| { | |
| "paper_id": "S10-1038", | |
| "header": { | |
| "generated_with": "S2ORC 1.0.0", | |
| "date_generated": "2023-01-19T15:28:12.208898Z" | |
| }, | |
| "title": "UNPMC: Na\u00efve Approach to Extract Keyphrases from Scientific Articles", | |
| "authors": [ | |
| { | |
| "first": "Jungyeul", | |
| "middle": [], | |
| "last": "Park", | |
| "suffix": "", | |
| "affiliation": { | |
| "laboratory": "", | |
| "institution": "Universit\u00e9 de Nantes Nantes", | |
| "location": { | |
| "country": "France" | |
| } | |
| }, | |
| "email": "[email protected]" | |
| }, | |
| { | |
| "first": "Jong", | |
| "middle": [ | |
| "Gun" | |
| ], | |
| "last": "Lee", | |
| "suffix": "", | |
| "affiliation": { | |
| "laboratory": "", | |
| "institution": "Universit\u00e9 de Nantes Nantes", | |
| "location": { | |
| "country": "France" | |
| } | |
| }, | |
| "email": "[email protected]" | |
| }, | |
| { | |
| "first": "Daille", | |
| "middle": [], | |
| "last": "B\u00e9atrice", | |
| "suffix": "", | |
| "affiliation": { | |
| "laboratory": "LIP6-CNRS, UPMC (", | |
| "institution": "", | |
| "location": { | |
| "settlement": "Paris 6) Paris", | |
| "country": "France" | |
| } | |
| }, | |
| "email": "[email protected]" | |
| }, | |
| { | |
| "first": "", | |
| "middle": [], | |
| "last": "Lina", | |
| "suffix": "", | |
| "affiliation": { | |
| "laboratory": "", | |
| "institution": "Universit\u00e9 de Nantes Nantes", | |
| "location": { | |
| "country": "France" | |
| } | |
| }, | |
| "email": "" | |
| } | |
| ], | |
| "year": "", | |
| "venue": null, | |
| "identifiers": {}, | |
| "abstract": "We describe our method for extracting keyphrases from scientific articles which we participate in the shared task of SemEval-2 Evaluation Exercise. Even though general-purpose term extractors along with linguistically-motivated analysis allow us to extract elaborated morphosyntactic variation forms of terms, a na\u00efve statistic approach proposed in this paper is very simple and quite efficient for extracting keyphrases especially from wellstructured scientific articles. Based on the characteristics of keyphrases with section information, we obtain 18.34% for f-measure using top 15 candidates. We also show further improvement without any complications and we discuss this at the end of the paper.", | |
| "pdf_parse": { | |
| "paper_id": "S10-1038", | |
| "_pdf_hash": "", | |
| "abstract": [ | |
| { | |
| "text": "We describe our method for extracting keyphrases from scientific articles which we participate in the shared task of SemEval-2 Evaluation Exercise. Even though general-purpose term extractors along with linguistically-motivated analysis allow us to extract elaborated morphosyntactic variation forms of terms, a na\u00efve statistic approach proposed in this paper is very simple and quite efficient for extracting keyphrases especially from wellstructured scientific articles. Based on the characteristics of keyphrases with section information, we obtain 18.34% for f-measure using top 15 candidates. We also show further improvement without any complications and we discuss this at the end of the paper.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Abstract", | |
| "sec_num": null | |
| } | |
| ], | |
| "body_text": [ | |
| { | |
| "text": "Key phrases are a set of words to capture the main topic of the document. Since key phrases contain the substance of the document, they are used in the large spectrum of areas; from applications which explicitly use key phrases such as automatic indexing, documents classification and search engine optimization in information retrieval, to applications which implicitly use key phrases such as summarization and question-answering systems. During the last decade, many previous works have dealt with the various methods for automatically extracting key phrases (e.g., Frank et al., 1999; Barker and Corrnacchia, 2000; Turney, 2003; Medelyan and Witten, 2006; Nguyen and Kan, 2007; Wan and Xiao, 2008) .", | |
| "cite_spans": [ | |
| { | |
| "start": 569, | |
| "end": 588, | |
| "text": "Frank et al., 1999;", | |
| "ref_id": "BIBREF1" | |
| }, | |
| { | |
| "start": 589, | |
| "end": 618, | |
| "text": "Barker and Corrnacchia, 2000;", | |
| "ref_id": null | |
| }, | |
| { | |
| "start": 619, | |
| "end": 632, | |
| "text": "Turney, 2003;", | |
| "ref_id": "BIBREF5" | |
| }, | |
| { | |
| "start": 633, | |
| "end": 659, | |
| "text": "Medelyan and Witten, 2006;", | |
| "ref_id": "BIBREF3" | |
| }, | |
| { | |
| "start": 660, | |
| "end": 681, | |
| "text": "Nguyen and Kan, 2007;", | |
| "ref_id": "BIBREF4" | |
| }, | |
| { | |
| "start": 682, | |
| "end": 701, | |
| "text": "Wan and Xiao, 2008)", | |
| "ref_id": "BIBREF6" | |
| } | |
| ], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Introduction 1", | |
| "sec_num": "1" | |
| }, | |
| { | |
| "text": "The task of extracting key phrases would be considered as a subtask of extracting terminology if key phrases are a kind of terms. Typical approaches for automatically extracting terms use linguistic preprocessing which involves morphosyntactic analysis such as part-of-speech tagging and phrase chunking, and statistical postprocessing such as log likelihood which compares the term frequencies in a document against their expected frequencies derived in a bigger text. Besides, extracting terms prefers syntactically plausible noun phrases (NPs) which are mainly multiwords terms. Kim and Kan (2009) report that most of key phrases are often simple words than less often compound words 2 .", | |
| "cite_spans": [ | |
| { | |
| "start": 582, | |
| "end": 600, | |
| "text": "Kim and Kan (2009)", | |
| "ref_id": "BIBREF2" | |
| } | |
| ], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Introduction 1", | |
| "sec_num": "1" | |
| }, | |
| { | |
| "text": "The task for extracting key phrases tend to include analyzing the document structure. Especially, extracting key phrases from well-structured scientific articles should consider cross-section information (Nguyen and Kan, 2007) . This information has been explored to assess the suitability of features during learning in Kim and Kan (2009) .", | |
| "cite_spans": [ | |
| { | |
| "start": 204, | |
| "end": 226, | |
| "text": "(Nguyen and Kan, 2007)", | |
| "ref_id": "BIBREF4" | |
| }, | |
| { | |
| "start": 321, | |
| "end": 339, | |
| "text": "Kim and Kan (2009)", | |
| "ref_id": "BIBREF2" | |
| } | |
| ], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Introduction 1", | |
| "sec_num": "1" | |
| }, | |
| { | |
| "text": "Extracting key phrases, however, is more than to extracting terminology or analyzing the document structure. While terms are words which appear in specific contexts and analyse concept structures in domains of human activity, key phrases are words that capture the key idea of documents. In addition, while terms usually occur in the given document more often than we would expect to occur, key phrases do not necessarily occur frequently or key phrases do not occur at all in the document. Consequently, the task for extracting key phrases should not be considered as the subtask of extracting terminology and we are not able to directly apply general-purpose term extractors to extract key phrases.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Introduction 1", | |
| "sec_num": "1" | |
| }, | |
| { | |
| "text": "In this paper, we describe our method for \"Automatic Keyphrase Extraction from Scientific Ar-ticles\", the shared task of SemEval-2 Evaluation Exercise which we participated in. Although term extractors along with linguisticallymotivated analysis allow us to extract even elaborated morpho-syntactic variation forms of terms, the na\u00efve statistic approach proposed in this paper is very simple and quite efficient for extracting keyphrases especially from well-structured scientific articles. In a nutshell, our method is based on empirical rules without any linguisticallymotivated preprocessing. Empirical rules are obtained from the analysis of the characteristics of keyphrases by observing training data.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Introduction 1", | |
| "sec_num": "1" | |
| }, | |
| { | |
| "text": "The remaining of this paper is organized as follows: Section 2 explains the characteristics of keyphrases in scientific articles. Section 3 and 4 detail our na\u00efve statistic approach and experiment, respectively. We conclude this paper and discuss a further improvement in Section 6.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Introduction 1", | |
| "sec_num": "1" | |
| }, | |
| { | |
| "text": "In this section, we investigate the characteristics of keyphrases in training data. Table 1 shows statistics of training data. In Table 1 , D-author means the keyphrases assigned by authors, D-reader the keyphrases assigned by readers, and D-combined the combined keyphrases assigned by both of authors and readers. We measure the distribution of word length of key phrases in training data and present it in Figure 1 . Over half of key phrases are two-word key phrases in both author-and reader-assigned key phrases. Differently with Kim and Kan (2009) which they reported that most of key phrases are often simple words than less often compound words, only 29.7% and 17.7% of key phrases are one-word key phrases. There are also more than four-word key phrases which hold 4.3% and 7.2% of author and reader assigned key phrases, respectively.", | |
| "cite_spans": [ | |
| { | |
| "start": 535, | |
| "end": 553, | |
| "text": "Kim and Kan (2009)", | |
| "ref_id": "BIBREF2" | |
| } | |
| ], | |
| "ref_spans": [ | |
| { | |
| "start": 84, | |
| "end": 91, | |
| "text": "Table 1", | |
| "ref_id": "TABREF1" | |
| }, | |
| { | |
| "start": 130, | |
| "end": 137, | |
| "text": "Table 1", | |
| "ref_id": "TABREF1" | |
| }, | |
| { | |
| "start": 409, | |
| "end": 417, | |
| "text": "Figure 1", | |
| "ref_id": "FIGREF0" | |
| } | |
| ], | |
| "eq_spans": [], | |
| "section": "Characteristics of Keyphrases in Scientific Articles", | |
| "sec_num": "2" | |
| }, | |
| { | |
| "text": "In which section do keyphrases occur frequently?", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Occurrences of keyphrases", | |
| "sec_num": "2.2" | |
| }, | |
| { | |
| "text": "To answer this question, we count the number of occurrences of keyphrases of each section. Due to the variation of the naming of the section, we divide sections into title and abstract, introduction, conclusion, and the rest including references. Table 2 and 3 show the number of occurrences and the accumulative number of unique occurrences of keyphrases in each section, respectively. We also show the accumulative number of words in each section in Table 4 . Including the rest sections exponentially diminishes the ratio of the number of gold keyphrases to the number of candidate keyphrases. Note that m words produce n\u22121 i=0 (m \u2212 i) candidate keyphrases for up to n-word keyphrases by supposing that candidate keyphrases are simple n-word terms.", | |
| "cite_spans": [], | |
| "ref_spans": [ | |
| { | |
| "start": 247, | |
| "end": 254, | |
| "text": "Table 2", | |
| "ref_id": "TABREF4" | |
| }, | |
| { | |
| "start": 452, | |
| "end": 459, | |
| "text": "Table 4", | |
| "ref_id": "TABREF6" | |
| } | |
| ], | |
| "eq_spans": [], | |
| "section": "Occurrences of keyphrases", | |
| "sec_num": "2.2" | |
| }, | |
| { | |
| "text": "Note also that both author-and reader-assigned keyphrases hold only 75.49% and 89.44%, respectively. Even some keyphrases are different with surface forms in the document and our na\u00efve method with no linguistic intervention is not able to recognize them. For example, one of readerassigned keyphrases distributed real-time embedded system for C-41 actually appears as distributed real-time and embedded (DRE) systems. ", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Occurrences of keyphrases", | |
| "sec_num": "2.2" | |
| }, | |
| { | |
| "text": "From training data, we observe and decide the followings:", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Methodology", | |
| "sec_num": "3" | |
| }, | |
| { | |
| "text": "\u2022 More than four-word keyphrases hold only 4.3% and 7.2% of whole keyphrases. We decide that our approach limits the word length as three for extracting keyphrases. Thus we extract only up to three-word keyphrases. This choice might lead the performance degradation of our method because we explicitly exclude more than four-word keyphrases.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Methodology", | |
| "sec_num": "3" | |
| }, | |
| { | |
| "text": "\u2022 Keysections hold 65.19% and 70.29% of keyphrases. We decide that our approach limits keysections from which we extract keyphrases. Including the rest sections may improve recall, but probably diminish precision since the rest sections occupy over 70% of the document.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Methodology", | |
| "sec_num": "3" | |
| }, | |
| { | |
| "text": "\u2022 Almost half of keyphrases occur coincidentally in keysections and the rest sections. We decide that our approach limits coincident keyphrases in both of them. This decision is made empirically and improve precision.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Methodology", | |
| "sec_num": "3" | |
| }, | |
| { | |
| "text": "The following procedure explains and details our approach for extracting keyphrases.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Methodology", | |
| "sec_num": "3" | |
| }, | |
| { | |
| "text": "\u2022 Extract up to three-word terms from keysections as candidate keyphrases. \u2022 Filter them out if they contain one or more of stop words or non-content-containing words (see Table 5 for non-content-containing words). \u2022 Count the number of occurrences of extracted terms from each keysection.", | |
| "cite_spans": [], | |
| "ref_spans": [ | |
| { | |
| "start": 172, | |
| "end": 179, | |
| "text": "Table 5", | |
| "ref_id": "TABREF7" | |
| } | |
| ], | |
| "eq_spans": [], | |
| "section": "Methodology", | |
| "sec_num": "3" | |
| }, | |
| { | |
| "text": "\u2022 Check the coincidence whether candidate keyphrases occurs in more than two keysections. If so, we assign weight. \u2022 Calculate a score for candidate keyphrases and list them by order of the score.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Methodology", | |
| "sec_num": "3" | |
| }, | |
| { | |
| "text": "This section shows the experiment results with training and test data.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Experiment results", | |
| "sec_num": "4" | |
| }, | |
| { | |
| "text": "To optimize our results, we use various thresholds for the number of n-word keyphrases and weight. We try to find the (i : j : k) pattern which means i one-word, j two-word, and K threeword keyphrases to produce the best results. We also try to find the threshold for weight d to calculate the score as follows: if keyphrases appear in more than two keysections, score = d * # of total occurences, otherwise score = # of total occurences. Table 6 shows our best results for training data where (i : j : k) = (3 : 9 : 3) and d = 2. Empirically, we found these thresholds from training data by iterating several possibilities 4 . Table 7 shows our test data results published by organizers of the shared task of SemEval-2 Evaluation Exercise. section, abstract, introduction, conclusion, reference, future work, figure, paper, result, laboratory, university Verb present, how, introduce, become, improve, find, help, improve, consider, call, yield, allow, give, assume Adverb always, formally, necessarily, successfully, previously, usually,mainly, final, essentially, ultinately, commonly, severely, significantly, dramatically, clearly, still, well, who, whose, whom, which, whether, therefore, Other POSs that, this, those, these, many, several, more, over, less, behind, above, below, each, few, different, under, both, within, through, prior, various, better, following, between, possible, via, before,even, such, if, new, show, important, simple, good, tranditional, current, varying, necessary, previous, clear ", | |
| "cite_spans": [ | |
| { | |
| "start": 741, | |
| "end": 1515, | |
| "text": "section, abstract, introduction, conclusion, reference, future work, figure, paper, result, laboratory, university Verb present, how, introduce, become, improve, find, help, improve, consider, call, yield, allow, give, assume Adverb always, formally, necessarily, successfully, previously, usually,mainly, final, essentially, ultinately, commonly, severely, significantly, dramatically, clearly, still, well, who, whose, whom, which, whether, therefore, Other POSs that, this, those, these, many, several, more, over, less, behind, above, below, each, few, different, under, both, within, through, prior, various, better, following, between, possible, via, before,even, such, if, new, show, important, simple, good, tranditional, current, varying, necessary, previous, clear", | |
| "ref_id": null | |
| } | |
| ], | |
| "ref_spans": [ | |
| { | |
| "start": 439, | |
| "end": 446, | |
| "text": "Table 6", | |
| "ref_id": "TABREF8" | |
| }, | |
| { | |
| "start": 628, | |
| "end": 635, | |
| "text": "Table 7", | |
| "ref_id": "TABREF9" | |
| } | |
| ], | |
| "eq_spans": [], | |
| "section": "Training data", | |
| "sec_num": "4.1" | |
| }, | |
| { | |
| "text": "In this paper, we described our simple method for extracting keyphrases from scientific articles which we participate in the shared task of SemEval-2 Evaluation Exercise. The na\u00efve approach was proposed. This approach turned out very simple and quite efficient for extracting keyphrases from well-structured scientific articles. Based on learning the distribution of keyphrases with section information, we obtain 18.34% for fmeasure using top 15 candidates. Our na\u00efve approach still has much room for improvement. For example, we are able to improve the result for same test data up to 20.71% and 25.55% for f-measure using top 15 candidates simply by adding the rest sections and normalizing the number of occurrences of terms from each section 5 . 5 The result is not improved only by adding the rest sections.", | |
| "cite_spans": [ | |
| { | |
| "start": 751, | |
| "end": 752, | |
| "text": "5", | |
| "ref_id": null | |
| } | |
| ], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Conclusion and Discussion", | |
| "sec_num": "5" | |
| }, | |
| { | |
| "text": "Moreover, our n-word terms based extraction can be benefited by linguistic preprocessing such as normalizing surface forms. Handcrafted regular expression rules along with part-of-speech tagging and phrase chunking would be also introduced to improve candidate selection. We have not explored thoroughly feature engineering, neither. For example, more fine-grained section information and weight re-assignment might help filter out irrelevant candidates. We leave these possibilities for future work.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Conclusion and Discussion", | |
| "sec_num": "5" | |
| }, | |
| { | |
| "text": "UNPMC means the collaborative team from Laboratoire d'Informatique de Nantes Atlantique of the Universit\u00e9 de Nantes and Laboratoire d'Informatique de Paris 6 of the Universit\u00e9 Pierre et Marie Curie.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "In training data, only 23.4% of keyphrases, however, are single words.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "We denote title and abstract as A, introduction as I, conclusion as C, and the rest sections including references as Other.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "These thresholds will be more examined in future work.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "", | |
| "sec_num": null | |
| } | |
| ], | |
| "back_matter": [], | |
| "bib_entries": { | |
| "BIBREF0": { | |
| "ref_id": "b0", | |
| "title": "Using noun phrase heads to extract document keyphrases", | |
| "authors": [ | |
| { | |
| "first": "Ken", | |
| "middle": [], | |
| "last": "Barker", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Nadia", | |
| "middle": [], | |
| "last": "Cornacchia", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2000, | |
| "venue": "Proceedings of the 13th Biennial Conference of the Canadian Society on Computational Studies of Intelligence: Advances in Artificial Intelligence", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "40--52", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Ken Barker and Nadia Cornacchia. 2000. Using noun phrase heads to extract document keyphrases. In Proceedings of the 13th Biennial Conference of the Canadian Soci- ety on Computational Studies of Intelligence: Advances in Artificial Intelligence, pages 40-52. May 14-17, 2000. Montr\u00e9al, Quebec, Canada.", | |
| "links": null | |
| }, | |
| "BIBREF1": { | |
| "ref_id": "b1", | |
| "title": "Domain-Specific Keyphrase Extraction", | |
| "authors": [ | |
| { | |
| "first": "Eibe", | |
| "middle": [], | |
| "last": "Frank", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Gordon", | |
| "middle": [ | |
| "W" | |
| ], | |
| "last": "Paynter", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Ian", | |
| "middle": [ | |
| "H" | |
| ], | |
| "last": "Witten", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Carl", | |
| "middle": [], | |
| "last": "Gutwin", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Craig", | |
| "middle": [ | |
| "G" | |
| ], | |
| "last": "Nevill-Manning", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 1999, | |
| "venue": "Proceedings of the 16th International Joint Conference on Artificial Intelligence", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "668--673", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Eibe Frank , Gordon W. Paynter , Ian H. Witten , Carl Gutwin, and Craig G. Nevill-Manning. 1999. Domain-Specific Keyphrase Extraction. In Proceedings of the 16th Inter- national Joint Conference on Artificial Intelligence, pages 668-673. July 31-August 6, 1999. Stockholm, Sweden.", | |
| "links": null | |
| }, | |
| "BIBREF2": { | |
| "ref_id": "b2", | |
| "title": "Re-examining Automatic Keyphrase Extraction Approaches in Scientific Articles", | |
| "authors": [ | |
| { | |
| "first": "Nam", | |
| "middle": [], | |
| "last": "Su", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Min-Yen", | |
| "middle": [], | |
| "last": "Kim", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "", | |
| "middle": [], | |
| "last": "Kan", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2009, | |
| "venue": "Proceedings of the Workshop on Multiword Expressions: Identification, Interpretation, Disambiguation and Applications", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "9--12", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Su Nam Kim and Min-Yen Kan. 2009. Re-examining Auto- matic Keyphrase Extraction Approaches in Scientific Ar- ticles. In Proceedings of the Workshop on Multiword Ex- pressions: Identification, Interpretation, Disambiguation and Applications (MWE 2009), ACL-IJCNLP 2009, pages 9-12. August 6, 2009. Singapore.", | |
| "links": null | |
| }, | |
| "BIBREF3": { | |
| "ref_id": "b3", | |
| "title": "Thesaurus based automatic keyphrase indexing", | |
| "authors": [ | |
| { | |
| "first": "Olena", | |
| "middle": [], | |
| "last": "Medelyan", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Ian", | |
| "middle": [ | |
| "H" | |
| ], | |
| "last": "Witten", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2006, | |
| "venue": "Proceedings of the 6th ACM/IEEE-CS joint conference on Digital libraries", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "296--297", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Olena Medelyan and Ian H. Witten. 2006. Thesaurus based automatic keyphrase indexing. In Proceedings of the 6th ACM/IEEE-CS joint conference on Digital libraries, pages 296-297. June 11-15, 2006. Chapel Hill, NC, USA.", | |
| "links": null | |
| }, | |
| "BIBREF4": { | |
| "ref_id": "b4", | |
| "title": "Key phrase Extraction in Scientific Publications", | |
| "authors": [ | |
| { | |
| "first": "Thuy", | |
| "middle": [ | |
| "Dung" | |
| ], | |
| "last": "Nguyen", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Min-Yen", | |
| "middle": [], | |
| "last": "Kan", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2007, | |
| "venue": "Asian Digital Libraries. Looking Back 10 Years and Forging New Frontiers", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "317--326", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Thuy Dung Nguyen and Min-Yen Kan. 2007. Key phrase Extraction in Scientific Publications. Asian Digital Li- braries. Looking Back 10 Years and Forging New Fron- tiers, pages 317-326. Springer Berlin, Heidelberg.", | |
| "links": null | |
| }, | |
| "BIBREF5": { | |
| "ref_id": "b5", | |
| "title": "Coherent keyphrase extraction via Web mining", | |
| "authors": [ | |
| { | |
| "first": "D", | |
| "middle": [], | |
| "last": "Peter", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "", | |
| "middle": [], | |
| "last": "Turney", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2003, | |
| "venue": "Proceedings of the 18th International Joint Conference on Artificial Intelligence", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "434--439", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Peter D. Turney. 2003. Coherent keyphrase extraction via Web mining. In Proceedings of the 18th International Joint Conference on Artificial Intelligence, pages 434- 439. August 9-15, 2003. Acapulco, Mexico.", | |
| "links": null | |
| }, | |
| "BIBREF6": { | |
| "ref_id": "b6", | |
| "title": "CollabRank: towards a collaborative approach to single-document keyphrase extraction", | |
| "authors": [ | |
| { | |
| "first": "Xiaojun", | |
| "middle": [], | |
| "last": "Wan", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Jianguo", | |
| "middle": [], | |
| "last": "Xiao", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2008, | |
| "venue": "Proceedings of the 22nd International Conference on Computational Linguistics", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "969--976", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Xiaojun Wan and Jianguo Xiao. 2008. CollabRank: towards a collaborative approach to single-document keyphrase extraction. In Proceedings of the 22nd International Conference on Computational Linguistics (Coling 2008), pages 969-976. 18-22 August, 2008. Manchester, UK.", | |
| "links": null | |
| } | |
| }, | |
| "ref_entries": { | |
| "FIGREF0": { | |
| "num": null, | |
| "uris": null, | |
| "type_str": "figure", | |
| "text": "Word length of keyphrases in training data" | |
| }, | |
| "FIGREF1": { | |
| "num": null, | |
| "uris": null, | |
| "type_str": "figure", | |
| "text": "Coincidence of keyphrases" | |
| }, | |
| "TABREF1": { | |
| "num": null, | |
| "type_str": "table", | |
| "text": "", | |
| "content": "<table/>", | |
| "html": null | |
| }, | |
| "TABREF4": { | |
| "num": null, | |
| "type_str": "table", | |
| "text": "Number of occurrences of keyphrases in each section", | |
| "content": "<table><tr><td/><td>D-author</td><td>D-reader</td></tr><tr><td>Total</td><td>563 (100.0%)</td><td>1,865 (100.0%)</td></tr><tr><td>Title and Abstract</td><td>277 (49.20%)</td><td>802 (43.00%)</td></tr><tr><td>'+' Introduction</td><td>317 (56.30%)</td><td>937 (50.24%)</td></tr><tr><td>'+' Conclusion</td><td>367 (65.19%)</td><td>1,311 (70.29%)</td></tr><tr><td>'+' Other</td><td>425 (75.49%)</td><td>1,668 (89.44%)</td></tr></table>", | |
| "html": null | |
| }, | |
| "TABREF5": { | |
| "num": null, | |
| "type_str": "table", | |
| "text": "", | |
| "content": "<table><tr><td>: Accumulative number of unique occur-</td></tr><tr><td>rences of keyphrases in each section</td></tr></table>", | |
| "html": null | |
| }, | |
| "TABREF6": { | |
| "num": null, | |
| "type_str": "table", | |
| "text": "Number of words in training data and gold data (D-reader)", | |
| "content": "<table><tr><td>2.3 Coincidence of keyphrases</td></tr><tr><td>Figure 2 shows the coincidence of keyphrases 3 .</td></tr><tr><td>Almost half of keyphrases (58.44% and 45.74%</td></tr><tr><td>for author-and reader-assigned keyphrases, re-</td></tr><tr><td>spectively) occur coincidentally in keysections</td></tr><tr><td>and the rest sections. Keysections hold 65.19%</td></tr><tr><td>and 70.29% of keyphrases and the rest sections</td></tr><tr><td>besides keysections hold 68.74% and 64.88% of</td></tr><tr><td>whole keyphrases. Note that the rest sections oc-</td></tr><tr><td>cupy over 70% of the document on the average.</td></tr></table>", | |
| "html": null | |
| }, | |
| "TABREF7": { | |
| "num": null, | |
| "type_str": "table", | |
| "text": "Example of (heuristically obtained) non-content-containing terms", | |
| "content": "<table><tr><td colspan=\"2\">AUTHOR.STEM.FINAL</td><td/><td/><td/></tr><tr><td># Gold: 559</td><td>Match</td><td>Precision</td><td>Recall</td><td>F-score</td></tr><tr><td>Top 05</td><td>43</td><td>5.97%</td><td>7.69%</td><td>6.72%</td></tr><tr><td>Top 10</td><td>101</td><td>7.01%</td><td>18.07%</td><td>10.10%</td></tr><tr><td>Top 15</td><td>139</td><td>6.44%</td><td>24.87%</td><td>10.23%</td></tr><tr><td colspan=\"2\">READER.STEM.FINAL</td><td/><td/><td/></tr><tr><td># Gold: 1824</td><td>Match</td><td>Precision</td><td>Recall</td><td>F-score</td></tr><tr><td>Top 05</td><td>118</td><td>16.39%</td><td>6.47%</td><td>9.28%</td></tr><tr><td>Top 10</td><td>249</td><td>17.29%</td><td>13.65%</td><td>15.26%</td></tr><tr><td>Top 15</td><td>361</td><td>16.71%</td><td>19.79%</td><td>18.12%</td></tr><tr><td colspan=\"2\">COMBINED.STEM.FINAL</td><td/><td/><td/></tr><tr><td># Gold: 2223</td><td>Match</td><td>Precision</td><td>Recall</td><td>F-score</td></tr><tr><td>Top 05</td><td>143</td><td>19.86%</td><td>6.43%</td><td>9.71%</td></tr><tr><td>Top 10</td><td>309</td><td>21.46%</td><td>13.90%</td><td>16.87%</td></tr><tr><td>Top 15</td><td>441</td><td>20.42%</td><td>19.84%</td><td>20.13%</td></tr></table>", | |
| "html": null | |
| }, | |
| "TABREF8": { | |
| "num": null, | |
| "type_str": "table", | |
| "text": "", | |
| "content": "<table><tr><td/><td colspan=\"3\">: Training data results</td></tr><tr><td colspan=\"2\">READER.STEM.FINAL</td><td/><td/></tr><tr><td># Gold: 1204</td><td>Precision</td><td>Recall</td><td>Fscore</td></tr><tr><td>Top 05</td><td>13.80%</td><td>5.73%</td><td>8.10%</td></tr><tr><td>Top 10</td><td>15.10%</td><td>12.54%</td><td>13.70%</td></tr><tr><td>Top 15</td><td>14.47%</td><td>18.02%</td><td>16.05%</td></tr><tr><td colspan=\"2\">COMBINED.STEM.FINAL</td><td/><td/></tr><tr><td># Gold: 1466</td><td>Precision</td><td>Recall</td><td>Fscore</td></tr><tr><td>Top 05</td><td>18.00%</td><td>6.14%</td><td>9.16%</td></tr><tr><td>Top 10</td><td>19.00%</td><td>12.96%</td><td>15.41%</td></tr><tr><td>Top 15</td><td>18.13%</td><td>18.55%</td><td>18.34%</td></tr></table>", | |
| "html": null | |
| }, | |
| "TABREF9": { | |
| "num": null, | |
| "type_str": "table", | |
| "text": "Test data results", | |
| "content": "<table/>", | |
| "html": null | |
| } | |
| } | |
| } | |
| } |