| { | |
| "paper_id": "O13-1011", | |
| "header": { | |
| "generated_with": "S2ORC 1.0.0", | |
| "date_generated": "2023-01-19T08:04:08.801551Z" | |
| }, | |
| "title": "An Exploration of Textual Entailment and Reading Comprehension for Chinese and English", | |
| "authors": [ | |
| { | |
| "first": "Wei-Jie", | |
| "middle": [], | |
| "last": "Huang", | |
| "suffix": "", | |
| "affiliation": { | |
| "laboratory": "", | |
| "institution": "National Chengchi University {100753014", | |
| "location": { | |
| "postCode": "101753028" | |
| } | |
| }, | |
| "email": "" | |
| }, | |
| { | |
| "first": "Po-Cheng", | |
| "middle": [], | |
| "last": "Lin", | |
| "suffix": "", | |
| "affiliation": { | |
| "laboratory": "", | |
| "institution": "National Chengchi University {100753014", | |
| "location": { | |
| "postCode": "101753028" | |
| } | |
| }, | |
| "email": "" | |
| }, | |
| { | |
| "first": "Chao-Lin", | |
| "middle": [], | |
| "last": "Liu", | |
| "suffix": "", | |
| "affiliation": { | |
| "laboratory": "", | |
| "institution": "National Chengchi University {100753014", | |
| "location": { | |
| "postCode": "101753028" | |
| } | |
| }, | |
| "email": "" | |
| } | |
| ], | |
| "year": "", | |
| "venue": null, | |
| "identifiers": {}, | |
| "abstract": "Research on text entailment studies the logical relationships between statements. We employed linguistic information at the lexical, syntactic, and semantic levels to build heuristics and machine-learning based models for algorithmic judgment of text entailment relationships. Methods proposed in this paper achieved relatively very good performances in the RITE task for both traditional and simplified Chinese entailment problems in NTCIR-10. We extended our work and attempted to automatically answer questions in reading comprehension tests in Chinese and English used in elementary and middle schools. To make the automatic answering more feasible, we manually selected statements which were relevant to the test items before we ran the text entailment component. Experimental results indicated that it was then possible to find the answers better than 50% of the time for one out of four multiple-choice items.", | |
| "pdf_parse": { | |
| "paper_id": "O13-1011", | |
| "_pdf_hash": "", | |
| "abstract": [ | |
| { | |
| "text": "Research on text entailment studies the logical relationships between statements. We employed linguistic information at the lexical, syntactic, and semantic levels to build heuristics and machine-learning based models for algorithmic judgment of text entailment relationships. Methods proposed in this paper achieved relatively very good performances in the RITE task for both traditional and simplified Chinese entailment problems in NTCIR-10. We extended our work and attempted to automatically answer questions in reading comprehension tests in Chinese and English used in elementary and middle schools. To make the automatic answering more feasible, we manually selected statements which were relevant to the test items before we ran the text entailment component. Experimental results indicated that it was then possible to find the answers better than 50% of the time for one out of four multiple-choice items.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Abstract", | |
| "sec_num": null | |
| } | |
| ], | |
| "body_text": [ | |
| { | |
| "text": "EQUATION", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [ | |
| { | |
| "start": 0, | |
| "end": 8, | |
| "text": "EQUATION", | |
| "ref_id": "EQREF", | |
| "raw_str": "T 1 \u548c T 2 \u5206\u5225\u4f5c\u70ba\u53e5\u5c0d\u65b7\u8a5e\u6216\u5206\u8a5e\u5f8c\u7684\u8a5e\u5f59\u96c6\u5408\uff0c\u5176\u4e2d\u4ee5 T 2 \u4f5c\u70ba\u6587\u672c\uff0c\u8a08\u7b97\u5169 \u500b\u53e5\u5b50\u7684\u8a5e\u5f59\u91cd\u758a\u6bd4\u4f8b\uff0c\u5982\u4e0b\u65b9\u516c\u5f0f(1)\uff0cT 1 \u548c T 2 \u5206\u5225\u70ba\u5169\u500b\u53e5\u5b50\u65b7\u8a5e\u6216\u5206\u8a5e\u5f8c\u7684\u8a5e\u5f59\u96c6 \u5408\uff0c\u900f\u904e\u8a72\u516c\u5f0f\u8a08\u7b97\u5169\u500b\u53e5\u5b50\u9593\u76f8\u540c\u8a5e\u5f59\u7684\u6bd4\u4f8b\uff0c\u4ee5 0 \u81f3 1 \u8868\u793a\u76f8\u540c\u6bd4\u4f8b\u7684\u9ad8\u4f4e\u3002 (1) \u4f46\u516c\u5f0f(1)\u9700\u8981\u8a5e\u5f59\u5b8c\u5168\u76f8\u540c\u624d\u6703\u7d0d\u5165\u8a08\u7b97\uff0c\u5982\u6b64\u4e00\u4f86\u53ef\u80fd\u6703\u6f0f\u6389\u90e8\u5206\u7684\u7e2e\u5beb\u8a5e\u5f59\u3001\u540c \u7fa9\u8a5e\u5f59\u6216\u56e0\u70ba\u5404\u7a2e\u539f\u56e0\u88ab\u65b7\u8a5e\u5668\u65b7\u958b\u7684\u60c5\u6cc1\uff0c\u56e0\u6b64\u6211\u5011\u4fee\u6539\u4e86\u516c\u5f0f(1)\uff0c\u52a0\u5165\u8a5e\u5f59\u90e8\u5206\u76f8\u540c \u7684\u8a08\u7b97\uff0c\u4e5f\u628a\u8fd1\u7fa9\u8a5e\u7684\u5224\u65b7[12][15]\u52a0\u5165\u5230\u516c\u5f0f(1)\u7684\u4fee\u6539\uff0c\u4f7f\u4e4b\u6210\u70ba\u516c\u5f0f(2) (2) 3.2 \u5be6\u9ad4\u540d\u8a5e\u5224\u65b7 \u5982\u679c\u53ea\u4f7f\u7528\u8a5e\u5f59\u7684\u8986\u84cb\u6bd4\u4f8b\u4f86\u8868\u793a\u53e5\u5b50\u9593\u7684\u63a8\u8ad6\u95dc\u4fc2\uff0c\u6211\u5011\u50c5\u80fd\u638c\u63e1\u53e5\u5b50\u8868\u9762\u7684\u8cc7\u8a0a\u542b \u91cf\uff0c\u800c\u7121\u6cd5\u4e86\u89e3\u53e5\u5b50\u6240\u8868\u9054\u7684\u5167\u5bb9\uff0c\u56e0\u6b64\u900f\u904e\u5be6\u9ad4\u540d\u8a5e\u6a19\u8a18\uff0c\u5c07\u53e5\u5b50\u4e2d\u7684\u4eba\u540d\u3001\u5730\u540d\u548c\u7d44 \u7e54\u540d\u64f7\u53d6\u51fa\u4f86\uff0c\u4e26\u628a\u9019\u4e9b\u6a19\u8a18\u51fa\u4f86\u7684\u8a5e\u5f59\u8996\u70ba\u91cd\u8981\u7684\u8cc7\u8a0a\uff0c\u5c07\u6709\u52a9\u65bc\u5224\u5225\u53e5\u5b50\u9593\u7684\u63a8\u8ad6\u95dc \u4fc2\u3002 \u6211\u5011\u5c07\u4e0a\u8ff0\u7684\u5047\u8a2d\u52a0\u5165\u4e00\u500b\u51fd\u6578\uff0c\u8abf\u6574\u63a8\u8ad6\u95dc\u4fc2\u7684\u8a08\u7b97\uff0c\u5982\u516c\u5f0f(3)\uff0cNE t2 \u70ba t 2 \u4e2d\u64f7\u53d6 \u51fa\u7684\u5be6\u9ad4\u540d\u8a5e\uff0c\u5176\u4e2d t \u70ba T \u53e5\u5b50\u7d93\u7531\u65b7\u8a5e\u6216\u5206\u8a5e\u5f8c\u7684\u96c6\u5408\uff1bf NEPenalty \u6703\u5224\u65b7 NE t2 \u4e2d\u7684\u5143\u7d20 \u6703\u88ab\u5305\u542b\u65bc t 1 \u8207\u5426\uff0c\u7576\u6709\u5143\u7d20\u4e0d\u5305\u542b\u5728 t 1 \u6642\uff0c\u5247\u7d66\u4e88\u4e00\u6b21\u7bc4\u570d 0 \u81f3 1 \u7684 \u03b1 \u61f2\u7f70\u5206\u6578\u3002\u56e0\u6b64 \u63a8\u8ad6\u95dc\u4fc2\u7684\u5224\u65b7\u52a0\u5165\u8a72\u51fd\u5f0f\uff0c\u8b8a\u6210\u516c\u5f0f(4)\u3002 \u03b1 (3)", | |
| "eq_num": "(4)" | |
| } | |
| ], | |
| "section": "", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "Proceedings of the Twenty-Fifth Conference on Computational Linguistics and Speech Processing(ROCLING 2013)", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "", | |
| "sec_num": null | |
| } | |
| ], | |
| "back_matter": [], | |
| "bib_entries": { | |
| "BIBREF0": { | |
| "ref_id": "b0", | |
| "title": "Textual Entailment Through Extended Lexical Overlap", | |
| "authors": [ | |
| { | |
| "first": "Rod", | |
| "middle": [], | |
| "last": "Adams", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2006, | |
| "venue": "Proceedings of the Second PASCAL Challenges Workshop on Recognising Textual Entailment", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "128--133", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Rod Adams, \"Textual Entailment Through Extended Lexical Overlap\", Proceedings of the Second PASCAL Challenges Workshop on Recognising Textual Entailment, pp. 128-133, 2006.", | |
| "links": null | |
| }, | |
| "BIBREF1": { | |
| "ref_id": "b1", | |
| "title": "The PASCAL Recognising Textual Entailment Challenge", | |
| "authors": [ | |
| { | |
| "first": "Oren", | |
| "middle": [], | |
| "last": "Ido Dagon", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Bernardo", | |
| "middle": [], | |
| "last": "Glickman", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "", | |
| "middle": [], | |
| "last": "Magnini", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2006, | |
| "venue": "", | |
| "volume": "3944", | |
| "issue": "", | |
| "pages": "177--190", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Ido Dagon, Oren Glickman and Bernardo Magnini, \"The PASCAL Recognising Textual Entailment Challenge\", Machine Learning Challenges. Lecture Notes in Computer Science, 3944, pp. 177-190, Springer, 2006.", | |
| "links": null | |
| }, | |
| "BIBREF2": { | |
| "ref_id": "b2", | |
| "title": "The WEKA Data Mining Software: An Update", | |
| "authors": [ | |
| { | |
| "first": "Mark", | |
| "middle": [], | |
| "last": "Hall", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Eibe", | |
| "middle": [], | |
| "last": "Frank", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Geoffrey", | |
| "middle": [], | |
| "last": "Holmes", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Bernhard", | |
| "middle": [], | |
| "last": "Pfahringer", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Peter", | |
| "middle": [], | |
| "last": "Reutemann", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Ian", | |
| "middle": [ | |
| "H" | |
| ], | |
| "last": "Witten", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2009, | |
| "venue": "", | |
| "volume": "11", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Mark Hall, Eibe Frank, Geoffrey Holmes, Bernhard Pfahringer, Peter Reutemann, and Ian H. Witten, \"The WEKA Data Mining Software: An Update\", SIGKDD Explorations, 11(1), 2009.", | |
| "links": null | |
| }, | |
| "BIBREF3": { | |
| "ref_id": "b3", | |
| "title": "Team SKL's Strategy and Experience in RITE2", | |
| "authors": [ | |
| { | |
| "first": "Shohei", | |
| "middle": [], | |
| "last": "Hattori", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Satoshi", | |
| "middle": [], | |
| "last": "Sato", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2013, | |
| "venue": "Proceedings of NTCIR-10 Workshop Meeting", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "435--442", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Shohei Hattori, Satoshi Sato, \"Team SKL's Strategy and Experience in RITE2\", Proceedings of NTCIR-10 Workshop Meeting, pp. 435-442, 2013.", | |
| "links": null | |
| }, | |
| "BIBREF4": { | |
| "ref_id": "b4", | |
| "title": "A Practical Guide to Support Vector Classification", | |
| "authors": [ | |
| { | |
| "first": "Chih-Wei", | |
| "middle": [], | |
| "last": "Hsu", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Chih-Chung", | |
| "middle": [], | |
| "last": "Chang", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Chih Jen", | |
| "middle": [], | |
| "last": "Lin", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2010, | |
| "venue": "", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Chih-Wei Hsu, Chih-Chung Chang and Chih Jen Lin, A Practical Guide to Support Vector Classification. Retrieved from website: http://www.csie.ntu.edu.tw/~cjlin/papers/guide/guide.pdf, 2010.", | |
| "links": null | |
| }, | |
| "BIBREF5": { | |
| "ref_id": "b5", | |
| "title": "The WHUTE System in NTCIR-10 RITE Task", | |
| "authors": [ | |
| { | |
| "first": "Hongmial", | |
| "middle": [], | |
| "last": "Han Ren", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Chen", | |
| "middle": [], | |
| "last": "Wu", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Donghong", | |
| "middle": [], | |
| "last": "Lv", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Jing", | |
| "middle": [], | |
| "last": "Ji", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "", | |
| "middle": [], | |
| "last": "Wan", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2013, | |
| "venue": "Proceedings of NTCIR-10 Workshop Meeting", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "560--565", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Han Ren, Hongmial Wu, Chen Lv, Donghong Ji, and Jing Wan, \"The WHUTE System in NTCIR-10 RITE Task\", Proceedings of NTCIR-10 Workshop Meeting, pp. 560-565, 2013.", | |
| "links": null | |
| }, | |
| "BIBREF6": { | |
| "ref_id": "b6", | |
| "title": "Foundations of Statistical Natural Language Processing", | |
| "authors": [ | |
| { | |
| "first": "Chris", | |
| "middle": [], | |
| "last": "Manning", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Hinrich", | |
| "middle": [], | |
| "last": "Schutze", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 1999, | |
| "venue": "", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Chris Manning and Hinrich Schutze, Foundations of Statistical Natural Language Processing, MIT Press. Cambridge, MA: May 1999.", | |
| "links": null | |
| }, | |
| "BIBREF7": { | |
| "ref_id": "b7", | |
| "title": "Overview of NTCIR-9 RITE: Recognizing Inference in Text", | |
| "authors": [ | |
| { | |
| "first": "Hideki", | |
| "middle": [], | |
| "last": "Shima", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Hiroshi", | |
| "middle": [], | |
| "last": "Kanayama", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Cheng-Wei", | |
| "middle": [], | |
| "last": "Lee", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Chuan-Jie", | |
| "middle": [], | |
| "last": "Lin", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Teruko", | |
| "middle": [], | |
| "last": "Mitamura", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Yusuke", | |
| "middle": [], | |
| "last": "Miyao", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Shuming", | |
| "middle": [], | |
| "last": "Shi", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Koichi", | |
| "middle": [], | |
| "last": "Takeda", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2011, | |
| "venue": "Proceedings of NTCIR-9 Workshop Meeting", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "291--301", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Hideki Shima, Hiroshi Kanayama, Cheng-Wei Lee, Chuan-Jie Lin, Teruko Mitamura, Yusuke Miyao, Shuming Shi and Koichi Takeda, \"Overview of NTCIR-9 RITE: Recognizing Inference in Text,\" Proceedings of NTCIR-9 Workshop Meeting, pp. 291-301, 2011.", | |
| "links": null | |
| }, | |
| "BIBREF9": { | |
| "ref_id": "b9", | |
| "title": "NTCIR(NII Test Collection for IR Systems", | |
| "authors": [], | |
| "year": null, | |
| "venue": "", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "NTCIR(NII Test Collection for IR Systems) Project http://research.nii.ac.jp/ntcir/index-en.html", | |
| "links": null | |
| }, | |
| "BIBREF10": { | |
| "ref_id": "b10", | |
| "title": "Microsoft Research Paraphrase Corpus", | |
| "authors": [], | |
| "year": null, | |
| "venue": "", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Microsoft Research Paraphrase Corpus, http://research.microsoft.com/en-us/downloads/607d14d9-20cd-47e3-85bc-a2f65cd28042/", | |
| "links": null | |
| }, | |
| "BIBREF11": { | |
| "ref_id": "b11", | |
| "title": "WEKA", | |
| "authors": [], | |
| "year": null, | |
| "venue": "", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "WEKA, http://www.cs.waikato.ac.nz/ml/weka/", | |
| "links": null | |
| } | |
| }, | |
| "ref_entries": { | |
| "TABREF0": { | |
| "html": null, | |
| "num": null, | |
| "type_str": "table", | |
| "text": ":\u65e5\u672c\u6642\u9593 2011 \u5e74 3 \u65e5 11 \u65e5\uff0c\u65e5\u672c\u5bae\u57ce\u7e23\u767c\u751f\u82ae\u6c0f\u898f\u6a21 9.0 \u5f37\u9707\uff0c\u9020\u6b7b\u50b7\u5931\u8e64\u7d04 3 \u842c\u591a\u4eba\u3002 \u5047\u8a2d:\u65e5\u672c\u6642\u9593 2011 \u5e74 3 \u65e5 11 \u65e5\uff0c\u65e5\u672c\u5bae\u57ce\u7e23\u767c\u751f\u82ae\u6c0f\u898f\u6a21 9.0 \u5f37\u9707\u3002 Recognizing Textual Entailment(RTE)[2]\u548c Recognizing Inference in Text(RITE)[8]\u5247\u70ba \u76ee\u524d\u70ba\u6587\u5b57\u860a\u6db5\u6240\u8209\u8fa6\u7684\u76f8\u95dc\u7af6\u8cfd\uff0c\u8a72\u6bd4\u8cfd\u5c07\u53e5\u5c0d\u5206\u985e\u70ba Yes \u6216 No \u5169\u7a2e\u63a8\u8ad6\u7684\u7d50\u679c\uff1b\u4ee5 \u4e0b\u9762\u9019\u7d44\u53e5\u5c0d\u70ba\u4f8b\uff0c\u300e\u5c3c\u6cca\u723e\u6bdb\u6d3e\u53db\u4e82\u4efd\u5b50\u5728\u65b0\u570b\u738b\u5927\u58fd\u524d\u5915\u767c\u52d5\u653b\u64ca\u300f\u8207\u300e\u5c3c\u6cca\u723e\u6bdb\u6d3e \u53db\u4e82\u4efd\u5b50\u5728\u65b0\u570b\u738b\u83ef\u8a95\u524d\u5915\u767c\u52d5\u653b\u64ca\u300f\uff0c\u524d\u53e5\u8207\u5f8c\u53e5\u5dee\u5225\u65bc\u300c\u5927\u58fd\u300d\u8207\u300c\u83ef\u8a95\u300d\uff0c\u4f46\u5169\u53e5 \u7684\u542b\u7fa9\u662f\u76f8\u540c\u7684\uff0c\u56e0\u6b64\u6211\u5011\u671f\u5f85\u7cfb\u7d71\u5224\u5225\u8a72\u53e5\u5c0d\u6709\u63a8\u8ad6\u7684\u95dc\u4fc2\uff0c\u4e26\u5f97\u5230 (Parse Trees) \u3001 POSes(Parts-Of-Speech) \u52d5 \u8a5e \u6a19 \u8a18 \u548c \u8a5e \u5f59 \u4f9d \u8cf4 \u95dc \u4fc2 (Word Dependency) [7]\u505a\u70ba\u8a13\u7df4\u6a21\u578b\u7684\u7279\u5fb5\u96c6\u5408\uff0c\u4e26\u63a1\u7528\u4e09\u7a2e\u4e0d\u540c\u7684\u5206\u985e\u6f14\u7b97\u6cd5\u8a13\u7df4\u5206\u985e\u6a21\u578b\uff0c \u5206\u5225\u662f\u652f\u6301\u5411\u91cf\u6a5f(Support Vector Machines, SVMs)[5]\u3001\u6c7a\u7b56\u6a39(Decision Trees)\u8207\u7dda\u6027\u56de\u6b78 (Linear Regression)[3]\uff0c\u900f\u904e\u4e0d\u540c\u985e\u578b\u7684\u5206\u985e\u5668\u7372\u5f97\u63a8\u8ad6\u95dc\u4fc2\u7684\u7d50\u679c\u3002 Question Answering(QA)\u3001Information Retrieval(IR)\u3001multi-document Summarization \u7b49\u7b49\u3002 Proceedings of the Twenty-Fifth Conference on Computational Linguistics and Speech Processing (ROCLING 2013) \u5716\u4e00\u3001\u7d93\u9a57\u6cd5\u5247\u5f0f\u63a8\u8ad6\u7cfb\u7d71\u67b6\u69cb\u8207\u6d41\u7a0b RITE \u5247\u662f NTCIR(NACSIS Test Collections for IR)\u570b\u969b\u8cc7\u8a0a\u6aa2\u7d22\u8a55\u4f30\u7af6\u8cfd\u7684\u5176\u4e2d\u4e00\u9805\u5b50 \u4efb\u52d9\uff0c\u8207 RTE \u4e0d\u540c\u7684\u662f\uff0cRecognizing Inference in Text (RITE-1) \u7af6\u8cfd\u958b\u59cb\u91dd\u5c0d\u4e2d\u6587\u8a9e\u53e5\u63a8 \u8ad6\u7684\u7814\u7a76\u8b70\u984c\u63d0\u4f9b\u8a55\u4f30\u7684\u5e73\u53f0\uff0c\u76ee\u7684\u662f\u70ba\u4e86\u8b93\u4e2d\u6587\u6bcd\u8a9e\u4f7f\u7528\u8005\u4e5f\u80fd\u5c08\u6ce8\u5230\u6b64\u8b70\u984c\u4e0a\u3002", | |
| "content": "<table><tr><td>1 \u7dd2\u8ad6 \u5728\u81ea\u7136\u8a9e\u8a00\u8655\u7406\u7684\u9818\u57df\u4e2d\uff0c\u8b93\u96fb\u8166\u80fd\u5920\u7406\u89e3\u4eba\u985e\u4f7f\u7528\u7684\u8a9e\u8a00\uff0c\u9032\u800c\u5e36\u7d66\u4eba\u985e\u4fbf\u5229\u7684\u751f\u6d3b\uff0c \u662f\u8a72\u9818\u57df\u7684\u7814\u7a76\u8005\u4e00\u76f4\u8ffd\u6c42\u7684\u76ee\u6a19\uff0c\u5176\u4e2d\u6587\u5b57\u860a\u6db5 Textual Entailment(TE)\u4fbf\u662f\u4e00\u500b\u76f8\u7576\u91cd \u8981\u7684\u8b70\u984c\uff0c\u85c9\u7531\u6587\u5b57\u860a\u6db5\u7684\u6280\u8853\u53ef\u4ee5\u5ef6\u4f38\u5230\u5f88\u591a\u61c9\u7528\u65b9\u9762\uff0c\u4f8b\u5982\u5728\u554f\u7b54\u7cfb\u7d71\u3001\u4fe1\u606f\u62bd\u53d6\u3001 \u95b1\u8b80\u7406\u89e3\u7b49\u7b49\u90fd\u6709\u5f88\u5927\u7684\u76ca\u52a9\uff0c\u800c\u6240\u8b02\u7684\u6587\u5b57\u860a\u6db5\u5c31\u662f\u8b93\u96fb\u8166\u81ea\u52d5\u5224\u65b7\u5169\u500b\u53e5\u5b50\u662f\u5426\u5177\u6709 \u63a8\u5c0e\u7684\u95dc\u4fc2\uff0c\u5728\u6587\u5b57\u860a\u6db5\u7684\u6846\u67b6\u4e2d\uff0c\u6211\u5011\u5c07\u53e5\u5c0d\u500b\u5225\u4ee5\u6587\u672c( )\u548c\u5047\u8a2d( )\u4f5c\u70ba\u5206\u5225\uff0c\u4ee5\u4e0b \u9762\u7684\u53e5\u5c0d\u70ba\u4f8b\uff0c\u6587\u672c\u5373\u53ef\u4ee5\u63a8\u5c0e\u81f3\u5047\u8a2d\uff0c\u56e0\u70ba\u5047\u8a2d\u6240\u64c1\u6709\u7684\u8cc7\u8a0a\u90fd\u5305\u542b\u65bc\u6587\u672c\u5167\u3002\u540c\u6642\uff0c Yes \u7684\u63a8\u8ad6\u7d50\u679c\u3002 \u6211\u5011\u5728\u5224\u65b7\u53e5\u5b50\u7684\u63a8\u8ad6\u95dc\u4fc2\u4e0a\u5206\u70ba\u5169\u500b\u505a\u6cd5\u4f5c\u70ba\u5224\u5225\u7684\u4f9d\u64da\uff1b\u7b2c\u4e00\u500b\u65b9\u6cd5\u662f\u4f7f\u7528\u7d93\u9a57 \u6cd5\u5247\u5f0f\u7684\u63a8\u8ad6\u6a21\u578b\uff0c\u8a72\u6a21\u578b\u5c07\u53ef\u80fd\u6703\u5f71\u97ff\u5230\u6587\u5b57\u860a\u6db5\u7684\u7279\u5fb5\u8cc7\u8a0a\u64f7\u53d6\u4e0b\u4f86\uff0c\u4e26\u5229\u7528\u52a0\u6e1b\u5206 \u7684\u6a5f\u5236\uff0c\u5c07\u4e4b\u5f62\u6210\u4e00\u500b\u8a08\u7b97\u516c\u5f0f\uff0c\u4f8b\u5982\u6211\u5011\u8a8d\u70ba\u7576\u5169\u500b\u53e5\u5b50\u7684\u8a5e\u5f59\u8986\u84cb[1]\u6bd4\u4f8b\u5920\u9ad8\uff0c\u67d0\u65b9 \u9762\u4e5f\u4ee3\u8868\u8457\u53e5\u5c0d\u9593\u5177\u6709\u76f8\u540c\u7684\u8cc7\u8a0a\u91cf\uff0c\u56e0\u6b64\u5728\u516c\u5f0f\u4e2d\uff0c\u8a5e\u5f59\u8986\u84cb\u7684\u6bd4\u4f8b\u5c31\u4ee5\u52a0\u5206\u7684\u65b9\u5f0f\u4f86 \u8655\u7406\uff1b\u800c\u53e5\u5c0d\u9593\u7684\u5426\u5b9a\u8a5e\u6578\u91cf\u5982\u679c\u4e0d\u4e00\u6a23\uff0c\u53e5\u5b50\u7684\u542b\u7fa9\u4e5f\u53ef\u80fd\u5927\u76f8\u9015\u5ead\uff0c\u56e0\u6b64\u7576\u5426\u5b9a\u8a5e\u7684 \u6578\u91cf\u4e0d\u540c\u6642\u7cfb\u7d71\u5247\u4ee5\u6e1b\u5206\u7684\u65b9\u5f0f\u8655\u7406\uff0c\u85c9\u7531\u9019\u4e9b\u7279\u5fb5\u7684\u52a0\u6e1b\u5206\u8a08\u7b97\u6700\u5f8c\u6211\u5011\u53ef\u4ee5\u5224\u5225\u6240\u5f97 \u7684\u5206\u6578\u662f\u5426\u6709\u8d85\u904e\u63a8\u8ad6\u7684\u9580\u6abb\u503c\uff0c\u518d\u4ee5\u6b64\u4f5c\u70ba\u5224\u5225\u63a8\u8ad6\u7684\u4f9d\u64da\u3002 \u7b2c\u4e8c\u500b\u65b9\u6cd5\u662f\u4f7f\u7528\u6a5f\u5668\u5b78\u7fd2\u7684\u65b9\u5f0f\uff0c\u9664\u4e86\u85c9\u7531\u7b2c\u4e00\u500b\u65b9\u6cd5\u6240\u8490\u96c6\u5230\u7684\u7279\u5fb5\u8cc7\u8a0a\uff0c\u6211\u5011 \u4e5f\u5c07\u5256 \u6790 \u6a39 \u6211\u5011\u5229\u7528\u4ee5\u4e0a\u8ff0\u5efa\u69cb\u7684\u63a8\u8ad6\u6a21\u578b\u53c3\u52a0 NTCIR-10[10]\u570b\u969b\u8cc7\u8a0a\u8a55\u4f30\u7af6\u8cfd\uff0c\u5728\u6587\u672c\u860a\u6db5 RITE \u7c21\u9ad4\u4e2d\u6587\u8207\u7e41\u9ad4\u4e2d\u6587\u5169\u500b\u5206\u9805\u7372\u5f97\u7b2c\u4e8c\u540d\u3002\u5176\u4e2d\u4f5c\u70ba\u7e41\u9ad4\u4e2d\u6587\u53ca\u7c21\u9ad4\u4e2d\u6587\u63a8\u8ad6\u8a55\u5206\u6a19 \u6e96\u7684 Macro-F1 \u5206\u5225\u70ba 67.07%\u548c 68.09%\u3002 \u672c\u7bc7\u8ad6\u6587\u65bc\u7b2c\u4e8c\u7bc0\u4ecb\u7d39\u95dc\u65bc\u53e5\u5b50\u860a\u6db5\u7684\u76f8\u95dc\u7af6\u8cfd\uff0c\u7b2c\u4e09\u7bc0\u4ecb\u7d39\u7d93\u9a57\u6cd5\u5247\u5f0f\u63a8\u8ad6\u6a21\u578b\u4ee5 \u53ca\u6211\u5011\u6240\u8490\u96c6\u8a8d\u70ba\u5c0d\u6587\u5b57\u860a\u6db5\u6709\u5e6b\u52a9\u7684\u8a9e\u6587\u7279\u5fb5\u8cc7\u8a0a\uff0c\u4e26\u65bc\u7b2c\u56db\u7bc0\u5448\u73fe\u5be6\u9a57\u7684\u7d50\u679c\u548c\u7d50 \u8ad6\uff1b\u7b2c\u4e94\u7bc0\u4ecb\u7d39\u6a5f\u5668\u5b78\u7fd2\u7684\u65b9\u6cd5\u5305\u542b\u8490\u96c6\u65b0\u7684\u7279\u5fb5\u3001\u7279\u5fb5\u7684\u64f7\u53d6\uff1b\u7b2c\u516d\u7bc0\u5247\u662f\u5be6\u6e2c\u6211\u5011\u7684 \u6f14\u7b97\u6cd5\u7684\u5be6\u9a57\u7d50\u679c\u3002\u7b2c\u4e03\u7bc0\u6211\u5011\u5c07\u524d\u9762\u6240\u5efa\u69cb\u7684\u7d93\u9a57\u6cd5\u5247\u5f0f\u63a8\u8ad6\u6a21\u578b\u548c\u6a5f\u5668\u5b78\u7fd2\u6a21\u578b\u61c9\u7528 \u5728\u95b1\u8b80\u7406\u89e3\u7684\u61c9\u7528\u4e0a\uff0c\u7b2c\u516b\u7bc0\u5247\u5448\u73fe\u95b1\u8b80\u7406\u89e3\u5be6\u9a57\u7684\u7d50\u679c\u53ca\u7d50\u8ad6\uff0c\u6700\u5f8c\u7b2c\u4e5d\u7bc0\u70ba\u7d50\u8ad6\u4ee5\u53ca \u672a\u4f86\u5c55\u671b\u3002 2 Textual Entailment \u80cc\u666f\u8cc7\u8a0a 2.1 \u76f8\u95dc\u7af6\u8cfd RTE \u662f\u57fa\u65bc\u82f1\u6587\u8a9e\u6599\u5c0d\u8a9e\u53e5\u63a8\u8ad6\u7684\u76f8\u95dc\u7af6\u8cfd\uff0c\u5f9e 2005 \u5e74\u958b\u59cb\uff0c\u7531 First Recognition Textual Entailment(RTE-1)\u6240\u8209\u8fa6\u7684\u7b2c\u4e00\u6b21\u6bd4\u8cfd\uff0c\u4e26\u91dd\u5c0d\u82f1\u6587\u8a9e\u53e5\u63a8\u8ad6\u63d0\u4f9b\u8a55\u4f30\u7684\u5e73\u53f0\uff0c\u4f7f\u5f97\u53e5\u5b50 \u7684\u63a8\u8ad6\u95dc\u4fc2\u9010\u6f38\u53d7\u5230\u91cd\u8996\uff0c\u800c\u96a8\u5f8c RTE \u7684\u7af6\u8cfd\u4e5f\u589e\u52a0\u4e86\u8a31\u591a\u95dc\u65bc\u8a9e\u610f\u63a8\u8ad6\u7684\u76f8\u95dc\u61c9\u7528\uff0c\u4f8b \u5728 RITE-2 \u7684\u7af6\u8cfd\u4e2d\uff0c\u6211\u5011\u767c\u73fe\u591a\u6578\u7684\u968a\u4f0d\u5728\u7814\u7a76\u6587\u5b57\u860a\u6db5\u6642\uff0c\u90fd\u6709\u4f7f\u7528\u8a5e\u5f59\u7684\u8986\u84cb\u6bd4\u7387 \u8207\u53e5\u5b50\u8868\u9762\u76f8\u4f3c\u5ea6[4]\u4f5c\u70ba\u5224\u5225\u6587\u5b57\u860a\u6db5\u7684\u91cd\u8981\u7279\u5fb5\uff0c\u7136\u800c\u50c5\u50c5\u9019\u4e9b\u65b9\u6cd5\u4e26\u4e0d\u8db3\u4ee5\u5224\u5225\u6587\u5b57 \u7684\u860a\u6db5\u95dc\u4fc2\uff0c\u56e0\u6b64\u67d0\u4e9b\u65b9\u6cd5\u5982 Wu[6]\u6240\u63d0\u51fa\u7684 LCS Similarity \u7528\u4f86\u5224\u5225 \u53ca \u53e5\u5c0d\u7684\u6700\u9577\u76f8 \u540c\u5b57\u4e32\uff0c\u7576\u4f5c\u5224\u5225\u860a\u542b\u7684\u4f9d\u64da\uff0c\u6216\u662f Hattori[4]\u5229\u7528\u53e5\u5b50\u8868\u9762\u76f8\u4f3c\u5ea6\u548c\u53e5\u610f\u76f8\u4f3c\u5ea6\u7684\u9ad8\u4f4e\uff0c \u7d44\u5408\u6210\u4e00\u500b 2x2 \u7684\u77e9\u9663\u4f5c\u70ba\u5224\u5225\u7684\u7b56\u7565\uff0c\u56e0\u6b64\u53ef\u4ee5\u9032\u4e00\u6b65\u7684\u5206\u6790 2x2 \u56db\u7a2e\u60c5\u6cc1\u7684\u7d44\u5408\u6703\u5728 \u4ec0\u9ebc\u60c5\u6cc1\u4e0b\u767c\u751f\uff0c\u4f8b\u5982\u7576\u8868\u9762\u76f8\u4f3c\u5ea6\u5f88\u9ad8\u4f46\u53e5\u610f\u76f8\u4f3c\u5ea6\u537b\u5f88\u4f4e\u6642\uff0c\u53ef\u4ee5\u731c\u60f3\u53e5\u5c0d\u4e2d\u53ef\u80fd\u6709 \u4e0d\u540c\u6578\u91cf\u7684\u5426\u5b9a\u8a5e\u5b58\u5728\uff1b\u6211\u5011\u53c3\u8003 RITE-1 \u7af6\u8cfd\u4e2d\u5177\u6709\u9ad8\u6548\u80fd\u7684\u65b9\u6cd5\u4e26\u642d\u914d\u6211\u5011\u81ea\u5df1\u7684\u65b9 \u6cd5\uff0c\u5efa\u69cb\u51fa\u5224\u5225\u6587\u5b57\u860a\u6db5\u7684\u6a21\u578b\u3002 3 \u7d93\u9a57\u6cd5\u5247\u5f0f\u63a8\u8ad6\u6a21\u578b\u8207\u7279\u5fb5\u4ecb\u7d39 \u7d93\u9a57\u6cd5\u5247\u5f0f(Heuristics)\u63a8\u8ad6\u6a21\u578b\u7684\u7cfb\u7d71\u67b6 \u69cb\u8207\u904b\u884c\u6d41\u7a0b\u5982\u5716\u4e00\u6240\u793a\uff0c\u9996\u5148\u5c07\u8a9e\u6599\u8b80 \u5165\u7cfb\u7d71\u5f8c\uff0c\u900f\u904e\u6578\u5b57\u8f49\u63db\u6a21\u7d44\u5c07\u6578\u5b57\u6b63\u898f \u5316\uff0c\u63a5\u8457\u9032\u884c\u4e2d\u6587\u65b7\u8a5e\u6216\u82f1\u6587\u5206\u8a5e[7]\uff0c\u4e26 \u6a19\u8a18\u5be6\u9ad4\u540d\u8a5e[10]\u8207\u89e3\u6790\u53e5\u6cd5\u7d50\u69cb\uff0c\u6700\u5f8c \u901a\u904e\u6211\u5011\u63d0\u51fa\u7684\u8a08\u7b97\u65b9\u6cd5\u8207\u9580\u6abb\u503c\u8a2d\u5b9a\uff0c \u8a08\u7b97\u63a8\u8ad6\u95dc\u4fc2\u7684\u8a55\u5206\uff0c\u7531 0 \u81f3 1\uff0c\u4e26\u6839\u64da\u9580 \u6abb\u503c\u7372\u5f97\u6b32\u5224\u65b7\u7684\u6587\u5b57\u860a\u6db5\u95dc\u4fc2\uff0c\u800c\u8a73\u7d30 \u7684\u7279\u5fb5\u6211\u5011\u5c07\u5728 3.1 \u7bc0\u81f3 3.5 \u7bc0\u4f5c\u4ecb\u7d39\u3002 3.1 \u8a5e\u5f59\u8986\u84cb\u6bd4\u4f8b \u5728\u8a55\u4f30\u4e00\u500b\u53e5\u5b50\u7684\u610f\u7fa9\u662f\u5426\u80fd\u63a8\u8ad6\u81f3\u53e6\u4e00\u500b\u53e5\u5b50\u6642\uff0c\u6211\u5011\u8a8d\u70ba\u53e5\u5b50\u4e2d\u6bcf\u4e00\u500b\u8a5e\u5f59\u90fd\u4ee3\u8868\u4e00 \u9805\u8cc7\u8a0a\uff0c\u7576\u5169\u500b\u53e5\u5b50\u88e1\u76f8\u540c\u7684\u8a5e\u5f59\u6bd4\u4f8b\u5920\u9ad8\u6642\uff0c\u901a\u5e38\u4ee3\u8868\u9019\u5169\u500b\u53e5\u5b50\u64c1\u6709\u76f8\u540c\u7684\u8cc7\u8a0a\u91cf\uff0c \u56e0\u6b64\u5177\u6709\u63a8\u8ad6\u7684\u95dc\u4fc2\u3002 \u5982 2.2 \u6587\u737b\u63a2\u8a0e \u6211\u5011\u4ee5</td></tr><tr><td>\u6211\u5011\u4e5f\u5229\u7528\u6587\u5b57\u860a\u6db5\u7684\u6280\u8853\u61c9\u7528\u5728\u95b1\u8b80\u6e2c\u9a57\u7684\u81ea\u52d5\u7b54\u984c\u4e0a\uff0c\u5982\u679c\u53ef\u4ee5\u5224\u5225\u95b1\u8b80\u6e2c\u9a57\u7684\u9078\u9805</td></tr><tr><td>\u8207\u672c\u6587\u5177\u6709\u63a8\u8ad6\u7684\u95dc\u4fc2\uff0c\u5247\u9593\u63a5\u53ef\u4ee5\u5224\u5225\u8a72\u9078\u9805\u70ba\u7b54\u6848\u7684\u6a5f\u7387\u8f03\u5927\uff0c\u8b93\u7cfb\u7d71\u80fd\u5920\u81ea\u52d5\u7b54\u984c\u3002</td></tr></table>" | |
| }, | |
| "TABREF1": { | |
| "html": null, | |
| "num": null, | |
| "type_str": "table", | |
| "text": "Proceedings of the Twenty-Fifth Conference on Computational Linguistics and Speech Processing (ROCLING 2013) \u5be6\u9a57\u8a9e\u6599 \u6211\u5011\u7d93\u7531\u53c3\u8207 NTCIR \u7684\u7af6\u8cfd\uff0c\u53d6\u5f97 RITE \u7684\u8a13\u7df4(Dev.)\u8207\u6e2c\u8a66(Test)\u4e2d\u6587\u8a9e\u6599\u96c6\uff0c\u8a9e\u6599\u70ba\u63a8 \u8ad6\u95dc\u4fc2\u4e8c\u5143\u5206\u985e(Binary Classification)\u3002\u5716\u4e09\u70ba\u4e2d\u6587\u4e8c\u5143\u5206\u985e\u7684\u8cc7\u6599\u5167\u5bb9\uff0c\u6bcf\u7b46\u8cc7\u6599\u7686\u6709 \u4e00\u500b\u7de8\u865f\u8a18\u9304\uff0c\u4e26\u5305\u542b\u5169\u500b\u53e5\u5b50-t 1 \u8207 t 2 \uff0c\u800c label \u4ee3\u8868\u7684\u662f t 1 \u7684\u5167\u5bb9\u662f\u5426\u80fd\u63a8\u8ad6\u51fa t 2 \u4e2d\u7684 \u5047\u8a2d\uff0cY \u8868\u793a\u6210\u7acb\uff0cN \u5247\u53cd\u4e4b\u3002\u6211\u5011\u53d6\u5f97\u4e86\u548c NTCIR-10 RITE-2 \u7684\u8a13\u7df4\u8207\u6e2c\u8a66\u8a9e\u6599\uff0c\u8868\u4e00 \u70ba\u8a13\u7df4\u8207\u6e2c\u8a66\u8a9e\u6599\u96c6\u7684\u6578\u91cf\u7d71\u8a08\u3002 \u82f1\u6587\u8a9e\u6599\u6211\u5011\u5247\u63a1\u7528 Microsoft Research Paraphrase Corpus(MSR Corpus)[12]\uff0cMSR \u65bc 2004 \u5e74\u7531 Quirk \u7b49\u4eba\u63d0\u51fa\uff0c\u8a9e\u6599\u96c6\u5171\u5305\u542b 5801 \u500b\u82f1\u6587\u53e5\u5c0d\uff0c\u4e26\u4e14\u6a19\u8a18\u5169\u500b\u53e5\u5b50\u4e4b\u9593\u662f\u5426 \u76f8\u95dc\u806f\u3002 Proceedings of the Twenty-Fifth Conference on Computational Linguistics and Speech Processing (ROCLING 2013) \u5716\u516d\u3001\u6a5f\u5668\u5b78\u7fd2\u63a8\u8ad6\u7cfb\u7d71\u67b6\u69cb \u8a66\u8a9e\u6599\u900f\u904e\u7d93\u9a57\u6cd5\u5247\u5f0f\u63a8\u8ad6\u6a21\u578b\u7684\u7cfb\u7d71\u6548\u80fd\u7d9c\u5408\u6307\u6a19\u3002\u6211\u5011\u89c0\u5bdf MSR \u6e2c\u8a66\u8a9e\u6599\u7684\u5be6\u9a57\u7d50 \u679c\uff0c\u5f9e\u8868\u4e94\u53ef\u4ee5\u770b\u51fa\u5be6\u9ad4\u540d\u8a5e\u932f\u4f4d\u7684\u61f2\u7f70\u53c3\u6578 \u03b4\uff0cC15 \u81f3 C17 \u7686\u70ba\u6700\u4f4e\u7684\u61f2\u7f70\u5206\u6578 1.0\uff0c\u6240\u4ee5 \u53ef\u5f97\u77e5\u8a72\u7279\u5fb5\u5c0d\u65bc\u63a8\u8ad6\u95dc\u4fc2\u7684\u5f71\u97ff\u4e0d\u5927\uff0c\u56e0\u6b64\u4e5f\u9593\u63a5\u5c0d\u5426\u5b9a\u63a8\u8ad6\u95dc\u4fc2\u5224\u5b9a\u8f03\u5dee\u7684\u60c5\u5f62\u767c \u751f\uff0c\u4f46\u4ecd\u80fd\u9054\u5230\u4e0d\u932f\u7684\u6e96\u78ba\u7387\u3002 Proceedings of the Twenty-Fifth Conference on Computational Linguistics and Speech Processing (ROCLING 2013) Proceedings of the Twenty-Fifth Conference on Computational Linguistics and Speech Processing (ROCLING 2013) Proceedings of the Twenty-Fifth Conference on Computational Linguistics and Speech Processing (ROCLING 2013)", | |
| "content": "<table><tr><td>Proceedings of the Twenty-Fifth Conference on Computational Linguistics and Speech Processing (ROCLING 2013) Proceedings of the Twenty-Fifth Conference on Computational Linguistics and Speech Processing (ROCLING 2013) Proceedings of the Twenty-Fifth Conference on Computational Linguistics and Speech Processing (ROCLING 2013) Proceedings of the Twenty-Fifth Conference on Computational Linguistics and Speech Processing (ROCLING 2013) Proceedings of the Twenty-Fifth Conference on Computational Linguistics and Speech Processing (ROCLING 2013)</td></tr><tr><td>3.3 \u5426\u5b9a\u8a5e\u5224\u65b7 \u5373\u4f7f\u5169\u500b\u53e5\u5b50\u64c1\u6709\u9ad8\u6bd4\u4f8b\u7684\u8a5e\u5f59\u8986\u84cb\u548c\u5be6\u9ad4\u540d\u7a31\u76f8\u540c\uff0c\u4f46\u53e5\u5b50\u9593\u5e38\u56e0\u70ba\u5b58\u5728\u5426\u5b9a\u8a5e\u800c\u4f7f\u53e5 \u610f\u5927\u70ba\u6539\u8b8a\uff0c\u9032\u800c\u9020\u6210\u932f\u8aa4\u7684\u63a8\u8ad6\u5224\u65b7\uff0c\u56e0\u6b64\u6211\u5011\u589e\u52a0\u7cfb\u7d71\u5c0d\u5426\u5b9a\u8a5e\u7684\u64f7\u53d6\uff0c\u4e26\u8a2d\u8a08\u7c21\u55ae \u7684\u898f\u5247\u5224\u65b7\u5426\u5b9a\u8a5e\u5c0d\u8a08\u7b97\u63a8\u8ad6\u95dc\u4fc2\u7684\u5f71\u97ff\uff0c\u6240\u8b02\u7684\u5426\u5b9a\u8a5e\u6211\u5011\u4ee5\u5426\u5b9a\u8a5e\u8fad\u5178\u4f5c\u70ba\u4f9d\u64da\uff0c\u4f8b \u5982\u8fad\u5178\u4e2d\uff1a\u300c\u7121\u300d\u3001\u300c\u672a\u300d\u3001\u300c\u4e0d\u300d\u3001\u300c\u6c92\u6709\u300d\u2026\u8996\u70ba\u5426\u5b9a\u8a5e\uff0c\u4e26\u85c9\u7531\u53e5\u5b50\u4e2d\u7684\u5426\u5b9a\u8a5e\u96c6\uff0c \u9069\u7576\u5730\u8abf\u6574\u63a8\u8ad6\u95dc\u4fc2\u7684\u8a55\u5206\u3002 \u6211\u5011\u8a8d\u70ba\u5169\u500b\u53e5\u5b50\u82e5\u5305\u542b\u4e0d\u540c\u6578\u91cf\u7684\u5426\u5b9a\u8a5e\u6642\uff0c\u8f03\u5bb9\u6613\u6709\u4e0d\u540c\u7684\u610f\u7fa9\u7522\u751f\uff0c\u800c\u964d\u4f4e\u63a8 \u8ad6\u95dc\u4fc2\u7684\u53ef\u80fd\u6027\uff0c\u56e0\u6b64\u518d\u5ea6\u52a0\u5165\u4e00\u500b\u51fd\u5f0f\u91dd\u5c0d\u5426\u5b9a\u8a5e\u505a\u63a8\u8ad6\u5206\u6578\u7684\u8abf\u6574\uff0c\u5982\u4e0b\u65b9\u516c\u5f0f(5)\u6240 \u793a\u3002Negation \u8868\u793a\u53e5\u5b50\u7576\u4e2d\u5305\u542b\u7684\u5426\u5b9a\u8a5e\u96c6\u5408\uff0c\u03b2 \u70ba\u5426\u5b9a\u8a5e\u6578\u91cf\u4e0d\u76f8\u7b49\u6642\u7528\u4ee5\u8abf\u6574\u7684\u61f2\u7f70 \u5206\u6578\uff0c\u5176\u503c\u4ecb\u65bc 0 \u81f3 1\uff0c\u4e26\u5c07\u63a8\u8ad6\u95dc\u4fc2\u7684\u5224\u65b7\u5ef6\u4f38\u6210\u516c\u5f0f(6)\u3002 \u03b2 (5) (6) 3.4 \u53cd\u7fa9\u8a5e\u5224\u65b7 \u9664\u4e86\u5426\u5b9a\u8a5e\u5916\uff0c\u53e5\u5b50\u4e4b\u9593\u82e5\u5b58\u5728\u53cd\u7fa9\u8a5e[12]\uff0c\u6211\u5011\u8a8d\u70ba\u9019\u6a23\u662f\u66f4\u52a0\u986f\u793a\u5169\u500b\u53e5\u5b50\u4e4b\u9593\u53ef\u80fd \u4e0d\u5177\u6709\u63a8\u8ad6\u7684\u95dc\u4fc2\uff0c\u56e0\u6b64\u6211\u5011\u5617\u8a66\u5206\u6790\u53e5\u5b50\u4e4b\u9593\u7684\u53cd\u7fa9\u8a5e\u5305\u542b\u72c0\u6cc1\uff0c\u82e5\u5305\u542b\u53cd\u7fa9\u8a5e\uff0c\u5247\u7d66 \u4e88\u8f03\u91cd\u7684\u61f2\u7f70\u5206\u6578\uff0c\u5927\u5e45\u8abf\u6574\u63a8\u8ad6\u95dc\u4fc2\u7684\u5224\u65b7\u3002\u516c\u5f0f(7)\u986f\u793a\u53cd\u7fa9\u8a5e\u5224\u65b7\u7684\u51fd\u5f0f\uff0cAntonym \u8868\u793a\u4e00\u500b\u8a5e\u5f59\u7684\u53cd\u7fa9\u8a5e\u96c6\u5408\uff0c\u03b3 \u5247\u662f\u53cd\u7fa9\u8a5e\u5b58\u5728\u6642\u7684\u61f2\u7f70\u5206\u6578\uff0c\u5176\u503c\u70ba 1 \u81f3 2\uff0c\u800c\u5224\u65b7\u63a8\u8ad6 \u95dc\u4fc2\u7684\u516c\u5f0f\u5247\u8b8a\u6210\u516c\u5f0f(8)\u3002 (7) (8) 3.5 \u5be6\u9ad4\u540d\u8a5e\u932f\u4f4d \u4e3b\u8a5e\u8207\u53d7\u8a5e\u4f4d\u7f6e\u53ef\u80fd\u5f71\u97ff\u53e5\u5b50\u7684\u8a9e\u610f\uff0c\u56e0\u6b64\u6211\u5011\u5728\u524d\u8655\u7406\u4fbf\u6a19\u8a18\u51fa\u5be6\u9ad4\u540d\u8a5e\u7684\u7d22\u5f15\uff0c\u4e26\u4e14 \u6211\u5011\u8a8d\u70ba\u7576\u63a8\u8ad6\u5206\u6578\u8f03\u9ad8\u6642\uff0c\u4ee3\u8868\u53e5\u5b50\u4e4b\u9593\u7684\u8a5e\u5f59\u4f7f\u7528\u975e\u5e38\u76f8\u8fd1\uff0c\u6b64\u6642\u82e5\u5be6\u9ad4\u540d\u8a5e\u767c\u751f\u932f \u4f4d\uff0c\u5247\u8f03\u5bb9\u6613\u5f71\u97ff\u5169\u500b\u53e5\u5b50\u8a9e\u610f\u7684\u76f8\u4f3c\u7a0b\u5ea6\uff0c\u5982\u5716\u4e8c\uff0c\u56e0\u6b64\u589e\u52a0\u4e00\u500b\u51fd\u5f0f\u5224\u65b7\u7d22\u5f15\u503c\u7684\u8fe5 \u7570\uff0c\u85c9\u4ee5\u8abf\u6574\u63a8\u8ad6\u95dc\u4fc2\u7684\u8a55\u5206\uff0c\u5982\u516c\u5f0f(9)\u3002\u516c\u5f0f\u4e2d i \u4ee3\u8868\u5be6\u9ad4\u540d\u8a5e\u65bc\u53e5\u5b50\u4e2d\u7684\u4f4d\u7f6e\uff0cm \u548c n \u70ba NE_Order \u7684\u7d22\u5f15\u503c\uff0c\u03b4 \u70ba\u7bc4\u570d 1 \u5230 2 \u7684\u61f2\u7f70\u5206\u6578\uff0c\u03bb \u70ba\u4f7f\u7528\u8a72\u51fd\u5f0f\u7684\u63a8\u8ad6\u5206\u6578\u9580\u6abb\u503c\u3002 \u900f\u904e\u4e0a\u8ff0\u7684\u5404\u7a2e\u8a9e\u8a00\u8cc7\u8a0a\u7684\u4f7f\u7528\uff0c\u6700\u5f8c\u5408\u4f75\u6210\u4e00\u9805\u63a8\u8ad6\u95dc\u4fc2\u7684\u8a08\u7b97\u516c\u5f0f(11)\uff0c\u5c07\u63a8\u8ad6\u95dc\u4fc2 \u7684\u7a0b\u5ea6\u4ee5 0 \u81f3 1 \u7684\u5206\u6578\u986f\u793a\u9ad8\u4f4e\uff0c\u6211\u5011\u9810\u671f\u8a72\u65b9\u6cd5\u80fd\u6709\u6548\u5730\u5224\u5b9a\u8a9e\u53e5\u9593\u7684\u63a8\u8ad6\u95dc\u4fc2\u3002 t 1 \uff1a\u53f0\u7063\u51fa\u53e3\u81f3\u5370\u5ea6\u6210\u9577 28.6% t 2 \uff1a\u5370\u5ea6\u5f9e\u53f0\u7063\u51fa\u53e3\u6210\u9577\u7387\u53ef\u9054 28.6% \u53f0\u7063\uff1a \u5370\u5ea6\uff1a \u53f0\u7063\uff1a \u5370\u5ea6\uff1a \u5716\u4e8c\u3001\u5be6\u9ad4\u540d\u8a5e\u4f4d\u7f6e\u6bd4\u5c0d\u7bc4\u4f8b (9) (10) (11) 4 \u7d93\u9a57\u6cd5\u5247\u5f0f\u63a8\u8ad6\u6a21\u578b\u5be6\u6e2c 4.1 \u8868\u4e00\u3001\u4e2d\u6587\u8a13\u7df4\u8a9e\u6599\u96c6\u7d71\u8a08 \u4f86\u6e90 NTCIR-10 RITE-2 MSR \u8a9e\u8a00 \u7e41\u9ad4\u4e2d\u6587 \u82f1\u6587 \u985e\u5225 Dev. Test Dev. Test Y 716 479 2753 1147 N 605 402 1323 578 \u7e3d\u548c 1321 881 4076 1725 4.2 \u63a8\u8ad6\u6a21\u578b\u9580\u6abb\u503c\u8207\u7279\u5fb5\u53c3\u6578\u9078\u5b9a \u70ba\u4e86\u6700\u4f73\u5316\u63a8\u8ad6\u7cfb\u7d71\u7684\u6548\u679c\uff0c\u6211\u5011\u900f\u904e RITE-2\u3001MSR \u53ca RTE \u4e09\u7a2e\u4e0d\u540c\u7684\u8a13\u7df4\u8a9e\u6599\u5f9e\u5be6\u9a57 \u88e1\u8a2d\u5b9a\u6240\u6709\u53c3\u6578\u7d44\u5408\u85c9\u7531\u6548\u80fd\u7684\u8b8a\u5316\u4ee5\u4eba\u5de5\u7684\u65b9\u5f0f\u8a2d\u5b9a\u53c3\u6578\uff0c\u8abf\u6574\u4e2d\u82f1\u6587\u63a8\u8ad6\u6a21\u578b\u7684\u5404\u9805 \u53c3\u6578\u8207\u9580\u6abb\u503c\u4ee5\u5c0b\u6c42\u6e96\u78ba\u7387\u7684\u6975\u5927\u503c\uff0c\u85c9\u4ee5\u5206\u6790\u53c3\u6578\u7d44\u5408\u5c0d\u65bc\u55ae\u9805\u63a8\u8ad6\u7684\u6548\u679c\uff0c\u6240\u8b02\u7684\u55ae \u9805\u63a8\u8ad6\u5373\u662f\u5728\u5224\u65b7\u6587\u5b57\u860a\u6db5\u95dc\u4fc2\u6642\uff0c\u50c5\u5224\u65b7\u5177\u6709\u860a\u6db5\u95dc\u4fc2\u6216\u4e0d\u5177\u6709\u860a\u6db5\u95dc\u4fc2\u5169\u7a2e\uff1b\u6700\u5f8c\u6211 \u5011\u4ee5\u6e96\u78ba\u7387\u8f03\u4f73\u7684\u53c3\u6578\u8a2d\u5b9a\u91dd\u5c0d\u6e2c\u8a66\u8a9e\u6599\u9032\u884c\u63a8\u8ad6\u7cfb\u7d71\u7684\u8a55\u4f30\uff0c\u4e0d\u904e\u7919\u65bc\u7248\u9762\u9650\u5236\uff0c\u672c\u7bc7 \u8ad6\u6587\u4e2d\u6587\u8a9e\u6599\u53ea\u7bc0\u9304 RITE-2 \u7e41\u9ad4\u8a9e\u6599\u4f5c\u70ba\u4ee3\u8868\uff0c\u800c\u82f1\u6587\u8a9e\u6599\u5247\u4ee5 MSR \u4f5c\u70ba\u4ee3\u8868\uff0c\u5176\u5b83\u8a73 \u7d30\u7684\u5be6\u9a57\u7d50\u679c\u53ef\u53c3\u7167\u9ec3\u744b\u6770\u78a9\u58eb\u8ad6\u6587[13]\u3002 \u8868\u56db\u5217\u51fa\u7e41\u9ad4\u4e2d\u6587\u8a13\u7df4\u8a9e\u6599\u7684\u53c3\u6578\u641c\u5c0b\u7d50\u679c\uff0c\u7531\u65bc\u641c\u5c0b\u7684\u7d50\u679c\u904e\u591a\uff0c\u56e0\u6b64\u5728\u9019\u88e1\u50c5\u5217 \u51fa\u8f03\u4f73\u7684\u5e7e\u7d44\u53c3\u6578\u8a2d\u5b9a\u8207\u8a13\u7df4\u8a9e\u6599\u7684\u6e96\u78ba\u7387\uff0c\u5176\u4e2d\u7de8\u865f E \u4ee3\u8868\u63a8\u8ad6\u6210\u7acb\u7684\u9580\u6abb\u503c\u3002\u800c\u8868\u4e94 \u5247\u5217\u51fa\u82f1\u6587\u8a13\u7df4\u8a9e\u6599-MSR \u7684\u53c3\u6578\u641c\u5c0b\u7d50\u679c\uff0c\u540c\u6a23\u5730\u50c5\u5217\u51fa\u8f03\u4f73\u7684\u5e7e\u7d44\u8a2d\u5b9a\u8207\u6e96\u78ba\u7387\uff0c\u6211 \u5011\u5c07\u6e96\u78ba\u7387(Acc)\u8207 Macro-F1 \u5b9a\u7fa9\u5982\u4e0b\u516c\u5f0f\u3002 \u6e96\u78ba\u7387 \u63a8\u8ad6\u7d50\u679c\u6b63\u78ba\u500b\u6578 \u8a9e\u6599\u500b\u6578 \u7cbe\u78ba\u7387 \u63a8\u8ad6\u7d50\u679c\u55ae\u9805\u6b63\u78ba\u500b\u6578 \u63a8\u8ad6\u7d50\u679c\u55ae\u9805\u500b\u6578 \u5716\u4e09\u3001\u4e8c\u5143\u5206\u985e\u8cc7\u6599\u96c6 \u63a8\u8ad6\u7d50\u679c\u55ae\u9805\u6b63\u78ba\u500b\u6578 5.2 POSes \u52d5\u8a5e\u6a19\u8a18 \u8868\u4e03\u3001\u82f1\u6587\u7279\u5fb5\u96c6\u7de8\u865f\u8868 \u985e\u6f14\u7b97\u6cd5\u9032\u884c\u4e2d\u82f1\u6587\u6e2c\u8a66\u8a9e\u6599\u7684\u6548\u80fd\u8a55 6.3 \u5be6\u9a57\u8a2d\u8a08\u3001\u6f14\u7b97\u6cd5\u8207\u53c3\u6578\u7684\u9078\u5b9a\u548c\u7d50\u679c 7.2 \u5f9e\u77ed\u6587\u7be9\u9078\u76f8\u95dc\u53e5 \u7d93\u9a57\u6cd5\u5247\u5f0f\u63a8\u8ad6\u6a21\u578b\u7684\u53c3\u6578\u8a2d\u5b9a\uff0c\u4e2d\u6587\u7684\u90e8\u4efd\u6211\u5011\u63a1\u7528 NTCIR-10 RITE-2 \u7af6\u8cfd\u6642\u7684\u6700 \u63a5\u8457\u89c0\u5bdf\u5716\u5341\u4e09\uff0c\u82f1\u6587\u8a9e\u6599\u63a1\u7528\u77ed\u6587\u904e\u6ffe\u7684\u65b9\u6cd5\u4f86\u9032\u884c\u5be6\u9a57\u6bd4\u8f03\uff0c\u5982\u540c\u4e2d\u6587\u95b1\u8b80\u6e2c\u9a57 \u53ec\u56de\u7387 \u53c3\u8003\u7b54\u6848\u4e2d\u7684\u55ae\u9805\u500b\u6578 Macro-F1 \u8868\u56db\u3001RITE-2 \u7e41\u9ad4\u4e2d\u6587\u8a13\u7df4\u8a9e\u6599\u53c3\u6578\u8a2d\u5b9a \u7de8\u865f E \u03b1 \u03b2 \u03b3 \u03bb \u03b4 Acc C1 0.54 0.1 0.27 1.8 0.85 1.9 73.05% C2 0.56 0.08 0.25 1.0 0.85 1.8 73.13% C3 0.56 0.08 0.25 1.7 0.85 1.8 73.20% \u8868\u4e94\u3001MSR \u8a13\u7df4\u8a9e\u6599\u53c3\u6578\u8a2d\u5b9a \u7de8\u865f E \u03b1 \u03b2 \u03b3 \u03bb \u03b4 Acc C13 0.47 0.05 0.13 1.3 0.55 1.2 71.07% C14 0.47 0.05 0.17 1.3 0.55 1.0 71.12% C15 0.49 0.05 0.14 1.2 0.55 1.0 71.15% C16 0.49 0.05 0.17 1.2 0.55 1.0 71.17% C17 0.49 0.05 0.20 1.2 0.55 1.0 71.20% 4.3 \u5be6\u6e2c\u7d50\u679c \u6839\u64da\u4e0a\u8ff0\u9019\u4e9b\u8a13\u7df4\u8a9e\u6599\u7684\u53c3\u6578\u8abf\u6574\uff0c\u9032\u884c\u6e2c\u8a66\u8a9e\u6599\u7684\u5be6\u9a57\uff0c\u5206\u6790\u7d93\u9a57\u6cd5\u5247\u5f0f\u63a8\u8ad6\u6a21\u578b\u7d93\u7531 \u53c3\u6578\u8abf\u6821\u5f8c\u7684\u6548\u80fd\u8207\u55ae\u9805\u63a8\u8ad6\u80fd\u529b\u3002 \u6211\u5011\u4f7f\u7528\u8868\u56db\u7684\u53c3\u6578\u9032\u884c RITE-2 \u7e41\u9ad4\u4e2d\u6587\u6e2c\u8a66\u8a9e\u6599\u7684\u63a8\u8ad6\u95dc\u4fc2\u9810\u6e2c\uff0c\u4e26\u4e14\u52a0\u5165\u8fd1\u7fa9 \u8a5e\u7684\u5224\u5b9a\uff0c\u89c0\u5bdf\u662f\u5426\u80fd\u63d0\u5347\u63a8\u8ad6\u6548\u679c\uff0c\u6700\u5f8c\u91dd\u5c0d\u9810\u6e2c\u7684\u7d50\u679c\u9032\u884c\u5206\u6790\uff0c\u8a08\u7b97\u55ae\u9805\u7b54\u6848\u7684\u6e96 \u78ba\u7387\u8207\u53ec\u56de\u7387\u3002\u5716\u56db\u5247\u70ba RITE-2 \u7e41\u9ad4\u4e2d\u6587\u6e2c\u8a66\u8a9e\u6599\u4f7f\u7528\u8fd1\u7fa9\u8a5e\u7684\u6548\u80fd\u6bd4\u8f03\uff0c\u5f9e\u5716\u4e2d\u7684\u7d50 \u679c\u986f\u793a\u8fd1\u7fa9\u8a5e\u5728 RITE-2 \u7684\u6e2c\u8a66\u8a9e\u6599\u4e2d\u80fd\u63d0\u5347\u4e0d\u5c11\u7cfb\u7d71\u6548\u80fd\uff0c\u800c\u6211\u5011\u4e5f\u6709\u5c0d RITE-1 \u6e2c\u8a66\u8a9e \u6599\u9032\u884c\u5be6\u9a57\u5176\u7d50\u679c\u5247\u662f\u7565\u5fae\u7684\u4e0b\u964d\uff0c\u7919\u65bc\u7248\u9762\u6240\u4ee5\u7701\u7565\u5176\u7d50\u679c\uff0c\u56e0\u6b64\u6211\u5011\u8a8d\u70ba\u8fd1\u7fa9\u8a5e\u5728\u63a8 \u8ad6\u95dc\u4fc2\u7684\u5224\u65b7\u662f\u5426\u5177\u6709\u5e6b\u52a9\uff0c\u56e0\u8a9e\u6599\u7279\u6027\u7684\u4e0d\u540c\u800c\u6709\u6240\u5dee\u7570\u3002 \u5716\u56db\u3001\u7d93\u9a57\u6cd5\u5247\u5f0f\u63a8\u8ad6\u6a21\u578b\u8fd1\u7fa9\u8a5e\u6548\u80fd\u6bd4\u8f03\uff1aRITE-2 \u7e41\u9ad4\u4e2d\u6587\u8a9e\u6599 \u6700\u5f8c\u900f\u904e\u76f8\u540c\u7684\u63a8\u8ad6\u6a21\u578b\uff0c\u4f7f\u7528 MSR \u82f1\u6587\u8a13\u7df4\u8a9e\u6599\u7684\u53c3\u6578\u8a2d\u5b9a\u5c0d\u8a9e\u6599\u9810\u6e2c\u63a8\u8ad6\u7d50 \u679c\uff0c\u85c9\u4ee5\u77ad\u89e3\u76f8\u540c\u7684\u8a9e\u8a00\u6a21\u578b\u662f\u5426\u53ef\u4ee5\u5957\u7528\u5728\u4e0d\u540c\u7684\u8a9e\u6599\u4e2d\u7684\u63a8\u8ad6\u95dc\u4fc2\u5224\u65b7\uff0c\u5716\u4e94\u986f\u793a\u6e2c C1 C2 C3 \u7121\u8fd1\u7fa9\u8a5e Macro-F1 65.79% 65.73% 65.55% \u8fd1\u7fa9\u8a5e Macro-F1 66.79% 67.46% 67.12% \u7121\u8fd1\u7fa9\u8a5e Accuracy 66.29% 66.29% 65.95% \u8fd1\u7fa9\u8a5e Accuracy 67.76% 68.56% 67.99% 58.00% 63.00% 68.00% 73.00% \u5716\u4e94\u3001 \u7d93\u9a57\u6cd5\u5247\u5f0f\u63a8\u8ad6\u6a21\u578b\u7cfb\u7d71\u6548\u80fd\uff1aMSR \u6e2c\u8a66\u8a9e\u6599 \u7d93\u7531\u591a\u7d44\u4e2d\u6587\u8207\u82f1\u6587\u8a9e\u6599\u5be6\u9a57\uff0c\u53ef\u4ee5\u767c\u73fe\u6211\u5011\u63d0\u51fa\u7684\u51fd\u5f0f\u7d44\u6210\u7d93\u9a57\u6cd5\u5247\u5f0f\u63a8\u8ad6\u7cfb\u7d71\u8207 NTCIR-9\u3001NTCIR-10 \u7af6\u8cfd\u6210\u7e3e\u76f8\u6bd4\uff0c\u5728\u4e2d\u6587\u8a9e\u6599\u4e2d\u4ecd\u5c6c\u65bc\u4e0d\u932f\u7684\u6548\u679c\u3002\u82f1\u6587\u7684\u5be6\u9a57\u7d50\u679c \u5247\u4ecd\u6709\u9032\u6b65\u7a7a\u9593\uff0c\u5169\u7a2e\u63a8\u8ad6\u80fd\u529b\u90fd\u9700\u8981\u5c31\u73fe\u6709\u7684\u51fd\u5f0f\u9032\u884c\u6539\u5584\uff0c\u4ee5\u63d0\u5347\u82f1\u6587\u8a9e\u53e5\u7684\u63a8\u8ad6\u6548 \u679c\u3002\u5f9e\u9019\u4e9b\u5be6\u9a57\u53ef\u4ee5\u5f97\u77e5\u672a\u4f86\u6211\u5011\u9700\u8981\u767c\u5c55\u66f4\u591a\u51fd\u5f0f\u4f86\u5224\u5b9a\u5426\u5b9a\u7684\u63a8\u8ad6\u95dc\u4fc2\uff0c\u5c24\u5176\u662f\u91dd\u5c0d \u8a9e\u53e5\u9593\u7684\u53cd\u7fa9\u3001\u7368\u7acb\u8207\u77db\u76fe\u7b49\u73fe\u8c61\u9700\u8981\u8655\u7406\u3002 5 \u6a5f\u5668\u5b78\u7fd2\u65b9\u6cd5 \u6a5f\u5668\u5b78\u7fd2\u6f14\u7b97\u6cd5\u5efa\u69cb\u7684\u63a8\u8ad6\u6a21\u578b\u7cfb\u7d71\u67b6\u69cb\u5982\u5716 \u516d\u6240\u793a\uff0c\u540c\u6a23\u4f7f\u7528\u4e0a\u4e00\u7bc0\u7684\u5143\u4ef6\u9032\u884c\u524d\u8655\u7406\uff0c \u63a5\u8457\u64f7\u53d6\u6211\u5011\u8a8d\u70ba\u53ef\u4ee5\u589e\u52a0\u63a8\u8ad6\u6548\u679c\u7684\u8a9e\u6587\u8cc7 \u8a0a\uff0c\u505a\u70ba\u8a13\u7df4\u6a21\u578b\u7684\u7279\u5fb5\u96c6\u5408\uff1b\u6700\u5f8c\u6211\u5011\u63a1\u7528 \u4e09\u7a2e\u4e0d\u540c\u7684\u5206\u985e\u6f14\u7b97\u6cd5\u8a13\u7df4\u5206\u985e\u6a21\u578b\uff0c\u5206\u5225\u662f \u652f\u6301\u5411\u91cf\u6a5f(Support Vector Machines, SVMs)\u3001 Weka J48 \u6c7a\u7b56\u6a39(J48 Decision Trees)\u8207 Weka \u7dda \u6027\u56de\u6b78(Linear Regression)[13]\uff0c\u900f\u904e\u4e0d\u540c\u985e\u578b \u7684\u5206\u985e\u5668\u7372\u5f97\u63a8\u8ad6\u95dc\u4fc2\u7684\u7d50\u679c\u3002 \u524d\u4e00\u5c0f\u7bc0\u8aaa\u660e\u4e86\u7d93\u9a57\u6cd5\u5247\u5f0f\u63a8\u8ad6\u6a21\u578b\u6240\u4f7f\u7528\u7684\u51fd\u5f0f\uff0c\u6211\u5011\u91dd\u5c0d\u9019\u4e9b\u51fd\u5f0f\u9032\u884c\u6578\u503c\u5316\u7684 \u8f49\u63db\uff0c\u505a\u70ba\u8a13\u7df4\u63a8\u8ad6\u6a21\u578b\u7684\u7279\u5fb5\uff1b\u9019\u4e9b\u7279\u5fb5\u5305\u542b\u8a5e\u5f59\u8986\u84cb\u6bd4\u4f8b\u3001\u5be6\u9ad4\u540d\u8a5e\u6578\u91cf\u3001\u5be6\u9ad4\u540d\u8a5e \u76f8\u4f3c\u5ea6\u3001\u5be6\u9ad4\u540d\u8a5e\u932f\u4f4d\u6578\u91cf\u3001\u53e5\u5b50\u9577\u5ea6\u3001\u5426\u5b9a\u8a5e\u6578\u91cf\u3001\u8fd1\u7fa9\u8a5e\u6578\u91cf\u3001\u53cd\u7fa9\u8a5e\u6578\u91cf\u7b49\u9805\u76ee\u3002 \u9664\u6b64\u4e4b\u5916\uff0c\u6211\u5011\u5e0c\u671b\u52a0\u6df1\u63a8\u8ad6\u6a21\u578b\u5c0d\u8a9e\u6cd5\u7d50\u69cb\u7684\u8a8d\u8b58\uff0c\u56e0\u6b64\u52a0\u5165\u5256\u6790\u6a39\u5206\u6790\u3001POSes \u52d5\u8a5e \u6a19\u8a18\u8207\u8a5e\u5f59\u4f9d\u8cf4\u95dc\u4fc2\u7b49\u5143\u7d20\uff0c\u8a08\u7b97\u5176\u76f8\u4f3c\u5ea6\u505a\u70ba\u7279\u5fb5\uff0c\u5e0c\u671b\u63d0\u9ad8\u63a8\u8ad6\u6a21\u578b\u7684\u80fd\u529b\u3002 5.1 \u5256\u6790\u6a39\u5206\u6790 \u6211\u5011\u900f\u904e\u53f2\u4e39\u4f5b\u5256\u6790\u5668(Stanford Parser)[9]\u53d6\u5f97\u53e5\u5b50\u7684\u5256\u6790\u6a39\uff0c\u4e26\u4e14\u6211\u5011\u8a8d\u70ba\u4f7f\u7528\u6574\u500b\u5256 \u6790\u6a39\u5206\u6790\u53e5\u6cd5\u7d50\u69cb\u76f8\u4f3c\u5ea6\u5bb9\u6613\u589e\u52a0\u8a08\u7b97\u7684\u96e3\u5ea6\uff0c\u56e0\u70ba\u53e5\u5b50\u4e4b\u9593\u53ef\u80fd\u50c5\u6709\u90e8\u5206\u7684\u7d50\u69cb\u5177\u6709\u5171 \u901a\u6027\u5373\u53ef\u5177\u5099\u63a8\u8ad6\u7684\u95dc\u4fc2\uff0c\u56e0\u6b64\u4ee5\u6bcf\u4e00\u500b\u7bc0\u9ede\u505a\u70ba\u6839\u7bc0\u9ede(ROOT)\u64f7\u53d6\u5176\u4e0b\u5c64\u7bc0\u9ede\u5f62\u6210\u7684 \u5b50\u6a39\uff0c\u4f7f\u7528\u9019\u4e9b\u5b50\u6a39\u4f86\u8a08\u7b97\u5169\u500b\u53e5\u5b50\u7d50\u69cb\u7684\u76f8\u4f3c\u7a0b\u5ea6\u3002 C15 C16 C17 Macro-F1 64.82% 64.67% 64.67% Accuracy 72.46% 72.23% 72.12% 50.00% 55.00% 60.00% 65.00% 70.00% POSes \u6a19\u8a18\u7531\u53f2\u4e39\u4f5b\u5256\u6790\u5668\u7372\u5f97\uff0c\u6211\u5011\u8a8d\u70ba\u52d5\u8a5e\u5728\u53e5\u5b50\u4e2d\u626e\u6f14\u8f03\u91cd\u8981\u7684\u89d2\u8272\uff0c\u56e0\u5176\u6307\u51fa\u6574 \u500b\u53e5\u5b50\u7684\u4e8b\u4ef6\u8207\u52d5\u4f5c\u610f\u5716\uff0c\u56e0\u6b64\u7279\u610f\u5c07\u88ab\u6a19\u8a3b\u6210\u52d5\u8a5e\u7684\u8a5e\u5f59\u6293\u53d6\u51fa\u4f86\uff0c\u4ee5\u5169\u500b\u53e5\u5b50\u500b\u5225\u7684 \u52d5\u8a5e\u6578\u91cf\u8207\u76f8\u4f3c\u5ea6\u505a\u70ba\u7279\u5fb5[15]\uff0c\u4e26\u671f\u671b\u8b93\u5206\u985e\u5668\u5b78\u7fd2\u52d5\u8a5e\u4f7f\u7528\u5728\u63a8\u8ad6\u95dc\u4fc2\u4e0a\u7684\u5f71\u97ff\u529b\u3002 5.3 \u8a5e\u5f59\u4f9d\u8cf4\u95dc\u4fc2 \u53f2\u4e39\u4f5b\u5256\u6790\u5668\u4ea6\u80fd\u6839\u64da\u5256\u6790\u6a39\u7684\u751f\u6210\uff0c\u7522\u751f\u8a5e\u5f59\u4e4b\u9593\u4f9d\u8cf4\u7684\u95dc\u4fc2(Stanford Dependencies)\uff0c \u6211\u5011\u5c07\u4f9d\u8cf4\u95dc\u4fc2\u4e2d\u7684\u8a5e\u5f59\u505a\u70ba\u7bc0\u9ede\uff0c\u5c07\u53e5\u5b50\u4e2d\u7684\u8a5e\u5f59\u95dc\u4fc2\u8996\u70ba\u4e00\u500b\u6709\u5411\u5716(Directed Graph)\uff0c\u4e26\u5316\u505a\u77e9\u9663\u5f62\u5f0f\u5982\u5716\u4e03\u3002 \u6211\u5011\u767c\u73fe\u4e00\u500b\u77e9\u9663\u5167\u53ef\u4ee5\u986f\u793a\u7684\u8cc7\u8a0a\u4e26\u4e0d\u5145\u6c9b\uff0c\u5982\u6b64\u7a00\u758f\u7684\u77e9\u9663\u4e2d\uff0c\u6211\u5011\u96e3\u4ee5\u627e\u5230\u53e5 \u5b50\u4e4b\u9593\u5305\u542b\u76f8\u540c\u95dc\u4fc2\u7684\u8a5e\u5f59\u7d44\u5408\uff0c\u56e0\u6b64\u4ee5\u76f8\u9130\u77e9\u9663(Adjacency Matrix)\u7684\u6982\u5ff5\u505a\u9032\u4e00\u6b65\u7684\u904b \u7b97\uff1b\u4f8b\u5982\u4e00\u500b\u77e9\u9663 M\uff0c\u53ef\u4ee5\u7d93\u7531\u77e9\u9663\u76f8\u4e58\u7372\u5f97\u7bc0\u9ede\u5230\u7bc0\u9ede\u4e4b\u9593\u79fb\u52d5\u6240\u9700\u8981\u7684\u6b65\u6578\uff0c\u56e0\u6b64 \u8a08\u7b97 M 3 \u4fbf\u80fd\u77ad\u89e3\u4efb\u4e00\u500b\u7bc0\u9ede\u904e\u7a0b\u7d93\u7531\u5169\u500b\u7bc0\u9ede\uff0c\u6240\u8207\u5176\u4ed6\u7bc0\u9ede\u7684\u9593\u63a5\u4f9d\u8cf4\u95dc\u4fc2\u3002\u6211\u5011\u5c07 \u9019\u6a23\u7684\u79fb\u52d5\u8996\u70ba\u4f9d\u8cf4\u95dc\u4fc2\u7684\u5ef6\u4f38\uff0c\u5982\u6b64\u80fd\u627e\u51fa\u66f4\u591a\u6f5b\u5728\u7684\u8a5e\u5f59\u4f9d\u8cf4\u95dc\u4fc2\uff0c\u4e26\u4e14\u5c07\u4e0d\u540c\u79fb\u52d5 \u6b65\u6578\u7684\u77e9\u9663\u7d50\u679c\u806f\u96c6\uff0c\u7372\u5f97\u66f4\u8c50\u5bcc\u7684\u4f9d\u8cf4\u95dc\u4fc2\u3002\u5716\u516b\u4fbf\u662f\u5716\u4e03\u7684\u77e9\u9663\u8a08\u7b97\u4efb\u4e00\u500b\u7bc0\u9ede\u7d93\u7531 \u56db\u500b\u4ee5\u5167\u7684\u7bc0\u9ede\u6240\u5f62\u6210\u7684\u76f4\u63a5\u6216\u9593\u63a5\u4f9d\u8cf4\u95dc\u4fc2\u8868\uff0c\u6211\u5011\u900f\u904e\u9019\u6a23\u7684\u77e9\u9663\uff0c\u5206\u6790\u53e5\u5b50\u4e4b\u9593\u8a5e \u5f59\u4f9d\u8cf4\u95dc\u4fc2\u76f8\u4f3c\u7684\u7a0b\u5ea6\uff0c\u4e26\u4ee5\u8a72\u6578\u503c\u505a\u70ba\u4e00\u9805\u7279\u5fb5\u3002 6 \u6a5f\u5668\u5b78\u7fd2\u65b9\u6cd5\u5be6\u6e2c 6.1 \u5be6\u9a57\u8a9e\u6599\u8207\u8a2d\u8a08 \u6211\u5011\u4f9d\u7167\u7d93\u9a57\u6cd5\u5247\u5f0f\u63a8\u8ad6\u6a21\u578b\u6240\u4f7f\u7528\u7684\u8a9e\u8a00\u8cc7 \u8a0a\u62bd\u53d6\u7279\u5fb5\uff0c\u4e26\u63d0\u51fa\u5982\u5256\u6790\u6a39\u7d50\u69cb\u53ca\u8a5e\u5f59\u4f9d\u8cf4 \u95dc\u4fc2\u7b49\u8a9e\u6cd5\u7d50\u69cb\u7279\u5fb5\uff0c\u5e0c\u671b\u589e\u52a0\u63a8\u8ad6\u95dc\u4fc2\u7684\u5206 \u985e\u80fd\u529b\u3002\u63a5\u8457\u4ee5 SVM\u3001J48 \u548c\u7dda\u6027\u56de\u6b78\u7b49\u6f14\u7b97 \u6cd5\u8a13\u7df4\u5206\u985e\u6a21\u578b\uff0c\u4e26\u4ee5\u8caa\u5a6a\u5f0f\u641c\u5c0b\u5404\u500b\u8a9e\u6599\u7684 \u7279\u5fb5\u7d44\u5408\u8207\u5176\u5206\u985e\u6548\u679c\uff0c\u6700\u5f8c\u7d93\u7531\u6311\u9078\u51fa\u4f86\u7684 \u7279\u5fb5\u7d44\u5408\u9032\u884c\u5206\u985e\u6f14\u7b97\u6cd5\u8a55\u6bd4\uff0c\u518d\u4ee5\u6307\u5b9a\u7684\u5206 \u539f\u53e5\uff1a1997 \u5e74\u9999\u6e2f\u56de\u6b78\u4e2d\u570b 1997 \u5e74 \u9999 \u6e2f \u56de \u6b78 \u4e2d \u570b ROOT 1997 0 0 0 1 0 0 \u5e74 0 0 0 1 0 0 \u9999\u6e2f 0 0 0 1 0 0 \u56de\u6b78 0 0 0 0 0 1 \u4e2d\u570b 0 0 0 1 0 0 ROOT 0 0 0 0 0 0 \u5716\u4e03\u3001\u8a5e\u5f59\u4f9d\u8cf4\u95dc\u4fc2\u77e9\u9663 M \u539f\u53e5\uff1a1997 \u5e74\u9999\u6e2f\u56de\u6b78\u4e2d\u570b 1997 \u5e74 \u9999 \u6e2f \u56de \u6b78 \u4e2d \u570b ROOT 1997 0 0 0 1 0 1 \u5e74 0 0 0 1 0 1 \u9999\u6e2f 0 0 0 1 0 1 \u56de\u6b78 0 0 0 0 0 1 E1 E2 E3 E4 E5 E6 E7 E8 E9 E10 E11 E12 E13 E14 \u4f30\u8207\u6307\u5b9a\u7279\u5fb5\u5c0d\u63a8\u8ad6\u95dc\u4fc2\u5224\u65b7\u7684\u6548\u80fd\u6bd4 \u8f03\u3002\u4e0d\u904e\u7919\u65bc\u7248\u9762\u9650\u5236\uff0c\u672c\u7bc7\u8ad6\u6587\u4e2d\u6587 \u8a9e\u6599\u53ea\u7bc0\u9304 RITE-2 \u7e41\u9ad4\u8a9e\u6599\u4f5c\u70ba\u4ee3\u8868\uff0c \u800c\u82f1\u6587\u8a9e\u6599\u5247\u4ee5 MSR \u4f5c\u70ba\u4ee3\u8868\uff0c\u5176\u5b83\u8a73 \u7d30\u7684\u5be6\u9a57\u7d50\u679c\u53ef\u53c3\u7167\u9ec3\u744b\u6770\u78a9\u58eb\u8ad6\u6587 [13]\u3002 \u70ba\u4e86\u77ad\u89e3\u5404\u7a2e\u7279\u5fb5\u7d44\u5408\u7684\u5206\u985e\u6548\u679c\uff0c\u6211\u5011\u63a1\u7528\u8caa\u5a6a\u5f0f\u7684\u7279\u5fb5\u7d44\u5408\u641c\u5c0b\uff0c\u6e2c\u8a66\u8a13\u7df4\u8a9e\u6599 \u70ba\u4e86\u77ad\u89e3\u4e09\u7a2e\u5206\u985e\u5668\u5728\u63a8\u8ad6\u95dc\u4fc2\u5224\u65b7\u4e0a\u7684\u6548\u679c\uff0c\u6211\u5011\u6839\u64da\u4e0a\u4e00\u5c0f\u7bc0\u7372\u5f97\u7684\u7279\u5fb5\u7d44\u5408\uff0c\u900f\u904e SVM\u3001J48 \u53ca\u7dda\u6027\u56de\u6b78\u7b49\u6f14\u7b97\u6cd5\u9032\u884c\u5206\u985e\u5668\u7684\u6548\u80fd\u8a55\u4f30\uff0cSVM \u8207 J48 \u6f14\u7b97\u6cd5\u4ee5\u5341\u7b49\u5206\u7684\u5faa \u74b0\u4f30\u8a08\u6e96\u78ba\u503c\u70ba\u8a55\u4f30\u6307\u6a19\uff0c\u7dda\u6027\u56de\u6b78\u6f14\u7b97\u6cd5\u5247\u518d\u4ee5\u8a13\u7df4\u8a9e\u6599\u6e2c\u8a66\uff0c\u8a2d\u5b9a\u9580\u6abb\u503c\u70ba 0.5 \u5c0d\u63a8 \u8ad6\u95dc\u4fc2\u5206\u985e\uff0c\u8a55\u4f30\u5176\u6e96\u78ba\u503c\u3002\u6211\u5011\u5c07\u4f9d\u64da\u5404\u7a2e\u5206\u985e\u6a21\u578b\u5728\u8a13\u7df4\u8a9e\u6599\u7684\u6548\u679c\uff0c\u5728\u4e0d\u540c\u985e\u578b\u7684 \u8a9e\u6599\u4e2d\u63a1\u7528\u6307\u5b9a\u7684\u5206\u985e\u6f14\u7b97\u6cd5\u9032\u884c\u63a8\u8ad6\u95dc\u4fc2\u7684\u5206\u985e\u3002 \u5716\u4e5d\u548c\u5716\u5341\u5206\u5225\u70ba RITE-2 \u7e41\u9ad4\u4e2d\u6587\u53ca MSR \u8a13\u7df4\u8a9e\u6599\u5728\u4e0d\u540c\u5206\u985e\u6a21\u578b\u4e0b\uff0c\u4ee5\u6e96\u78ba\u7387\u8f03 \u4f73\u8a2d\u5b9a\uff0c\u800c\u82f1\u6587\u5247\u662f\u5f9e\u5be6\u9a57\u4e2d\u85c9\u7531\u6548\u80fd\u7684\u8b8a\u5316\u4ee5\u4eba\u5de5\u7684\u65b9\u5f0f\u8a2d\u5b9a\u53c3\u6578\uff0c\u5982\u8868\u5341\u4e00\u6240\u793a\u3002\u6a5f \u7684\u6548\u679c\uff0c\u539f\u59cb\u7684\u82f1\u6587\u8a9e\u6599\u900f\u904e\u7d93\u9a57\u6cd5\u5247\u5f0f\u63a8\u8ad6\u6a21\u578b\u90fd\u80fd\u9054\u5230 30%\u4ee5\u4e0a\u7684\u57fa\u672c\u6548\u80fd\uff0c\u800c\u63a1\u7528 \u9664\u4e86\u5c07\u554f\u984c\u53ca\u9078\u9805\u8f49\u63db\u70ba\u76f4\u8ff0\u53e5\u4f86\u9032\u884c\u63a8\u8ad6\u95dc\u4fc2\u7684\u5224\u65b7\u4e4b\u5916\uff0c\u4e00\u7bc7\u77ed\u6587\u4e2d\u53ef\u80fd\u540c\u6642\u6558\u8ff0\u76f8 \u5668\u5b78\u7fd2\u7684\u5206\u985e\u6a21\u578b\u5247\u7531 RITE-2 \u7e41\u9ad4\u4e2d\u6587\u8a13\u7df4\u8a9e\u6599\u53ca MSR \u82f1\u6587\u8a13\u7df4\u8a9e\u6599\uff0c\u9078\u53d6\u9069\u7576\u7684\u7279\u5fb5 \u77ed\u6587\u904e\u6ffe\u5f8c\uff0c\u5247\u5927\u7d04\u90fd\u80fd\u63d0\u5347\u5341\u5230\u4e8c\u5341\u500b\u767e\u5206\u9ede\uff0c\u8aaa\u660e\u77ed\u6587\u904e\u6ffe\u5728\u589e\u5f37\u63a8\u8ad6\u7cfb\u7d71\u5224\u65b7\u95b1\u8b80 \u7576\u591a\u7a2e\u7684\u4e8b\u5be6\u8207\u52d5\u4f5c\uff0c\u56e0\u6b64\u6bcf\u4e00\u9053\u554f\u984c\u7684\u80cc\u5f8c\u5f80\u5f80\u90fd\u50c5\u6709\u5229\u7528\u5230\u77ed\u6587\u4e2d\u90e8\u5206\u7684\u9673\u8ff0\u53e5\u5b50\u4f86 \u8a13\u7df4\u5206\u985e\u6a21\u578b\uff0c\u63a5\u8457\u9032\u884c\u95b1\u8b80\u6e2c\u9a57\u4e2d\u77ed\u6587\u8207\u6bcf\u4e00\u500b\u9078\u9805\u7684\u63a8\u8ad6\u95dc\u4fc2\u5224\u65b7\u3002\u8868\u5341\u4e8c\u986f\u793a\u4e2d\u6587 \u6e2c\u9a57\u7b54\u6848\u6642\u5177\u6709\u826f\u597d\u7684\u529f\u6548\uff0c\u672a\u4f86\u53ef\u4ee5\u91dd\u5c0d\u6b64\u90e8\u5206\u767c\u5c55\u81ea\u52d5\u5316\u7684\u8655\u7406\u65b9\u6cd5\u904e\u6ffe\u77ed\u6587\u3002 \u56de\u7b54\u3002 \u70ba\u4e86\u77ad\u89e3\u7d93\u7531\u77ed\u6587\u5167\u5bb9\u6311\u9078\u9069\u7576\u7684\u53e5\u5b50\u5f8c\uff0c\u5c0d\u6307\u5b9a\u554f\u984c\u56de\u7b54\u7684\u63a8\u8ad6\u6548\u679c\uff0c\u6211\u5011\u9996\u5148\u63a1 \u95b1\u8b80\u6e2c\u9a57\u63a1\u7528 SVM \u6f14\u7b97\u6cd5\u7684\u7279\u5fb5\u96c6\uff0c\u8868\u5341\u4e09\u70ba\u4f7f\u7528\u7dda\u6027\u56de\u6b78\u4e4b\u7279\u5fb5\u96c6\uff0c\u8868\u5341\u56db\u70ba\u82f1\u6587\u4e4b \u63a5\u8457\u4f7f\u7528\u6a5f\u5668\u5b78\u7fd2\u6f14\u7b97\u6cd5\u8a13\u7df4\u5206\u985e\u6a21\u578b\u5224\u65b7\u95b1\u8b80\u6e2c\u9a57\u4e2d\u6bcf\u500b\u9078\u9805\u7684\u63a8\u8ad6\u95dc\u4fc2\uff0c\u5728\u4e2d\u6587 \u7279\u5fb5\u96c6\u3002 \u95b1\u8b80\u6e2c\u9a57\uff0c\u6211\u5011\u4ee5\u4e0a\u4e00\u5c0f\u7bc0\u7684\u7279\u5fb5\u96c6\uff0c\u63a1\u7528 SVM \u53ca\u7dda\u6027\u56de\u6b78\u5169\u7a2e\u6f14\u7b97\u6cd5\u505a\u63a8\u8ad6\u95dc\u4fc2\u7684\u5206 \u7528\u4eba\u5de5\u7684\u65b9\u5f0f\u9032\u884c\u77ed\u6587\u7684\u904e\u6ffe\uff0c\u4f9d\u64da\u984c\u7d44\u4e2d\u6bcf\u4e00\u9053\u554f\u984c\uff0c\u5c0d\u77ed\u6587\u63a1\u53d6\u904e\u6ffe\uff0c\u6311\u9078\u5176\u4e2d\u8207\u6b64 \u8868\u5341\u4e09\u3001\u4e2d\u6587\u95b1\u8b80\u6e2c\u9a57\u7279\u5fb5\u96c6 -\u7dda\u6027\u56de\u6b78 \u8868\u5341\u4e8c\u3001\u4e2d\u6587\u95b1\u8b80\u6e2c\u9a57\u7279\u5fb5\u96c6 -SVM \u985e\u3002\u5716\u5341\u56db\u53ca\u5716\u5341\u4e94\u70ba\u4e2d\u6587\u95b1\u8b80\u6e2c\u9a57\u7684\u6548\u80fd\u6bd4\u8f03\uff0c\u7531\u5716\u8868\u89c0\u5bdf\u5f97\u77e5\uff0c\u77ed\u6587\u904e\u6ffe\u5728\u95b1\u8b80\u6e2c\u9a57 \u9053\u554f\u984c\u76f8\u95dc\u7684\u53e5\u5b50\uff0c\u5f62\u6210\u4e00\u500b\u8f03\u5c0f\u7684\u53e5\u5b50\u96c6\u5408\u4f86\u5c0d\u554f\u984c\u53ca\u9078\u9805\u7684\u7d44\u5408\u5224\u65b7\u63a8\u8ad6\u95dc\u4fc2\u3002 \u4e2d\u5224\u65b7\u63a8\u8ad6\u95dc\u4fc2\u662f\u4e00\u9805\u975e\u5e38\u6709\u6548\u7528\u7684\u6b65\u9a5f\u3002\u7136\u800c\u4f7f\u7528\u6a5f\u5668\u5b78\u7fd2\u5206\u985e\u6a21\u578b\u7684\u95b1\u8b80\u6e2c\u9a57\u6548\u679c\u5247 \u4e2d\u6240\u6709\u7684\u7279\u5fb5\u7d44\u5408\uff0c\u7531 LibSVM \u8207 Weka \u5c07\u8a13\u7df4\u8a9e\u6599\u81ea\u52d5\u5207\u70ba\u5341\u500b\u7b49\u5206(10-fold)\uff0c\u5728 SVM \u53ca J48 \u6f14\u7b97\u6cd5\u7684\u5206\u985e\u4e0b\u9032\u884c\u5faa\u74b0\u4f30\u8a08(Cross-Validation)\uff0c\u627e\u5c0b\u6e96\u78ba\u7387\u6975\u5927\u503c\u7684\u7279\u5fb5\u7d44\u5408\uff0c \u800c\u7dda\u6027\u56de\u6b78\u5247\u518d\u6b21\u4f7f\u7528\u8a13\u7df4\u8a9e\u6599\u505a\u70ba\u8a55\u4f30\u8a9e\u6599\uff0c\u8a2d\u5b9a\u9580\u6abb\u503c\u70ba 0.5 \u627e\u5c0b\u6e96\u78ba\u7387\u6700\u5927\u503c\uff0c\u6700 \u5f8c\u5c07\u7372\u5f97\u7684\u7279\u5fb5\u7d44\u5408\u9032\u884c\u6e2c\u8a66\u8a9e\u6599\u7684\u5be6\u9a57\u8a55\u4f30\u3002\u8868\u516d\u8207\u8868\u4e03\u5177\u6709\u7de8\u865f\u5f62\u5f0f\u7684\u4e2d\u82f1\u6587\u7279\u5fb5 \u4f73\u7684\u7279\u5fb5\u7d44\u5408\u9032\u884c\u63a8\u8ad6\u95dc\u4fc2\u7684\u5206\u985e\u7d50\u679c\uff0cM1 \u81f3 M3 \u53ef\u53c3\u7167\u8868\u516b\uff0c\u70ba\u7e41\u9ad4\u4e2d\u6587\u7684\u7279\u5fb5\u7d44\u5408\uff1b \u4e0d\u5982\u7d93\u9a57\u6cd5\u5247\u5f0f\u63a8\u8ad6\u6a21\u578b\u4f86\u7684\u6709\u6548\u679c\u3002 \u6211\u5011\u5e0c\u671b\u5148\u900f\u904e\u4eba\u5de5\u904e\u6ffe\u7684\u5f62\u5f0f\uff0c\u9032\u884c\u90e8\u5206\u5be6\u9a57\u4f86\u9a57\u8b49\u9019\u6a23\u7684\u5de5\u4f5c\u5177\u6709\u4e00\u5b9a\u6210\u6548\uff0c\u63a5 M4 \u81f3 M6 \u53ef\u53c3\u7167\u8868\u4e5d\u70ba MSR \u82f1\u6587\u8a9e\u6599\u7684\u7279\u5fb5\u7d44\u5408\uff1b\u5f9e\u7e41\u9ad4\u4e2d\u6587\u8207 MSR \u5169\u7a2e\u8a9e\u6599\u7684\u7d50\u679c \u8457\u518d\u767c\u5c55\u76f8\u95dc\u7684\u81ea\u52d5\u5316\u6280\u8853\u8207\u65b9\u6cd5\uff0c\u5982\u5224\u5b9a\u77ed\u6587\u8207\u554f\u984c\u7684\u95dc\u9023\u6027\u3001\u4e2d\u5fc3\u8a5e\u5f59\u6216\u95dc\u9375\u5b57\u641c\u5c0b\uff0c \u89c0\u5bdf\uff0c\u4f7f\u7528\u7dda\u6027\u56de\u6b78\u6f14\u7b97\u6cd5\u9032\u884c\u63a8\u8ad6\u95dc\u4fc2\u5206\u985e\u6642\uff0c\u5e73\u5747\u4e0a\u90fd\u80fd\u7372\u5f97\u8f03\u4f73\u7684\u6e96\u78ba\u7387\uff0c\u5373\u4f7f\u5728 \u85c9\u4ee5\u63d0\u6607\u63a8\u8ad6\u7cfb\u7d71\u5728\u95b1\u8b80\u6e2c\u9a57\u4e2d\u7684\u6548\u80fd\u3002 SVM \u53ca J48 \u5206\u985e\u6a21\u578b\u80fd\u7372\u5f97\u6700\u9ad8\u6e96\u78ba\u7387\u7684\u7279\u5fb5\u7d44\u5408\uff0c\u900f\u904e\u7dda\u6027\u56de\u6b78\u6f14\u7b97\u6cd5\u7684\u4f7f\u7528\uff0c\u76f8\u8f03\u65bc \u4e2d\u570b 0 0 0 1 0 1 ROOT 0 0 0 0 0 \u96c6\u3002 \u5169\u7a2e\u6f14\u7b97\u6cd5\u7684\u6700\u9ad8\u6e96\u78ba\u7387\u50c5\u6709\u4e9b\u5fae\u7684\u4e0b\u8dcc\uff0c\u4ecd\u80fd\u9054\u5230\u4e0d\u932f\u7684\u6548\u679c 0 \u5716\u516b\u3001\u7d93\u904e\u4e94\u6b65\u7684\u8a5e\u5f59\u4f9d\u8cf4\u95dc\u4fc2\uff0c \u8868\u516d\u3001\u4e2d\u6587\u7279\u5fb5\u96c6\u7de8\u865f\u8868 F1 F2 F3 F4 F5 F6 F7 F8 F9 F10 F11 F12 F13 F14 F15 F16 F17 6.2 \u7279\u5fb5\u9078\u53d6 \u63a5\u8457\u5c55\u958b\u4e09\u7a2e\u5206\u985e\u6f14\u7b97\u6cd5\u5728\u5404\u7a2e\u8a9e\u6599\u7684\u7279\u5fb5\u7d44\u5408\u641c\u5c0b\u3002\u6211\u5011\u7531\u4e09\u7a2e\u5206\u985e\u6f14\u7b97\u6cd5\u7684\u7d50\u679c\u4e2d\u641c \u5c0b\u5404\u7a2e\u8a9e\u6599\u4e2d\u6e96\u78ba\u7387\u8f03\u4f73\u7684\u7279\u5fb5\u7d44\u5408\uff0c\u8868\u516b\u8868\u4e5d\u986f\u793a\u5728\u4e0d\u540c\u8a9e\u6599\u8207\u5206\u985e\u6f14\u7b97\u6cd5\u4e2d\u7372\u5f97\u8f03\u4f73 \u679c\u3002 \u8868\u516b\u3001RITE-2 \u7e41\u9ad4\u4e2d\u6587\u8a13\u7df4\u8a9e\u6599\u7279\u5fb5\u7d44\u5408\u641c\u5c0b SVM \u7de8\u865f \u7279\u5fb5\u7d44\u5408\u7de8\u865f Accuracy M1 F1, F2, F3, F4, F5, F6, F8, F9, F12, F14 71.99% J48 M2 F1, F2, F3, F5, F7, F8, F12, F13, F15 71.78% \u7dda\u6027\u56de\u6b78 M3 F1,F3,F4,F5,F6,F7,F8,F9,F10,F11,F12,F13,F14,F15,F16,F17 72.98% \u8868\u4e5d\u3001MSR \u8a13\u7df4\u8a9e\u6599\u7279\u5fb5\u7d44\u5408\u641c\u5c0b SVM \u7de8\u865f \u7279\u5fb5\u7d44\u5408\u7de8\u865f M4 E1, E6, E9, E12 70.93% J48 M5 E1, E6, E8, E10, E12, E14 71.82% \u7dda\u6027\u56de\u6b78 M6 E1,E2,E3,E4,E5,E6,E7,E9,E10,E11,E12,E13,E14 72.45% \u5716\u5341\u4e00\u3001\u76f4\u8ff0\u53e5\u8f49\u63db\u7bc4\u4f8b \u5c07\u554f\u53e5\u53ca\u9078\u9805\u901a\u904e\u4eba\u5de5\u7684\u65b9\u5f0f\u8f49\u63db\u6210\u56db\u500b\u76f4\u8ff0 \u53e5\uff0c\u518d\u63a1\u7528\u63a8\u8ad6\u7cfb\u7d71\u9032\u884c\u77ed\u6587\u8207\u56db\u500b\u76f4\u8ff0\u53e5\u7684\u63a8 \u8ad6\u95dc\u4fc2\u5224\u5b9a\uff0c\u5982\u5716\u5341\u4e00\u70ba\u4e2d\u6587\u95b1\u8b80\u6e2c\u9a57\u76f4\u8ff0\u53e5\u8f49 \u63db\u7684\u4f8b\u5b50\u3002 \u51fa\u7684\u65b9\u6cd5\u3002 \u5716\u5341\u4e09\u3001\u82f1\u6587\u95b1\u8b80\u6e2c\u9a57\u6e96\u78ba\u7387 -\u7d93\u9a57\u6cd5\u5247\u5f0f\u63a8\u8ad6\u6a21\u578b \u5f62\u5f0f\u8207\u6e2c\u9a57\u8a2d\u8a08\uff0c\u9700\u8981\u66f4\u591a\u7684\u8a9e\u6599\u4f86\u9a57\u8b49\u6211\u5011\u63d0 \u63a8\u8ad6\u95dc\u4fc2\u5224\u65b7\uff0c\u4ee5\u6578\u503c\u6700\u9ad8\u7684\u9078\u9805\u505a\u70ba\u7b54\u6848\u3002 \u4e26\u7121\u6cd5\u6709\u6548\u986f\u793a\u771f\u6b63\u5728\u5c0f\u5b78\u4e00\u5e74\u7d1a\u95b1\u8b80\u6e2c\u9a57\u7684 \u8b80\u7406\u89e3\u7684\u90e8\u5206\u518d\u627e\u51fa\u6709\u7528\u7684\u8a9e\u8a00\u7279\u5fb5\u85c9\u4ee5\u63d0\u5347\u7b54\u984c\u7684\u6e96\u78ba\u7387\u3002 \u7387\u503c\uff0c\u4ee5\u9078\u9805\u4e2d\u6a5f\u7387\u503c\u6700\u9ad8\u7684\u505a\u70ba\u7b54\u6848\uff0c\u6b64\u5916\u6211\u5011\u5728\u4e2d\u6587\u7684\u90e8\u5206\u4e5f\u52a0\u5165\u7dda\u6027\u56de\u6b78\u6f14\u7b97\u6cd5\u7684 \u8a9e\u6599\u7684\u6578\u91cf\u975e\u5e38\u7a00\u5c11\uff0c\u56e0\u6b64\u6211\u5011\u8a8d\u70ba\u9019\u6a23\u7684\u6548\u679c \u7387\u4e0a\u5347\u5341\u5e7e\u500b\u767e\u5206\u9ede\uff0c\u6211\u5011\u5e0c\u671b\u672a\u4f86\u80fd\u5920\u81ea\u52d5\u5316\u7684\u5b8c\u6210\u95b1\u8b80\u6e2c\u9a57\u524d\u8655\u7406\u7684\u90e8\u5206\uff0c\u4e26\u91dd\u5c0d\u95b1 \u9ad8\u8005\u70ba\u8a72\u554f\u984c\u7684\u6700\u4f73\u7b54\u6848\uff1b\u800c\u6a5f\u5668\u5b78\u7fd2\u5206\u985e\u6a21\u578b\u5247\u7531 SVM \u6f14\u7b97\u6cd5\uff0c\u8f38\u51fa\u5176\u63a8\u8ad6\u95dc\u4fc2\u7684\u6a5f \u6211\u5011\u8a8d\u70ba\u548c\u8a9e\u6599\u7684\u6578\u91cf\u5177\u6709\u76f8\u7576\u7684\u95dc\u4fc2\uff0c\u4e00\u5e74\u7d1a \u76f4\u8ff0\u53e5\u548c\u7be9\u9078\u76f8\u95dc\u53e5\u7684\u90e8\u5206\u76ee\u524d\u4ecd\u662f\u4ee5\u4eba\u5de5\u8655\u7406\uff0c\u5176\u4e2d\u5728\u7be9\u9078\u76f8\u95dc\u53e5\u7684\u90e8\u5206\u5c31\u8db3\u4ee5\u8b93\u6e96\u78ba \u6cd5\u5247\u5f0f\u63a8\u8ad6\u6a21\u578b\u4e2d\uff0c\u6211\u5011\u4ee5\u5404\u500b\u9078\u9805\u901a\u904e\u8a08\u7b97\u5f8c\u7684\u63a8\u8ad6\u5206\u6578\u70ba\u8a55\u91cf\u6307\u6a19\uff0c\u9078\u53d6\u5176\u4e2d\u5206\u6578\u6700 \u6b65\u3002\u800c\u4e00\u5e74\u7d1a\u8a9e\u6599\u4e26\u672a\u5728\u77ed\u6587\u904e\u6ffe\u4e2d\u767c\u63ee\u529f\u6548\uff0c \u95dc\u4fc2\u7684\u6e96\u78ba\u7387\uff0c\u56e0\u70ba\u67d0\u4e9b\u7279\u5fb5\u53ef\u80fd\u53ea\u5c0d\u90e8\u5206\u7684\u8a9e\u6599\u6709\u6548\uff0c\u800c\u5728\u95b1\u8b80\u7406\u89e3\u7684\u90e8\u5206\uff0c\u5728\u554f\u984c\u8f49 \u7d71-\u7d93\u9a57\u6cd5\u5247\u5f0f\u63a8\u8ad6\u6a21\u578b\u8207\u6a5f\u5668\u5b78\u7fd2\u5206\u985e\u6a21\u578b\uff0c\u5224\u65b7\u95b1\u8b80\u6e2c\u9a57\u4e2d\u6700\u4f73\u7684\u56de\u7b54\u9078\u9805\u3002\u5728\u7d93\u9a57 \u9810\u671f\u6709\u8f03\u591a\u7684\u9032\u6b65\u5e45\u5ea6\uff0c\u50c5\u5728\u56db\u5e74\u7d1a\u6709\u4e9b\u5fae\u7684\u9032 \u932f\u7684\u6548\u679c\u3002\u82f1\u6587\u7684\u7d50\u679c\u5247\u4ecd\u6709\u7684\u9032\u6b65\u7a7a\u9593\uff0c\u6211\u5011\u8a8d\u70ba\u8a9e\u6599\u7684\u4e0d\u540c\u8a9e\u8a00\u7279\u6027\uff0c\u8db3\u4ee5\u5f71\u97ff\u63a8\u8ad6 \u6211\u5011\u5c07\u8a9e\u6599\u5206\u70ba\u4e09\u7a2e\u985e\u5225\uff0c\u539f\u59cb\u8a9e\u6599\u3001\u554f\u53e5\u91cd\u7d44\u53ca\u77ed\u6587\u904e\u6ffe\uff0c\u4e26\u5206\u5225\u63a1\u7528\u5169\u7a2e\u63a8\u8ad6\u7cfb \u7b54\u95b1\u8b80\u6e2c\u9a57\u7684\u554f\u984c\uff1b\u800c\u76f4\u8ff0\u53e5\u8f49\u63db\u5247\u8f03\u4e0d\u5982\u6211\u5011 \u6211\u5011\u63d0\u51fa\u7684\u63a8\u8ad6\u7cfb\u7d71\u8207 NTCIR-9\u3001NTCIR-10 \u7af6\u8cfd\u6210\u7e3e\u76f8\u6bd4\uff0c\u5728\u4e2d\u6587\u8a9e\u6599\u4e2d\u4ecd\u5c6c\u65bc\u4e0d Accuracy 7 \u95b1\u8b80\u7406\u89e3\u7684\u5be6\u9a57\u6e96\u5099 \u672c\u7bc0\u5c07\u4ecb\u7d39\u6587\u5b57\u860a\u6db5\u5728\u95b1\u8b80\u6e2c\u9a57\u4e2d\u7684\u61c9\u7528\uff0c\u85c9\u7531\u524d\u9762\u7bc0\u6b21\u5224\u5225\u6587\u5b57\u860a\u6db5\u95dc\u4fc2\u6240\u5efa\u69cb\u7684\u6a21 7.1 \u554f\u984c\u8f49\u76f4\u8ff0\u53e5 \u5728\u524d\u9762\u5be6\u9a57\u6240\u4f7f\u7528\u7684\u63a8\u8ad6\u7cfb\u7d71\u4e2d\uff0c\u6240\u6709\u7684\u8a9e\u6599\u90fd \u662f\u4ee5\u5169\u500b\u76f4\u8ff0\u53e5\u9032\u884c\u63a8\u8ad6\u95dc\u4fc2\u7684\u5224\u65b7\uff0c\u800c\u5728\u95b1\u8b80 \u6e2c\u9a57\u4e2d\uff0c\u70ba\u4e86\u76f4\u63a5\u63d0\u5347\u63a8\u8ad6\u95dc\u4fc2\u7684\u6548\u679c\uff0c\u6211\u5011\u4e5f 8.2 \u5be6\u9a57\u8a2d\u8a08\u3001\u8a9e\u6599\u7684\u4f7f\u7528\u65b9\u5f0f \u5728\u95b1\u8b80\u6e2c\u9a57\u7684\u5be6\u9a57\u4e2d\uff0c\u6211\u5011\u63a1\u7528\u524d\u9762\u5169\u7a2e\u4e0d\u540c\u7684\u63a8 \u8ad6\u7cfb\u7d71\u9032\u884c\u6548\u80fd\u8a55\u4f30\uff0c\u4e26\u5c07\u8a9e\u6599\u63a1\u7528\u4e0d\u540c\u7684\u65b9\u5f0f\u9032 \u884c\u4eba\u5de5\u8f49\u63db\u6216\u904e\u6ffe\uff0c\u4ee5\u5617\u8a66\u6b64\u65b9\u6cd5\u5728\u95b1\u8b80\u6e2c\u9a57\u4e2d\u7684 \u6548\u679c\u3002 \u8a9e\u8a00 E \u03b1 \u03b2 \u03b3 \u03bb \u4e2d\u6587 \u82f1\u6587 0.47 0.0 0.26 1.3 0.6 \u5716\u5341\u4e8c\u3001\u4e2d\u6587\u95b1\u8b80\u6e2c\u9a57\u6e96\u78ba\u7387-\u7d93\u9a57\u6cd5\u5247\u5f0f\u63a8\u8ad6\u6a21\u578b \u7684\u8a9e\u6599\u5916\uff0c\u90fd\u986f\u793a\u4e86\u6b64\u65b9\u6cd5\u6709\u52a9\u65bc\u63a8\u8ad6\u7cfb\u7d71\u6b63\u78ba\u56de \u7684\u63a8\u8ad6\u7cfb\u7d71\u4f5c\u95b1\u8b80\u6e2c\u9a57\u7684\u81ea\u52d5\u7b54\u984c\uff0c\u5728\u56db\u9078\u4e00\u7684\u95b1\u8b80\u6e2c\u9a57\u4e2d\u4e5f\u53ef\u4ee5\u7372\u5f97\u7d04 50%\u7684\u6e96\u78ba\u7387\u3002 1.2 \u6027\u56de\u6b78\u7684\u65b9\u6cd5\uff0c\u4e2d\u6587\u548c\u82f1\u6587\u8a9e\u6599\u7684\u6e96\u78ba\u7387\u5206\u5225\u53ef\u9054 72.98%\u548c 72.54%\uff1b\u800c\u57fa\u65bc\u4e0a\u8ff0\u5efa\u69cb\u597d \u6e2c\u9a57\u4e2d\u63a8\u8ad6\u95dc\u4fc2\u7684\u5224\u65b7\u662f\u8f03\u6709\u5e6b\u52a9\u7684\uff0c\u9664\u4e86\u4e00\u5e74\u7d1a 0.57 0.28 0.24 2.0 0.85 2.0 \u5f0f\u6a21\u578b\u7684\u65b9\u6cd5\u4e0a\uff0c\u4e2d\u6587\u548c\u82f1\u6587\u8a9e\u6599\u7684\u6e96\u78ba\u7387\u5206\u5225\u53ef\u9054 68.56%\u548c 72.23%\uff1b\u63a1\u7528\u6a5f\u5668\u5b78\u7fd2\u7dda \u6cd5\u5f8c\uff0c\u7531\u4e2d\u6587\u7684\u7d50\u679c\u53ef\u4ee5\u767c\u73fe\uff0c\u77ed\u6587\u904e\u6ffe\u5c0d\u65bc\u95b1\u8b80 \u03b4 \u80fd\u8a55\u4f30\uff0c\u6700\u5f8c\u4e26\u5229\u7528\u524d\u9762\u5efa\u69cb\u597d\u7684\u63a8\u8ad6\u7cfb\u7d71\u61c9\u7528\u65bc\u95b1\u8b80\u6e2c\u9a57\u7684\u81ea\u52d5\u7b54\u984c\u4e0a\u9762\uff0c\u5728\u7d93\u9a57\u6cd5\u5247 \u4e2d\uff0c\u4ecd\u53ef\u4ee5\u7372\u5f97\u7d04 37%\u7684\u6548\u679c\uff0c\u800c\u5728\u5957\u7528\u9069\u7576\u7684\u65b9 \u8868\u5341\u4e00\u3001\u95b1\u8b80\u6e2c\u9a57\u5be6\u9a57\u53c3\u6578\u8a2d\u5b9a \u8b80\u6e2c\u9a57\u4e2d\uff0c\u6211\u5011\u7684\u63a8\u8ad6\u7cfb\u7d71\u5373\u4f7f\u5728\u9ad8\u5e74\u7d1a\u7684\u8a9e\u6599 \u8a9e\u6599\u7684\u6587\u5b57\u860a\u6db5\u95dc\u4fc2\uff0c\u4e5f\u63a1\u7528\u6a5f\u5668\u5b78\u7fd2\u7684\u65b9\u6cd5\u900f\u904e SVM\u3001J48 \u53ca\u7dda\u6027\u56de\u6b78\u7b49\u6f14\u7b97\u6cd5\u9032\u884c\u6548 \u5b57\u860a\u6db5\u5728\u95b1\u8b80\u6e2c\u9a57\u4e2d\u7684\u61c9\u7528\uff0c\u672a\u4f86\u53ef\u4ee5\u5c07\u6b64\u61c9\u7528\u63a8\u5ee3\u81f3\u5be6\u52d9\u7684\u6559\u80b2\u8cc7\u8a0a\u7cfb\u7d71\u3002 \u7684\u6548\u80fd\u5716\u8868\uff0c\u5f9e\u5404\u5e74\u7d1a\u7684\u7d50\u679c\u986f\u793a\uff0c\u5728\u56db\u9078\u4e00\u7684\u95b1 \u672c\u7814\u7a76\u5229\u7528\u6703\u5f71\u97ff\u6587\u5b57\u860a\u6db5\u7684\u7279\u5fb5\u8cc7\u8a0a\uff0c\u5efa\u69cb\u7d93\u9a57\u6cd5\u5247\u5f0f\u6a21\u578b\u7528\u4ee5\u5224\u5225 RITE\u3001RTE\u3001MSR \u578b\uff0c\u4f5c\u70ba\u63a8\u8ad6\u95b1\u8b80\u6e2c\u9a57\u7b54\u6848\u7684\u4f9d\u64da\uff1b\u6211\u5011\u5728 7.1 \u8207 7.2 \u5c0f\u7bc0\u4ecb\u7d39\u5be6\u9a57\u7684\u524d\u8655\u7406\u3002\u5e0c\u671b\u900f\u904e\u6587 \u5716\u4e5d\u3001\u5206\u985e\u6a21\u578b\u6e96\u78ba\u7387\u6bd4\u8f03\uff1aRITE-2 \u7e41\u9ad4\u4e2d\u6587\u8a13\u7df4\u8a9e\u6599 \u5716\u5341\u3001\u5206\u985e\u6a21\u578b\u6e96\u78ba\u7387\u6bd4\u8f03\uff1aMSR \u8a13\u7df4\u8a9e\u6599 J48 69.46% 71.78% 66.34% \u7dda\u6027\u56de\u6b78 71.01% 71.23% 72.98% J48 70.80% 71.82% 69.89% \u7dda\u6027\u56de\u6b78 72.06% 71.57% 72.45% 8.1 \u5be6\u9a57\u8a9e\u6599\u7684\u4f86\u6e90\u3001\u6578\u91cf \u6211\u5011\u8490\u96c6\u4e2d\u82f1\u6587\u7684\u95b1\u8b80\u6e2c\u9a57\u8a9e\u6599\u96c6\uff0c\u4e2d\u6587\u7684\u90e8\u5206\u4ee5\u570b\u5c0f\u5b69\u7ae5\u95b1\u8b80\u6e2c\u9a57\u70ba\u4e3b\uff0c\u82f1\u6587\u5247\u8490 \u96c6\u570b\u4e2d\u7684\u95b1\u8b80\u6e2c\u9a57\uff0c\u4e26\u4e14\u6211\u5011\u4f9d\u7167\u5e74\u7d1a\u5c07\u8a9e\u6599\u5206\u985e\uff0c\u76f8\u95dc\u7684\u7d71\u8a08\u5982\u8868\u5341\uff0c\u8a9e\u6599\u5167\u5bb9\u90fd\u4ee5\u4e00 \u8a9e\u6599\u5c6c\u65bc\u4e09\u500b\u9078\u9805\uff0c\u4e26\u4e14\u6bcf\u4e00\u9053\u984c\u76ee\u7684\u7b54\u6848\u90fd\u70ba\u55ae\u4e00\u9078\u9805\u3002 \u5716\u5341\u4e8c\u8207\u5716\u5341\u4e09\u5206\u5225\u70ba\u4e2d\u6587\u53ca\u82f1\u6587\u95b1\u8b80\u6e2c\u9a57 9 \u7d50\u8ad6 \u7bc7\u77ed\u6587\u8207\u6578\u5247\u984c\u76ee\u7d44\u6210\uff0c\u6bcf\u4e00\u9053\u984c\u76ee\u90fd\u5305\u542b\u4e00\u500b\u554f\u984c\u8207\u56db\u500b\u9078\u9805\uff0c\u50c5\u6709\u570b\u5c0f\u4e09\u5e74\u7d1a\u7684\u4e2d\u6587 \u570b\u5c0f\u4e94\u5e74\u7d1a 44 \u570b\u5c0f\u516d\u5e74\u7d1a \u65b7\u3002 \u5728\u56db\u9078\u4e00\u500b\u95b1\u8b80\u6e2c\u9a57\u4e2d\u53ef\u4ee5\u7372\u5f97 50%\u5de6\u53f3\u7684\u6e96\u78ba\u7387\u3002 \u6211\u5011\u4f9d\u5e8f\u5c0d\u539f\u59cb\u8a9e\u6599\u3001\u76f4\u8ff0\u53e5\u8f49\u63db\u8207\u77ed\u6587\u904e\u6ffe\u7684\u4e09\u7a2e\u5f62\u5f0f\u8a9e\u6599\u9032\u884c\u95b1\u8b80\u6e2c\u9a57\u7684\u63a8\u8ad6\u95dc\u4fc2\u5224 \u6e96\u78ba\u7387\u5728\u4e0d\u540c\u5e74\u7d1a\u8a9e\u6599\u4e2d\u90fd\u80fd\u7372\u5f97\u7d04\u5341\u4e94\u5230\u4e8c\u5341\u500b\u767e\u5206\u9ede\u7684\u9032\u6b65\uff0c\u662f\u500b\u76f8\u7576\u4e0d\u932f\u7684\u6548\u80fd\uff0c \u9996\u5148\u63a1\u7528\u7d93\u9a57\u6cd5\u5247\u5f0f\u63a8\u8ad6\u6a21\u578b\u5c0d\u4e2d\u82f1\u6587\u95b1\u8b80\u6e2c\u9a57\u9032\u884c\u5be6\u9a57\uff0c\u6240\u4f7f\u7528\u7684\u53c3\u6578\u5982\u4e0a\u4e00\u7bc0\u6240\u793a\uff0c \u65b9\u6cd5\u5c0d\u95b1\u8b80\u6e2c\u9a57\u6587\u672c\u524d\u8655\u7406\uff0c\u5f9e\u5716\u5341\u516d\u7684\u7d50\u679c\u53ef\u4ee5\u767c\u73fe\u7d93\u7531\u77ed\u6587\u904e\u6ffe\u5f8c\uff0c\u95b1\u8b80\u6e2c\u9a57\u7684\u56de\u7b54 86 8.3 \u5be6\u9a57\u7d50\u679c \u800c\u82f1\u6587\u95b1\u8b80\u6e2c\u9a57\u4e2d\uff0c\u6211\u5011\u50c5\u4f7f\u7528 SVM \u6f14\u7b97\u6cd5\u9032\u884c\u90e8\u5206\u7684\u5be6\u9a57\uff0c\u4e26\u50c5\u63a1\u7528\u77ed\u6587\u904e\u6ffe\u7684 \u6e96\u78ba\u7387\u7684\u7279\u5fb5\u7d44\u5408\uff0c\u6211\u5011\u5c07\u900f\u904e\u9019\u4e9b\u7279\u5fb5\u7d44\u5408\u6bd4\u8f03\u4e09\u7a2e\u5206\u985e\u6f14\u7b97\u6cd5\u5728\u63a8\u8ad6\u95dc\u4fc2\u5224\u65b7\u4e0a\u7684\u6548 M1 M2 M3 SVM 71.99% 69.87% 70.40% 60.00% 65.00% 70.00% 75.00% M4 M5 M6 SVM 70.93% 68.38% 68.03% 65.00% 70.00% 8 \u95b1\u8b80\u6e2c\u9a57\u7b54\u984c\u5be6\u6e2c \u8868\u5341\u3001\u95b1\u8b80\u6e2c\u9a57\u8a9e\u6599\u96c6\u7d71\u8a08 75.00% \u4e2d\u6587\u95b1\u8b80\u6e2c\u9a57 \u82f1\u6587\u95b1\u8b80\u6e2c\u9a57 \u8868\u5341\u56db\u3001\u82f1\u6587\u95b1\u8b80\u6e2c\u9a57\u7279\u5fb5\u96c6 -SVM \u6211\u5011\u900f\u904e\u4e0a\u8ff0\u6240\u5efa\u69cb\u7684\u7d93\u9a57\u6cd5\u5247\u5f0f\u63a8\u8ad6\u6a21\u578b\u548c \u5e74\u7d1a \u6578\u91cf \u6578\u91cf \u6a5f\u5668\u5b78\u7fd2\u6a21\u578b\u5206\u5225\u5c0d\u4e2d\u82f1\u6587\u95b1\u8b80\u6e2c\u9a57\u9032\u884c\u7b54\u984c \u570b\u5c0f\u4e00\u5e74\u7d1a 21 \u570b\u4e2d\u4e00\u5e74\u7d1a 260 60.00% \u6548\u80fd\u7684\u8a55\u4f30\uff0c\u4e26\u4ecb\u7d39\u8a9e\u6599\u4f86\u6e90\u3001\u5be6\u9a57\u8a2d\u8a08\u53ca\u5448 \u73fe\u5be6\u9a57\u7d50\u679c\u3002 \u570b\u5c0f\u4e09\u5e74\u7d1a 39 \u570b\u4e2d\u4e8c\u5e74\u7d1a 468 \u570b\u5c0f\u56db\u5e74\u7d1a 40 498 \u570b\u4e2d\u4e09\u5e74\u7d1a \u5716\u5341\u56db\u3001\u4e2d\u6587\u95b1\u8b80\u6e2c\u9a57\u6e96\u78ba\u7387-SVM \u5716\u5341\u4e94\u3001\u4e2d\u6587\u95b1\u8b80\u6e2c\u9a57\u6e96\u78ba\u7387-\u7dda\u6027\u56de\u6b78 \u5716\u5341\u516d\u3001\u82f1\u6587\u95b1\u8b80\u6e2c\u9a57\u6e96\u78ba\u7387-SVM</td></tr></table>" | |
| } | |
| } | |
| } | |
| } |