64\uff0c\u6709\u5b78\u8005\u63d0\u51fa\u4e00\u500b\u89e3\u6c7a\u65b9 \u6cd5\u4f86\u5e73\u8861\u8a9e\u53e5\u9577\u5ea6\u7684\u5f71\u97ff\u3002\u70ba\u4e86\u65b9\u4fbf\u89e3\u91cb\u6b64\u65b9\u6cd5\uff0c\u6211\u5011\u5c07\u5f0f(16)\u91cd\u65b0\u6539\u5beb\u5982\u4e0b\uff1a 1 1 ) , ( ' ) 1 )( , ( ' ) , ( k S w c k S w c S w Sim \uf02b \uf02b \uf03d (21) \u5176\u4e2d c'(w,S)\u70ba avgsl S b b S w c S w c | | 1 ) , ( ) , ( ' \uf02b \uf02d \uf03d (22) \u7576\u4ee3\u975e\u76e3\u7763\u5f0f\u65b9\u6cd5\u4e4b\u6bd4\u8f03\u65bc\u7bc0\u9304\u5f0f\u8a9e\u97f3\u6458\u8981 13 \u7576\u91cd\u65b0\u6539\u5beb\u70ba\u5f0f(21)\u5f8c\uff0c\u6709\u5b78\u8005\u63d0\u51fa\u4f7f\u7528\u65b0\u7684\u6587\u4ef6\u76f8\u4f3c\u5ea6 Sim'(w,S)\uff0c\u5176\u5b9a\u7fa9\u5982\u4e0b\uff1a \uf0ef \uf0ee \uf0ef \uf0ed \uf0ec \uf03d \uf03e \uf02b \uf02b \uf02b \uf02b \uf03d 0 ) , ( ' 0 0 ) , ( ' ) ) , ( ' ( ) 1 )( ) , ( ' ( ) , ( ' 1 1 S w c S w c if k S w c k S w c S w Sim \uf064 \uf064 (23) \u5176\u4e2d\u03b4\u70ba\u4e00\u5b9a\u503c\uff0c\u5247\u65b0\u7684\u6392\u5e8f\u516c\u5f0f\u70ba(BM25L)\uff1a \uf0e5 \uf0ce \uf0d7 \uf0d7 \uf03d S w B w IDF S w Sim D w F B D S L BM ) , ( ) , ( ' ) , ( ) , , ( 25 (24) \u6b64\u65b0\u7684\u6587\u4ef6\u76f8\u4f3c\u5ea6\u4e0d\u50c5\u4fdd\u7559\u539f\u6709 BM25 \u7684\u5169\u9ede\u826f\u597d\u7279\u6027\uff0c(\u5373\u7576 c'(w,S)=0 \u6642 Sim'(w,S)=0\uff1b \u53e6\u5916\uff0cc'(w,S)\u8207 BM25L \u7686\u5448\u55ae\u8abf\u905e\u589e\uff0c\u4e26\u4e14 BM25L \u6709\u6f38\u9032\u6700\u5927\u503c(Asymptotic maximal))\uff0c \u540c\u6642\u4e5f\u56e0\u6b64\u6709\u4e86\u4e00\u500b\u6b63\u4e0b\u754c(positive lower bound)\u7684\u7279\u6027(\u5373\u5c0d\u65bc c'(w,S)>0\uff0c\u81f3\u5c11\u90fd\u6703\u6709 ) /( ) 1 ( 1 1 \uf064 \uf064 \uf02b \uf02b k k )\uff0c\u6b64\u7279\u6027\u53ef\u4ee5\u5e73\u8861\u8a9e\u53e5\u9577\u5ea6\u4e4b\u5f71\u97ff\uff0c\u4e0d\u6703\u56e0\u70ba\u8a9e\u53e5\u904e\u9577\u800c\u5f71\u97ff\u8b8a\u5927 \u4e14\u4e0d\u6703\u7279\u5225\u5bb9\u6613\u504f\u597d\u77ed\u8a9e\u53e5\u3002 \u4e00\u65b9\u9762\uff0cLv & Zhai (2011a)\u767c\u73fe\u4e0d\u53ea\u539f\u59cb BM25 \u6392\u5e8f\u516c\u5f0f\u6703\u904e\u5ea6\u61f2\u7f70\u9577\u8a9e\u53e5\uff0c\u5c31\u9023 \u5176\u4ed6\u7684\u6392\u5e8f\u516c\u5f0f\u90fd\u6703\u6709\u4e00\u6a23\u7684\u60c5\u5f62\uff0c\u56e0\u6b64\u4ed6\u5011\u66f4\u9032\u4e00\u6b65\u5730\u63d0\u51fa\u4e00\u822c\u5316\u7684\u65b9\u6cd5\u4f86\u89e3\u6c7a\u6b64\u554f \u984c\uff0c\u4e5f\u5c31\u662f\u8981\u4fdd\u8b49\u53ea\u51fa\u73fe\u4e00\u6b21\u7684\u8a5e\u5f59\u5728\u9577\u8a9e\u53e5\u4e2d\u81f3\u5c11\u6703\u6709\u4e00\u5b9a\u7684\u8ca2\u737b\u5ea6\uff0c\u70ba\u4e86\u9054\u5230\u6b64\u76ee \u7684\uff0c\u4ed6\u5011\u5c31\u5728\u539f\u59cb BM25 \u516c\u5f0f\u4e2d\u7684\u6587\u4ef6\u76f8\u4f3c\u5ea6 Sim(w,S)\u88e1\u52a0\u5165\u4e00\u500b\u5e38\u6578\u503c\uff0c\u4e14\u53cd\u6587\u4ef6\u983b \u51fd\u6578 IDF(w,B)\u4e5f\u6709\u5c0f\u4fee\u6539\uff0c\u5247\u65b0\u7684\u6392\u5e8f\u516c\u5f0f\u70ba(BM25+\uff0c(Lv & Zhai, 2011b))\uff1a \uf0e5 \uf0ce \uf02b \uf02b \uf0d7 \uf0d7 \uf03d \uf02b S w B w IDF S w Sim D w F B D S BM ) , ( ) , ( ) , ( ) , , ( 25 (25) \uf064 \uf02b \uf02b \uf02d \uf02b \uf02b \uf03d \uf02b ) | | 1 ( ) , ( ) 1 )( , ( ) , ( 1 1 avgsl S b b k S w c k S w c S w Sim (26) ) ( 1 | | ) , ( w n B B w IDF \uf02b \uf03d \uf02b (27) \u5176\u4e2d\u03b4\u70ba\u4e00\u500b\u56fa\u5b9a\u503c\u3002 4.3 BM25T \u5728 4.1 \u5c0f\u7bc0\u6240\u4ecb\u7d39\u7684 BM25 \u6392\u5e8f\u516c\u5f0f\u4e2d\u6709\u4e09\u500b\u9700\u8981\u8a2d\u5b9a\u7684\u53c3\u6578(k 1 , k 2 , b)\uff0c\u4e14\u6240\u6709\u7684\u8a5e\u5f59\u5171 \u4eab\u540c\u4e00\u7d44\u8a2d\u5b9a\uff0c\u4f46\u5176\u5be6\u6bcf\u500b\u8a5e\u5f59\u61c9\u8a72\u8981\u6839\u64da\u4e0d\u540c\u7684\u91cd\u8981\u6027\u800c\u8a2d\u8a08\u4e0d\u540c\u7684\u53c3\u6578\u503c\u3002\u7531\u65bc\u6587 \u4ef6\u76f8\u4f3c\u5ea6 Sim(w,S) \u662f BM25 \u516c\u5f0f\u4e2d\u6700\u91cd\u8981\u7684\u6392\u5e8f\u56e0\u5b50\uff0c\u6240\u4ee5\u53c3\u6578 k 1 \u7684\u8a2d\u8a08\u5c31\u66f4\u986f\u91cd\u8981\u3002 Lv & Zhai (2012) \u8a8d\u70ba\u7d93\u9577\u5ea6\u6b63\u898f\u5316\u7684\u8a5e\u983b\u8ca2\u737b\u5ea6\u61c9\u8a72\u8981\u8207\u6709\u8f03\u9ad8\u7684\u9577\u5ea6\u6b63\u898f\u5316\u8a5e\u983b\u7684 \u6587\u7ae0\u6578\u6210\u6b63\u6bd4\uff0c\u56e0\u6b64\u4ed6\u5011\u4f7f\u7528\u5c0d\u6578\u908f\u8f2f (Log-logistic)\u65b9\u6cd5\u4f86\u8a08\u7b97\u6bcf\u500b\u8a5e\u5f59\u6240\u5c0d\u61c9\u4e0d\u540c\u7684 \u5289\u58eb\u5f18 \u7b49 \u53c3\u6578 k 1 \uff0c\u9996\u5148\u5b9a\u7fa9\u4e00\u500b\u83c1\u82f1\u96c6(Elite set)C w \uff0c\u610f\u5373\u6240\u6709\u5305\u542b\u8a5e\u5f59 w \u7684\u8a9e\u53e5\u96c6\u5408\uff0c\u5247\u8a5e\u5f59 w \u7684\u53c3\u6578 k 1 '\u4e4b\u5b9a\u7fa9\u5982\u4e0b\uff1a \uf028 \uf029 2 ' 1 ) ( ) 1 ) ' , ( ' log( min arg ' 1 1 \uf0f7 \uf0f7 \uf0f7 \uf0f8 \uf0f6 \uf0e7 \uf0e7 \uf0e7 \uf0e8 \uf0e6 \uf02b \uf02d \uf03d \uf0e5 \uf0ce w n S w c g k w C S k k (28) \uf0ef \uf0ee \uf0ef \uf0ed \uf0ec \uf03d \uf0b9 \uf02d \uf03d 1 1 1 ) log( 1 1 1 1 1 1 1 k k if k k k g k (29) \u5176\u4e2d c'(w,S')\u8207\u5f0f(22)\u76f8\u540c\uff0c\u6211\u5011\u5c07 k 1 \u7684\u7bc4\u570d\u8a2d\u5b9a\u5728 0.1 \u5230 10 \u4e4b\u9593(\u6bcf\u6b21\u589e\u52a0 0.1)\uff0c\u900f\u904e \u5f0f(28)\u6211\u5011\u53ef\u627e\u5230\u6bcf\u4e00\u500b\u8a5e\u5f59 w \u7684\u6700\u4f73\u53c3\u6578 k 1 ' \uff0c\u5c07\u5f0f(28)\u6240\u6c42\u5f97\u7684\u53c3\u6578\u5e36\u56de\u539f\u59cb\u7684 BM25 \u6392\u5e8f\u516c\u5f0f\uff0c\u4fbf\u53ef\u5f97\u5230\u65b0\u7684\u6392\u5e8f\u516c\u5f0f(BM25T\uff0c(Lv & Zhai, 2012))\uff1a \uf0e5 \uf0ce \uf0d7 \uf0d7 \uf03d S w T B w IDF S w Sim D w F B D S T BM ) , ( ) , ( ) , ( ) , , ( 25 (30) ) | | 1 ( ' ) , ( ) 1 ' )( , ( ) , ( 1 1 avgsl S b b k S w c k S w c S w Sim T \uf02b \uf02d \uf02b \uf02b \uf03d (31) \u8868 1. \u5be6\u9a57\u8a9e\u6599\u7d71\u8a08\u8cc7\u8a0a [",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "BM25L and BM25+",
                "sec_num": "4.2"
            },
            {
                "text": ") | ( ) 1 ( ) | ( ) | ( B w P S w P S w P JM \uf06c \uf06c \uf02d \uf02b \uf03d (32) \u5176\u4e2d\u03bb\u70ba\u7dda\u6027\u7d50\u5408\u53c3\u6578\uff0c\u5728\u5be6\u9a57\u8a2d\u5b9a\u4e2d\u662f\u5f9e 0.1 \u5230 0.9(\u6bcf\u6b21\u589e\u52a0 0.1)\u3002(ii) Dirichlet \u5e73\u6ed1 \u5316\u4e3b\u8981\u662f\u6839\u6e90\u65bc\u8c9d\u5f0f\u5e73\u6ed1(Bayesian Smoothing)\u800c\u4f86\u7684\uff0c\u5b83\u5047\u8a2d\u8a9e\u8a00\u6a21\u578b\u6709\u500b\u4e8b\u524d(Prior) \u6a5f\u7387\uff0c\u800c\u6b64\u4e8b\u524d\u6a5f\u7387\u7684\u5206\u5e03\u525b\u597d\u5c31\u662f Dirichlet \u5206\u5e03\uff0c\u56e0\u6b64 Dirichlet \u5e73\u6ed1\u5316\u516c\u5f0f\u53ef\u5b9a\u7fa9\u5982 \u4e0b(Zhai & Lafferty, 2001b)\uff1a \uf06d \uf06d \uf02b \uf0d7 \uf02b \uf03d | | ) | ( ) | ( ) | ( S B w P S w c S w P",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "BM25L and BM25+",
                "sec_num": "4.2"
            },
            {
                "text": "\u5176\u4e2d\u03bc\u70ba Dirichlet \u53c3\u6578\uff0c\u5728\u5be6\u9a57\u8a2d\u5b9a\u4e2d\u7684\u7bc4\u570d\u70ba 1 \u5230 100(\u6bcf\u6b21\u589e\u52a0 1)\u3002(iii) Add-delta \u5e73 \u6ed1\u5316\u662f\u4e00\u500b\u7c21\u55ae\u5e73\u6ed1\u5316\u6280\u8853\uff0c\u5176\u539f\u7406\u5c31\u662f\u52a0\u5165\u4e00\u9ede\u9ede\u503c\uff0c\u4f7f\u6c92\u6709\u51fa\u73fe\u904e\u7684\u8a5e\u5f59\u4e4b\u6a5f\u7387\u4e0d \u70ba\u96f6\uff0c\u5176\u516c\u5f0f\u5b9a\u7fa9\u5982\u4e0b(Lv & Zhai, 2014)\uff1a | | | | ) | ( ) | ( F Delta V S S w c S w P \uf0d7 \uf02b \uf02b \uf03d \uf064 \uf064 (34) \u5176\u4e2d\u03b4\u70ba\u53ef\u8abf\u53c3\u6578\uff0c\u5728\u5be6\u9a57\u8a2d\u5b9a\u7bc4\u570d\u70ba 0.1 \u5230 1(\u6bcf\u6b21\u589e\u52a0 0.1)\uff0c\u800c|V F |\u70ba\u865b\u64ec\u76f8\u95dc\u56de\u994b\u6587 \u4ef6(\u5728\u6b64\u70ba 15 \u7bc7)\u4e2d\u4e0d\u540c\u8a5e\u5f59\u7684\u500b\u6578\u3002\u4e09\u7a2e\u5e73\u6ed1\u5316\u6280\u8853\u65bc\u95dc\u806f\u6a21\u578b(RM)\u7684\u8a9e\u97f3(\u6587\u5b57)\u6458\u8981 \u7d50\u679c\u5982\u8868 4 \u6240\u793a\uff0c\u7121\u8ad6\u5728 TD \u6216 SD \u7684\u60c5\u6cc1\u4e0b\uff0c\u5176\u4e2d\u8868\u73fe\u6700\u4f73\u70ba Add-delta \u5e73\u6ed1\u5316\uff0c\u5176\u6b21 \u662f Dirichlet \u5e73\u6ed1\u5316\uff0c\u6700\u5dee\u7684\u662f Jelinek-Mercer \u5e73\u6ed1\u5316\u3002Add-delta \u5e73\u6ed1\u5316\u8868\u73fe\u6bd4\u8f03\u597d\u7684\u539f \u56e0\u662f\u56e0\u70ba\u5229\u7528\u5230\u76f8\u95dc\u56de\u994b\u6587\u4ef6\u4e2d\u4e0d\u540c\u8a5e\u5f59\u7684\u500b\u6578(|V F |)\u7684\u8cc7\u8a0a\uff0c\u4f7f\u4e4b\u80fd\u8b93\u5171\u540c\u51fa\u73fe\u5728\u8a9e\u53e5 \u8207\u76f8\u95dc\u56de\u994b\u6587\u4ef6\u4e2d\u7684\u8a5e\u5f59 w \u6709\u6bd4\u8f03\u9ad8\u7684\u6a5f\u7387(\u76f8\u8f03\u65bc\u6c92\u6709\u5171\u540c\u51fa\u73fe\u7684\u8a5e\u5f59)\uff0c\u56e0\u6b64\u5728\u4f30\u6e2c \u95dc\u806f\u6a21\u578b\u6642\u80fd\u66f4\u5177\u6709\u9451\u5225\u80fd\u529b(\u5340\u5206\u51fa\u91cd\u8981\u4e14\u5171\u540c\u51fa\u73fe\u5728\u8a9e\u53e5\u8207\u76f8\u95dc\u56de\u994b\u6587\u4ef6\u7684\u8a5e\u5f59\u8207 \u4e00 \u822c \u6027 \u4e14 \u4e0d \u91cd \u8981 \u7684 \u8a5e \u5f59 ) \uff0c \u800c \u8b93 \u6458 \u8981 \u6548 \u80fd \u8b8a \u5f97 \u66f4 \u597d \uff0c \u5c24 \u5176 \u662f TD \u7684 \u60c5 \u6cc1 \u4e0b \u76f8 \u8f03 \u65bc Jelinek-Mercer \u5e73\u6ed1\u5316\u5728 ROUGE-2 \u7684\u7d55\u5c0d\u9032\u6b65\u7387\u6709 5%\u4e4b\u591a\uff0c\u9019\u662f\u76f8\u7576\u986f\u8457\u7684\u3002\u4f46\u5728 SD \u7684\u60c5\u6cc1\u4e0b\uff0c\u96d6\u7136 Add-delta \u5e73\u6ed1\u5316\u9084\u662f\u6703\u6bd4 Dirichlet \u5e73\u6ed1\u5316\u53ca Jelinek-",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "BM25L and BM25+",
                "sec_num": "4.2"
            }
        ],
        "back_matter": [],
        "bib_entries": {
            "BIBREF0": {
                "ref_id": "b0",
                "title": "Using Lexical Chains for Text Summarization",
                "authors": [
                    {
                        "first": "R",
                        "middle": [],
                        "last": "Barzilay",
                        "suffix": ""
                    },
                    {
                        "first": "M",
                        "middle": [],
                        "last": "Elhadad",
                        "suffix": ""
                    }
                ],
                "year": 1997,
                "venue": "Proceedings of ACL Workshop on Intelligent Scalable Text Summarization",
                "volume": "",
                "issue": "",
                "pages": "10--17",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Barzilay, R., & Elhadad, M. (1997). Using Lexical Chains for Text Summarization. In Proceedings of ACL Workshop on Intelligent Scalable Text Summarization, 10-17.",
                "links": null
            },
            "BIBREF1": {
                "ref_id": "b1",
                "title": "Machine-made Index for Technical Literature -an Experiment",
                "authors": [
                    {
                        "first": "P",
                        "middle": [],
                        "last": "Baxendale",
                        "suffix": ""
                    }
                ],
                "year": 1958,
                "venue": "IBM Journal of Research and Development",
                "volume": "2",
                "issue": "4",
                "pages": "354--361",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Baxendale, P. (1958). Machine-made Index for Technical Literature -an Experiment. IBM Journal of Research and Development, 2(4), 354-361.",
                "links": null
            },
            "BIBREF2": {
                "ref_id": "b2",
                "title": "Ranking through Clustering: An Integrated Approach to Multi-Document Summarization",
                "authors": [
                    {
                        "first": "X.-Y",
                        "middle": [],
                        "last": "Cai",
                        "suffix": ""
                    },
                    {
                        "first": "W.-J",
                        "middle": [],
                        "last": "Li",
                        "suffix": ""
                    }
                ],
                "year": 2013,
                "venue": "IEEE Transactions on Audio, Speech and Language Processing",
                "volume": "21",
                "issue": "7",
                "pages": "1424--1433",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Cai, X.-Y., & Li, W.-J. (2013). Ranking through Clustering: An Integrated Approach to Multi-Document Summarization. IEEE Transactions on Audio, Speech and Language Processing, 21(7), 1424-1433.",
                "links": null
            },
            "BIBREF3": {
                "ref_id": "b3",
                "title": "The Use of MMR Diversity-based Reranking for Reordering Documents and Producing Summaries",
                "authors": [
                    {
                        "first": "J",
                        "middle": [],
                        "last": "Carbonell",
                        "suffix": ""
                    },
                    {
                        "first": "J",
                        "middle": [],
                        "last": "Goldstein",
                        "suffix": ""
                    }
                ],
                "year": 1998,
                "venue": "Proceedings of the 21th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval",
                "volume": "",
                "issue": "",
                "pages": "335--336",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Carbonell, J., & Goldstein, J. (1998). The Use of MMR Diversity-based Reranking for Reordering Documents and Producing Summaries. In Proceedings of the 21th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, 335-336.",
                "links": null
            },
            "BIBREF4": {
                "ref_id": "b4",
                "title": "A Probabilistic Generative Framework for Extractive Broadcast News Speech Summarization",
                "authors": [
                    {
                        "first": "Y.-T",
                        "middle": [],
                        "last": "Chen",
                        "suffix": ""
                    },
                    {
                        "first": "B",
                        "middle": [],
                        "last": "Chen",
                        "suffix": ""
                    },
                    {
                        "first": "H.-M",
                        "middle": [],
                        "last": "Wang",
                        "suffix": ""
                    }
                ],
                "year": 2009,
                "venue": "IEEE Transactions on Audio, Speech and Language Processing",
                "volume": "17",
                "issue": "1",
                "pages": "95--106",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Chen, Y.-T., Chen, B., & Wang, H.-M. (2009). A Probabilistic Generative Framework for Extractive Broadcast News Speech Summarization. IEEE Transactions on Audio, Speech and Language Processing, 17(1), 95-106.",
                "links": null
            },
            "BIBREF5": {
                "ref_id": "b5",
                "title": "Sentence Modeling for Extractive Speech Summarization",
                "authors": [
                    {
                        "first": "B",
                        "middle": [],
                        "last": "Chen",
                        "suffix": ""
                    },
                    {
                        "first": "H.-C",
                        "middle": [],
                        "last": "Chang",
                        "suffix": ""
                    },
                    {
                        "first": "K.-Y",
                        "middle": [],
                        "last": "Chen",
                        "suffix": ""
                    }
                ],
                "year": 2013,
                "venue": "Proceedings of the International Conference on Multimedia & Expo (ICME)",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {
                    "DOI": [
                        "10.1109/ICME.2013.6607518"
                    ]
                },
                "num": null,
                "urls": [],
                "raw_text": "Chen, B., Chang, H.-C., & Chen, K.-Y. (2013). Sentence Modeling for Extractive Speech Summarization. In Proceedings of the International Conference on Multimedia & Expo (ICME). doi: 10.1109/ICME.2013.6607518",
                "links": null
            },
            "BIBREF6": {
                "ref_id": "b6",
                "title": "Enhancing Query Formulation for Spoken Document Retrieval",
                "authors": [
                    {
                        "first": "B",
                        "middle": [],
                        "last": "Chen",
                        "suffix": ""
                    },
                    {
                        "first": "Y.-W",
                        "middle": [],
                        "last": "Chen",
                        "suffix": ""
                    },
                    {
                        "first": "K.-Y",
                        "middle": [],
                        "last": "Chen",
                        "suffix": ""
                    },
                    {
                        "first": "H.-M",
                        "middle": [],
                        "last": "Wang",
                        "suffix": ""
                    },
                    {
                        "first": "K.-T",
                        "middle": [],
                        "last": "Yu",
                        "suffix": ""
                    }
                ],
                "year": 2014,
                "venue": "Journal of Information Science and Engineering",
                "volume": "30",
                "issue": "3",
                "pages": "553--569",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Chen, B., Chen, Y.-W., Chen, K.-Y., Wang, H.-M., & Yu, K.-T. (2014). Enhancing Query Formulation for Spoken Document Retrieval. Journal of Information Science and Engineering, 30(3), 553-569.",
                "links": null
            },
            "BIBREF7": {
                "ref_id": "b7",
                "title": "Text Summarization via Hidden Markov Models",
                "authors": [
                    {
                        "first": "J.-M",
                        "middle": [],
                        "last": "Conroy",
                        "suffix": ""
                    },
                    {
                        "first": "D.-P",
                        "middle": [],
                        "last": "Leary",
                        "suffix": ""
                    }
                ],
                "year": 2001,
                "venue": "Proceedings of the 24th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval",
                "volume": "",
                "issue": "",
                "pages": "406--407",
                "other_ids": {
                    "DOI": [
                        "10.1145/383952.384042"
                    ]
                },
                "num": null,
                "urls": [],
                "raw_text": "Conroy, J.-M., & O'Leary, D.-P. (2001). Text Summarization via Hidden Markov Models. In Proceedings of the 24th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, 406-407. doi: 10.1145/383952.384042",
                "links": null
            },
            "BIBREF8": {
                "ref_id": "b8",
                "title": "LexRank: Graph-based Lexical Centrality as Salience in Text Summarization",
                "authors": [
                    {
                        "first": "G",
                        "middle": [],
                        "last": "Erkan",
                        "suffix": ""
                    },
                    {
                        "first": "D",
                        "middle": [
                            "R"
                        ],
                        "last": "Radev",
                        "suffix": ""
                    }
                ],
                "year": 2004,
                "venue": "Journal of Artificial Intelligent Research",
                "volume": "22",
                "issue": "1",
                "pages": "457--479",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Erkan, G., & Radev, D. R. (2004). LexRank: Graph-based Lexical Centrality as Salience in Text Summarization. Journal of Artificial Intelligent Research, 22(1), 457-479.",
                "links": null
            },
            "BIBREF9": {
                "ref_id": "b9",
                "title": "Identifying Agreement and Disagreement in Conversational Speech: Use of Bayesian Networks to Model Pragmatic Dependencies",
                "authors": [
                    {
                        "first": "M",
                        "middle": [],
                        "last": "Galley",
                        "suffix": ""
                    },
                    {
                        "first": "K",
                        "middle": [],
                        "last": "Mckeown",
                        "suffix": ""
                    },
                    {
                        "first": "J",
                        "middle": [],
                        "last": "Hirschberg",
                        "suffix": ""
                    },
                    {
                        "first": "E",
                        "middle": [],
                        "last": "Shriberg",
                        "suffix": ""
                    }
                ],
                "year": 2004,
                "venue": "Proceedings of the 42 nd Annual Meeting of the Association for Computational Linguistics (ACL)",
                "volume": "",
                "issue": "",
                "pages": "669--676",
                "other_ids": {
                    "DOI": [
                        "10.3115/1218955.1219040"
                    ]
                },
                "num": null,
                "urls": [],
                "raw_text": "Galley, M., McKeown, K., Hirschberg, J., & Shriberg, E. (2004). Identifying Agreement and Disagreement in Conversational Speech: Use of Bayesian Networks to Model Pragmatic Dependencies. In Proceedings of the 42 nd Annual Meeting of the Association for Computational Linguistics (ACL), 669-676. doi: 10.3115/1218955.1219040",
                "links": null
            },
            "BIBREF10": {
                "ref_id": "b10",
                "title": "Generic Text Summarization using Relevance Measure and Latent Semantic Analysis",
                "authors": [
                    {
                        "first": "Y",
                        "middle": [],
                        "last": "Gong",
                        "suffix": ""
                    },
                    {
                        "first": "X",
                        "middle": [],
                        "last": "Liu",
                        "suffix": ""
                    }
                ],
                "year": 2001,
                "venue": "Proceedings of the 24 th annual International ACM SIGIR Conference on Research and Development in Information Retrieval",
                "volume": "",
                "issue": "",
                "pages": "19--25",
                "other_ids": {
                    "DOI": [
                        "10.1145/383952.383955"
                    ]
                },
                "num": null,
                "urls": [],
                "raw_text": "Gong, Y., & Liu, X. (2001). Generic Text Summarization using Relevance Measure and Latent Semantic Analysis. In Proceedings of the 24 th annual International ACM SIGIR Conference on Research and Development in Information Retrieval, 19-25. doi: 10.1145/383952.383955",
                "links": null
            },
            "BIBREF11": {
                "ref_id": "b11",
                "title": "Parsimonious Language Models for Information Retrieval",
                "authors": [
                    {
                        "first": "D",
                        "middle": [],
                        "last": "Hiemstra",
                        "suffix": ""
                    },
                    {
                        "first": "S",
                        "middle": [],
                        "last": "Robertson",
                        "suffix": ""
                    },
                    {
                        "first": "H",
                        "middle": [],
                        "last": "Zaragoza",
                        "suffix": ""
                    }
                ],
                "year": 2004,
                "venue": "Proceedings of the 27 th annual international ACM SIGIR conference on Research and development in information retrieval",
                "volume": "",
                "issue": "",
                "pages": "178--185",
                "other_ids": {
                    "DOI": [
                        "10.1145/1008992.1009025"
                    ]
                },
                "num": null,
                "urls": [],
                "raw_text": "Hiemstra, D., Robertson, S., & Zaragoza, H. (2004). Parsimonious Language Models for Information Retrieval. In Proceedings of the 27 th annual international ACM SIGIR conference on Research and development in information retrieval, 178-185. doi: 10.1145/1008992.1009025",
                "links": null
            },
            "BIBREF13": {
                "ref_id": "b13",
                "title": "A probabilistic model of information retrieval: development and comparative experiments. Information Processing and Management",
                "authors": [
                    {
                        "first": "K",
                        "middle": [
                            "S"
                        ],
                        "last": "Jones",
                        "suffix": ""
                    },
                    {
                        "first": "S",
                        "middle": [],
                        "last": "Walker",
                        "suffix": ""
                    },
                    {
                        "first": "S",
                        "middle": [
                            "E"
                        ],
                        "last": "Robertson",
                        "suffix": ""
                    }
                ],
                "year": 2000,
                "venue": "",
                "volume": "36",
                "issue": "",
                "pages": "779--808",
                "other_ids": {
                    "DOI": [
                        "10.1016/S0306-4573(00)00015-7"
                    ]
                },
                "num": null,
                "urls": [],
                "raw_text": "Jones, K. S., Walker, S., & Robertson, S. E. (2000). A probabilistic model of information retrieval: development and comparative experiments. Information Processing and Management, 36(6), 779-808. doi: 10.1016/S0306-4573(00)00015-7",
                "links": null
            },
            "BIBREF14": {
                "ref_id": "b14",
                "title": "Ranking Explanatory Sentences for Opinion Summarization",
                "authors": [
                    {
                        "first": "H.-D",
                        "middle": [],
                        "last": "Kim",
                        "suffix": ""
                    },
                    {
                        "first": "M",
                        "middle": [
                            "G"
                        ],
                        "last": "Castellanos",
                        "suffix": ""
                    },
                    {
                        "first": "M",
                        "middle": [],
                        "last": "Hsu",
                        "suffix": ""
                    },
                    {
                        "first": "C",
                        "middle": [],
                        "last": "Zhai",
                        "suffix": ""
                    },
                    {
                        "first": "U",
                        "middle": [],
                        "last": "Dayal",
                        "suffix": ""
                    },
                    {
                        "first": "R",
                        "middle": [],
                        "last": "Ghosh",
                        "suffix": ""
                    }
                ],
                "year": 2013,
                "venue": "Proceedings of the 36 th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR",
                "volume": "",
                "issue": "",
                "pages": "1069--1072",
                "other_ids": {
                    "DOI": [
                        "10.1145/2484028.2484172"
                    ]
                },
                "num": null,
                "urls": [],
                "raw_text": "Kim, H.-D., Castellanos, M. G., Hsu, M., Zhai, C., Dayal, U., & Ghosh, R. (2013). Ranking Explanatory Sentences for Opinion Summarization. In Proceedings of the 36 th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR), 1069-1072. doi: 10.1145/2484028.2484172",
                "links": null
            },
            "BIBREF15": {
                "ref_id": "b15",
                "title": "Summarization as Feature Selection for Text Categorization",
                "authors": [
                    {
                        "first": "A",
                        "middle": [],
                        "last": "Kolcz",
                        "suffix": ""
                    },
                    {
                        "first": "V",
                        "middle": [],
                        "last": "Prabakarmurthi",
                        "suffix": ""
                    },
                    {
                        "first": "J",
                        "middle": [],
                        "last": "Kalita",
                        "suffix": ""
                    }
                ],
                "year": 2001,
                "venue": "Proceedings of the tenth International Conference on Information and Knowledge Management",
                "volume": "",
                "issue": "",
                "pages": "365--370",
                "other_ids": {
                    "DOI": [
                        "10.1145/502585.502647"
                    ]
                },
                "num": null,
                "urls": [],
                "raw_text": "Kolcz, A., Prabakarmurthi, V., & Kalita, J. (2001). Summarization as Feature Selection for Text Categorization. In Proceedings of the tenth International Conference on Information and Knowledge Management, 365-370. doi: 10.1145/502585.502647",
                "links": null
            },
            "BIBREF16": {
                "ref_id": "b16",
                "title": "Multi-document Summary Generation using Informative and Event Words",
                "authors": [
                    {
                        "first": "J.-J",
                        "middle": [],
                        "last": "Kuo",
                        "suffix": ""
                    },
                    {
                        "first": "H.-H",
                        "middle": [],
                        "last": "Chen",
                        "suffix": ""
                    }
                ],
                "year": 2006,
                "venue": "Journal of ACM Transactions on Asian Language Information Processing",
                "volume": "7",
                "issue": "1",
                "pages": "550--557",
                "other_ids": {
                    "DOI": [
                        "10.1145/1330291.1330294"
                    ]
                },
                "num": null,
                "urls": [],
                "raw_text": "Kuo, J.-J., & Chen, H.-H. (2006). Multi-document Summary Generation using Informative and Event Words. Journal of ACM Transactions on Asian Language Information Processing, 7(1), 550-557. doi: 10.1145/1330291.1330294",
                "links": null
            },
            "BIBREF17": {
                "ref_id": "b17",
                "title": "A Trainable Document Summarizer",
                "authors": [
                    {
                        "first": "J",
                        "middle": [],
                        "last": "Kupiec",
                        "suffix": ""
                    },
                    {
                        "first": "J",
                        "middle": [],
                        "last": "Pedersen",
                        "suffix": ""
                    },
                    {
                        "first": "F",
                        "middle": [],
                        "last": "Chen",
                        "suffix": ""
                    }
                ],
                "year": 1995,
                "venue": "Proceedings of the 18 th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval",
                "volume": "",
                "issue": "",
                "pages": "68--73",
                "other_ids": {
                    "DOI": [
                        "10.1145/215206.215333"
                    ]
                },
                "num": null,
                "urls": [],
                "raw_text": "Kupiec , J., Pedersen, J., & Chen, F. (1995). A Trainable Document Summarizer. In Proceedings of the 18 th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, 68-73. doi: 10.1145/215206.215333",
                "links": null
            },
            "BIBREF18": {
                "ref_id": "b18",
                "title": "Relevance-based Language Models",
                "authors": [
                    {
                        "first": "V",
                        "middle": [],
                        "last": "Lavrenko",
                        "suffix": ""
                    },
                    {
                        "first": "W.-B",
                        "middle": [],
                        "last": "Croft",
                        "suffix": ""
                    }
                ],
                "year": 2001,
                "venue": "Proceedings of the 24th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval",
                "volume": "",
                "issue": "",
                "pages": "120--127",
                "other_ids": {
                    "DOI": [
                        "10.1145/383952.383972"
                    ]
                },
                "num": null,
                "urls": [],
                "raw_text": "Lavrenko, V., & Croft, W.-B. (2001). Relevance-based Language Models. In Proceedings of the 24th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, 120-127. doi: 10.1145/383952.383972",
                "links": null
            },
            "BIBREF19": {
                "ref_id": "b19",
                "title": "Improved Speech Summarization with Multiple-hypothesis Representations and Kullback-Leibler Divergence Measures",
                "authors": [
                    {
                        "first": "S.-H",
                        "middle": [],
                        "last": "Lin",
                        "suffix": ""
                    },
                    {
                        "first": "B",
                        "middle": [],
                        "last": "Chen",
                        "suffix": ""
                    }
                ],
                "year": 2009,
                "venue": "Proceeding of the 10th Annual Conference of the International Speech Communication Association (Interspeech)",
                "volume": "",
                "issue": "",
                "pages": "1847--1850",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Lin, S.-H., & Chen, B. (2009). Improved Speech Summarization with Multiple-hypothesis Representations and Kullback-Leibler Divergence Measures. In Proceeding of the 10th Annual Conference of the International Speech Communication Association (Interspeech), 1847-1850.",
                "links": null
            },
            "BIBREF20": {
                "ref_id": "b20",
                "title": "A Survey on Speech Summarization Techniques",
                "authors": [
                    {
                        "first": "S.-H",
                        "middle": [],
                        "last": "Lin",
                        "suffix": ""
                    },
                    {
                        "first": "B",
                        "middle": [],
                        "last": "Chen",
                        "suffix": ""
                    }
                ],
                "year": 2010,
                "venue": "The Association for Computational Linguistics and Chinese Language Processing Newsletter",
                "volume": "21",
                "issue": "",
                "pages": "4--16",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Lin, S.-H., & Chen, B. (2010). A Survey on Speech Summarization Techniques. The Association for Computational Linguistics and Chinese Language Processing Newsletter, 21(1), 4-16.",
                "links": null
            },
            "BIBREF21": {
                "ref_id": "b21",
                "title": "Multi-document Summarization via Budgeted Maximization of Submodular Functions",
                "authors": [
                    {
                        "first": "H",
                        "middle": [],
                        "last": "Lin",
                        "suffix": ""
                    },
                    {
                        "first": "J",
                        "middle": [],
                        "last": "Bilmes",
                        "suffix": ""
                    }
                ],
                "year": 2010,
                "venue": "Proceeding of NAACL HLT",
                "volume": "",
                "issue": "",
                "pages": "912--920",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Lin, H., & Bilmes, J. (2010). Multi-document Summarization via Budgeted Maximization of Submodular Functions. In Proceeding of NAACL HLT, 912-920.",
                "links": null
            },
            "BIBREF22": {
                "ref_id": "b22",
                "title": "Leveraging Kullback-Leibler Divergence Measures and Information-Rich Cues for Speech Summarization",
                "authors": [
                    {
                        "first": "S.-H",
                        "middle": [],
                        "last": "Lin",
                        "suffix": ""
                    },
                    {
                        "first": "Y.-M",
                        "middle": [],
                        "last": "Yeh",
                        "suffix": ""
                    },
                    {
                        "first": "B",
                        "middle": [],
                        "last": "Chen",
                        "suffix": ""
                    }
                ],
                "year": 2011,
                "venue": "IEEE Transactions on Audio, Speech and Language Processing",
                "volume": "19",
                "issue": "4",
                "pages": "871--882",
                "other_ids": {
                    "DOI": [
                        "10.1109/TASL.2010.2066268"
                    ]
                },
                "num": null,
                "urls": [],
                "raw_text": "Lin, S.-H., Yeh, Y.-M. , & Chen, B. (2011). Leveraging Kullback-Leibler Divergence Measures and Information-Rich Cues for Speech Summarization. IEEE Transactions on Audio, Speech and Language Processing, 19(4), 871-882. doi: 10.1109/TASL.2010.2066268",
                "links": null
            },
            "BIBREF23": {
                "ref_id": "b23",
                "title": "ROUGE: Recall-oriented Understudy for Gisting Evaluation",
                "authors": [
                    {
                        "first": "C.-Y",
                        "middle": [],
                        "last": "Lin",
                        "suffix": ""
                    }
                ],
                "year": 2003,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Lin, C.-Y. (2003). ROUGE: Recall-oriented Understudy for Gisting Evaluation. [Online]. Retrieved from http://haydn.isi.edu/ROUGE/.",
                "links": null
            },
            "BIBREF24": {
                "ref_id": "b24",
                "title": "Spoken Language Understanding: Systems for Extracting Semantic Information from Speech",
                "authors": [
                    {
                        "first": "Y",
                        "middle": [],
                        "last": "Liu",
                        "suffix": ""
                    },
                    {
                        "first": "D",
                        "middle": [],
                        "last": "Hakkani-Tur",
                        "suffix": ""
                    }
                ],
                "year": 2011,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {
                    "DOI": [
                        "10.1002/9781119992691.ch13"
                    ]
                },
                "num": null,
                "urls": [],
                "raw_text": "Liu, Y., & Hakkani-Tur, D. (2011). Speech Summarization. In G. Turand & R. D. Mori (Eds), Spoken Language Understanding: Systems for Extracting Semantic Information from Speech. West Sussex, U.K.: Wiley. doi: 10.1002/9781119992691.ch13",
                "links": null
            },
            "BIBREF25": {
                "ref_id": "b25",
                "title": "Effective Pseudo-relevance Feedback for Language Modeling in Extractive Speech Summarization",
                "authors": [
                    {
                        "first": "S.-H",
                        "middle": [],
                        "last": "Liu",
                        "suffix": ""
                    },
                    {
                        "first": "K.-Y",
                        "middle": [],
                        "last": "Chen",
                        "suffix": ""
                    },
                    {
                        "first": "Y.-L",
                        "middle": [],
                        "last": "Hsieh",
                        "suffix": ""
                    },
                    {
                        "first": "B",
                        "middle": [],
                        "last": "Chen",
                        "suffix": ""
                    },
                    {
                        "first": "H.-M",
                        "middle": [],
                        "last": "Wang",
                        "suffix": ""
                    },
                    {
                        "first": "H.-C",
                        "middle": [],
                        "last": "Yen",
                        "suffix": ""
                    },
                    {
                        "first": "W.-L",
                        "middle": [],
                        "last": "Hsu",
                        "suffix": ""
                    }
                ],
                "year": 2014,
                "venue": "Proceedings of the IEEE International Conference on \u7576\u4ee3\u975e\u76e3\u7763\u5f0f\u65b9\u6cd5\u4e4b\u6bd4\u8f03\u65bc\u7bc0\u9304\u5f0f\u8a9e\u97f3\u6458\u8981 25",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Liu, S.-H., Chen, K.-Y., Hsieh, Y.-L., Chen, B., Wang, H.-M., Yen, H.-C., & Hsu, W.-L. (2014). Effective Pseudo-relevance Feedback for Language Modeling in Extractive Speech Summarization. In Proceedings of the IEEE International Conference on \u7576\u4ee3\u975e\u76e3\u7763\u5f0f\u65b9\u6cd5\u4e4b\u6bd4\u8f03\u65bc\u7bc0\u9304\u5f0f\u8a9e\u97f3\u6458\u8981 25",
                "links": null
            },
            "BIBREF27": {
                "ref_id": "b27",
                "title": "The Automatic Creation of Literature Abstracts",
                "authors": [
                    {
                        "first": "H",
                        "middle": [
                            "P"
                        ],
                        "last": "Luhn",
                        "suffix": ""
                    }
                ],
                "year": 1958,
                "venue": "IBM Journal of Research and Development",
                "volume": "2",
                "issue": "2",
                "pages": "159--165",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Luhn, H. P. (1958). The Automatic Creation of Literature Abstracts. IBM Journal of Research and Development, 2(2), 159-165.",
                "links": null
            },
            "BIBREF28": {
                "ref_id": "b28",
                "title": "When Documents Are Very Long, BM25 Fails!",
                "authors": [
                    {
                        "first": "Y",
                        "middle": [],
                        "last": "Lv",
                        "suffix": ""
                    },
                    {
                        "first": "C.-X",
                        "middle": [],
                        "last": "Zhai",
                        "suffix": ""
                    }
                ],
                "year": 2011,
                "venue": "Proceedings of the 34 th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR",
                "volume": "",
                "issue": "",
                "pages": "1103--1104",
                "other_ids": {
                    "DOI": [
                        "10.1145/2009916.2010070"
                    ]
                },
                "num": null,
                "urls": [],
                "raw_text": "Lv, Y., & Zhai, C.-X. (2011a). When Documents Are Very Long, BM25 Fails! In Proceedings of the 34 th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR), 1103-1104. doi: 10.1145/2009916.2010070",
                "links": null
            },
            "BIBREF29": {
                "ref_id": "b29",
                "title": "Lower-bounding Term Frequency Normalization",
                "authors": [
                    {
                        "first": "Y",
                        "middle": [],
                        "last": "Lv",
                        "suffix": ""
                    },
                    {
                        "first": "C.-X",
                        "middle": [],
                        "last": "Zhai",
                        "suffix": ""
                    }
                ],
                "year": 2011,
                "venue": "Proceeding of the 20th ACM International Conference on Information and Knowledge Management (CIKM)",
                "volume": "",
                "issue": "",
                "pages": "7--16",
                "other_ids": {
                    "DOI": [
                        "10.1145/2063576.2063584"
                    ]
                },
                "num": null,
                "urls": [],
                "raw_text": "Lv, Y., & Zhai, C.-X. (2011b). Lower-bounding Term Frequency Normalization, In Proceeding of the 20th ACM International Conference on Information and Knowledge Management (CIKM), 7-16. doi: 10.1145/2063576.2063584",
                "links": null
            },
            "BIBREF30": {
                "ref_id": "b30",
                "title": "A Log-logistic Model-based Interpretation of TF Normalization of BM25",
                "authors": [
                    {
                        "first": "Y",
                        "middle": [],
                        "last": "Lv",
                        "suffix": ""
                    },
                    {
                        "first": "C.-X",
                        "middle": [],
                        "last": "Zhai",
                        "suffix": ""
                    }
                ],
                "year": 2012,
                "venue": "Proceedings of European Conference on Information Retrieval",
                "volume": "",
                "issue": "",
                "pages": "244--255",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Lv, Y., & Zhai, C.-X. (2012). A Log-logistic Model-based Interpretation of TF Normalization of BM25. In Proceedings of European Conference on Information Retrieval (ECIR), 244-255.",
                "links": null
            },
            "BIBREF31": {
                "ref_id": "b31",
                "title": "Revisiting the Divergence Minimization Feedback Model",
                "authors": [
                    {
                        "first": "Y",
                        "middle": [],
                        "last": "Lv",
                        "suffix": ""
                    },
                    {
                        "first": "C.-X",
                        "middle": [],
                        "last": "Zhai",
                        "suffix": ""
                    }
                ],
                "year": 2014,
                "venue": "Proceeding of the 23rd ACM International Conference on Information and Knowledge Management (CIKM)",
                "volume": "",
                "issue": "",
                "pages": "1863--1866",
                "other_ids": {
                    "DOI": [
                        "10.1145/2661829.2661900"
                    ]
                },
                "num": null,
                "urls": [],
                "raw_text": "Lv, Y., & Zhai, C.-X. (2014). Revisiting the Divergence Minimization Feedback Model, In Proceeding of the 23rd ACM International Conference on Information and Knowledge Management (CIKM), 1863-1866. doi: 10.1145/2661829.2661900",
                "links": null
            },
            "BIBREF32": {
                "ref_id": "b32",
                "title": "Advances in Automatic Text Summarization",
                "authors": [
                    {
                        "first": "I",
                        "middle": [],
                        "last": "Mani",
                        "suffix": ""
                    },
                    {
                        "first": "M. -T",
                        "middle": [],
                        "last": "Maybury",
                        "suffix": ""
                    }
                ],
                "year": 1999,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Mani , I., & Maybury, M. -T. (1999) . Advances in Automatic Text Summarization. Cambridge, MA: MIT Press.",
                "links": null
            },
            "BIBREF33": {
                "ref_id": "b33",
                "title": "A Study of Global Inference Algorithms in Multi-document Summarization",
                "authors": [
                    {
                        "first": "R",
                        "middle": [],
                        "last": "Mcdonald",
                        "suffix": ""
                    }
                ],
                "year": 2007,
                "venue": "Proceedings of the 29 th European Conference on Information Retrieval",
                "volume": "",
                "issue": "",
                "pages": "557--564",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "McDonald, R. (2007). A Study of Global Inference Algorithms in Multi-document Summarization, In Proceedings of the 29 th European Conference on Information Retrieval, 557-564.",
                "links": null
            },
            "BIBREF34": {
                "ref_id": "b34",
                "title": "TextRank Bringing Order into Texts",
                "authors": [
                    {
                        "first": "R",
                        "middle": [],
                        "last": "Mihalcea",
                        "suffix": ""
                    },
                    {
                        "first": "P",
                        "middle": [],
                        "last": "Tarau",
                        "suffix": ""
                    }
                ],
                "year": 2004,
                "venue": "Proceedings of Empirical Method in Natural Language Processing",
                "volume": "",
                "issue": "",
                "pages": "404--411",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Mihalcea, R., & Tarau, P. (2004). TextRank Bringing Order into Texts. In Proceedings of Empirical Method in Natural Language Processing (EMNLP 2004), 404-411.",
                "links": null
            },
            "BIBREF35": {
                "ref_id": "b35",
                "title": "Extractive Summarization of Meeting Recordings",
                "authors": [
                    {
                        "first": "G",
                        "middle": [],
                        "last": "Murray",
                        "suffix": ""
                    },
                    {
                        "first": "S",
                        "middle": [],
                        "last": "Renals",
                        "suffix": ""
                    },
                    {
                        "first": "J",
                        "middle": [],
                        "last": "Carletta",
                        "suffix": ""
                    }
                ],
                "year": 2005,
                "venue": "Proceedings of the Conference of the International Speech Communication Association (Interspeech)",
                "volume": "",
                "issue": "",
                "pages": "593--596",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Murray, G., Renals, S., & Carletta, J. (2005). Extractive Summarization of Meeting Recordings. In Proceedings of the Conference of the International Speech Communication Association (Interspeech), 593-596.",
                "links": null
            },
            "BIBREF36": {
                "ref_id": "b36",
                "title": "Automatic Summarization. Foundations and Trends in Information Retrieval",
                "authors": [
                    {
                        "first": "A",
                        "middle": [],
                        "last": "Nenkova",
                        "suffix": ""
                    },
                    {
                        "first": "K",
                        "middle": [],
                        "last": "Mckeown",
                        "suffix": ""
                    }
                ],
                "year": 2011,
                "venue": "",
                "volume": "5",
                "issue": "",
                "pages": "103--233",
                "other_ids": {
                    "DOI": [
                        "10.1561/1500000015"
                    ]
                },
                "num": null,
                "urls": [],
                "raw_text": "Nenkova, A., & McKeown, K. (2011). Automatic Summarization. Foundations and Trends in Information Retrieval, 5(2-3), 103-233. doi : 10.1561/1500000015",
                "links": null
            },
            "BIBREF37": {
                "ref_id": "b37",
                "title": "Speech Technology and Information Access",
                "authors": [
                    {
                        "first": "M",
                        "middle": [],
                        "last": "Ostendorf",
                        "suffix": ""
                    }
                ],
                "year": 2008,
                "venue": "IEEE Signal Processing Magazine",
                "volume": "25",
                "issue": "3",
                "pages": "150--152",
                "other_ids": {
                    "DOI": [
                        "10.1109/MSP.2008.918685"
                    ]
                },
                "num": null,
                "urls": [],
                "raw_text": "Ostendorf, M. (2008) Speech Technology and Information Access. IEEE Signal Processing Magazine, 25(3), 150-152. doi: 10.1109/MSP.2008.918685",
                "links": null
            },
            "BIBREF38": {
                "ref_id": "b38",
                "title": "Constructing Literature Abstracts by Computer Techniques and Prospects",
                "authors": [
                    {
                        "first": "C.-D",
                        "middle": [],
                        "last": "Paice",
                        "suffix": ""
                    }
                ],
                "year": 1990,
                "venue": "Journal of Information Processing and Management",
                "volume": "26",
                "issue": "1",
                "pages": "171--186",
                "other_ids": {
                    "DOI": [
                        "10.1016/0306-4573(90)90014-S"
                    ]
                },
                "num": null,
                "urls": [],
                "raw_text": "Paice, C.-D. (1990). Constructing Literature Abstracts by Computer Techniques and Prospects. Journal of Information Processing and Management, 26(1), 171-186. doi: 10.1016/0306-4573(90)90014-S",
                "links": null
            },
            "BIBREF39": {
                "ref_id": "b39",
                "title": "A Critical Reassessment of Evaluation Baselines for Speech Summarization",
                "authors": [
                    {
                        "first": "G",
                        "middle": [],
                        "last": "Penn",
                        "suffix": ""
                    },
                    {
                        "first": "X",
                        "middle": [],
                        "last": "Zhu",
                        "suffix": ""
                    }
                ],
                "year": 2008,
                "venue": "Proceedings of Annual Meeting of the Association for Computational Linguistics",
                "volume": "",
                "issue": "",
                "pages": "470--478",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Penn, G. , & Zhu, X. (2008). A Critical Reassessment of Evaluation Baselines for Speech Summarization. In Proceedings of Annual Meeting of the Association for Computational Linguistics, 470-478.",
                "links": null
            },
            "BIBREF40": {
                "ref_id": "b40",
                "title": "The Probabilistic Relevance Framework: BM25 and Beyond",
                "authors": [
                    {
                        "first": "S",
                        "middle": [],
                        "last": "Robertson",
                        "suffix": ""
                    },
                    {
                        "first": "H",
                        "middle": [],
                        "last": "Zaragoza",
                        "suffix": ""
                    }
                ],
                "year": 2008,
                "venue": "Foundations and Trends in Information Retrieval",
                "volume": "3",
                "issue": "4",
                "pages": "333--389",
                "other_ids": {
                    "DOI": [
                        "10.1561/1500000019"
                    ]
                },
                "num": null,
                "urls": [],
                "raw_text": "Robertson, S. & Zaragoza, H. (2008). The Probabilistic Relevance Framework: BM25 and Beyond. Foundations and Trends in Information Retrieval, 3(4), 333-389. doi: 10.1561/1500000019",
                "links": null
            },
            "BIBREF42": {
                "ref_id": "b42",
                "title": "Document Summarization using Conditional Random Fields",
                "authors": [
                    {
                        "first": "D",
                        "middle": [],
                        "last": "Shen",
                        "suffix": ""
                    },
                    {
                        "first": "J.-T",
                        "middle": [],
                        "last": "Sun",
                        "suffix": ""
                    },
                    {
                        "first": "H",
                        "middle": [],
                        "last": "Li",
                        "suffix": ""
                    },
                    {
                        "first": "Q",
                        "middle": [],
                        "last": "Yang",
                        "suffix": ""
                    },
                    {
                        "first": "Z",
                        "middle": [],
                        "last": "Chen",
                        "suffix": ""
                    }
                ],
                "year": 2007,
                "venue": "Proceedings of the 20th International Joint Conference on Artificial Intelligence (IJCAI)",
                "volume": "",
                "issue": "",
                "pages": "2862--2867",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Shen, D., Sun, J.-T., Li, H., Yang, Q., & Chen, Z. (2007). Document Summarization using Conditional Random Fields. In Proceedings of the 20th International Joint Conference on Artificial Intelligence (IJCAI), 2862-2867.",
                "links": null
            },
            "BIBREF43": {
                "ref_id": "b43",
                "title": "A Robust Practical Text Summarization",
                "authors": [
                    {
                        "first": "T",
                        "middle": [],
                        "last": "Strzalkowski",
                        "suffix": ""
                    },
                    {
                        "first": "J",
                        "middle": [],
                        "last": "Wand",
                        "suffix": ""
                    },
                    {
                        "first": "B",
                        "middle": [],
                        "last": "Wise",
                        "suffix": ""
                    }
                ],
                "year": 1998,
                "venue": "Proceedings of AAAI Conference on Artificial Intelligence Spring Symposium on Intelligent Text Summarization",
                "volume": "",
                "issue": "",
                "pages": "26--33",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Strzalkowski, T., Wand, J., & Wise, B. (1998). A Robust Practical Text Summarization. In Proceedings of AAAI Conference on Artificial Intelligence Spring Symposium on Intelligent Text Summarization, 26-33.",
                "links": null
            },
            "BIBREF44": {
                "ref_id": "b44",
                "title": "Multi-document Summarization using Cluster-based Link Analysis",
                "authors": [
                    {
                        "first": "X",
                        "middle": [],
                        "last": "Wan",
                        "suffix": ""
                    },
                    {
                        "first": "J",
                        "middle": [],
                        "last": "Yang",
                        "suffix": ""
                    }
                ],
                "year": 2008,
                "venue": "Proceedings of the 31 st Annual International ACM SIGIR Conference on Research and Development in Information Retrieval",
                "volume": "",
                "issue": "",
                "pages": "299--306",
                "other_ids": {
                    "DOI": [
                        "10.1145/1390334.1390386"
                    ]
                },
                "num": null,
                "urls": [],
                "raw_text": "Wan, X., & Yang, J. (2008). Multi-document Summarization using Cluster-based Link Analysis. In Proceedings of the 31 st Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, 299-306. doi: 10.1145/1390334.1390386",
                "links": null
            },
            "BIBREF45": {
                "ref_id": "b45",
                "title": "MATBN: A Mandarin Chinese broadcast news corpus",
                "authors": [
                    {
                        "first": "H.-M",
                        "middle": [],
                        "last": "Wang",
                        "suffix": ""
                    },
                    {
                        "first": "B",
                        "middle": [],
                        "last": "Chen",
                        "suffix": ""
                    },
                    {
                        "first": "J.-W",
                        "middle": [],
                        "last": "Kuo",
                        "suffix": ""
                    },
                    {
                        "first": "S.-S",
                        "middle": [],
                        "last": "Cheng",
                        "suffix": ""
                    }
                ],
                "year": 2005,
                "venue": "International Journal of Computational Linguistics and Chinese Language Processing",
                "volume": "10",
                "issue": "2",
                "pages": "219--236",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Wang, H.-M., Chen, B., Kuo, J.-w., & Cheng, S.-S. (2005). MATBN: A Mandarin Chinese broadcast news corpus. International Journal of Computational Linguistics and Chinese Language Processing, 10(2), 219-236.",
                "links": null
            },
            "BIBREF46": {
                "ref_id": "b46",
                "title": "Ultra Summarization: a Statistical Approach to Generating Highly Condensed Non-extractive Summaries",
                "authors": [
                    {
                        "first": "M",
                        "middle": [],
                        "last": "Witbrock",
                        "suffix": ""
                    },
                    {
                        "first": "V",
                        "middle": [],
                        "last": "Mittal",
                        "suffix": ""
                    }
                ],
                "year": 1999,
                "venue": "Proceedings of the 22th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR",
                "volume": "",
                "issue": "",
                "pages": "315--316",
                "other_ids": {
                    "DOI": [
                        "10.1145/312624.312748"
                    ]
                },
                "num": null,
                "urls": [],
                "raw_text": "Witbrock, M. , & Mittal, V. (1999). Ultra Summarization: a Statistical Approach to Generating Highly Condensed Non-extractive Summaries. In Proceedings of the 22th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR), 315-316. doi: 10.1145/312624.312748",
                "links": null
            },
            "BIBREF47": {
                "ref_id": "b47",
                "title": "Model-based feedback in the language modeling approach to information retrieval",
                "authors": [
                    {
                        "first": "C.-X",
                        "middle": [],
                        "last": "Zhai",
                        "suffix": ""
                    },
                    {
                        "first": "J",
                        "middle": [],
                        "last": "Lafferty",
                        "suffix": ""
                    }
                ],
                "year": 2001,
                "venue": "Proceeding of the tenth International Conference on Information and Knowledge Management (CIKM)",
                "volume": "",
                "issue": "",
                "pages": "403--410",
                "other_ids": {
                    "DOI": [
                        "10.1145/502585.502654"
                    ]
                },
                "num": null,
                "urls": [],
                "raw_text": "Zhai, C.-X., & Lafferty, J. (2001a). Model-based feedback in the language modeling approach to information retrieval. In Proceeding of the tenth International Conference on Information and Knowledge Management (CIKM), 403-410. doi: 10.1145/502585.502654",
                "links": null
            },
            "BIBREF48": {
                "ref_id": "b48",
                "title": "A Study of Smoothing Methods for Language Models Applied to Ad Hoc Information Retrieval",
                "authors": [
                    {
                        "first": "C.-X",
                        "middle": [],
                        "last": "Zhai",
                        "suffix": ""
                    },
                    {
                        "first": "J",
                        "middle": [],
                        "last": "Lafferty",
                        "suffix": ""
                    }
                ],
                "year": 2001,
                "venue": "Proceedings of the 24th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval",
                "volume": "",
                "issue": "",
                "pages": "334--342",
                "other_ids": {
                    "DOI": [
                        "10.1145/383952.384019"
                    ]
                },
                "num": null,
                "urls": [],
                "raw_text": "Zhai, C.-X., & Lafferty, J. (2001b). A Study of Smoothing Methods for Language Models Applied to Ad Hoc Information Retrieval. In Proceedings of the 24th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, 334-342. doi: 10.1145/383952.384019",
                "links": null
            },
            "BIBREF49": {
                "ref_id": "b49",
                "title": "Statistical Language Models for Information Retrieval: A Critical Review. Foundations and Trends in Information Retrieval",
                "authors": [
                    {
                        "first": "C.-X",
                        "middle": [],
                        "last": "Zhai",
                        "suffix": ""
                    }
                ],
                "year": 2008,
                "venue": "",
                "volume": "2",
                "issue": "",
                "pages": "137--213",
                "other_ids": {
                    "DOI": [
                        "10.1561/1500000008"
                    ]
                },
                "num": null,
                "urls": [],
                "raw_text": "Zhai, C.-X. (2008). Statistical Language Models for Information Retrieval: A Critical Review. Foundations and Trends in Information Retrieval, 2(3), 137-213. doi: 10.1561/1500000008",
                "links": null
            },
            "BIBREF50": {
                "ref_id": "b50",
                "title": "Speech Summarization without Lexical Features for Mandarin Broadcast News",
                "authors": [
                    {
                        "first": "J",
                        "middle": [],
                        "last": "Zhang",
                        "suffix": ""
                    },
                    {
                        "first": "P",
                        "middle": [],
                        "last": "Fung",
                        "suffix": ""
                    }
                ],
                "year": 2007,
                "venue": "Proceedings of NAACL HLT, Companion Volume",
                "volume": "",
                "issue": "",
                "pages": "213--216",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Zhang, J., & Fung, P. (2007). Speech Summarization without Lexical Features for Mandarin Broadcast News. In Proceedings of NAACL HLT, Companion Volume, 213-216.",
                "links": null
            },
            "BIBREF51": {
                "ref_id": "b51",
                "title": "Extractive Speech Summarization using Shallow Rhetorical Structure Modeling",
                "authors": [
                    {
                        "first": "J.-J",
                        "middle": [],
                        "last": "Zhang",
                        "suffix": ""
                    },
                    {
                        "first": "H.-Y",
                        "middle": [],
                        "last": "Chan",
                        "suffix": ""
                    },
                    {
                        "first": "P",
                        "middle": [],
                        "last": "Fung",
                        "suffix": ""
                    }
                ],
                "year": 2010,
                "venue": "IEEE Transactions on Audio, Speech and Language Processing",
                "volume": "18",
                "issue": "6",
                "pages": "1147--1157",
                "other_ids": {
                    "DOI": [
                        "10.1109/TASL.2009.2030951"
                    ]
                },
                "num": null,
                "urls": [],
                "raw_text": "Zhang, J.-J., Chan, H.-Y., & Fung, P. (2010). Extractive Speech Summarization using Shallow Rhetorical Structure Modeling. IEEE Transactions on Audio, Speech and Language Processing, 18(6), 1147-1157. doi: 10.1109/TASL.2009.2030951",
                "links": null
            }
        },
        "ref_entries": {
            "TABREF3": {
                "content": "<table><tr><td>6</td><td>\u5289\u58eb\u5f18 \u7b49</td></tr><tr><td colspan=\"2\">\u5982\uff1a\u97f3\u8abf(Intonation)\u3001\u97f3\u9ad8(Pitch)\u3001\u97f3\u5f37(Power)\u3001\u8a9e\u8005\u767c\u8072\u6301\u7e8c\u6642\u9593(Duration)\u3001\u8a9e\u8005\u8aaa</td></tr><tr><td colspan=\"2\">\u8a71\u901f\uf961(Rate)\u3001\u8a9e\u8005(Speaker)\u3001\u60c5\u611f(Emotion)\u548c\u8aaa\u8a71\u6642\u5834\u666f(Environment)\u7b49\u8cc7\u8a0a\uff0c\u9019\u4e9b\u90fd</td></tr><tr><td colspan=\"2\">\u662f\u5f9e\u4e8b\u8a9e\u97f3\u6587\u4ef6\u6458\u8981\u6642\u53ef\u4ee5\u5584\u52a0\u5229\u7528\u7684\u8a9e\u53e5\u7279\u5fb5\u8cc7\u8a0a(Liu &amp; Hakkani-Tur, 2011)\u3002</td></tr><tr><td colspan=\"2\">3. \u4f7f \u7528 \u8a9e \u8a00 \u6a21 \u578b \u65bc \u8a9e \u97f3 \u6587 \u4ef6 \u6458 \u8981 \u8a9e\u8a00\u6a21\u578b\u7684\u7814\u7a76\u8207\u767c\u5c55\u6700\u65e9\u662f\u6e90\u81ea\u65bc\u8a9e\u97f3\u8fa8\u8b58\u53ca\u81ea\u7136\u8a9e\u8a00\u8655\u7406\u3002\u8a9e\u8a00\u6a21\u578b\u65e8\u5728\u63cf\u8ff0\u8a9e\u8a00</td></tr><tr><td colspan=\"2\">\u4e2d\u7684\u6240\u6709\u8a5e\u5f59\u4e4b\u9593\u5171\u540c\u51fa\u73fe\u8207\u76f8\u9130\u8cc7\u8a0a\u7684\u95dc\u4fc2\u3002\u5176\u5047\u8a2d\u4eba\u985e\u8a9e\u8a00\u751f\u6210(Human Language</td></tr><tr><td colspan=\"2\">Generation)\u662f\u4e00\u500b\u96a8\u6a5f\u904e\u7a0b\uff0c\u800c\u8a9e\u8a00\u6a21\u578b\u5c31\u662f\u5728\u6a21\u64ec\u5982\u4f55\u7531\u8a5e\u5f59\u69cb\u6210\u7247\u8a9e\u3001\u8a9e\u53e5\u3001\u6bb5\u843d</td></tr><tr><td colspan=\"2\">\u6216 \u8005 \u6587 \u4ef6 \u4e4b \u904e \u7a0b \u7684 \u6a5f \u7387 \u6a21 \u578b \uff0c \u6545 \u53c8 \u7a31 \u70ba \u751f \u6210 \u5f0f \u8a9e \u8a00 \u6a21 \u578b (Generative Language</td></tr><tr><td colspan=\"2\">Modeling)(Zhai, 2008)\u3002\u6700\u7c21\u55ae\u7684\u8a9e\u8a00\u6a21\u578b\u70ba\u55ae\u9023\u8a9e\u8a00\u6a21\u578b(Unigram Language Model,</td></tr><tr><td colspan=\"2\">ULM)\uff0c\u5b83\u4e0d\u8003\u616e\u8a5e\u5f59\u4e4b\u9593\u7684\u9806\u5e8f\u95dc\u4fc2\uff0c\u53ea\u500b\u5225\u8003\u616e\u6bcf\u4e00\u500b\u8a5e\u672c\u8eab\u51fa\u73fe\u7684\u6a5f\u7387\u3002\u8f03\u70ba\u8907\u96dc</td></tr><tr><td colspan=\"2\">\u4e14\u5e38\u88ab\u4f7f\u7528\u7684\u8a9e\u8a00\u6a21\u578b\u70ba N-\u9023\u8a9e\u8a00\u6a21\u578b\uff0c\u901a\u5e38 N \u70ba 2 \u6216 3(\u5373\u4e8c\u9023\u6216\u4e09\u9023\u8a9e\u8a00\u6a21\u578b)\uff0c</td></tr><tr><td colspan=\"2\">\u5176\u8003\u616e\u5169\u500b\u8a5e\u5f59\u6216\u4e09\u500b\u8a5e\u5f59\u4e4b\u9593\u7684\u5171\u540c\u51fa\u73fe\u8207\u7dca\u9023\u7684\u9806\u5e8f\u95dc\u4fc2\u3002\u503c\u5f97\u4e00\u63d0\u7684\u662f\uff0c\u55ae\u9023\u8a9e</td></tr><tr><td colspan=\"2\">\u8a00\u6a21\u578b\u548c N-\u9023\u8a9e\u8a00\u6a21\u578b\u7684\u4e3b\u8981\u512a\u9ede\u4e4b\u4e00\u662f\uff1a\u5b83\u5011\u50c5\u9700\u4f7f\u7528\u8a13\u7df4\u8a9e\u6599\u4f86\u4f30\u6e2c\u6bcf\u4e00\u500b\u8a5e\u672c\u8eab</td></tr><tr><td colspan=\"2\">\u51fa\u73fe\u7684\u6a5f\u7387\u5206\u4f48\uff0c\u6216\u8005\u8a5e\u5f59\u4e4b\u9593\u7684\u5171\u540c\u51fa\u73fe\u8207\u9130\u8fd1\u95dc\u4fc2\u7684\u6a5f\u7387\u5206\u4f48\uff0c\u4e26\u4e0d\u9700\u8981\u984d\u5916\u7684\u4eba</td></tr><tr><td>\u5de5\u6a19\u8a18\u8cc7\u8a0a\uff0c\u56e0\u6b64\u8a9e\u8a00\u6a21\u578b\u662f\u5c6c\u65bc\u57fa\u65bc\u975e\u76e3\u7763\u5f0f\u6a5f\u5668\u5b78\u7fd2\u4e4b\u6a21\u578b\u6280\u8853\u3002</td><td/></tr><tr><td colspan=\"2\">\u5728\u904e\u53bb\u5e7e\uf98e\u4e2d\uff0c\u8a9e\u8a00\u6a21\u578b\u5728\u8cc7\u8a0a\u6aa2\u7d22\u4efb\u52d9\u4e2d\u5df2\u88ab\u5ee3\u6cdb\u5730\u61c9\u7528\u4e14\u6709\u4e0d\u932f\u7684\u5be6\u52d9\u6210\u6548</td></tr><tr><td colspan=\"2\">(Zhai, 2008)\uff1b\u4f46\u5c31\u6211\u5011\u6240\u77e5\uff0c\u5728\u8a9e\u97f3\u6587\u4ef6\u6458\u8981\u7684\u4efb\u52d9\u4e0a\uff0c\u95dc\u65bc\u4f7f\u7528\u8a9e\u8a00\u6a21\u578b\u7684\u7814\u7a76\u662f\u76f8</td></tr><tr><td colspan=\"2\">\u5c0d\u8f03\u5c11\u7684\u3002\u672c\u8ad6\u6587\u5c07\u85c9\u7531\u8a9e\u8a00\u6a21\u578b\u7684\u4f7f\u7528\u4f86\u9032\u884c\u6458\u8981\u8a9e\u53e5\u9078\u53d6\uff0c\u5176\u57fa\u672c\u65b9\u6cd5\u70ba\u4f7f\u7528\u8a9e\u53e5</td></tr><tr><td colspan=\"2\">\u8a9e\u8a00\u6a21\u578b\u751f\u6210\u6587\u4ef6\u7684\u6587\u4ef6\u76f8\u4f3c\u5ea6\u91cf\u503c(Document Likelihood Measure, DLM)(Chen et al.,</td></tr><tr><td colspan=\"2\">2009) \u3002\u6b64\u5916\uff0c\u672c\u7ae0\u7b2c 2 \u5c0f\u7bc0\u6211\u5011\u5c07\u95e1\u8ff0\u5982\u4f55\u4f7f\u7528\u57fa\u65bc\u95dc\u806f\u6027\u8cc7\u8a0a\u4f86\u6539\u9032\u8a9e\u53e5\u6a21\u578b\u4e4b\u4f30\u6e2c\uff0c</td></tr><tr><td>\u4f7f\u5176\u5f97\u4ee5\u66f4\u7cbe\u6e96\u7684\u4ee3\u8868\u8a9e\u53e5\u7684\u8a9e\u610f\u5167\u5bb9\u3002</td><td/></tr><tr><td colspan=\"2\">\u6211\u5011\u53ef\u4ee5\u628a\u8a9e\u97f3\u6587\u4ef6\u6458\u8981\u4efb\u52d9\u8996\u70ba\u662f\u8cc7\u8a0a\u6aa2\u7d22\u7684\u554f\u984c\u3002\u4e00\u822c\u4f86\u8aaa\uff0c\u8cc7\u8a0a\u6aa2\u7d22(Information</td></tr><tr><td colspan=\"2\">Retrieval, IR)\u65e8\u5728\u5c0b\u627e\u76f8\u95dc\u6587\u4ef6(Relevant Document)\u4f86\u56de\u61c9\u4f7f\u7528\u8005\u6240\u9001\u51fa\u7684\u67e5\u8a62(Query)</td></tr><tr><td colspan=\"2\">\u6216\u8cc7\u8a0a\u9700\u6c42(Information Need)\u3002\u540c\u6a23\u5730\uff0c\u5728\u5f9e\u4e8b\u8a9e\u97f3\u6587\u4ef6\u6458\u8981\u6642\uff0c\u6211\u5011\u53ef\u5c07\u6bcf\u4e00\u7bc7\u88ab\u6458</td></tr><tr><td colspan=\"2\">\u8981\u6587\u4ef6\u8996\u70ba\u662f\u67e5\u8a62\uff0c\u800c\u6587\u4ef6\u4e2d\u7684\u8a9e\u53e5(Sentence)\u8996\u70ba\u5019\u9078\u8cc7\u8a0a\u55ae\u5143(Candidate Information</td></tr><tr><td colspan=\"2\">Unit)\uff1b\u64da\u6b64\uff0c\u6211\u5011\u53ef\u4ee5\u5047\u8a2d\u5728\u88ab\u6458\u8981\u6587\u4ef6\u4e2d\uff0c\u8207\u5176\u6108\u76f8\u95dc\u7684\u8a9e\u53e5\u6108\u6709\u53ef\u80fd\u662f\u53ef\u7528\u4f86\u4ee3\u8868</td></tr><tr><td>\u6587\u4ef6\u4e3b\u65e8\u6216\u4e3b\u984c\u4e4b\u6458\u8981\u8a9e\u53e5\u3002</td><td/></tr><tr><td colspan=\"2\">\u7576\u7d66\u4e88\u4e00\u7bc7\u88ab\u6458\u8981\u6587\u4ef6 D \u6642\uff0c\u6587\u4ef6\u4e2d\u6bcf\u4e00\u8a9e\u53e5 S \u7684\u4e8b\u5f8c\u6a5f\uf961 P(S|D)\u53ef\u4ee5\u7528\u4f86\u8868\u793a\u8a9e</td></tr><tr><td colspan=\"2\">)\u3002 \u53e5 S \u5c0d\u65bc\u6587\u4ef6 D \u7684\u91cd\u8981\u6027\u3002\u7576\u4f7f\u7528\u8a9e\u8a00\u6a21\u578b\u4f86\u8a08\u7b97 P(S|D)\u6642\uff0c\u6211\u5011\u900f\u904e\u8c9d\u6c0f\u5b9a\u7406(Bayes'</td></tr><tr><td colspan=\"2\">\u6b64\u5916\uff0c\u6587\u5b57\u6587\u4ef6\u6240\u8981\u5f37\u8abf\u7684\u662f\u600e\u9ebc\u8aaa(What-is-said)\uff0c\u800c\u8a9e\u97f3\u6587\u4ef6\u64c1\u6709\u8a31\u591a\u7d14\u6587\u5b57\u6587</td></tr><tr><td colspan=\"2\">\u4ef6\u6240\u6c92\u6709\u7684\u8cc7\u8a0a\uff0c\u901a\u5e38\u9664\u4e86\u600e\u9ebc\u8aaa\uff0c\u66f4\u5f37\u8abf\u7684\u662f\u5982\u4f55\u8aaa(How-is-said)(Penn &amp; Zhu, 2008)\uff0c</td></tr><tr><td colspan=\"2\">\u660e\u986f\u5730\uff0c\u8a9e\u97f3\u662f\u591a\u5a92\u9ad4\u5167\u6db5\u4e2d\u6700\u5177\u8cc7\u8a0a\u7684\u6210\u5206\u4e4b\u4e00\uff0c\u4e5f\u56e0\u6b64\u8a9e\u97f3\u6587\u4ef6\u6458\u8981\u7684\u76f8\u95dc\u7814\u7a76\u901a</td></tr><tr><td colspan=\"2\">\u5e38\u5f9e\u591a\u5a92\u9ad4\u8a9e\u97f3\u8a0a\u865f\u4e2d\u8403\u53d6\u8c50\u5bcc\u7684\u97fb\u5f8b\u8cc7\u8a0a(Prosodic Information)\u4f86\u5224\u65b7\u8a9e\u53e5\u7684\u91cd\u8981\u6027\uff0c</td></tr></table>",
                "text": "",
                "type_str": "table",
                "num": null,
                "html": null
            },
            "TABREF6": {
                "content": "<table><tr><td>\u8a13\u7df4\u96c6 2001/11/07-2002/01/22 185 129.4 326.0 20.0 28.8% 38.0% 5. \u5be6\u9a57\u8a9e\u6599\u53ca\u8a55\u4f30\u65b9\u6cd5 (Dataset and Evaluation Method) \u8a9e\u6599\u6642\u9593 2002/01/23-2002/08/22 \u6e2c\u8a66\u96c6 \u6587\u4ef6\u500b\u6578 20 \u6587\u4ef6\u5e73\u5747\u6301\u7e8c\u5e7e\u79d2 141.2 \u6587\u4ef6\u5e73\u5747\u8a5e\u500b\u6578 290.3 \u6587\u4ef6\u5e73\u5747\u8a9e\u53e5\u500b\u6578 23.3 \u6587\u4ef6\u5e73\u5747\u5b57\u932f\u8aa4\uf961 (Character Error Rate, CER) 29.8% \u6587\u4ef6\u5e73\u5747\u8a5e\u932f\u8aa4\uf961 (Word Error Rate, WER) 39.4% 5.1 \u5be6\u9a57\u8a9e\u6599 (Dataset) \u672c \uf941 \u6587 \u5be6 \u9a57 \u8a9e \u6599 \u5eab \u70ba \u516c \u8996 \u65b0 \u805e \u8a9e \u6599 (Mandarin Chinese Broadcast News Corpus, MATBN)(Wang, Chen, Kuo &amp; Cheng, 2005)\uff0c\u662f\u7531\u4e2d\u592e\u7814\u7a76\u9662\u8cc7\u8a0a\u79d1\u5b78\u7814\u7a76\u6240\u8017\u6642\u4e09\uf98e\u8207 \u516c\u5171\u96fb\u8996\u53f0\u5408\u4f5c\uf93f\u88fd\u4e26\u6574\u7406\u7684\u4e2d\u6587\u65b0\u805e\u8a9e\u6599\uff0c\u5176\uf93f\u88fd\u5167\u5bb9\u70ba\u6bcf\u5929\u4e00\u500b\u5c0f\u6642\u7684\u516c\u8996\u665a\u9593\u65b0 \u805e\u6df1\u5ea6\u5831\u5c0e\u3002\u6211\u5011\u62bd\u53d6\u5176\u4e2d\u7531 2001 \uf98e 11 \u6708\u5230 2002 \uf98e 8 \u6708\u7e3d\u5171 205 \u5247\u65b0\u805e\u5831\u5c0e\uff0c\u5340\u5206\u6210 \u8a13\u7df4\u96c6(\u5171 185 \u5247\u65b0\u805e)\u4ee5\u53ca\u6e2c\u8a66\u96c6(\u5171 20 \u5247\u65b0\u805e)\u5169\u90e8\u5206\uff0c\u5176\u8a73\u7d30\u7684\u7d71\u8a08\u8cc7\u8a0a\u5982\u8868 1 \u6240\u793a\u3002 \u5168\u90e8 205 \u5247\u8a9e\u97f3\u6587\u4ef6\u9577\u5ea6\u7d04\u70ba 7.5 \u5c0f\u6642\uff0c\u6211\u5011\u5148\u505a\u4eba\u5de5\u5207\u97f3\uff0c\u5207\u51fa\u771f\u6b63\u542b\u6709\u8b1b\u8a71\u5167\u5bb9\u7684 \u97f3\u8a0a\u6bb5\u843d\uff0c\u518d\u7d93\u7531\u8a9e\u97f3\u8fa8\u8b58\u5668\u81ea\u52d5\u7522\u751f\u51fa\u7684\u8a9e\u97f3\u8fa8\u8b58\u7d50\u679c\u7a31\u4e4b\u70ba\u8a9e\u97f3\u6587\u4ef6(Spoken Document, SD)\uff0c\u56e0\u6b64\u8a9e\u97f3\u6587\u4ef6\u4e2d\u53ea\u5305\u542b\u6709\u8a9e\u97f3\u8fa8\u8b58\u932f\u8aa4\u4e4b\u96dc\u8a0a\uff1b\u53e6\u4e00\u65b9\u9762\uff0c\u6211\u5011\u5c07\u6b64 205 \u5247\u8a9e\u97f3\u6587\u4ef6\u85c9\u7531\u4eba\u5de5\u807d\u5beb\u7684\u65b9\u5f0f\uff0c\u7522\u751f\u51fa\u6c92\u6709\u8fa8\u8b58\u932f\u8aa4\u7684\u6b63\u78ba\u6587\u5b57\u8a9e\u6599\uff0c\u6211\u5011\u7a31\u4e4b\u70ba\u6587 \u5b57\u6587\u4ef6(Text Document, TD)\uff0c\u6bcf\u5247\u6587\u5b57\u6587\u4ef6\u518d\u7d93\u7531\u4e09\u4f4d\u5c08\u5bb6\u6a19\u8a18\u6458\u8981\u8a9e\u53e5\uff0c\u6211\u5011\u5c07\u6b64\u6a19 \u8a18\u7684\u4eba\u5de5\u6458\u8981\u505a\u70ba\u8a9e\u97f3\u6587\u4ef6\u8207\u6587\u5b57\u6587\u4ef6\u7684\u6b63\u78ba\u6458\u8981\u7b54\u6848\u3002\u85c9\u7531\u6bd4\u8f03\u8a9e\u97f3\u6587\u4ef6\u548c\u6587\u5b57\u6587\u4ef6 \u7684\u6458\u8981\u6548\u80fd\uff0c\u6211\u5011\u53ef\u4ee5\u89c0\u5bdf\u8a9e\u97f3\u8fa8\u8b58\u932f\u8aa4\u5c0d\u65bc\u5404\u7a2e\u6458\u8981\u65b9\u6cd5\u4e4b\u5f71\u97ff\u3002\u672c\u7814\u7a76\u7684\u80cc\u666f\u8a9e\u8a00 \u6a21\u578b\u8a13\u7df4\u8a9e\u6599\u53d6\u6750\u81ea 2001 \u5230 2002 \uf98e\u7684\u4e2d\u592e\u793e\u65b0\u805e\u6587\u5b57\u8a9e\u6599(Central News Agency, CNA)\uff0c \u4e26\u4e14\u4ee5 SRI \u8a9e\u8a00\u6a21\u578b\u5de5\u5177\u8a13\u7df4\u51fa\u7d93\u5e73\u6ed1\u5316\u7684\u55ae\u9023\u8a9e\u8a00\u6a21\u578b\uff0c\u6211\u5011\u5047\u8a2d\u6b64\u55ae\u9023\u8a9e\u8a00\u6a21\u578b\u70ba \u660e\u78ba\u5ea6\u4e2d\u7684\u975e\u76f8\u95dc\u8cc7\u8a0a\u4e4b\u4f86\u6e90\u3002\u53e6\u5916\uff0c\u672c\u8ad6\u6587\u8490\u96c6 2002 \uf98e\u4e2d\u592e\u901a\u8a0a\u793e\u7684\u7d04\u5341\u842c\u5247\u540c\u6642\u671f \u65b0\u805e\u6587\u5b57\u6587\u4ef6\u505a\u70ba\u5efa\uf9f7\u95dc\u806f\u6a21\u578b\u6642\u7684\u6aa2\u7d22\u6a19\u7684(Chen et al., 2013)\uff0c\u95dc\u65bc\u8a9e\u53e5 S \u7684\u865b\u64ec\u76f8\u95dc \u6587\u4ef6(\u6700\u9ad8\u6392\u5e8f\u6587\u4ef6)\u7bc7\u6578\u70ba 15(\u4e5f\u5c31\u662f|D Top |=15)\u3002 5.2 \u8a55\u4f30\u65b9\u6cd5 (Evaluation Method) \u81ea\u52d5\u6458\u8981\u7684\u8a55\u4f30\u65b9\u6cd5\u4e3b\u8981\u6709\u5169\u7a2e\uff0c\u4e00\u70ba\u4e3b\u89c0\u4eba\u70ba\u8a55\u4f30\uff0c\u53e6\u4e00\u70ba\u5ba2\u89c0\u81ea\u52d5\u8a55\u4f30\uff1b\u524d\u8005\u70ba\u8acb \u5e7e\u4f4d\u6e2c\u8a66\u4eba\u54e1\u4f86\u70ba\u7cfb\u7d71\u6240\u7522\u751f\u7684\u6458\u8981\u505a\u8a55\u4f30\uff0c\u7d66\u5206\u7684\u7bc4\u570d\u70ba 1-5 \u5206\uff0c\u5f8c\u8005\u5247\u662f\u9810\u5148\u8acb\u5e7e \u4f4d\u6e2c\u8a66\u8005\u4f9d\u64da\u4e8b\u5148\u5b9a\u7fa9\u597d\u7684\u6458\u8981\u6bd4\u4f8b\u6311\u9078\u51fa\u9069\u5408\u7684\u6458\u8981\u8a9e\u53e5\uff0c\u7cfb\u7d71\u6240\u7522\u751f\u7684\u6458\u8981\u53e5\u5b50\u5c07 \u8207\u6e2c\u8a66\u8005\u6240\u6311\u9078\u51fa\u7684\u53e5\u5b50\u8a08\u7b97\u53ec\u56de\uf961\u5c0e\u5411\u7684\u8981\u9ede\u8a55\u4f30(Recall-Oriented Understudy for Gisting Evaluation, ROUGE)(Lin, 2003)\u3002\u7531\u65bc\u4e3b\u89c0\u4eba\u70ba\u8a55\u4f30\u975e\u5e38\u8017\u6642\u8017\u529b\uff0c\u6240\u4ee5\u76ee\u524d\u591a\u6578 \u81ea\u52d5\u6458\u8981\u65b9\u6cd5\u7686\u63a1\u7528\u53ec\u56de\uf961\u5c0e\u5411\u7684\u8981\u9ede\u8a55\u4f30\u505a\u70ba\u6587\u4ef6\u6458\u8981\u7684\u8a55\u4f30\u65b9\u5f0f\uff0c\u672c\u8ad6\u6587\u4ea6\u63a1\u7528\u6b64 \u7a2e\u8a55\u4f30\u65b9\u5f0f\u3002ROUGE \u65b9\u6cd5\u662f\u8a08\u7b97\u81ea\u52d5\u6458\u8981\u7d50\u679c\u8207\u4eba\u5de5\u6458\u8981\u4e4b\u9593\u7684\u91cd\u758a\u55ae\u4f4d\u5143(Units)\u6578\u76ee \u5360\u53c3\u8003\u6458\u8981(Reference Summary)\u9577\u5ea6(\u55ae\u4f4d\u5143\u7e3d\u500b\u6578)\u7684\u6bd4\u4f8b\u3002\u4f30\u8a08\u7684\u55ae\u4f4d\u53ef\u4ee5\u662f N-\u9023\u8a5e (N-gram)\u3001\u8a5e\u5e8f\u5217(Word Sequences)\uff0c\u5982\uff1a\u6700\u9577\u76f8\u540c\u8a5e\u5e8f\u5217\u6216\u8a5e\u6210\u5c0d(Word Pairs)\u3002\u7531\u65bc\u6b64 \u65b9\u6cd5\u662f\u63a1\u7528\u55ae\u4f4d\u5143\u6bd4\u5c0d\u7684\u65b9\u5f0f\uff0c\u4e0d\u6703\u7522\u751f\u8a9e\u53e5\u908a\u754c\u5b9a\u7fa9\u7684\u554f\u984c\uff0c\u4e26\u4e14\u9069\u5408\u65bc\u591a\u4efd\u4eba\u5de5\u6458 \u8981\u7684\u8a55\u4f30\u3002\u5176\u8a55\u4f30\u7684\u5206\u6578\u6709\u4e09\u7a2e\uff0cROUGE-1 (\u55ae\u9023\u8a5e Unigram)\u3001ROUGE-2 (\u96d9\u9023\u8a5e Bigram) \u548c ROUGE-L (\u6700\u9577\u5171\u540c\u7247\u6bb5 Longest Common Subsequence) \u5206\u6578\uff0cROUGE-1 \u662f\u8a55\u4f30\u81ea\u52d5 \u6458\u8981\u7684\u8a0a\u606f\u91cf\uff0cROUGE-2 \u662f\u8a55\u4f30\u81ea\u52d5\u6458\u8981\u7684\u6d41\u66a2\u6027\uff0cROUGE-L \u662f\u6700\u9577\u5171\u540c\u5b57\uf905\uff0c\u672c\uf941 \u6587\u5e0c\u671b\u89c0\u5bdf\u6458\u8981\u7684\u6d41\u66a2\u6027\uff0c\u56e0\u6b64\uff0c\u5be6\u9a57\u6578\u64da\u4e3b\u8981\u662f\u4ee5 ROUGE-2 \u5206\u6578\u70ba\u4e3b\u3002\u672c\u8ad6\u6587\u6240\u8a2d \u5289\u58eb\u5f18 \u7b49 \u5f59\u505a\u70ba\u5224\u65b7\u6458\u8981\u6bd4\u4f8b\u7684\u55ae\u5143\u3002\u5728\u6311\u9078\u6458\u8981\u8a9e\u53e5\u904e\u7a0b\u4e2d\uff0c\u82e5\u9078\u5230\u67d0\u8a9e\u53e5\u4e2d\u7684\u67d0\u500b\u8a5e\u5f59\u6642\u5c31 \u5df2\u7d93\u525b\u597d\u9054\u5230\u6458\u8981\u6bd4\u4f8b\uff0c\u70ba\u4e86\u4fdd\u6301\u8a9e\u53e5\u8a9e\u610f\u5b8c\u6574\u6027\uff0c\u6b64\u8a9e\u53e5\u5269\u4e0b\u7684\u8a5e\u5f59\u4e5f\u6703\u88ab\u6311\u9078\u6210\u70ba \u6458\u8981\u3002 6. \u5be6\u9a57\u7d50\u679c (Experimental Results) 6.1 \u57fa\u790e\u5be6\u9a57\u7d50\u679c (Baseline Experiments) \u9996\u5148\uff0c\u6211\u5011\u6bd4\u8f03\u6587\u4ef6\u76f8\u4f3c\u5ea6\u91cf\u503c(DLM)\u8207\u6578\u500b\u975e\u76e3\u7763\u5f0f\u6458\u8981\u65b9\u6cd5\u4e4b\u6458\u8981\u6210\u6548\uff0c\u5305\u542b\u6709\u6700 \u9577\u8a9e\u53e5\u6458\u8981(Longest Sentence, LS)\u3001\u9996\u53e5\u6458\u8981(LEAD)(Penn &amp; Zhu, 2008)\u3001\u5411\u91cf\u7a7a\u9593\u6a21\u578b (Vector Space Model, VSM)(Gong &amp;. Liu, 2001)\u3001\u6f5b\u85cf\u8a9e\u610f\u5206\u6790(Latent Semantic Analysis, LSA)(Gong &amp;. Liu, 2001)\u3001\u6700\u5927\u908a\u969b\u95dc\u806f(Maximal Marginal Relevance, MMR)(Carbonell &amp; Goldstein, 1998)\u3001\u99ac\u53ef\u592b\u96a8\u6a5f\u6f2b\u6b65(Markov Random Walk, MRW)(Wan &amp; Yang, 2008)\u3001\u6b21 \u6a21(Submodularity)(Lin &amp; Bilmes, 2010)\u4ee5\u53ca\u6574\u6578\u7dda\u6027\u898f\u5283(Integer Linear Programming, ILP)(McDonald, 2007)\u3002\u4e00\u822c\u4f86\u8aaa\uff0c\u6587\u4ef6\u4e2d\u9577\u53e5\u53ef\u80fd\u860a\u542b\u6709\u8f03\u8c50\u5bcc\u7684\u4e3b\u984c\u8cc7\u8a0a\uff0c\u56e0\u6b64\u4f9d\u64da \u6587\u4ef6\u4e2d\u8a9e\u53e5\u9577\u5ea6\u505a\u6392\u5e8f\u5f8c\uff0c\u4f9d\u5e8f\u9078\u53d6\u6700\u9577\u8a9e\u53e5\u505a\u70ba\u6458\u8981\u7d50\u679c\u662f\u4e00\u7a2e\u7c21\u55ae\u7684\u6458\u8981\u65b9\u6cd5\u3002\u9664 \u6b64\u4e4b\u5916\uff0c\u4e5f\u6709\u5b78\u8005\u7814\u7a76\u767c\u73fe\uff0c\u6587\u4ef6\u5e38\u4ee5\u958b\u9580\u898b\u5c71\u6cd5\u7684\u65b9\u5f0f\u4f86\u63d0\u9ede\u51fa\u4e3b\u984c\uff0c\u56e0\u6b64\u6587\u4ef6\u958b\u982d \u7684\u524d\u5e7e\u500b\u8a9e\u53e5\u7d93\u5e38\u662f\u5177\u4ee3\u8868\u6027\u7684\u8a9e\u53e5\uff0c\u9996\u53e5\u6458\u8981\u5373\u662f\u4ee5\u6b64\u6982\u5ff5\u51fa\u767c\uff0c\u9078\u53d6\u524d\u5e7e\u53e5\u8a9e\u53e5\u4f86 \u5f62\u6210\u6574\u500b\u6587\u4ef6\u7684\u6458\u8981\u3002\u6700\u9577\u8a9e\u53e5\u6458\u8981(LS)\u53ca\u9996\u53e5\u6458\u8981(LEAD)\u90fd\u50c5\u9069\u7528\u5728\u4e00\u90e8\u5206\u5177\u6709\u7279\u6b8a \u7d50\u69cb\u7684\u6587\u4ef6\u4e0a\uff0c\u56e0\u6b64\u5b83\u5011\u7684\u7f3a\u9ede\u5c31\u662f\u6709\u5176\u4fb7\u9650\u6027\u3002\u53e6\u5916\uff0c\u5411\u91cf\u7a7a\u9593\u6a21\u578b\u662f\u628a\u6587\u4ef6\u548c\u8a9e\u53e5 \u5206\u5225\u8996\u70ba\u4e00\u500b\u5411\u91cf\uff0c\u4e26\u4f7f\u7528\u8a5e\u983b-\u53cd\u6587\u4ef6\u983b(TF-IDF)\u7279\u5fb5\u4f86\u8a08\u7b97\u6bcf\u4e00\u7dad\u5ea6\u7684\u6b0a\u91cd\u503c\uff0c\u6587\u4ef6 \u8207\u8a9e\u53e5\u9593\u7684\u95dc\u806f\u6027\u662f\u85c9\u7531\u9918\u5f26\u76f8\u4f3c\u5ea6\u91cf\u503c\u4f86\u4f30\u6e2c\uff0c\u7576\u8a9e\u53e5\u5206\u6578\u8f03\u9ad8\u6642\uff0c\u5247\u8d8a\u6709\u6a5f\u6703\u6210\u70ba \u6b64 \u6587\u4ef6 \u7684\u6458\u8981 \u3002\u6f5b \u85cf\u8a9e\u610f \u5206\u6790 \u662f\u5728\u5411 \u91cf\u7a7a \u9593\u7684\u5047 \u8a2d\u4e0b \u66f4\u9032\u4e00 \u6b65\u5730 \u4f7f\u7528\u5947 \u7570\u503c \u5206\u89e3 (Singular Value Decomposition, SVD)\u4f86\u627e\u5230\u53ef\u80fd\u7684\u6f5b\u85cf\u8a9e\u610f\u7a7a\u9593\uff0c\u4f7f\u4e4b\u80fd\u5728\u8003\u91cf\u6f5b\u85cf\u8a9e \u610f\u7684\u60c5\u6cc1\u4e0b\u9032\u884c\u6587\u4ef6\u8207\u8a9e\u53e5\u7684\u95dc\u806f\u6027\u91cf\u6e2c\u3002\u6700\u5927\u908a\u969b\u95dc\u806f\u53ef\u8996\u70ba\u662f\u5411\u91cf\u7a7a\u9593\u6a21\u578b\u7684\u4e00\u500b \u5ef6\u4f38\uff0c\u5728\u505a\u8a9e\u53e5\u6392\u5e8f\u6642\u8003\u91cf\u4e86\u5197\u9918\u6027\u4ee5\u9054\u5230\u66f4\u597d\u7684\u6458\u8981\u7d50\u679c\u3002\u99ac\u53ef\u592b\u96a8\u6a5f\u6f2b\u6b65(MRW)\u7684 \u6982\u5ff5\u662f\u628a\u6587\u4ef6\u4e2d\u7684\u8a9e\u53e5\u8996\u70ba\u4e00\u500b\u7db2\u969b\u7db2\u8def\uff0c\u6587\u4ef6\u4e2d\u7684\u8a9e\u53e5\u4ee3\u8868\u7db2\u8def\u4e2d\u7684\u7bc0\u9ede\uff0c\u7bc0\u908a\u6b0a\u91cd \u503c\u662f\u5169\u500b\u7bc0\u9ede\u4e4b\u9593\u7684\u8a9e\u5f59\u76f8\u4f3c\u5ea6\uff0c\u901a\u5e38\u662f\u900f\u904e\u7bc0\u9ede\u7684\u5167\u5206\u652f\u5ea6(Indegree)\u8207\u5916\u5206\u652f\u5ea6 (Outdegree)\u4e26\u63a1\u7528\u9918\u5f26(Cosine)\u4f30\u6e2c\u6cd5\u6c42\u5f97\uff0c\u6240\u4ee5\u99ac\u53ef\u592b\u96a8\u6a5f\u6f2b\u6b65\u4e3b\u8981\u662f\u4f9d\u8cf4\u8f03\u4e00\u822c\u5316\u7684 \u8cc7\u8a0a\uff0c\u4f8b\u5982\uff1a\u6709\u6982\u5ff5\u6027\u7684\u7db2\u969b\u7db2\u8def\uff0c\u800c\u4e0d\u662f\u8003\u616e\u5340\u57df\u6027\u7684\u7279\u5fb5(\u4f8b\u5982\uff1a\u6bcf\u500b\u8a9e\u53e5)\uff0c\u56e0\u6b64 \u5982\u679c\u6709\u4e00\u500b\u8a9e\u53e5\u8ddf\u5176\u4ed6\u8a9e\u53e5\u5f88\u76f8\u4f3c\u7684\u8a71\uff0c\u5247\u53ef\u4ee5\u4ee3\u8868\u6458\u8981\u4f7f\u4e4b\u4f86\u63cf\u8ff0\u6587\u4ef6\u4e2d\u7684\u4e3b\u65e8(Wan &amp; Yang, 2008)\u3002\u6b21\u6a21\u662f\u4e00\u500b\u8caa\u5a6a(Greedy) \u7684\u8a9e\u53e5\u9078\u53d6\u65b9\u6cd5\uff0c\u56e0\u5176\u6eff\u8db3\u6b21\u6a21\u7684\u7279\u6027\uff0c\u610f\u5373 \u6bcf\u9078\u53d6\u4e00\u8a9e\u53e5\u5c31\u6703\u6709\u56de\u5831\u6e1b\u5c11(Diminished Return)\u7684\u6548\u61c9\uff0c\u56e0\u6b64\u6b21\u6a21\u5177\u6709\u4e00\u500b\u8fd1\u4f3c\u6700\u4f73\u89e3 (Near-Optimal)(Lin &amp; Bilmes, 2010)\u3002\u6574\u6578\u7dda\u6027\u898f\u5283\u662f\u4e00\u500b\u5168\u57df(Global)\u7684\u9650\u5236\u6027\u6700\u4f73\u5316 (Constraint Optimization)\u7684\u8a9e\u53e5\u9078\u53d6\u65b9\u6cd5(McDonald, 2007)\u3002 LEAD\u3001VSM\u3001LSA\u3001MMR \u7b49\u975e\u76e3\u7763\u5f0f\u6458\u8981\u65b9\u6cd5\u4f86\u5f97\u597d\u4e9b\uff1b\u56e0 LS \u8207 LEAD \u50c5\u9069\u7528\u65bc\u7279 \u5f71\u97ff\u3002 \u8868 2 \u70ba\u672c\u8ad6\u6587\u4e4b\u57fa\u790e\u5be6\u9a57\u7d50\u679c\u3002\u9996\u5148\uff0c\u5728 TD \u7684\u5be6\u9a57\u4e2d\uff0cDLM \u7684\u6458\u8981\u6548\u679c\u6bd4 LS\u3001 \u8868 2. \u57fa\u790e\u5be6\u9a57\u7d50\u679c [Table 2. Baseline experiments.] F-score ROUGE-1 ROUGE-2 ROUGE-L TD LS 0.225 0.098 0.183 LEAD 0.310 0.194 0.276 VSM 0.347 0.228 0.290 LSA 0.362 0.233 0.316 MMR 0.368 0.248 0.322 MRW 0.412 0.282 0.358 Submodularity 0.414 0.286 0.363 ILP 0.442 0.337 0.401 DLM 0.411 0.298 0.361 SD LS 0.181 0.044 0.138 LEAD 0.255 0.117 0.221 VSM 0.342 0.189 0.287 LSA 0.345 0.201 0.301 MMR 0.366 0.215 0.315 MRW 0.332 0.191 0.291 Submodularity 0.332 0.204 0.303 ILP 0.348 0.209 0.306 DLM 0.364 0.210 0.307 \u76f8\u8f03\u4e4b\u4e0b\uff0cDLM \u662f\u8f03\u5177\u4e00\u822c\u6027\u7684\u6458\u8981\u65b9\u6cd5\uff0c\u56e0\u6b64\u6bd4\u8f03\u4e0d\u6703\u53d7\u9650\u65bc\u6587\u7ae0\u7684\u7d50\u69cb\u4e4b\u5f71\u97ff\uff0c\u6545 \u6458\u8981\u6548\u80fd\u6bd4 LS \u4ee5\u53ca LEAD \u4f86\u5f97\u5f70\u986f\u3002DLM \u8207 VSM \u7686\u4f7f\u7528\u6dfa\u5c64\u7684\u8a5e\u5f59(\u8a5e\u983b)\u8cc7\u8a0a\uff0c\u4f46 \u7531\u65bc DLM \u662f\u8a08\u7b97\u8a9e\u53e5\u6a21\u578b\u8207\u6587\u4ef6\u6a21\u578b\u4e4b\u9593\u7684\u8ddd\u96e2\u95dc\u4fc2\uff0c\u5c0d\u65bc\u4ee3\u8868\u8a9e\u53e5\u8207\u6587\u4ef6\u7684\u8a9e\u8a00\u6a21 \u578b\uff0c\u6211\u5011\u8f03\u5bb9\u6613\u900f\u904e\u5404\u7a2e\u6280\u8853\u4f86\u9032\u884c\u6a21\u578b\u7684\u4f30\u8a08\u8207\u8abf\u9069\uff0c\u9032\u800c\u7372\u5f97\u8f03\u597d\u7684\u6458\u8981\u6210\u679c\u3002\u6574 \u6578\u7dda\u6027\u898f\u5283\u662f\u4e00\u500b\u5168\u57df\u9078\u64c7\u65b9\u6cd5\uff0c\u6240\u4ee5\u5728 TD \u4e0a\u53ef\u4ee5\u5f97\u5230\u6700\u597d\u7684\u6458\u8981\u6548\u80fd\u3002 \u80fd\uff0c\u7d50\u679c\u53cd\u800c\u662f MMR \u5f97\u5230\u6700\u597d\u7684\u6458\u8981\u6548\u80fd\uff0c\u53ef\u80fd\u7684\u539f\u56e0\u662f ILP \u53d7\u5230\u8a9e\u97f3\u8fa8\u8b58\u932f\u8aa4\u7684\u5f71 \u97ff\u6bd4\u8f03\u5927\uff0c\u9020\u6210\u5176\u6458\u8981\u7d50\u679c\u4e0d\u5f70\u3002 \u901a\u5e38\u8a9e\u97f3\u6587\u4ef6\u4e3b\u8981\u6703\u6709\u8a9e\u97f3\u8fa8\u8b58\u932f\u8aa4\u548c\u8a9e\u53e5\u908a\u754c\u5075\u6e2c\u932f\u8aa4\u7684\u554f\u984c\uff0c\u4f46\u6211\u5011\u6709\u5148\u7d93\u4eba \u5de5\u5207\u97f3\uff0c\u56e0\u6b64\u6452\u9664\u4e86\u8a9e\u53e5\u908a\u754c\u5075\u6e2c\u932f\u8aa4\u7684\u554f\u984c\uff0c\u85c9\u7531\u6bd4\u8f03 TD \u8207 SD \u4e4b\u5be6\u9a57\u7d50\u679c\uff0c\u6211\u5011 \u53ef\u4ee5\u89c0\u5bdf\u8a9e\u97f3\u8fa8\u8b58\u932f\u8aa4\u7387\u5c0d\u6458\u8981\u7d50\u679c\u7684\u5f71\u97ff\u6027\u3002\u6bd4\u8f03\u5404\u5f0f\u65b9\u6cd5\uff0cSD \u6bd4 TD \u4e0b\u964d\u4e86 1.9%~8.8%\u7684 ROUGE-2 \u6458\u8981\u6548\u80fd\uff0c\u7531\u6b64\u53ef\u77e5\u8a9e\u97f3\u8fa8\u8b58\u932f\u8aa4\u7387\u5c0d\u6458\u8981\u6548\u80fd\u662f\u6709\u986f\u8457\u7684\u5f71\u97ff \u6027\u3002\u70ba\u4e86\u6e1b\u7de9\u8a9e\u97f3\u8fa8\u8a8d\u932f\u8aa4\u7684\u554f\u984c\uff0c\u5728\u672a\u4f86\u6211\u5011\u5c07\u5617\u8a66\u4f7f\u7528\u97f3\u7bc0(Syllable)\u70ba\u55ae\u4f4d\u4f86\u5efa\u7acb \u8a9e\u53e5\u4ee5\u53ca\u6587\u4ef6\u6a21\u578b\uff1b\u6216\u5229\u7528\u8a5e\u5716(Word Graph)\u3001\u6df7\u6dc6\u7db2\u8def(Confusion Network)\u4f86\u542b\u62ec\u66f4\u591a \u7684\u53ef\u80fd\u6b63\u78ba\u5019\u9078\u8a5e\u5f59\u4ee5\u88e8\u76ca\u6a21\u578b\u4f30\u6e2c\uff1b\u66f4\u53ef\u5229\u7528\u97fb\u5f8b\u8cc7\u8a0a(Prosodic Information)\u7b49\u8072\u5b78\u7dda \u7d22\u4f86\u8f14\u52a9\u6e1b\u7de9\u8a9e\u97f3\u8fa8\u8a8d\u932f\u8aa4\u5c0d\u6458\u8981\u6548\u80fd\u7684\u5f71\u97ff\u3002 6.2 \u95dc\u806f\u6a21\u578b\u4e4b\u5be6\u9a57\u7d50\u679c (Experiments of Relevance Model) \u4f7f\u7528\u95dc\u806f\u6a21\u578b\u65bc\u8a9e\u53e5\u6a21\u578b\u4e4b\u5efa\u7acb\u6642\uff0c\u9700\u8981\u505a\u4e00\u6b21\u7684\u8cc7\u8a0a\u6aa2\u7d22\u4f86\u70ba\u6bcf\u500b\u8a9e\u53e5\u627e\u51fa\u865b\u64ec\u76f8\u95dc \u6587\u4ef6\uff0c\u7531\u540c\u6642\u671f\u7684\u65b0\u805e\u6587\u5b57\u6587\u4ef6(\u5171 101,268 \u7bc7)\u4e2d\u70ba\u6bcf\u4e00\u8a9e\u53e5\u9078\u53d6\u51fa 15 \u7bc7\u865b\u64ec\u76f8\u95dc\u6587\u4ef6\u3002 \u7531\u65bc\u6587\u4ef6\u4e2d\u7684\u8a9e\u53e5\u901a\u5e38\u76f8\u5c0d\u7c21\u77ed\uff0c\u56e0\u6b64\u7576\u4f7f\u7528\u6700\u5927\u5316\u76f8\u4f3c\u5ea6\u4f30\u6e2c\u5efa\u7acb\u8a9e\u53e5\u6a21\u578b\u6642\uff0c\u5bb9\u6613 \u906d\u9047\u8cc7\u6599\u7a00\u758f\u7684\u554f\u984c\uff0c\u4e0d\u5bb9\u6613\u7372\u5f97\u7cbe\u6e96\u7684\u6a21\u578b\uff0c\u6545\u6211\u5011\u671f\u671b\u8003\u616e\u984d\u5916\u7684\u95dc\u806f\u8cc7\u8a0a\u65bc\u8a9e\u97f3 \u6587\u4ef6\u6458\u8981\uff0c\u4ea6\u5373\u85c9\u7531\u865b\u64ec\u76f8\u95dc\u6587\u4ef6\u4f86\u91cd\u65b0\u4f30\u6e2c\u4e26\u5efa\u7acb\u8a9e\u53e5\u7684\u8a9e\u8a00\u6a21\u578b\uff0c\u80fd\u7372\u5f97\u9032\u4e00\u6b65\u5730 \u6458\u8981\u6210\u6548\u3002\u91cd\u65b0\u4f30\u6e2c\u5f8c\u7684\u95dc\u806f\u6a21\u578b\u5247\u53ef\u8207\u539f\u672c\u7684\u8a9e\u53e5\u6a21\u578b\u76f8\u7d50\u5408\u6216\u53d6\u4ee3\u4e4b\uff0c\u76f8\u7d50\u5408\u7684\u53c3 \u6578\u8abf\u6574\u5728\u672c\u5be6\u9a57\u4e2d\u662f\u63a1\u7528\u7d93\u9a57\u8a2d\u5b9a(Empirical Setting) \u3002\u5be6\u9a57\u7d50\u679c\u5982\u8868 3 \u6240\u793a\uff0c\u5728 TD \u8207 SD \u4e4b\u6458\u8981\u6210\u6548\u4e0a\uff0c\u4f7f\u7528\u95dc\u806f\u6a21\u578b(RM)\u3001\u7c21\u55ae\u6df7\u5408\u6a21\u578b(SMM)\u53ca\u4e09\u6df7\u5408\u6a21\u578b(TriMM)\u7686\u80fd\u6bd4 \u57fa\u790e\u7684 DLM \u5be6\u9a57\u8f03\u597d\uff0c\u5c24\u5176\u662f\u4e09\u6df7\u5408\u6a21\u578b(TriMM)\u76f8\u8f03\u65bc DLM \u5728 TD \u53ca SD \u7684 ROUGE-2 \u7d50\u679c\u4e0a\u80fd\u6709 5.2%\u8207 1.8%\u7684\u6539\u9032\u3002\u63a5\u8457\uff0c\u6211\u5011\u6bd4\u8f03\u4e0d\u540c\u95dc\u806f\u6a21\u578b\u7684\u6458\u8981\u6210\u6548\uff0c\u9996\u5148\u662f\u95dc \u73fe\u6bd4\u7c21\u55ae\u6df7\u5408\u6a21\u578b\u4f86\u5f97\u597d\uff0c\u4f46\u5728 SD \u4f3c\u4e4e\u5728 ROUGE-1 \u5c31\u6c92\u6bd4\u7c21\u55ae\u6df7\u5408\u6a21\u578b\u597d\uff0c\u4e0d\u904e SD \u7684 ROUGE-2 \u8ddf ROUGE-L \u90fd\u9084\u662f\u6bd4\u7c21\u55ae\u6df7\u5408\u6a21\u578b\u7684\u6548\u679c\u597d\u3002\u95dc\u806f\u6a21\u578b\u7684\u5047\u8a2d\u662f\u5f37\u8abf\u8a5e \u5f59 w \u8207\u8a9e\u53e5 S \u5728\u9019\u4e9b\u865b\u64ec\u76f8\u95dc\u6587\u4ef6\u4e2d\u540c\u6642\u51fa\u73fe\u4e4b\u95dc\u4fc2(\u53c3\u7167\u5f0f(5))\u4f86\u4f30\u6e2c\u6a21\u578b\uff0c\u800c\u7c21\u55ae\u6df7 \u5408\u6a21\u578b\u662f\u5f37\u8abf\u8a13\u7df4\u597d\u7684\u6a21\u578b\u80fd\u8b93\u6709\u7368\u7279\u6027\u7684\u8a5e\u5f59\u5f97\u5230\u66f4\u591a\u7684\u6a5f\u7387\u503c\u56e0\u800c\u8b93\u6a21\u578b\u5177\u6709\u9451\u5225 \u80fd\u529b\uff0c\u5169\u8005\u7686\u6709\u5176\u597d\u8655\u3002\u6700\u5f8c\uff0c\u4e09\u6df7\u5408\u6a21\u578b(TriMM)\u56e0\u8907\u96dc\u5316\u4e86\u7c21\u55ae\u6df7\u5408\u6a21\u578b(SMM)\uff0c \u984d\u5916\u591a\u8003\u91cf\u6587\u4ef6\u6a21\u578b\u7684\u5f71\u97ff\u529b\uff0c\u56e0\u6b64\u76f8\u8f03\u65bc\u95dc\u806f\u6a21\u578b\u53ca\u7c21\u55ae\u6df7\u5408\u6a21\u578b\u80fd\u5f97\u5230\u66f4\u4f73\u7684\u6458\u8981 \u6548\u80fd\uff0c\u4e09\u6df7\u5408\u6a21\u578b\u76f8\u8f03\u65bc\u95dc\u806f\u6a21\u578b\u5728 TD \u4e0a\u6709\u660e\u986f\u7684\u9032\u6b65\uff0c\u65bc ROUGE-2 \u7d50\u679c\u80fd\u6709 1.4% \u7684\u6539\u9032\uff0c\u4f46\u5728 SD \u4e0a\uff0c\u65bc ROUGE-2 \u7d50\u679c\u53ea\u6709\u5fae\u91cf\u7684 0.2%\u6539\u5584\u3002 \u5728\u95dc\u806f\u6a21\u578b\u7684\u76f8\u95dc\u5be6\u9a57\u4e2d\uff0c\u8a9e\u97f3\u8fa8\u8b58\u932f\u8aa4\u4e5f\u662f\u5f71\u97ff\u6458\u8981\u6548\u80fd\u975e\u5e38\u56b4\u91cd\uff0c\u5728\u4e09\u6df7\u5408\u6a21 \u578b\u7684\u6578\u64da\u4e2d\uff0cSD \u6bd4 TD \u5287\u70c8\u4e0b\u964d\u4e86 12.2%\u7684 ROUGE-2 \u6458\u8981\u6548\u80fd\uff0c\u5728\u672a\u4f86\u7814\u7a76\u4e2d\uff0c\u6211\u5011 \u8a8d\u70ba\u53ef\u4ee5\u4ee5\u6b21\u8a5e\u7d22\u5f15(Subword Indexing)\u7684\u65b9\u5f0f\u4f86\u5efa\u7acb\u95dc\u806f\u6a21\u578b\u4ee5\u6e1b\u7de9\u8a9e\u97f3\u8fa8\u8b58\u932f\u8aa4\u4e4b \u7576\u4ee3\u975e\u76e3\u7763\u5f0f\u65b9\u6cd5\u4e4b\u6bd4\u8f03\u65bc\u7bc0\u9304\u5f0f\u8a9e\u97f3\u6458\u8981 19 \u8868 3\u3001\u95dc\u806f\u6a21\u578b\u4e4b\u5be6\u9a57\u7d50\u679c [F-score ROUGE-1 ROUGE-2 ROUGE-L TD DLM 0.411 0.298 0.361 RM 0.450 0.336 0.400 SMM 0.436 0.325 0.385 TriMM 0.457 0.350 0.404 SD DLM 0.364 0.210 0.307 RM 0.374 0.226 0.321 SMM 0.375 0.221 0.314 TriMM 0.379 0.228 0.325 6.3 \u5e73 \u6ed1 \u5316 \u6280 \u8853 \u65bc \u95dc \u806f \u6a21 \u578b \u4e4b \u5be6 \u9a57 \u7d50 \u679c (Experiments of Smoothing Methods for Relevance Model) \u8a9e\u8a00\u6a21\u578b\u5728\u4f7f\u7528\u6642\u6703\u9047\u5230\u8cc7\u6599\u7a00\u758f\u7684\u554f\u984c\uff0c\u901a\u5e38\u7684\u89e3\u6c7a\u65b9\u6cd5\u70ba\u66ff\u8a9e\u8a00\u6a21\u578b\u505a\u5e73\u6ed1\u5316 (Smoothing)\uff0c\u6211\u5011\u5c07\u63a2\u8a0e\u5e73\u6ed1\u5316\u6280\u8853\u65bc\u8a9e\u8a00\u6a21\u578b\u5728\u8a9e\u97f3(\u6587\u5b57)\u6458\u8981\u7d50\u679c\u4e0a\u7684\u5f71\u97ff\uff0c\u5728\u672c \u5c0f\u7bc0\u4e2d\u6211\u5011\u4ee5\u95dc\u806f\u6a21\u578b(RM\uff0c\u53c3\u8003 3.2.1 \u5c0f\u7bc0)\u70ba\u4f8b 1 \uff0c\u63a1\u7528\u4e09\u7a2e\u4e0d\u540c\u7684\u5e73\u6ed1\u5316\u6280\u8853\u65bc\u95dc\u806f \u6a21\u578b\u4e2d\uff0c\u7b2c\u4e00\u70ba Jelinek-Mercer \u5e73\u6ed1\u5316\uff0c\u7b2c\u4e8c\u70ba Dirichlet \u5e73\u6ed1\u5316\uff0c\u7b2c\u4e09\u70ba Add-delta \u5e73\u6ed1 \u5316\uff0c\u8332\u5206\u5225\u5982\u4e0b(Zhai &amp; Lafferty, 2001b)\uff1a(i) Jelinek-Mercer \u5e73\u6ed1\u5316\u70ba\u6700\u7c21\u55ae\u7684\u8207\u80cc\u666f\u6a21 \u53e6\u4e00\u65b9\u9762\uff0c\u5728 \u5289\u58eb\u5f18 \u7b49 \u806f\u6a21\u578b(RM)\u8207\u7c21\u55ae\u6df7\u5408\u6a21\u578b(SMM)\u7684\u6bd4\u8f03\uff0c\u5f9e\u8868 3 \u7684\u5be6\u9a57\u7d50\u679c\u5f97\u77e5\u95dc\u806f\u6a21\u578b\u5728 TD \u4e0a\u8868 \u578b P(w|B)\u7dda\u6027\u7d50\u5408\u7684\u5e73\u6ed1\u5316\u6280\u8853\uff0c\u5176\u516c\u5f0f\u70ba\uff1a</td></tr><tr><td>\u5b9a\u7684\u6458\u8981\u6bd4\u4f8b\u70ba 10%\uff0c\u5176\u5b9a\u7fa9\u70ba\u6458\u8981\u6240\u542b\u8a5e\u5f59\u6578\u5360\u6574\u7bc7\u6587\u4ef6\u8a5e\u5f59\u6578\u7684\u6bd4\u4f8b\uff0c\u4e5f\u5c31\u662f\u4ee5\u8a5e \u6b8a\u6587\u7ae0\u7d50\u69cb\u4e0a\uff0c\u6240\u4ee5\u82e5\u88ab\u6458\u8981\u6587\u4ef6\u4e0d\u5177\u6709\u67d0\u7a2e\u7279\u6b8a\u7684\u6587\u7ae0\u7d50\u69cb\uff0c\u5176\u6458\u8981\u6548\u80fd\u5c31\u6703\u6709\u9650\u3002</td></tr></table>",
                "text": "SD \u7684\u5be6\u9a57\u4e2d\uff0cDLM \u540c\u6a23\u8f03\u512a\u65bc LS\u3001LEAD\u3001VSM\u3001LSA \u7b49\u4e4b\u6458\u8981 \u65b9\u6cd5\uff0c\u4f46 MMR \u7684\u7d50\u679c\u5247\u7a0d\u5fae\u8f03 DLM \u597d\u4e00\u9ede\uff0c\u6211\u5011\u8a8d\u70ba\u9019\u53ef\u80fd\u662f\u56e0\u70ba MMR \u6bd4\u8f03\u4e0d\u53d7\u5230 \u8a9e\u97f3\u8fa8\u8a8d\u932f\u8aa4\u7684\u5f71\u97ff\u3002\u4f46 MRW \u53ca\u6b21\u6a21\u4e5f\u53ef\u80fd\u662f\u53d7\u5230\u8a9e\u97f3\u8fa8\u8b58\u932f\u8aa4\u7684\u5f71\u97ff\u800c\u9020\u6210\u6458\u8981\u6548 \u80fd\u6e1b\u4f4e\uff0c\u751a\u81f3\u6bd4 DLM \u4f86\u5f97\u5dee\u3002\u51fa\u4e4e\u610f\u6599\u7684\u662f\u539f\u4ee5\u70ba ILP \u4e5f\u6703\u5728 SD \u4e2d\u5f97\u5230\u6700\u597d\u7684\u6458\u8981\u6548",
                "type_str": "table",
                "num": null,
                "html": null
            },
            "TABREF7": {
                "content": "<table><tr><td>20</td><td/><td/><td/><td>\u5289\u58eb\u5f18 \u7b49</td></tr><tr><td colspan=\"3\">\u8868 4. \u5e73\u6ed1\u5316\u6280\u8853\u65bc\u95dc\u806f\u6a21\u578b(RM)\u4e4b\u5be6\u9a57\u7d50\u679c</td><td/><td/></tr><tr><td/><td/><td/><td>F-score</td><td/></tr><tr><td colspan=\"2\">Relevance Model (RM)</td><td/><td/><td/></tr><tr><td/><td/><td>ROUGE-1</td><td>ROUGE-2</td><td>ROUGE-L</td></tr><tr><td/><td>Jelinek-Mercer</td><td>0.450</td><td>0.336</td><td>0.400</td></tr><tr><td>TD</td><td>Dirichlet</td><td>0.472</td><td>0.365</td><td>0.428</td></tr><tr><td/><td>Add-delta</td><td>0.493</td><td>0.386</td><td>0.441</td></tr><tr><td/><td>Jelinek-Mercer</td><td>0.374</td><td>0.226</td><td>0.321</td></tr><tr><td>SD</td><td>Dirichlet</td><td>0.401</td><td>0.254</td><td>0.349</td></tr><tr><td/><td>Add-delta</td><td>0.402</td><td>0.255</td><td>0.347</td></tr><tr><td/><td/><td/><td/><td>(33)</td></tr><tr><td>Dir</td><td/><td/><td/><td/></tr><tr><td colspan=\"5\">1 \u6211\u5011\u5be6\u9a57\u767c\u73fe\u4e0d\u540c\u7684\u5e73\u6ed1\u5316\u6280\u8853\u90fd\u6703\u5c0d\u9019\u4e09\u7a2e\u4e0d\u540c\u7684\u95dc\u806f\u6a21\u578b\u6709\u5e6b\u52a9\uff0c\u5728\u6458\u8981\u6210\u6548\u4e0a\u4e5f\u90fd\u6709\u660e</td></tr><tr><td colspan=\"3\">\u986f\u7684\u9032\u6b65\uff0c\u4e14\u95dc\u806f\u6a21\u578b(RM)\u6703\u6709\u6700\u5927\u7684\u9032\u6b65\u3002</td><td/><td/></tr></table>",
                "text": "",
                "type_str": "table",
                "num": null,
                "html": null
            },
            "TABREF8": {
                "content": "<table><tr><td>22</td><td>\u5289\u58eb\u5f18 \u7b49</td></tr><tr><td colspan=\"2\">6.4 \u63a5\u8457\u6211\u5011\u5c07\u7126\u9ede\u8f49\u79fb\u5230\u6a5f\u7387\u5f0f\u6392\u5e8f\u6a21\u578b\u4e0a\uff0c\u5728\u672c\u5c0f\u7bc0\u4e2d\u6211\u5011\u5c07\u6bd4\u8f03\u5404\u7a2e\u4e0d\u540c\u7684 BM25 \u6392 \u5e8f\u6a21\u578b\uff0c\u5305\u542b\u6709\u539f\u59cb BM25(\u53c3\u7167\u5f0f(14))\u3001BM25 E (\u53c3\u7167\u5f0f(18))\u3001BM25L(\u53c3\u7167\u5f0f(24))\u3001 \u8868 5. F-score</td></tr><tr><td colspan=\"2\">BM25+(\u53c3\u7167\u5f0f(25))\u53ca BM25T(\u53c3\u7167\u5f0f(30)) \u3002\u5176\u5be6\u9a57\u7d50\u679c\u5982\u8868 5 \u6240\u793a\uff0c\u5728 TD \u7684\u90e8\u5206\uff0cBM25 ROUGE-1 ROUGE-2 ROUGE-L</td></tr><tr><td colspan=\"2\">\u7684\u6458\u8981\u8868\u73fe\u5df2\u7d93\u5f88\u4e0d\u932f\uff0c\u751a\u81f3\u90fd\u6bd4\u5176\u4ed6\u826f\u597d\u767c\u5c55\u7684\u975e\u76e3\u7763\u5f0f\u6458\u8981\u65b9\u6cd5\u4f86\u5f97\u597d(\u8207\u8868 2 \u4e4b\u7d50 BM25 0.484 0.374 0.442</td></tr><tr><td colspan=\"2\">\u679c\u6bd4\u8f03) \uff0cBM25 E \u56e0\u5c11\u4e86\u4e00\u500b\u91cd\u8981\u56e0\u5b50\u4f86\u505a\u8a9e\u53e5\u6392\u5e8f\uff0c\u6240\u4ee5\u53ef\u9810\u671f\u5b83\u7684\u6458\u8981\u6548\u80fd\u6bd4 BM25 \u4f86\u5f97\u5dee\uff0c\u4f46\u8d85\u4e4e\u9810\u671f\u7684\u662f\u5728 ROUGE-2 \u5c07\u8fd1\u6709 16%\u7684\u5dee\u8ddd\u3002BM25L \u7684\u63d0\u51fa\u662f\u70ba\u4e86\u89e3\u6c7a\u904e BM25 E 0.352 0.210 0.294</td></tr><tr><td colspan=\"2\">\u5ea6\u61f2\u7f70\u9577\u8a9e\u53e5\u7684\u554f\u984c\uff0c\u5728\u672c\u5be6\u9a57\u4e2d\u7684 TD \u60c5\u6cc1\u4e0b\uff0c\u53ef\u770b\u51fa BM25L \u7684\u6458\u8981\u6548\u80fd\u6703\u6c92\u6709\u6bd4\u539f TD BM25L 0.480 0.365 0.434</td></tr><tr><td colspan=\"2\">\u59cb BM25 \u4f86\u7684\u597d\uff0c\u5176\u53ef\u80fd\u7684\u539f\u56e0\u662f\u5728\u8cc7\u8a0a\u6aa2\u7d22\u9818\u57df\u4e2d\uff0c\u78ba\u5be6\u6703\u6709\u5f88\u9577\u6587\u7ae0(Long Document) BM25+ 0.486 0.376 0.444</td></tr><tr><td colspan=\"2\">\u7684\u51fa\u73fe\uff0c\u61f2\u7f70\u9577\u6587\u7ae0\u6703\u6709\u4e00\u5b9a\u7684\u6548\u679c\uff0c\u4f46\u5728\u8a9e\u97f3\u6458\u8981\u4efb\u52d9\u4e2d\uff0c\u5f88\u9577\u8a9e\u53e5(Long Sentence) BM25T 0.463 0.352 0.419</td></tr><tr><td colspan=\"2\">\u7684\u51fa\u73fe\u5e7e\u4e4e\u662f\u4e0d\u592a\u53ef\u80fd\uff0c\u56e0\u6b64\u61f2\u7f70\u9577\u8a9e\u53e5\u5c31\u4e0d\u4e00\u5b9a\u6703\u5f97\u5230\u597d\u7684\u6458\u8981\u6548\u80fd\u3002BM25+\u4e5f\u662f\u70ba \u4e86\u89e3\u6c7a\u904e\u5ea6\u61f2\u7f70\u9577\u8a9e\u53e5\u7684\u554f\u984c\uff0c\u4f46\u66f4\u4e00\u822c\u5316\u5730\u4fdd\u8b49\u53ea\u51fa\u73fe\u4e00\u6b21\u7684\u8a5e\u5f59\u81f3\u5c11\u8981\u6709\u500b\u4e0b\u754c\uff0c BM25 0.390 0.247 0.338</td></tr><tr><td colspan=\"2\">\u56e0\u6b64 BM25+\u7684\u6458\u8981\u6548\u80fd\u6bd4 BM25L \u80fd\u66f4\u9032\u4e00\u6b65\u7684\u63d0\u5347\uff0c\u8207\u539f\u59cb BM25 \u53ca BM25L \u76f8\u6bd4\u8f03\uff0c BM25 E 0.279 0.151 0.250</td></tr><tr><td colspan=\"2\">\u5728 ROUGE-2 \u4e0a\u5206\u5225\u80fd\u6709 0.2%\u53ca 1.1%\u7684\u7d55\u5c0d\u9032\u6b65\u3002BM25T \u5f9e\u8a13\u7df4\u8cc7\u6599\u4e2d\u81ea\u52d5\u5b78\u7fd2\u8207\u8a5e SD BM25L 0.384 0.246 0.337</td></tr><tr><td colspan=\"2\">\u5f59\u76f8\u95dc\u7684\u53c3\u6578 k 1 \uff0c\u5728\u539f\u59cb\u7684\u6587\u737b\u88e1\u7528\u65bc\u8cc7\u8a0a\u6aa2\u7d22\u9818\u57df\u4e2d\u7684\u5be6\u9a57\u662f\u76f8\u5c0d\u6392\u5e8f\u516c\u5f0f\u4f86\u5f97\u597d(Lv BM25+ 0.383 0.242 0.335</td></tr><tr><td colspan=\"2\">&amp; Zhai, 2012)\uff0c\u4f46 BM25T \u7684\u6458\u8981\u6548\u80fd\u6709\u9ede\u51fa\u4e4e\u610f\u6599\u4e4b\u5916\uff0c\u6c92\u6709\u6bd4 BM25L \u8207 BM25+\u597d\uff0c \u751a\u81f3\u4e5f\u6703\u6bd4\u539f\u59cb BM25 \u6392\u5e8f\u516c\u5f0f\u4f86\u7684\u5dee\uff0c\u6211\u5011\u8a8d\u70ba\u6b64\u7a2e\u81ea\u52d5\u5b78\u7fd2\u53c3\u6578\u7684\u65b9\u6cd5\u53ef\u80fd\u662f\u8207\u8cc7 BM25T 0.382 0.238 0.332</td></tr><tr><td colspan=\"2\">\u6599\u76f8\u95dc\u7684\uff0c\u6216\u8a31\u53ef\u4ee5\u66ff\u63db\u53e6\u4e00\u5957\u8a13\u7df4\u8cc7\u6599\u96c6\u4f86\u91cd\u65b0\u5b78\u7fd2\u8a5e\u5f59\u76f8\u95dc\u7684\u53c3\u6578\uff0c\u4f46\u9019\u4e5f\u662f\u672a\u4f86 \u7684\u5de5\u4f5c\u4e4b\u4e00\u3002\u53e6\u4e00\u65b9\u9762\uff0c\u5728 SD \u7684\u5be6\u9a57\u90e8\u5206\uff0c\u539f\u59cb\u7684 BM25 \u6392\u5e8f\u516c\u5f0f\u9084\u662f\u7dad\u6301\u4e00\u5b9a\u7684\u6c34 7. \u7d50\u8ad6\u8207\u672a\u4f86\u5c55\u671b (Conclusions and Future work)</td></tr><tr><td colspan=\"2\">\u6e96\uff0c\u8207\u5176\u4ed6\u975e\u76e3\u7763\u5f0f\u6458\u8981\u65b9\u6cd5\u76f8\u6bd4\u9084\u662f\u6703\u6bd4\u8f03\u597d(\u53c3\u7167\u8868 2)\uff0cBM25L\u3001BM25+\u53ca BM25T \u672c\u8ad6\u6587\u4e3b\u8981\u6709\u4e09\u500b\u7814\u7a76\u8ca2\u737b\uff0c\u5176\u4e00\u70ba\u6709\u9451\u65bc\u95dc\u806f\u6027(Relevance)\u7684\u6982\u5ff5\u5728\u8cc7\u8a0a\u6aa2\u7d22\u9818\u57df\u4e2d</td></tr><tr><td colspan=\"2\">\u7684\u512a\u52e2\u5c31\u6c92\u90a3\u9ebc\u5927\uff0c\u5176\u6458\u8981\u6548\u80fd\u5728 ROUGE-2 \u4e0a\u6bd4\u539f\u59cb BM25 \u90fd\u8981\u4f86\u5f97\u5dee\uff0c\u78ba\u5be6\u8a9e\u97f3\u8fa8 \u5df2\u6709\u4e0d\u932f\u7684\u767c\u5c55\u6210\u679c\uff0c\u672c\u8ad6\u6587\u5617\u8a66\u7d50\u5408\u95dc\u806f\u6027\u8cc7\u8a0a\u4f86\u91cd\u65b0\u4f30\u6e2c\u4e26\u5efa\u7acb\u8a9e\u53e5\u7684\u8a9e\u8a00\u6a21\u578b\uff0c</td></tr><tr><td colspan=\"2\">\u8b58\u932f\u8aa4\u7684\u5f71\u97ff\u9084\u883b\u5927\u7684\uff0c\u539f\u672c BM25L\u3001BM25+\u53ca BM25T \u7684\u512a\u9ede\u90fd\u88ab\u932f\u8aa4\u8fa8\u8b58\u7684\u8a5e\u5f59\u6240 \u4e26\u9996\u6b21\u4f7f\u7528\u4e09\u6df7\u5408(Tri-Mixture Model, TriMM)\u6a21\u578b\uff0c\u4f7f\u5176\u5f97\u4ee5\u66f4\u7cbe\u6e96\u5730\u4ee3\u8868\u8a9e\u53e5\u7684\u8a9e\u610f</td></tr><tr><td colspan=\"2\">\u6d88\u5f4c\uff0c\u751a\u81f3 BM25L\u3001BM25+\u8207 BM25T \u7684\u6458\u8981\u7d50\u679c\u5e7e\u4e4e\u76f8\u5dee\u7121\u5e7e\u3002 \u5167\u5bb9\uff0c\u671f\u671b\u53ef\u589e\u9032\u81ea\u52d5\u6458\u8981\u4e4b\u6548\u80fd\uff0c\u5be6\u9a57\u7d50\u679c\u986f\u793a\u4e09\u6df7\u5408\u6a21\u578b\u53ef\u4ee5\u6709\u6700\u4f73\u7684\u6458\u8981\u6548\u80fd\u3002</td></tr><tr><td colspan=\"2\">\u5176\u4e8c\u70ba\u6709\u9451\u65bc\u8a9e\u8a00\u6a21\u578b\u8457\u91cd\u4f9d\u8cf4\u5e73\u6ed1\u5316\u6280\u8853\uff0c\u672c\u8ad6\u6587\u4e5f\u662f\u9996\u6b21\u6bd4\u8f03\u7814\u7a76\u4e0d\u540c\u5e73\u6ed1\u5316\u6280\u8853</td></tr><tr><td colspan=\"2\">\u6240\u4f30\u6e2c\u5f97\u7684\u8a9e\u8a00\u6a21\u578b\u5c0d\u8a9e\u97f3\u6587\u4ef6\u6458\u8981\u4efb\u52d9\u4e4b\u5f71\u97ff\uff0c\u6839\u64da\u5be6\u9a57\u7d50\u679c Add-delta \u5e73\u6ed1\u5316\u53ef\u4ee5\u9054</td></tr><tr><td colspan=\"2\">\u5230\u6700\u4f73\u6458\u8981\u6548\u679c\uff0c\u6240\u4ee5\u6211\u5011\u5efa\u8b70\u95dc\u806f\u6a21\u578b\u7684\u5e73\u6ed1\u5316\u6280\u8853\u61c9\u7576\u4f7f\u7528 Add-delta \u5e73\u6ed1\u5316\u4f86\u9054\u6210\u3002</td></tr><tr><td colspan=\"2\">\u6700\u5f8c\u70ba\u6211\u5011\u9996\u6b21\u63d0\u51fa\u4e26\u61c9\u7528\u591a\u7a2e\u6a5f\u7387\u5f0f\u8cc7\u8a0a\u6aa2\u7d22\u6392\u5e8f\u6a21\u578b\u65bc\u8a9e\u97f3\u6458\u8981\u4efb\u52d9\u4e0a\uff0c\u4e26\u4e14\u5f9e\u5be6</td></tr><tr><td>\u9a57\u7d50\u679c\u4e2d\u5f97\u77e5\u8207\u5176\u4ed6\u5e38\u898b\u7684\u975e\u76e3\u7763\u5f0f\u6458\u8981\u65b9\u6cd5\u76f8\u6bd4\u8f03\u80fd\u6709\u4e0d\u932f\u7684\u6458\u8981\u6548\u80fd\u3002</td><td/></tr><tr><td colspan=\"2\">Mercer \u5e73\u6ed1\u5316\u4f86\u5f97 \u672a\u4f86\uff0c\u6211\u5011\u7684\u7814\u7a76\u5c07\u6709\u4e09\u500b\u4e3b\u8981\u7684\u65b9\u5411\uff1a\u9996\u5148\uff0c\u591a\u7a2e\u6a5f\u7387\u5f0f\u6aa2\u7d22\u6392\u5e8f\u6a21\u578b\u9084\u662f\u9700\u8981</td></tr><tr><td colspan=\"2\">\u597d\uff0c\u4f46\u8207 Dirichlet \u5e73\u6ed1\u5316\u76f8\u6bd4\uff0c\u5176\u6458\u8981\u6548\u80fd\u5176\u5be6\u5df2\u7d93\u76f8\u5dee\u7121\u5e7e\uff0c\u53ef\u80fd\u7684\u539f\u56e0\u4e4b\u4e00\u9084\u662f\u56e0 \u7d93\u9a57\u53bb\u8abf\u6574\u4e0d\u78ba\u5b9a\u7684\u53c3\u6578\uff0c\u6211\u5011\u5c07\u9032\u4e00\u6b65\u7684\u7814\u7a76\u662f\u5426\u53ef\u4ee5\u91dd\u5c0d\u4e0d\u540c\u7684\u6587\u4ef6\u6216\u4e0d\u540c\u7684\u8a9e\u53e5</td></tr><tr><td colspan=\"2\">\u70ba\u8a9e\u97f3\u8fa8\u8b58\u932f\u8aa4\u7684\u5f71\u97ff\u6240\u9020\u6210\uff0c\u5728\u672a\u4f86\u7814\u7a76\u4e2d\uff0c\u6211\u5011\u5c07\u4ee5\u6b21\u8a5e\u7d22\u5f15(Subword Indexing)\u7684 \u7d66\u4e88\u9069\u7576\u7684\u6b0a\u91cd\u8abf\u6574\uff0c\u4ee5\u671f\u7372\u5f97\u66f4\u597d\u7684\u6458\u8981\u6210\u6548\uff1b\u7b2c\u4e8c\uff0c\u76ee\u524d\u95dc\u806f\u6a21\u578b\u50c5\u904b\u7528\u65bc\u91cd\u5efa\u8a9e</td></tr><tr><td colspan=\"2\">\u65b9\u5f0f\u4f86\u5efa\u7acb\u95dc\u806f\u6a21\u578b\u4ee5\u6e1b\u7de9\u6b64\u554f\u984c\u3002 \u53e5\u7684\u8a9e\u8a00\u6a21\u578b\uff0c\u6211\u5011\u5c07\u5617\u8a66\u4f7f\u7528\u88ab\u6458\u8981\u6587\u4ef6\u7684\u95dc\u806f\u8cc7\u8a0a\u4f86\u91cd\u65b0\u4f30\u6e2c\u4e26\u5efa\u7acb\u6587\u4ef6\u7684\u8a9e\u8a00\u6a21</td></tr><tr><td colspan=\"2\">\u578b\uff1b\u6700\u5f8c\uff0c\u6211\u5011\u5e0c\u671b\u5c07\u975e\u76e3\u7763\u5f0f\u6458\u8981\u65b9\u6cd5\u6240\u7522\u751f\u7684\u5206\u6578\u8996\u70ba\u4e00\u7a2e\u5177\u4ee3\u8868\u6027\u7684\u6458\u8981\u7279\u5fb5\u8cc7</td></tr><tr><td colspan=\"2\">\u8a0a\u4e26\u7d50\u5408\u65bc\u76e3\u7763\u5f0f\u6a5f\u5668\u5b78\u7fd2\u65b9\u6cd5(\u5982\u689d\u4ef6\u96a8\u6a5f\u5834\u57df(Conditional Random Fields, CRFs)\u6216\u6df1</td></tr><tr><td colspan=\"2\">\u5ea6\u985e\u795e\u7d93\u7db2\u7d61(Deep Neural Network Learning, DNN)\u7b49)\u4e2d\uff0c\u671f\u671b\u8a13\u7df4\u5f8c\u7684\u6a21\u578b\u80fd\u5920\u5728\u6587\u5b57</td></tr><tr><td>\u6587\u4ef6\u6458\u8981\u6216\u8a9e\u97f3\u6587\u4ef6\u6458\u8981\u4e0a\u7372\u5f97\u66f4\u597d\u7684\u8868\u73fe\u3002</td><td/></tr></table>",
                "text": "",
                "type_str": "table",
                "num": null,
                "html": null
            }
        }
    }
}