| { | |
| "paper_id": "W06-0119", | |
| "header": { | |
| "generated_with": "S2ORC 1.0.0", | |
| "date_generated": "2023-01-19T04:04:49.076235Z" | |
| }, | |
| "title": "BMM-based Chinese Word Segmentor with Word Support Model for the SIGHAN Bakeoff 2006", | |
| "authors": [ | |
| { | |
| "first": "Jia-Lin", | |
| "middle": [], | |
| "last": "Tsai", | |
| "suffix": "", | |
| "affiliation": { | |
| "laboratory": "", | |
| "institution": "Tung Nan Institute of Technology", | |
| "location": { | |
| "country": "Taiwan, R.O.C" | |
| } | |
| }, | |
| "email": "[email protected]" | |
| } | |
| ], | |
| "year": "", | |
| "venue": null, | |
| "identifiers": {}, | |
| "abstract": "This paper describes a Chinese word segmentor (CWS) for the third International Chinese Language Processing Bakeoff (SIGHAN Bakeoff 2006). We participate in the word segmentation task at the Microsoft Research (MSR) closed testing track. Our CWS is based on backward maximum matching with word support model (WSM) and contextual-based Chinese unknown word identification. From the scored results and our experimental results, it shows WSM can improve our previous CWS, which was reported at the SIGHAN Bakeoff 2005, about 1% of F-measure.", | |
| "pdf_parse": { | |
| "paper_id": "W06-0119", | |
| "_pdf_hash": "", | |
| "abstract": [ | |
| { | |
| "text": "This paper describes a Chinese word segmentor (CWS) for the third International Chinese Language Processing Bakeoff (SIGHAN Bakeoff 2006). We participate in the word segmentation task at the Microsoft Research (MSR) closed testing track. Our CWS is based on backward maximum matching with word support model (WSM) and contextual-based Chinese unknown word identification. From the scored results and our experimental results, it shows WSM can improve our previous CWS, which was reported at the SIGHAN Bakeoff 2005, about 1% of F-measure.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Abstract", | |
| "sec_num": null | |
| } | |
| ], | |
| "body_text": [ | |
| { | |
| "text": "A high-performance Chinese word segmentor (CWS) is a critical processing stage to produce an intermediate result for later processes, such as search engines, text mining, word spell checking, text-to-speech and speech recognition, etc. As per (Lin et al. 1993; Tsai et al. 2003; Tsai, 2005) , the bottleneck for developing a highperformance CWS is to comprise of high performance Chinese unknown word identification (UWI). It is because Chinese is written without any separation between words and more than 50% words of the Chinese texts in web corpus are out-of-vocabulary (Tsai et al. 2003) . In our report for the SIGHAN Bakeoff 2005 (Tsai, 2005) , we have shown that a highly performance of 99.1% F-measure can be achieved while a BMM-based CWS using a perfect system dictionary (Tsai, 2005) . A perfect system dictionary means all word types of the dictionary are extracted from training and testing gold standard corpus.", | |
| "cite_spans": [ | |
| { | |
| "start": 243, | |
| "end": 260, | |
| "text": "(Lin et al. 1993;", | |
| "ref_id": "BIBREF4" | |
| }, | |
| { | |
| "start": 261, | |
| "end": 278, | |
| "text": "Tsai et al. 2003;", | |
| "ref_id": "BIBREF8" | |
| }, | |
| { | |
| "start": 279, | |
| "end": 290, | |
| "text": "Tsai, 2005)", | |
| "ref_id": "BIBREF10" | |
| }, | |
| { | |
| "start": 574, | |
| "end": 592, | |
| "text": "(Tsai et al. 2003)", | |
| "ref_id": "BIBREF8" | |
| }, | |
| { | |
| "start": 617, | |
| "end": 636, | |
| "text": "SIGHAN Bakeoff 2005", | |
| "ref_id": null | |
| }, | |
| { | |
| "start": 637, | |
| "end": 649, | |
| "text": "(Tsai, 2005)", | |
| "ref_id": "BIBREF10" | |
| }, | |
| { | |
| "start": 783, | |
| "end": 795, | |
| "text": "(Tsai, 2005)", | |
| "ref_id": "BIBREF10" | |
| } | |
| ], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Introduction", | |
| "sec_num": "1" | |
| }, | |
| { | |
| "text": "Conventionally, there are four approaches to develop a CWS: (1) Dictionary-based approach (Cheng et al. 1999) , especial forward and backward maximum matching (Wong and Chan, 1996) ; (2) Linguistic approach based on syntax-semantic knowledge (Chen et al. 2002) ;", | |
| "cite_spans": [ | |
| { | |
| "start": 90, | |
| "end": 109, | |
| "text": "(Cheng et al. 1999)", | |
| "ref_id": "BIBREF1" | |
| }, | |
| { | |
| "start": 159, | |
| "end": 180, | |
| "text": "(Wong and Chan, 1996)", | |
| "ref_id": null | |
| }, | |
| { | |
| "start": 242, | |
| "end": 260, | |
| "text": "(Chen et al. 2002)", | |
| "ref_id": "BIBREF0" | |
| } | |
| ], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Introduction", | |
| "sec_num": "1" | |
| }, | |
| { | |
| "text": "(3) Statistical approach based on statistical language model (SLM) (Sproat and Shih, 1990; Teahan et al. 2000; Gao et al. 2003) ; and (4) Hybrid approach trying to combine the benefits of dictionary-based, linguistic and statistical approaches (Tsai et al. 2003; Ma and Chen, 2003) . In practice, statistical approaches are most widely used because their effective and reasonable performance.", | |
| "cite_spans": [ | |
| { | |
| "start": 67, | |
| "end": 90, | |
| "text": "(Sproat and Shih, 1990;", | |
| "ref_id": "BIBREF6" | |
| }, | |
| { | |
| "start": 91, | |
| "end": 110, | |
| "text": "Teahan et al. 2000;", | |
| "ref_id": "BIBREF7" | |
| }, | |
| { | |
| "start": 111, | |
| "end": 127, | |
| "text": "Gao et al. 2003)", | |
| "ref_id": "BIBREF3" | |
| }, | |
| { | |
| "start": 244, | |
| "end": 262, | |
| "text": "(Tsai et al. 2003;", | |
| "ref_id": "BIBREF8" | |
| }, | |
| { | |
| "start": 263, | |
| "end": 281, | |
| "text": "Ma and Chen, 2003)", | |
| "ref_id": "BIBREF5" | |
| } | |
| ], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Introduction", | |
| "sec_num": "1" | |
| }, | |
| { | |
| "text": "To develop UWI, there are three approaches: (1) Statistical approach, researchers use common statistical features, such as maximum entropy (Chieu et al. 2002) , association strength, mutual information, ambiguous matching, and multi-statistical features for unknown word detection and extraction; (2) Linguistic approach, three major types of linguistic rules (knowledge): morphology, syntax, and semantics, are used to identify unknown words; and (3) Hybrid approach, recently, one important trend of UWI follows a hybrid approach so as to take advantage of both merits of statistical and linguistic approaches. Statistical approaches are simple and efficient whereas linguistic approaches are effective in identifying low frequency unknown words (Chen et al. 2002) .", | |
| "cite_spans": [ | |
| { | |
| "start": 139, | |
| "end": 158, | |
| "text": "(Chieu et al. 2002)", | |
| "ref_id": "BIBREF2" | |
| }, | |
| { | |
| "start": 748, | |
| "end": 766, | |
| "text": "(Chen et al. 2002)", | |
| "ref_id": "BIBREF0" | |
| } | |
| ], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Introduction", | |
| "sec_num": "1" | |
| }, | |
| { | |
| "text": "To develop WSD, there are two major types of word segmentation ambiguities while there are no unknown word problems with them: (1) Overlap Ambiguity (OA). Take string C1C2C3 comprised of three Chinese characters C1, C2 and C3 as an example. If its segmentation can be either C1C2/C3 or C1/C2C3 depending on context meaning, the C1C2C3 is called an overlap ambiguity string (OAS), such as \"\u5c07\u8ecd(a general)/\u7528(use)\" and \"\u5c07(to get)/\u8ecd\u7528(for military use)\" (the symbol \"/\" indicates a word boundary). (2) Combination Ambiguity (CA). Take string C1C2 comprised of two Chinese characters C1 and C2 as an example. If its segmentation can be either C1/C2 or C1C2 depending on context meaning, the C1C2 is called a combination ambiguity string (CAS), such as \"\u624d(just)/ \u80fd(can)\" and \"\u624d\u80fd(ability).\" Besides the OA and CA problems, the other two types of word segmentation errors are caused by unknown word problems. They are: (1) Lack of unknown word (LUW), it means segmentation error occurred by lack of an unknown word in the system dictionary, and (2) Error identified word (EIW), it means segmentation error occurred by an error identified unknown words.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Introduction", | |
| "sec_num": "1" | |
| }, | |
| { | |
| "text": "The goal of this paper is to report the approach and experiment results of our backward maximum matching-based (BMM-based) CWS with word support model (WSM) for the SIGHAN Bakeoff 2006. In (Tsai, 2006) , WSM has been shown effectively to improve Chinese input system. In the third Bakeoff, our CWS is mainly addressed on improving its performance of OA/CA disambiguation by WSM. We show that WSM is able to improve our BMM-based CWS, which reported at the SIGHAN Bakeoff 2005, about 1% of F-measure.", | |
| "cite_spans": [ | |
| { | |
| "start": 189, | |
| "end": 201, | |
| "text": "(Tsai, 2006)", | |
| "ref_id": "BIBREF11" | |
| } | |
| ], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Introduction", | |
| "sec_num": "1" | |
| }, | |
| { | |
| "text": "The remainder of this paper is arranged as follows. In Section 2, we present the details of our BMM-based CWS comprised of WSM. In Section 3, we present the scored results of the CWS at the Microsoft Research closed track and give our experiment results and analysis. Finally, in Section 4, we give our conclusions and future research directions.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Introduction", | |
| "sec_num": "1" | |
| }, | |
| { | |
| "text": "From our work (Tsai et al. 2004) , the Chinese word segmentation performance of BMM technique is about 1% greater than that of forward maximum matching (FMM) technique. Thus, we adopt BMM technique as base to develop our CWS. In this Bakeoff, we use context-based Chinese unknown word identification (CCUWI) (Tsai, 2005) to resolve unknown word problem. The CCUWI uses template matching technique to extract unknown words from sentences. The context template includes triple context template (TCT) and word context template (WCT). The details of the CCUWI can be found in (Tsai, 2005) . In (Tsai, 2006) , we propose a new language model named word support model (WSM) and shown it can effectively perform homophone selection and word-syllable segmentation to improve Chinese input system. For this Bakeoff, we use WSM to resolve OA/CA problems.", | |
| "cite_spans": [ | |
| { | |
| "start": 14, | |
| "end": 32, | |
| "text": "(Tsai et al. 2004)", | |
| "ref_id": "BIBREF9" | |
| }, | |
| { | |
| "start": 308, | |
| "end": 320, | |
| "text": "(Tsai, 2005)", | |
| "ref_id": "BIBREF10" | |
| }, | |
| { | |
| "start": 572, | |
| "end": 584, | |
| "text": "(Tsai, 2005)", | |
| "ref_id": "BIBREF10" | |
| }, | |
| { | |
| "start": 590, | |
| "end": 602, | |
| "text": "(Tsai, 2006)", | |
| "ref_id": "BIBREF11" | |
| } | |
| ], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "BMM-based CWS with WSM", | |
| "sec_num": "2" | |
| }, | |
| { | |
| "text": "The two steps of our BMM-based CWS with WSM are as below:", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "BMM-based CWS with WSM", | |
| "sec_num": "2" | |
| }, | |
| { | |
| "text": "Step 1. Generate the BMM segmentation for the given Chinese sentence by system dictionary.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "BMM-based CWS with WSM", | |
| "sec_num": "2" | |
| }, | |
| { | |
| "text": "Step 2. Use WSM to resolve OA/CA problems for the BMM segmentation of Step 1. Now, we give a brief description of how we use WSM to resolve OA/CA problem. Firstly, we pre-collect OA/CA pattern-pairs (such as \"\u5c31/ \u662f\"-\"\u5c31\u662f\") by compare each training gold segmentation and its corresponding BMM segmentation. The pattern of OA/CA patternpairs can be a segmentation pattern, such as \"\u5c31/\u662f,\" or just a word, such as \"\u5c31\u662f.\" Secondly, for a BMM segmentation of Step 1, if one pattern matching (matching pattern) with at least one pattern of those pre-collected OA/CA pattern-pairs (matching OA/CA pattern-pairs), CWS will compute the word support degree for each pattern of the matching OA/CA pattern-pair. Finally, select out the pattern with maximum word support degree as its segmentation for the matching pattern. If the patterns of the matching OA/CA patternpair having the same word support degree, randomly select one to be its segmentation.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "BMM-based CWS with WSM", | |
| "sec_num": "2" | |
| }, | |
| { | |
| "text": "The details of WSM can be found in (Tsai, 2006) . Tables 1a and 1b show the details of MSR training and testing corpus for 2 nd (2005) and 3 rd (2006) bakeoff. From Table 1a and 1b, it indicates that MSR track of 3 rd bakeoff seems to be a more difficult WS task than that of 2 nd bakeoff, since (1) the training size of 2 nd bakeoff is two times as great as that of 3 rd bakeoff;", | |
| "cite_spans": [ | |
| { | |
| "start": 35, | |
| "end": 47, | |
| "text": "(Tsai, 2006)", | |
| "ref_id": "BIBREF11" | |
| } | |
| ], | |
| "ref_spans": [ | |
| { | |
| "start": 50, | |
| "end": 66, | |
| "text": "Tables 1a and 1b", | |
| "ref_id": null | |
| }, | |
| { | |
| "start": 165, | |
| "end": 173, | |
| "text": "Table 1a", | |
| "ref_id": null | |
| } | |
| ], | |
| "eq_spans": [], | |
| "section": "BMM-based CWS with WSM", | |
| "sec_num": "2" | |
| }, | |
| { | |
| "text": "(2) in training data, the word type number of 3 rd bakeoff is less than that of 2 nd bakeoff, and (3) in testing data, the word type number of 3 rd bakeoff is greater than that of 2 nd bakeoff. Table 2 shows the scored results of our CWS at the MSR_C track of this bakeoff. In Table 2 , the symbols a, b and c stand for the CWS with a, b and c system dictionary. The system dictionary \"a\" is the dictionary comprised of all word types found in the MSR training corpus. The system dictionary \"b\" is the dictionary comprised of \"a\" system dictionary and the word types found in the testing corpus by CCUWI with TCT knowledge. The system dictionary \"c\" is the dictionary comprised of \"a\" system dictionary and the word types found in the testing corpus by CCUWI with TCT and WCT knowledge. Table 3 is F-measure differences between the BMM-based CWS system and it with WSM and CCUWI using \"a\", \"b\" and \"c\" system dictionary in the MSR_C track.", | |
| "cite_spans": [], | |
| "ref_spans": [ | |
| { | |
| "start": 194, | |
| "end": 201, | |
| "text": "Table 2", | |
| "ref_id": null | |
| }, | |
| { | |
| "start": 277, | |
| "end": 284, | |
| "text": "Table 2", | |
| "ref_id": null | |
| }, | |
| { | |
| "start": 787, | |
| "end": 794, | |
| "text": "Table 3", | |
| "ref_id": null | |
| } | |
| ], | |
| "eq_spans": [], | |
| "section": "Scored Results and our Experiments", | |
| "sec_num": "3.1" | |
| }, | |
| { | |
| "text": "From Tables 2 and 3, we conclude that our CWS of 3 rd bakeoff improve the CWS of 2 nd bakeoff about 1.8% of F-measure. Among the 1.8% F-measure improvement, 1% is contributed by WSM for resolving OA/CA problems and the other 0.8% is contributed by CCUWI for resolving UWI problem. Table 3 . The F-measure improvement between the BMM-based CWS and it with WSM in the MSR_C track (OOV is 0.034) using a, b, and c system dictionary. Table 4 shows the F-measure and R OOV differences between each result of our CWS with a, b and c system dictionaries. From Table 4 , it indicates that the most contribution for increasing the overall performance (F-measure) of our CWS is occurred while our CWS comprised of WSM and CCUWI with TCT knowledge. Table 5 . The number of OAS (types), CAS (types), LUW (types) and EIW (types) for our CWS. Table 5 shows the distributions of four segmentation error types (OA, CA, LUW and EIW) for each result of our CWS with a and c system dictionaries. From Table 5 , it shows CCUWI with the knowledge of TCT and WCT can be used to optimize the LUW-EIW tradeoff. Moreover, it shows that WSM can effectively to reduce the number of OA/CA segmentation errors from 1,070 to 391.", | |
| "cite_spans": [], | |
| "ref_spans": [ | |
| { | |
| "start": 281, | |
| "end": 288, | |
| "text": "Table 3", | |
| "ref_id": null | |
| }, | |
| { | |
| "start": 430, | |
| "end": 437, | |
| "text": "Table 4", | |
| "ref_id": null | |
| }, | |
| { | |
| "start": 553, | |
| "end": 560, | |
| "text": "Table 4", | |
| "ref_id": null | |
| }, | |
| { | |
| "start": 738, | |
| "end": 745, | |
| "text": "Table 5", | |
| "ref_id": null | |
| }, | |
| { | |
| "start": 829, | |
| "end": 836, | |
| "text": "Table 5", | |
| "ref_id": null | |
| }, | |
| { | |
| "start": 982, | |
| "end": 989, | |
| "text": "Table 5", | |
| "ref_id": null | |
| } | |
| ], | |
| "eq_spans": [], | |
| "section": "Scored Results and our Experiments", | |
| "sec_num": "3.1" | |
| }, | |
| { | |
| "text": "In this paper, we have applied a BMM-based CWS comprised of a context-based UWI and word support model to the Chinese word segmentation. While we repeat the CWS with the MSR_C track data of 2 nd bakeoff, we obtained 96.3% F-measure, which is 0.8% greater than that (95.5%) of our CWS at 2 nd bakeoff. To sum up the results of this study, we have following conclusions and future directions:", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Conclusions and Future Directions", | |
| "sec_num": "4" | |
| }, | |
| { | |
| "text": "(1) UWI and OA/CA problems could be independent tasks for developing a CWS.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Conclusions and Future Directions", | |
| "sec_num": "4" | |
| }, | |
| { | |
| "text": "The experiment results of this study support this observation. It is because we found 1% improvement is stable contributed by WSM and the other 0.8% improvement is stable contributed by the CCUWI while the BMMbased CWS with difference a, b and c system dictionaries and different MSR_C training and testing data of 2 nd and 3 rd bakeoff. (2) About 89% of segmentation errors of our CWS caused by unknown word problem. In the 89%, we found 66% is LUW problem and 23% is EIW problem. This result indicates that the major target to improve our CWS is CCUWI. The result also supports that a high performance CWS is relied on a high performance Chinese UWI (Tsai, 2005) . (3) We will continue to expand our CWS with other unknown word identification techniques, especially applying n-gram extractor with the TCT and WCT template matching technique to improve our CCUWI for attending the fourth SIGHAN Bakeoff.", | |
| "cite_spans": [ | |
| { | |
| "start": 652, | |
| "end": 664, | |
| "text": "(Tsai, 2005)", | |
| "ref_id": "BIBREF10" | |
| } | |
| ], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Conclusions and Future Directions", | |
| "sec_num": "4" | |
| } | |
| ], | |
| "back_matter": [], | |
| "bib_entries": { | |
| "BIBREF0": { | |
| "ref_id": "b0", | |
| "title": "Unknown Word Extraction for Chinese Documents, Proceedings of 19 th COLING", | |
| "authors": [ | |
| { | |
| "first": "Keh-Jiann", | |
| "middle": [], | |
| "last": "Chen", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Wei-Yun", | |
| "middle": [], | |
| "last": "", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Ma", | |
| "middle": [], | |
| "last": "", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2002, | |
| "venue": "", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "169--175", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Chen, Keh-Jiann and Wei-Yun, Ma. 2002. Unknown Word Extraction for Chinese Documents, Pro- ceedings of 19 th COLING 2002, Taipei, 169- 175.", | |
| "links": null | |
| }, | |
| "BIBREF1": { | |
| "ref_id": "b1", | |
| "title": "A study on word-based and integral-bit Chinese text compression algorithms", | |
| "authors": [ | |
| { | |
| "first": "Kowk-", | |
| "middle": [], | |
| "last": "Cheng", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Gilbert", | |
| "middle": [ | |
| "H" | |
| ], | |
| "last": "Shing", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Kam-Fai", | |
| "middle": [], | |
| "last": "Yong", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "", | |
| "middle": [], | |
| "last": "Wong", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 1999, | |
| "venue": "JASIS", | |
| "volume": "50", | |
| "issue": "3", | |
| "pages": "218--228", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Cheng, Kowk-Shing, Gilbert H. Yong and Kam-Fai Wong.. 1999. A study on word-based and in- tegral-bit Chinese text compression algorithms. JASIS, 50(3): 218-228.", | |
| "links": null | |
| }, | |
| "BIBREF2": { | |
| "ref_id": "b2", | |
| "title": "Named Entity Recognition: A Maximum Entropy Approach Using Global Information", | |
| "authors": [ | |
| { | |
| "first": "H", | |
| "middle": [ | |
| "L" | |
| ], | |
| "last": "Chieu", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "H", | |
| "middle": [ | |
| "T" | |
| ], | |
| "last": "Ng", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2002, | |
| "venue": "Proceedings of 19 th COLING", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "190--196", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Chieu, H.L. and H.T. Ng. 2002. Named Entity Rec- ognition: A Maximum Entropy Approach Us- ing Global Information. Proceedings of 19 th COLING 2002, Taipei, 190-196.", | |
| "links": null | |
| }, | |
| "BIBREF3": { | |
| "ref_id": "b3", | |
| "title": "Improved Source-Channel Models for Chinese Word Segmentation", | |
| "authors": [ | |
| { | |
| "first": "Jianfeng", | |
| "middle": [], | |
| "last": "Gao", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Mu", | |
| "middle": [], | |
| "last": "Li", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Chang-Ning", | |
| "middle": [], | |
| "last": "Uang", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2003, | |
| "venue": "Proceedings of the 41st Annual Meeting of the Association for Computational Linguistics", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "272--279", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Gao, Jianfeng, Mu Li and Chang-Ning uang. 2003. Improved Source-Channel Models for Chinese Word Segmentation. Proceedings of the 41st Annual Meeting of the Association for Compu- tational Linguistics, 272-279.", | |
| "links": null | |
| }, | |
| "BIBREF4": { | |
| "ref_id": "b4", | |
| "title": "A preliminary study on unknown word problem in Chinese word segmentation", | |
| "authors": [ | |
| { | |
| "first": "Ming-Yu", | |
| "middle": [], | |
| "last": "Lin", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Tung-Hui", | |
| "middle": [], | |
| "last": "Chiang", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Keh-Yi", | |
| "middle": [], | |
| "last": "Su", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 1993, | |
| "venue": "", | |
| "volume": "6", | |
| "issue": "", | |
| "pages": "119--141", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Lin, Ming-Yu, Tung-Hui Chiang and Keh-Yi Su. 1993. A preliminary study on unknown word problem in Chinese word segmentation. ROCLING 6, 119-141.", | |
| "links": null | |
| }, | |
| "BIBREF5": { | |
| "ref_id": "b5", | |
| "title": "Introduction to CKIP Chinese Word Segmentation System for the First International Chinese Word Segmentation Bakeoff", | |
| "authors": [ | |
| { | |
| "first": "Wei-Yun", | |
| "middle": [], | |
| "last": "Ma", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Keh-Jiann", | |
| "middle": [], | |
| "last": "Chen", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2003, | |
| "venue": "Proceedings of ACL, Second SIGHAN Workshop on Chinese Language Processing", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "168--171", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Ma, Wei-Yun and Keh-Jiann Chen, 2003, \"Introduc- tion to CKIP Chinese Word Segmentation System for the First International Chinese Word Segmentation Bakeoff\", Proceedings of ACL, Second SIGHAN Workshop on Chinese Language Processing, pp168-171.", | |
| "links": null | |
| }, | |
| "BIBREF6": { | |
| "ref_id": "b6", | |
| "title": "A Statistical Method for Finding Word Boundaries in Chinese Text", | |
| "authors": [ | |
| { | |
| "first": "R", | |
| "middle": [], | |
| "last": "Sproat", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "C", | |
| "middle": [], | |
| "last": "Shih", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 1990, | |
| "venue": "Computer proceeding of Chinese and Oriental Language", | |
| "volume": "4", | |
| "issue": "4", | |
| "pages": "336--349", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Sproat, R. and C., Shih. 1990. A Statistical Method for Finding Word Boundaries in Chinese Text. Computer proceeding of Chinese and Oriental Language, 4(4):336 349.", | |
| "links": null | |
| }, | |
| "BIBREF7": { | |
| "ref_id": "b7", | |
| "title": "A compression-based algorithm for Chinese word segmentation", | |
| "authors": [ | |
| { | |
| "first": "W", | |
| "middle": [ | |
| "J" | |
| ], | |
| "last": "Teahan", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Yingying", | |
| "middle": [], | |
| "last": "Wen", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Rodger", | |
| "middle": [], | |
| "last": "Mcnad", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Ian", | |
| "middle": [], | |
| "last": "Witten", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2000, | |
| "venue": "Computational Linguistics", | |
| "volume": "26", | |
| "issue": "3", | |
| "pages": "375--393", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Teahan, W. J., Yingying Wen, Rodger McNad and Ian Witten. 2000. A compression-based algo- rithm for Chinese word segmentation. Compu- tational Linguistics, 26(3): 375-393.", | |
| "links": null | |
| }, | |
| "BIBREF8": { | |
| "ref_id": "b8", | |
| "title": "Chinese Word Auto-Confirmation Agent, Proceedings of ROCLING XV", | |
| "authors": [ | |
| { | |
| "first": "Jia-Lin", | |
| "middle": [], | |
| "last": "Tsai", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "C", | |
| "middle": [ | |
| "L" | |
| ], | |
| "last": "Sung", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "W", | |
| "middle": [ | |
| "L" | |
| ], | |
| "last": "Hsu", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2003, | |
| "venue": "", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "175--192", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Tsai, Jia-Lin, C.L., Sung and W.L., Hsu. 2003. Chi- nese Word Auto-Confirmation Agent, Pro- ceedings of ROCLING XV, Taiwan, 175-192.", | |
| "links": null | |
| }, | |
| "BIBREF9": { | |
| "ref_id": "b9", | |
| "title": "Auto-Generation of NVEF knowledge in Chinese", | |
| "authors": [ | |
| { | |
| "first": "Jia-Lin", | |
| "middle": [], | |
| "last": "Tsai", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "G", | |
| "middle": [], | |
| "last": "Hsieh", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "W", | |
| "middle": [ | |
| "L" | |
| ], | |
| "last": "Hsu", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2004, | |
| "venue": "Computational Linguistics and Chinese Language Processing", | |
| "volume": "9", | |
| "issue": "", | |
| "pages": "41--64", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Tsai, Jia-Lin, G., Hsieh and W.L., Hsu. 2004. Auto- Generation of NVEF knowledge in Chinese, Computational Linguistics and Chinese Lan- guage Processing, 9(1):41-64.", | |
| "links": null | |
| }, | |
| "BIBREF10": { | |
| "ref_id": "b10", | |
| "title": "A Study of Applying BTM Model on the Chinese Chunk Bracketing. Proceedings of IJCNLP, 6th International Workshop on Linguistically Interpreted Corpora", | |
| "authors": [ | |
| { | |
| "first": "Jia-Lin", | |
| "middle": [], | |
| "last": "Tsai", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2005, | |
| "venue": "", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Tsai, Jia-Lin. 2005. A Study of Applying BTM Model on the Chinese Chunk Bracketing. Pro- ceedings of IJCNLP, 6th International Work- shop on Linguistically Interpreted Corpora, Jeju Island.", | |
| "links": null | |
| }, | |
| "BIBREF11": { | |
| "ref_id": "b11", | |
| "title": "Using Word Support Model to Improve Chinese Input System", | |
| "authors": [ | |
| { | |
| "first": "Jia-Lin", | |
| "middle": [], | |
| "last": "Tsai", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2006, | |
| "venue": "Proceedings of ACL/COLING 2006", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Tsai, Jia-Lin. 2006. Using Word Support Model to Improve Chinese Input System. Proceedings of ACL/COLING 2006, Sydney.", | |
| "links": null | |
| }, | |
| "BIBREF12": { | |
| "ref_id": "b12", | |
| "title": "Chinese Word Segmentation. based on Maximum Matching and Word Binding Force. Proceedings of the 16th International conference on Computational linguistic", | |
| "authors": [ | |
| { | |
| "first": "Pak-Kwong", | |
| "middle": [], | |
| "last": "Wong", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Chorkin", | |
| "middle": [], | |
| "last": "Chanwong", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 1996, | |
| "venue": "", | |
| "volume": "1", | |
| "issue": "", | |
| "pages": "200--203", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Wong, Pak-Kwong and Chorkin ChanWong. 1996. Chinese Word Segmentation. based on Maxi- mum Matching and Word Binding Force. Pro- ceedings of the 16th International conference on Computational linguistic, 1:200-203.", | |
| "links": null | |
| } | |
| }, | |
| "ref_entries": {} | |
| } | |
| } |