{ "paper_id": "O09-1008", "header": { "generated_with": "S2ORC 1.0.0", "date_generated": "2023-01-19T08:11:18.210758Z" }, "title": "\u610f\u898b\u6301\u6709\u8005\u8fa8\u8b58 \u610f\u898b\u6301\u6709\u8005\u8fa8\u8b58 \u610f\u898b\u6301\u6709\u8005\u8fa8\u8b58 \u610f\u898b\u6301\u6709\u8005\u8fa8\u8b58\u4e4b\u7814\u7a76 \u4e4b\u7814\u7a76 \u4e4b\u7814\u7a76 \u4e4b\u7814\u7a76 A Study on Identification of Opinion Holders", "authors": [], "year": "", "venue": null, "identifiers": {}, "abstract": "The identification of opinion holders aims to extract entities that express opinions in opinion sentences. In this paper, the task of opinion holder identification is divided into two subtasks: the identification of author's opinions and the labeling of opinion holders. Support vector machine is adopted to identify author's opinions, and conditional random field model (CRF) is utilized to label opinion holders. The proposed method achieves an F-score 0.734 in NTCIR7 MOAT task at traditional Chinese side. The proposed method achieves the best performance among participants who adopted machine learning methods, and also this performance was close to the best performance in this task. In addition, the ambiguous markings of opinion holders are analyzed, and the best way to utilize the training instances with ambiguous markings is proposed.", "pdf_parse": { "paper_id": "O09-1008", "_pdf_hash": "", "abstract": [ { "text": "The identification of opinion holders aims to extract entities that express opinions in opinion sentences. In this paper, the task of opinion holder identification is divided into two subtasks: the identification of author's opinions and the labeling of opinion holders. Support vector machine is adopted to identify author's opinions, and conditional random field model (CRF) is utilized to label opinion holders. The proposed method achieves an F-score 0.734 in NTCIR7 MOAT task at traditional Chinese side. The proposed method achieves the best performance among participants who adopted machine learning methods, and also this performance was close to the best performance in this task. In addition, the ambiguous markings of opinion holders are analyzed, and the best way to utilize the training instances with ambiguous markings is proposed.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Abstract", "sec_num": null } ], "body_text": [], "back_matter": [], "bib_entries": { "BIBREF0": { "ref_id": "b0", "title": "Determining the sentiment of opinions", "authors": [ { "first": "S", "middle": [ "M" ], "last": "Kim", "suffix": "" }, { "first": "E", "middle": [], "last": "Hovy", "suffix": "" } ], "year": 2004, "venue": "Proceedings of the COLING conference", "volume": "", "issue": "", "pages": "1367--1374", "other_ids": {}, "num": null, "urls": [], "raw_text": "S. M. Kim and E. Hovy. \"Determining the sentiment of opinions.\" Proceedings of the COLING conference, pp.1367-1374 , 2004", "links": null }, "BIBREF1": { "ref_id": "b1", "title": "Opinion mining and sentiment analysis", "authors": [ { "first": "B", "middle": [], "last": "Pang", "suffix": "" }, { "first": "L", "middle": [], "last": "Lee", "suffix": "" }, { "first": ";", "middle": [ "M" ], "last": "Kim", "suffix": "" }, { "first": "E", "middle": [], "last": "Hovy", "suffix": "" } ], "year": 2004, "venue": "Proceedings of the COLING conference", "volume": "2", "issue": "", "pages": "1367--1374", "other_ids": {}, "num": null, "urls": [], "raw_text": "B. Pang and L. Lee. \"Opinion mining and sentiment analysis\" Foundations and Trends in Information Retrieval, Vol. 2, pp. 1-135, 2008S. M. Kim and E. Hovy. \"Determining the sentiment of opinions.\" Proceedings of the COLING conference, pp.1367-1374 , 2004", "links": null }, "BIBREF2": { "ref_id": "b2", "title": "Multilingual opinion holder identification using author and authority viewpoints", "authors": [ { "first": "Y", "middle": [], "last": "Seki", "suffix": "" }, { "first": "N", "middle": [], "last": "Kando", "suffix": "" }, { "first": "M", "middle": [], "last": "Aono", "suffix": "" } ], "year": 2009, "venue": "Journal of Information Processing and Management", "volume": "", "issue": "", "pages": "189--199", "other_ids": {}, "num": null, "urls": [], "raw_text": "Y. Seki, N. Kando and M. Aono. \"Multilingual opinion holder identification using author and authority viewpoints.\" Journal of Information Processing and Management, pp. 189-199, 2009 [co-training]", "links": null }, "BIBREF3": { "ref_id": "b3", "title": "Coarse-Fine opinion mining -WIA in NTCIR-7 MOAT task", "authors": [ { "first": "R", "middle": [], "last": "Xu", "suffix": "" }, { "first": "K", "middle": [ "F" ], "last": "Wong", "suffix": "" } ], "year": 2008, "venue": "Proceedings of the Seventh NTCIR Workshop", "volume": "", "issue": "", "pages": "307--313", "other_ids": {}, "num": null, "urls": [], "raw_text": "R. Xu and K. F. Wong. \"Coarse-Fine opinion mining -WIA in NTCIR-7 MOAT task.\" Proceedings of the Seventh NTCIR Workshop, pp. 307-313, 2008", "links": null }, "BIBREF4": { "ref_id": "b4", "title": "Extracting opinions, opinion holders, and topics expressed in online news media text", "authors": [ { "first": "S", "middle": [ "M" ], "last": "Kim", "suffix": "" }, { "first": "E", "middle": [], "last": "Hovy", "suffix": "" } ], "year": 2006, "venue": "Proceedings of the Workshop on Sentiment and Subjectivity in Text at the joint COLING-ACL conference", "volume": "", "issue": "", "pages": "1--8", "other_ids": {}, "num": null, "urls": [], "raw_text": "S. M. Kim and E. Hovy. \"Extracting opinions, opinion holders, and topics expressed in online news media text.\" Proceedings of the Workshop on Sentiment and Subjectivity in Text at the joint COLING-ACL conference, pp. 1-8, 2006", "links": null }, "BIBREF5": { "ref_id": "b5", "title": "Identifying opinion holders in opinion text from online newspapers", "authors": [ { "first": "Y", "middle": [], "last": "Kim", "suffix": "" }, { "first": "Y", "middle": [], "last": "Jung", "suffix": "" }, { "first": "S", "middle": [ "H" ], "last": "Myaeng", "suffix": "" } ], "year": 2007, "venue": "International Conference on Granular Computing", "volume": "", "issue": "", "pages": "699--702", "other_ids": {}, "num": null, "urls": [], "raw_text": "Y. Kim, Y. Jung and S. H. Myaeng. \"Identifying opinion holders in opinion text from online newspapers.\" International Conference on Granular Computing, pp. 699-702, 2007", "links": null }, "BIBREF6": { "ref_id": "b6", "title": "Extracting topic-related opinions and their targets in NTCIR-7", "authors": [ { "first": "Y", "middle": [], "last": "Kim", "suffix": "" }, { "first": "S", "middle": [], "last": "Kim", "suffix": "" }, { "first": "S", "middle": [ "H" ], "last": "Myaeng", "suffix": "" } ], "year": 2008, "venue": "Proceedings of the Seventh NTCIR Workshop", "volume": "", "issue": "", "pages": "247--254", "other_ids": {}, "num": null, "urls": [], "raw_text": "Y. Kim, S. Kim and S. H. Myaeng. \"Extracting topic-related opinions and their targets in NTCIR-7.\" Proceedings of the Seventh NTCIR Workshop, pp. 247-254, 2008", "links": null }, "BIBREF7": { "ref_id": "b7", "title": "Tornado in multilingual opinion analysis: a transductive learning approach for Chinese sentimental polarity recognition", "authors": [ { "first": "Y", "middle": [ "C" ], "last": "Wu", "suffix": "" }, { "first": "L", "middle": [ "W" ], "last": "Yang", "suffix": "" }, { "first": "J", "middle": [ "Y" ], "last": "Shen", "suffix": "" }, { "first": "L", "middle": [ "Y" ], "last": "Chen", "suffix": "" }, { "first": "S", "middle": [ "T" ], "last": "Wu", "suffix": "" } ], "year": 2008, "venue": "Proceedings of the Seventh NTCIR Workshop", "volume": "", "issue": "", "pages": "301--306", "other_ids": {}, "num": null, "urls": [], "raw_text": "Y. C. Wu, L. W. Yang, J. Y. Shen, L. Y. Chen and S. T. Wu. \"Tornado in multilingual opinion analysis: a transductive learning approach for Chinese sentimental polarity recognition.\" Proceedings of the Seventh NTCIR Workshop, pp. 301-306, 2008", "links": null }, "BIBREF8": { "ref_id": "b8", "title": "Identifying expressions of opinion in context", "authors": [ { "first": "E", "middle": [], "last": "Breck", "suffix": "" }, { "first": "Y", "middle": [], "last": "Choi", "suffix": "" }, { "first": "C", "middle": [], "last": "Cardie", "suffix": "" } ], "year": 2007, "venue": "Proceedings of the 20th International Joint Conference on Artificial Intelligence", "volume": "", "issue": "", "pages": "2683--2688", "other_ids": {}, "num": null, "urls": [], "raw_text": "E. Breck and Y. Choi and C. Cardie. \"Identifying expressions of opinion in context.\" Proceedings of the 20th International Joint Conference on Artificial Intelligence, pp. 2683-2688, 2007", "links": null }, "BIBREF9": { "ref_id": "b9", "title": "Identifying sources of opinions with conditional random fields and extraction patterns", "authors": [ { "first": "Y", "middle": [], "last": "Choi", "suffix": "" }, { "first": "C", "middle": [], "last": "Cardie", "suffix": "" }, { "first": "E", "middle": [], "last": "Riloff", "suffix": "" }, { "first": "S", "middle": [], "last": "Patwardhan", "suffix": "" } ], "year": 2005, "venue": "Proceedings of EMNLP conference", "volume": "", "issue": "", "pages": "355--362", "other_ids": {}, "num": null, "urls": [], "raw_text": "Y. Choi, C. Cardie, E. Riloff and S. Patwardhan. \"Identifying sources of opinions with conditional random fields and extraction patterns.\" Proceedings of EMNLP conference, pp. 355-362, 2005", "links": null }, "BIBREF10": { "ref_id": "b10", "title": "Detecting opinionated sentences by extracting context information", "authors": [ { "first": "X", "middle": [], "last": "Meng", "suffix": "" }, { "first": "H", "middle": [], "last": "Wang", "suffix": "" } ], "year": 2008, "venue": "Proceedings of the Seventh NTCIR Workshop", "volume": "", "issue": "", "pages": "268--271", "other_ids": {}, "num": null, "urls": [], "raw_text": "X. Meng and H. Wang. \"Detecting opinionated sentences by extracting context information.\" Proceedings of the Seventh NTCIR Workshop, pp. 268-271, 2008", "links": null }, "BIBREF11": { "ref_id": "b11", "title": "NLPR at Multilingual Opinion Analysis Task in NTCIR7", "authors": [ { "first": "K", "middle": [], "last": "Liu", "suffix": "" }, { "first": "J", "middle": [], "last": "Zhao", "suffix": "" } ], "year": 2008, "venue": "Proceedings of the Seventh NTCIR Workshop", "volume": "", "issue": "", "pages": "226--231", "other_ids": {}, "num": null, "urls": [], "raw_text": "K. Liu and J. Zhao. \"NLPR at Multilingual Opinion Analysis Task in NTCIR7.\" Proceedings of the Seventh NTCIR Workshop, pp. 226-231, 2008", "links": null }, "BIBREF13": { "ref_id": "b13", "title": "LIBSVM: a library for support vector machines", "authors": [ { "first": "C", "middle": [ "C" ], "last": "Chang", "suffix": "" }, { "first": "C", "middle": [ "J" ], "last": "Lin", "suffix": "" } ], "year": 2001, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "C. C. Chang and C. J. Lin. \"LIBSVM: a library for support vector machines\", 2001", "links": null }, "BIBREF14": { "ref_id": "b14", "title": "YALE: rapid prototyping for complex data mining tasks", "authors": [ { "first": "I", "middle": [], "last": "Mierswa", "suffix": "" }, { "first": "M", "middle": [], "last": "Wurst", "suffix": "" }, { "first": "R", "middle": [], "last": "Klinkenberg", "suffix": "" }, { "first": "M", "middle": [], "last": "Scholz", "suffix": "" }, { "first": "T", "middle": [], "last": "Euler", "suffix": "" } ], "year": 2006, "venue": "Proceedings of the 12th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining", "volume": "", "issue": "", "pages": "935--940", "other_ids": {}, "num": null, "urls": [], "raw_text": "I. Mierswa, M. Wurst, R. Klinkenberg, M. Scholz and T. Euler. \"YALE: rapid prototyping for complex data mining tasks\" In Proceedings of the 12th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pp. 935-940, 2006", "links": null }, "BIBREF15": { "ref_id": "b15", "title": "Conditional random fields: probabilistic models for segmenting and labeling sequence data", "authors": [ { "first": "J", "middle": [], "last": "Lafferty", "suffix": "" }, { "first": "A", "middle": [], "last": "Mccallum", "suffix": "" }, { "first": "F", "middle": [], "last": "Pereira", "suffix": "" } ], "year": 2001, "venue": "Proceedings of International Conference on Machine Learning", "volume": "", "issue": "", "pages": "282--289", "other_ids": {}, "num": null, "urls": [], "raw_text": "J. Lafferty, A. McCallum, F. Pereira. \"Conditional random fields: probabilistic models for segmenting and labeling sequence data\" In Proceedings of International Conference on Machine Learning, pp. 282-289, 2001", "links": null }, "BIBREF16": { "ref_id": "b16", "title": "CRF++: yet another CRF toolkit", "authors": [ { "first": "T", "middle": [], "last": "Kudo", "suffix": "" } ], "year": 2003, "venue": "", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "T. Kudo, \"CRF++: yet another CRF toolkit.\" http://crfpp.sourceforge.net/ , 2003", "links": null }, "BIBREF17": { "ref_id": "b17", "title": "Combining labeled and unlabeled data with co-training", "authors": [ { "first": "A", "middle": [], "last": "Blum", "suffix": "" }, { "first": "T", "middle": [], "last": "Mitchell", "suffix": "" } ], "year": 1998, "venue": "Conference on Computational Learning Theory", "volume": "", "issue": "", "pages": "92--100", "other_ids": {}, "num": null, "urls": [], "raw_text": "A. Blum and T. Mitchell. \"Combining labeled and unlabeled data with co-training.\" Conference on Computational Learning Theory, pp. 92-100, 1998", "links": null }, "BIBREF18": { "ref_id": "b18", "title": "Overview of multilingual opinion analysis task at NTCIR-7", "authors": [ { "first": "Y", "middle": [], "last": "Seki", "suffix": "" }, { "first": "D", "middle": [ "K" ], "last": "Evans", "suffix": "" }, { "first": "L", "middle": [ "W" ], "last": "Ku", "suffix": "" }, { "first": "L", "middle": [], "last": "Sun", "suffix": "" }, { "first": "H", "middle": [ "H" ], "last": "Chen", "suffix": "" }, { "first": "N", "middle": [], "last": "Kando", "suffix": "" } ], "year": 2008, "venue": "Proceedings of the Seventh NTCIR Workshop", "volume": "", "issue": "", "pages": "185--203", "other_ids": {}, "num": null, "urls": [], "raw_text": "Y. Seki, D. K. Evans, L. W. Ku, L. Sun, H. H. Chen and N. Kando. \"Overview of multilingual opinion analysis task at NTCIR-7.\" Proceedings of the Seventh NTCIR Workshop, pp.185-203, 2008", "links": null }, "BIBREF19": { "ref_id": "b19", "title": "Mining opinions from the web: beyond relevance retrieval", "authors": [ { "first": "L", "middle": [ "W" ], "last": "Ku", "suffix": "" }, { "first": "H", "middle": [ "H" ], "last": "Chen", "suffix": "" } ], "year": 2007, "venue": "Journal of American Society for Information Science and Technology", "volume": "", "issue": "", "pages": "1838--1850", "other_ids": {}, "num": null, "urls": [], "raw_text": "L. W. Ku and H. H. Chen. \"Mining opinions from the web: beyond relevance retrieval.\" Journal of American Society for Information Science and Technology, pp. 1838-1850, 2007", "links": null } }, "ref_entries": { "TABREF0": { "type_str": "table", "content": "
\u76ee\u6a19\u70ba\u6253\u68d2\u7403\u3002\u610f\u898b\u6301\u6709\u8005\u901a\u5e38\u6703\u4ee5\u4e00\u6216\u591a\u500b\u8a5e\u7684\u5f62\u5f0f\u51fa\u73fe\u5728\u610f\u898b\u53e5\u4e2d\uff0c\u6211\u5011\u5c07\u9019\u4e9b\u8a5e \u548c Wong[4] \u63d0\u51fa\u7684\u65b9\u6cd5\u662f\u5148\u89e3\u6c7a\u540c\u6307\u6d89\u554f\u984c\uff0c\u518d\u4f7f\u7528\u7d93\u9a57\u6cd5\u5247\u64f7\u53d6\u51fa\u610f\u898b\u6301\u6709\u8005\uff0c\u4f7f (\u4e8c)\u3001\u4f5c\u8005\u610f\u898b\u8fa8\u8b58 \u8868\u4e8c\u3001\u610f\u898b\u6301\u6709\u8005\u6a19\u8a18\u4f7f\u7528\u7684\u7279\u5fb5\u503c \u610f\u898b\u76f8\u95dc\u8cc7\u8a0a\u516d\u7a2e\u985e\u5225\uff0c\u8868\u4e00\u5217\u51fa\u4f5c\u8005\u610f\u898b\u8fa8\u8b58\u6240\u6709\u4f7f\u7528\u7684\u7279\u5fb5\u503c\uff0c\u5176\u4e2d\u8a5e\u6027\u3001\u6587\u53e5\u7d44 \u898b\u6301\u6709\u8005\u7684\u4ee3\u8868\u8a5e\uff0c\u624d\u63d0\u5831\u8a72\u53e5\u7684\u610f\u898b\u6301\u6709\u8005\u662f\u4f5c\u8005\u3002\u4f5c\u8005\u610f\u898b\u8fa8\u8b58\u5224\u65b7\u51fa\u7684\u975e\u4f5c\u8005\u610f NTCIR7 \u6e2c\u8a66\u96c6\u4e2d\u7684\u610f\u898b\u53e5\u5224\u5b9a\u5206\u70ba\u56b4\u683c\u610f\u898b\u53e5\u8207\u5bec\u9b06\u610f\u898b\u53e5\uff0c\u56b4\u683c\u610f\u898b\u53e5\u7684\u689d\u4ef6\u662f\u4e09 \u672c\u5be6\u9a57\u7684\u5be6\u9a57\u8a2d\u5b9a\u5171\u6709\u4e09\u7a2e\uff1a\u5c07\u6a19\u8a18\u6b67\u7570\u7684\u8a9e\u6599\u8996\u70ba\u4f5c\u8005\u610f\u898b\u3001\u5c07\u6a19\u8a18\u6b67\u7570\u7684\u8a9e\u6599\u8996\u70ba \u597d\u3002 \u8868\u4e94\u3001\u7d50\u679c\u5408\u4f75\u7b56\u7565\u5c0d\u610f\u898b\u6301\u6709\u8005\u8fa8\u8b58\u6548\u80fd\u7684\u5f71\u97ff \u4e94\u3001\u7d50\u8ad6\u8207\u672a\u4f86\u5c55\u671b |
(opinion target) \u56db\u500b\u8981\u7d20\u3002\u610f\u898b\u50be\u5411\u63cf\u8ff0\u6b64\u610f\u898b\u662f\u6b63\u9762\u3001\u4e2d\u7acb\u6216\u8ca0\u9762\uff0c\u610f\u898b\u5f37\u5ea6\u63cf\u8ff0\u6b64\u610f\u898b\u7684 \u8a9e\u6c23\u5f37\u5f31\uff0c\u8868\u8ff0\u6b64\u610f\u898b\u7684\u4eba\u6216\u7d44\u7e54\u7a31\u70ba\u610f\u898b\u6301\u6709\u8005\uff0c\u800c\u8a0e\u8ad6\u7684\u4e3b\u984c\u5247\u7a31\u70ba\u8a55\u8ad6\u76ee\u6a19\u3002\u4ee5 \u4f8b\u53e5 1 \u70ba\u4f8b\uff0c\u6b64\u53e5\u7684\u610f\u898b\u50be\u5411\u70ba\u6b63\u9762\u3001\u610f\u898b\u5f37\u5ea6\u70ba\u5f37\u70c8\u3001\u610f\u898b\u6301\u6709\u8005\u70ba\u738b\u5efa\u6c11\u3001\u8a55\u8ad6 \u7a31\u70ba\u610f\u898b\u6301\u6709\u8005\u7684\u4ee3\u8868\u8a5e\uff0c\u4f46\u6709\u6642\u610f\u898b\u6301\u6709\u8005\u4e0d\u6703\u4ee5\u8a5e\u7684\u5f62\u5f0f\u51fa\u73fe\u5728\u610f\u898b\u53e5\u4e2d\uff0c\u4f8b\u5982\u4f8b \u53e5 2 \u662f\u4f5c\u8005\u6839\u64da\u4f8b\u53e5 1\u300c\u738b\u5efa\u6c11\u300d\u7684\u610f\u898b\u63a8\u8ad6\u7684\u610f\u898b\uff0c\u4f8b\u53e5 2 \u7684\u610f\u898b\u6301\u6709\u8005\u70ba\u6587\u7ae0\u4f5c \u8005\u3002 \u4f8b\u53e5 1\uff1a\u738b\u5efa\u6c11\u975e\u5e38\u559c\u6b61\u6253\u68d2\u7403 \u4f8b\u53e5 2\uff1a\u738b\u5efa\u6c11\u61c9\u8a72\u4e5f\u559c\u6b61\u6253\u7db2\u7403 \u5728\u610f\u898b\u63a2\u52d8\u4e2d\uff0c\u610f\u898b\u6301\u6709\u8005\u8fa8\u8b58\u7684\u6280\u8853\u5c0d\u65bc\u4e86\u89e3\u6709\u54ea\u4e9b\u4eba\u6216\u7d44\u7e54\u5728\u8868\u8ff0\u610f\u898b\u3001\u67d0\u500b\u4eba\u6216 \u7d44\u7e54\u5728\u54ea\u4e9b\u8b70\u984c\u4e2d\u767c\u8868\u904e\u610f\u898b\u53ca\u5169\u500b\u4eba\u6216\u7d44\u7e54\u767c\u8868\u904e\u7684\u610f\u898b\u662f\u5426\u76f8\u4f3c\u7b49\u76f8\u95dc\u8cc7\u8a0a\u7279\u5225 \u91cd\u8981\u3002\u610f\u898b\u6301\u6709\u8005\u8fa8\u8b58\u53ef\u61c9\u7528\u65bc\u793e\u7fa4\u7db2\u8def\u5206\u6790\u4e2d\uff0c\u627e\u51fa\u793e\u7fa4\u7db2\u8def\u4e2d\u662f\u5426\u5b58\u5728\u8457\u4e00\u4e9b\u610f\u898b \u9818\u8896\uff0c\u4ed6\u5011\u7684\u610f\u898b\u5e38\u88ab\u5f15\u7528\uff0c\u4e5f\u6703\u5f71\u97ff\u5176\u4ed6\u4f7f\u7528\u8005\u7684\u610f\u898b\u3002\u610f\u898b\u6301\u6709\u8005\u8fa8\u8b58\u4e5f\u53ef\u4ee5\u61c9\u7528 \u5728\u610f\u898b\u554f\u7b54\u7cfb\u7d71\u4e2d\uff0c\u627e\u51fa\u67d0\u4e9b\u610f\u898b\u662f\u7531\u54ea\u4e9b\u610f\u898b\u6301\u6709\u8005\u63d0\u51fa\u7684\uff0c\u4e26\u9032\u800c\u85c9\u7531\u610f\u898b\u6301\u6709\u8005 \u7684\u6b0a\u5a01\u6027\u8207\u53ef\u9760\u5ea6\u4f86\u8f14\u52a9\u5224\u65b7\u7b54\u6848\u7684\u6b0a\u5a01\u6027\u8207\u53ef\u9760\u5ea6\u3002 \u610f\u898b\u6301\u6709\u8005\u8fa8\u8b58\u4e3b\u8981\u6709\u4e09\u5927\u6311\u6230\uff1a\u540c\u6307\u6d89\u89e3\u6790\u3001\u5de2\u72c0\u7d50\u69cb\u53ca\u8655\u7406\u6b67\u7570\u7684\u6a19\u8a18\u3002\u610f\u898b\u6301\u6709 \u8005 \u6709 \u6642 \u6703 \u4ee5 \u4ee3 \u8a5e (Anaphor) \u7684 \u5f62 \u5f0f \u51fa \u73fe \u5728 \u6587 \u53e5 \u4e2d \uff0c \u4e26 \u6307 \u6d89 \u5230 \u524d \u9762 \u7684 \u5148 \u884c \u8a5e (Antecedent) \uff0c\u4f8b\u5982\u4f8b\u53e5 3 \u4e2d\u7684\u300c\u96d9\u65b9\u300d\u5373\u662f\u6307\u6d89\u5230\u300c\u7f8e\u570b\u300d\u8207\u300c\u4e2d\u5171\u300d\u3002 \u4f8b\u53e5 3\uff1a \u64da\u5a92\u9ad4\u5831\u5c0e\uff0c\u7f8e\u570b\u5728\u8207\u4e2d\u5171\u8a0e\u8ad6\u7c3d\u7f72\u505c\u6b62\u4ee5\u6838\u6b66\u76f8\u4e92\u7784\u6e96\u5354\u8b70\u7684\u554f\u984c\uff0c\u8c9d\u80af\u8aaa\uff0c \u96d9\u65b9\u904e\u53bb\u5c31\u66fe\u8a0e\u8ad6\u6b64\u4e8b\uff0c\u524d\u4efb\u570b\u9632\u90e8\u9577\u88f4\u5229\u5728\u4e2d\u5171\u570b\u9632\u90e8\u9577\u9072\u6d69\u7530\u65bc\u4e00\u4e5d\u4e5d\u516d\u5e74 \u5341\u4e8c\u6708\u8a2a\u7f8e\u6642\u5c31\u66fe\u63d0\u8d77\uff0c\u5f8c\u4f86\u96d9\u65b9\u5728\u5176\u4ed6\u6703\u8b70\u4e2d\u4e5f\u66fe\u8a0e\u8ad6\u3002 \u610f\u898b\u53e5\u6709\u6642\u6703\u6709\u5de2\u72c0\u7d50\u69cb (nested structure) \uff0c\u4ee5\u4f8b\u53e5 3 \u70ba\u4f8b\uff0c\u6587\u7ae0\u4f5c\u8005\u5f15\u8ff0\u300c\u5a92\u9ad4\u5831 \u5c0e\u300d\u7684\u5167\u5bb9\uff0c\u300c\u5a92\u9ad4\u5831\u5c0e\u300d\u7684\u5167\u5bb9\u53c8\u5f15\u8ff0\u300c\u8c9d\u80af\u300d\u7684\u767c\u8a00\uff0c\u610f\u898b\u6301\u6709\u8005\u5e38\u6703\u662f\u5b50\u53e5\u7684\u4e3b \u8a5e\uff0c\u5224\u65b7\u610f\u898b\u6301\u6709\u8005\u662f\u54ea\u4e00\u5c64\u7d50\u69cb\u4e2d\u7684\u4e3b\u8a5e\u4e5f\u662f\u610f\u898b\u6301\u6709\u8005\u8fa8\u8b58\u7684\u4e00\u500b\u91cd\u8981\u8b70\u984c\u3002 \u6a19\u8a18\u610f\u898b\u6301\u6709\u8005\u8fa8\u8b58\u6240\u7528\u7684\u8a9e\u6599\u6642\uff0c\u6709\u6642\u6703\u51fa\u73fe\u6a19\u8a18\u6b67\u7570\uff0c\u4e0d\u540c\u7684\u6a19\u8a18\u8005\u53ef\u80fd\u6703\u8a8d\u70ba\u6587 \u53e5\u7684\u610f\u898b\u6301\u6709\u8005\u70ba\u4e0d\u540c\u7684\u5be6\u9ad4\u3002\u4ee5\u4f8b\u53e5 3 \u70ba\u4f8b\uff0c\u4e00\u4f4d\u6a19\u8a18\u8005\u8a8d\u70ba\u610f\u898b\u6301\u6709\u8005\u70ba\u300c\u570b\u9632 \u90e8\u767c\u8a00\u4eba\u8c9d\u80af/\u8c9d\u80af\u300d\uff0c\u53e6\u4e00\u4f4d\u6a19\u8a18\u8005\u537b\u8a8d\u70ba\u610f\u898b\u6301\u6709\u8005\u70ba\u300e\u7f8e\u300c\u4e2d\u300d/\u96d9\u65b9\u300f\uff0c\u5f9e\u6587\u53e5 \u5167\u5bb9\u4f86\u770b\uff0c\u610f\u898b\u6301\u6709\u8005\u70ba\u300c\u570b\u9632\u90e8\u767c\u8a00\u4eba\u8c9d\u80af/\u8c9d\u80af\u300d\uff0c\u4f46\u6df1\u7a76\u80cc\u5f8c\u7684\u610f\u7fa9\uff0c\u8c9d\u80af\u662f\u8f49 \u8ff0\u300e\u7f8e\u300c\u4e2d\u300d/\u96d9\u65b9\u300f\u7684\u610f\u898b\uff0c\u5169\u7a2e\u8aaa\u6cd5\u90fd\u6c92\u932f\uff0c\u7aef\u770b\u6a19\u8a18\u8005\u7684\u8a8d\u77e5\uff0c\u4e5f\u56e0\u6b64\u610f\u898b\u6301\u6709 \u8005\u53ef\u80fd\u88ab\u591a\u500b\u6a19\u8a18\u8005\u6a19\u8a18\u51fa\u4e0d\u540c\u7684\u7b54\u6848\u3002\uff0c\u5982\u4f55\u5229\u7528\u6a19\u8a18\u6b67\u7570\u7684\u8a9e\u6599\u4e5f\u662f\u610f\u898b\u6301\u6709\u8005\u8fa8 \u8b58\u7684\u4e00\u5927\u6311\u6230\u3002 \u4e8c\u3001\u76f8\u95dc\u7814\u7a76 Pang \u548c Lee[2] \u6574\u7406\u51fa\u610f\u898b\u63a2\u52d8\u9818\u57df\u4e2d\u91cd\u8981\u7684\u7814\u7a76\uff0c\u610f\u898b\u6301\u6709\u8005\u8fa8\u8b58\u7684\u7814\u7a76\u525b\u958b\u59cb\u8d77 \u6b65\uff0c\u7814\u7a76\u5718\u968a\u4f7f\u7528\u7684\u65b9\u6cd5\u4e3b\u8981\u53ef\u5206\u70ba\u4ee5\u7d93\u9a57\u6cd5\u5247 (heuristic rule) \u70ba\u57fa\u790e\u8207\u4ee5\u6a5f\u5668\u5b78\u7fd2\u70ba \u57fa\u790e\u5169\u7a2e\u3002 (\u4e00)\u3001\u4ee5\u7d93\u9a57\u6cd5\u5247\u70ba\u57fa\u790e\u7684\u65b9\u6cd5 \u4ee5\u7d93\u9a57\u6cd5\u5247\u70ba\u57fa\u790e\u7684\u65b9\u6cd5\u4e2d\uff0cYohei \u7b49\u4eba[3] \u5148\u4f7f\u7528\u540d\u8a5e\u7247\u8a9e\u8207\u8a9e\u6cd5\u7279\u5fb5\u503c\uff0c\u900f\u904e\u652f\u63f4 \u5411\u91cf\u6a5f\uff0c\u5c07\u610f\u898b\u6301\u6709\u8005\u5206\u70ba\u6587\u7ae0\u4f5c\u8005\u8207\u975e\u6587\u7ae0\u4f5c\u8005\uff0c\u63a5\u8457\u518d\u900f\u904e\u8a9e\u6cd5\u898f\u5247\uff0c\u9078\u51fa\u6700\u6709\u53ef \u80fd\u7684\u5177\u540d\u5be6\u9ad4\uff0c\u505a\u70ba\u7b54\u6848\u7684\u610f\u898b\u6301\u6709\u8005\uff0c\u4ed6\u5011\u4e3b\u8981\u5c08\u6ce8\u65bc\u8655\u7406\u82f1\u6587\u8207\u65e5\u6587\u7684\u8a9e\u6599\u3002Xu \u7528\u7684\u898f\u5247\u8207\u6a19\u9ede\u7b26\u865f\u3001\u9023\u63a5\u8a5e\u3001\u5b57\u9996 (prefix) \u3001\u5b57\u5c3e (suffix) \u8207\u8868\u8ff0\u95dc\u9375\u5b57\u76f8\u95dc\uff0cXu \u548c Wong \u7684\u65b9\u6cd5\u662f\u76ee\u524d\u4e2d\u6587\u610f\u898b\u6301\u6709\u8005\u8fa8\u8b58\u4e2d\u6548\u80fd\u6700\u4f73\u7684\uff0c\u5728 NTCIR7 \u591a\u8a9e\u610f\u898b\u5206\u6790 \u8a55\u6bd4\u9805\u76ee\u7684\u7e41\u9ad4\u4e2d\u6587\u8a9e\u6599\u4e0a\uff0c F \u503c\u53ef\u9054\u5230 0.825\u3002 (\u4e8c)\u3001\u4ee5\u6a5f\u5668\u5b78\u7fd2\u70ba\u57fa\u790e\u7684\u65b9\u6cd5 \u4ee5\u6a5f\u5668\u5b78\u7fd2\u70ba\u57fa\u790e\u7684\u65b9\u6cd5\u4e2d\uff0c\u8a31\u591a\u7814\u7a76\u5718\u968a\u4f7f\u7528\u6700\u5927\u71b5\u6cd5 (maximum entropy) \u3001\u652f\u63f4\u5411 \u91cf\u6a5f\u6f14\u7b97\u6cd5 (support vector machine algorithm) \u8207\u689d\u4ef6\u96a8\u6a5f\u57df\u6a21\u578b (conditional random field model) \u7b49\u5206\u985e\u5668\u89e3\u6c7a\u6b64\u554f\u984c\u3002 Kim \u548c Hovy[5] \u4ee5\u6700\u5927\u71b5\u6cd5\u5f9e\u65b0\u805e\u8a9e\u6599\u7684\u6587\u53e5\u64f7\u53d6\u51fa\u610f\u898b\u6301\u6709\u8005\u8207\u610f\u898b\u8a55\u8ad6\u76ee\u6a19\uff0c\u4ed6 \u5011\u5148\u627e\u51fa\u610f\u898b\u8a5e (opinion words) \u8207\u9032\u884c\u8a9e\u610f\u89d2\u8272\u6a19\u6ce8 (semantic role labeling) \uff0c\u518d\u627e\u51fa \u4ee3\u8868\u610f\u898b\u6301\u6709\u8005\u8207\u610f\u898b\u8a55\u8ad6\u76ee\u6a19\u7684\u8a9e\u610f\u89d2\u8272\u3002 dependency relation) \u7279\u5fb5\u503c\uff0c\u900f\u904e\u689d\u4ef6\u96a8\u6a5f\u57df\u6a21\u578b\u6a19 \u8a18\u51fa\u6700\u6709\u53ef\u80fd\u7684\u610f\u898b\u6301\u6709\u8005\u3002\u76f8\u5f62\u4e4b\u4e0b\uff0c Meng \u548c Wang[11] \u4f7f\u7528\u8a5e\u5f59\u3001\u8a5e\u6027\u53ca\u8868\u8ff0\u95dc \u9375\u5b57 (operator) \u7279\u5fb5\u503c\uff0c\u800c Liu \u548c Zhao[12] \u5247\u4f7f\u7528\u8a5e\u6027\u3001\u8a9e\u610f\u3001\u4f9d\u5b58\u95dc\u4fc2\u3001\u4f4d\u7f6e (position) \u53ca\u524d\u5f8c\u6587 (contextual) \u7279\u5fb5\u503c\uff0c\u900f\u904e\u689d\u4ef6\u96a8\u6a5f\u57df\u6a21\u578b\u6a19\u8a18\u51fa\u6700\u6709\u53ef\u80fd\u7684\u610f\u898b\u6301\u6709\u8005\u3002 \u4e09\u3001\u610f\u898b\u6301\u6709\u8005\u8fa8\u8b58\u65b9\u6cd5 \u672c\u7814\u7a76\u5c07\u610f\u898b\u6301\u6709\u8005\u8fa8\u8b58\u5206\u70ba\u4f5c\u8005\u610f\u898b\u8fa8\u8b58\u53ca\u610f\u898b\u6301\u6709\u8005\u6a19\u8a18\u5169\u500b\u4e3b\u8981\u5de5\u4f5c\u3002\u672c\u7814\u7a76\u63d0 \u51fa\u7684\u6d41\u7a0b\u5305\u62ec\u524d\u7f6e\u8655\u7406\u7a0b\u5e8f\u3001\u4f5c\u8005\u610f\u898b\u8fa8\u8b58\u7a0b\u5e8f\u3001\u610f\u898b\u6301\u6709\u8005\u6a19\u8a18\u7a0b\u5e8f\u3001\u5f8c\u7f6e\u8655\u7406\u7a0b\u5e8f \u53ca\u7d50\u679c\u5408\u4f75\u7a0b\u5e8f\u4e94\u500b\u90e8\u4efd\u3002 (\u4e00)\u3001\u91dd\u5c0d\u65b7\u8a5e\u8207\u8a5e\u6027\u6a19\u8a18\u7684\u7279\u6b8a\u8655\u7406 \u524d\u8655\u7406\u7a0b\u5e8f\u5305\u62ec\u65b7\u8a5e\u3001\u8a5e\u6027\u6a19\u8a18\u3001\u5177\u540d\u5be6\u9ad4\u8fa8\u8b58\u53ca\u7279\u5fb5\u503c\u64f7\u53d6\u3002\u672c\u5be6\u9a57\u4f7f\u7528\u7684\u662f\u7f85[13] \u7814\u767c\u7684\u65b7\u8a5e\u53ca\u8a5e\u6027\u6a19\u8a18\u7cfb\u7d71\uff0c\u70ba\u4e86\u80fd\u5920\u66f4\u6e96\u78ba\u7684\u65b7\u51fa\u8207\u610f\u898b\u6301\u6709\u8005\u76f8\u95dc\u7684\u5177\u540d\u5be6\u9ad4\uff0c\u6211 \u5011\u4fee\u6539\u65b7\u8a5e\u7cfb\u7d71\u7684\u4eba\u540d\u6a21\u7d44\u4e26\u5f15\u5165\u5b57\u5178\u8cc7\u8a0a\u3002\u6211\u5011\u767c\u73fe\u5916\u570b\u4eba\u540d\u5bb9\u6613\u51fa\u73fe\u65b7\u8a5e\u932f\u8aa4\uff0c\u6240 \u4ee5\u6211\u5011\u8457\u624b\u4fee\u6539\u65b7\u8a5e\u53ca\u8a5e\u6027\u6a19\u8a18\u7cfb\u7d71\u7684\u4eba\u540d\u6a21\u7d44\uff0c\u4f86\u8655\u7406\u65e5\u6587\u59d3\u540d\u9577\u5ea6\u8207\u4e2d\u6587\u59d3\u540d\u9577\u5ea6 \u4e0d\u540c\u7684\u554f\u984c\uff0c\u6211\u5011\u5728\u539f\u672c\u4eba\u540d\u6a21\u7d44\u4f7f\u7528\u7684\u59d3\u6c0f\u5217\u8868\u4e2d\u52a0\u5165\u65e5\u672c\u5e38\u898b\u59d3\u6c0f\uff0c\u540d\u5b57\u9577\u5ea6\u7684\u9650 \u5236\u4e5f\u5f9e\u5169\u500b\u5b57\u653e\u5bec\u70ba\u4e09\u500b\u5b57\uff0c\u4f7f\u5f97\u7cfb\u7d71\u80fd\u6b63\u78ba\u65b7\u51fa\u5982\u300c\u9ad8\u6751\u6b63\u5f65\u300d \u3001 \u300c\u5152\u7389\u6e90\u592a\u90ce\u300d \u3001 \u300c\u9234 \u6728\u300d\u7b49\u65e5\u672c\u4eba\u540d\u3002 \u6211\u5011\u53e6\u5916\u52a0\u5165\u4e86\u8077\u7a31\u540d\u3001\u8077\u696d\u540d\u3001\u65e5\u672c\u5e38\u898b\u59d3\u6c0f\u53ca\u53f0\u7063\u516c\u71df\u4f01\u696d\u5217\u8868\u7b49\u5b57\u5178\u3002\u672c\u7814\u7a76\u7684 \u5177\u540d\u5be6\u9ad4\u8fa8\u8b58\u662f\u4f7f\u7528\u67e5\u8a62\u8a5e\u5178\u7684\u65b9\u6cd5\uff0c\u5c07\u4eba\u540d\u3001\u5730\u540d\u53ca\u7d44\u7e54\u540d\u5206\u5225\u6a19\u4e0a\u6a19\u7c64\u3002 \u4f5c\u8005\u610f\u898b\u8fa8\u8b58\u7684\u76ee\u7684\u662f\u8fa8\u8b58\u610f\u898b\u53e5\u4e4b\u610f\u898b\u6301\u6709\u8005\u662f\u5426\u70ba\u6587\u7ae0\u4f5c\u8005\u3002\u672c\u7814\u7a76\u628a\u4f5c\u8005\u610f\u898b\u8fa8 \u8b58\u7684\u554f\u984c\u8996\u70ba\u4e8c\u5143\u5206\u985e\u554f\u984c (binary classification problem)\uff0c\u4f7f\u7528\u652f\u63f4\u5411\u91cf\u6a5f\u4f86\u8655\u7406\uff0c\u5be6 \u4f5c\u4e0a\u4f7f\u7528\u7684\u5957\u88dd\u8edf\u9ad4\u662f Chang \u548c Lin[14] \u958b\u767c\u7684 LIBSVM \u3002 \u8868\u4e00\u3001\u4f5c\u8005\u610f\u898b\u8fa8\u8b58\u4f7f\u7528\u7684\u7279\u5fb5\u503c \u7279\u5fb5\u503c\u985e\u5225 \u7279\u5fb5\u503c\u4ee3\u865f \u7279\u5fb5\u503c\u63cf\u8ff0 fHasI \u672c\u53e5\u6709\u6c92\u6709\u300c\u6211\u300d fHasWe \u672c\u53e5\u6709\u6c92\u6709\u300c\u6211\u5011\u300d fNumI \u672c\u53e5\u6709\u5e7e\u500b\u300c\u6211\u300d \u8a5e\u5f59\u76f8\u95dc\u8cc7\u8a0a fNumWe \u672c\u53e5\u6709\u5e7e\u500b\u300c\u6211\u5011\u300d fHasPronoun \u672c\u53e5\u6709\u6c92\u6709\u4ee3\u540d\u8a5e fHasManPronoun \u672c\u53e5\u6709\u6c92\u6709\u4eba\u7a31\u4ee3\u540d\u8a5e fNumPronoun \u672c\u53e5\u6709\u5e7e\u500b\u4ee3\u540d\u8a5e \u8a5e\u6027\u76f8\u95dc\u8cc7\u8a0a fNumManPronoun \u672c\u53e5\u6709\u5e7e\u500b\u4eba\u7a31\u4ee3\u540d\u8a5e fHasPer \u672c\u53e5\u6709\u6c92\u6709\u4eba\u540d\u8a5e fHasLoc \u672c\u53e5\u6709\u6c92\u6709\u5730\u540d\u8a5e fHasOrg \u672c\u53e5\u6709\u6c92\u6709\u7d44\u7e54\u540d\u8a5e fHasNa \u672c\u53e5\u6709\u6c92\u6709\u666e\u901a\u540d\u8a5e fHasNb \u672c\u53e5\u6709\u6c92\u6709\u5c08\u6709\u540d\u8a5e fHasNc \u672c\u53e5\u6709\u6c92\u6709\u5730\u65b9\u540d\u8a5e fNumLoc \u672c\u53e5\u6709\u5e7e\u500b\u5730\u540d\u8a5e fNumOrg \u672c\u53e5\u6709\u5e7e\u500b\u7d44\u7e54\u540d\u8a5e fNumPer \u672c\u53e5\u6709\u5e7e\u500b\u4eba\u540d\u8a5e fNumNa \u672c\u53e5\u6709\u5e7e\u500b\u666e\u901a\u540d\u8a5e fNumNb \u672c\u53e5\u6709\u5e7e\u500b\u5c08\u6709\u540d\u8a5e \u5177\u540d\u5be6\u9ad4\u8cc7\u8a0a fNumNc \u672c\u53e5\u6709\u5e7e\u500b\u5730\u65b9\u540d\u8a5e fHasExclamation \u672c\u53e5\u6709\u6c92\u6709\u9a5a\u5606\u865f\uff0c\u4f8b\u5982\uff1a\u300c\uff01\u300d\u6216\u300c!\u300d fHasQuestion \u672c\u53e5\u6709\u6c92\u6709\u554f\u865f\uff0c\u4f8b\u5982\uff1a\u300c\uff1f\u300d\u6216\u300c?\u300d fHasColon \u672c\u53e5\u6709\u6c92\u6709\u5192\u865f\uff0c\u4f8b\u5982\uff1a\u300c\uff1a\u300d\u6216\u300c:\u300d fHasLeftQuotation \u672c\u53e5\u6709\u6c92\u6709\u4e0a\u5f15\u865f\uff0c\u4f8b\u5982\uff1a\u300e\u300c\u300f\u6216\u300c\u3010\u300d \u6a19\u9ede\u7b26\u865f\u8cc7\u8a0a fHasRightQuotation \u672c\u53e5\u6709\u6c92\u6709\u4e0b\u5f15\u865f\uff0c\u4f8b\u5982\uff1a\u300e\u300d\u300f \u6216\u300c\u3011\u300d fNumChar \u672c\u53e5\u6709\u5e7e\u500b\u5b57 fNumWord \u672c\u53e5\u6709\u5e7e\u500b\u8a5e \u6587\u53e5\u7d44\u6210\u8cc7\u8a0a fNumSubsen \u672c\u53e5\u6709\u5e7e\u500b\u5b50\u53e5 \u610f\u898b\u76f8\u95dc\u8cc7\u8a0a fOperator \u672c\u53e5\u6709\u6c92\u6709\u67d0\u500b\u8868\u8ff0\u95dc\u9375\u5b57 \u7279\u5fb5\u503c\u985e\u5225 \u7279\u5fb5\u503c\u4ee3\u865f \u7279\u5fb5\u503c\u63cf\u8ff0 \u8a5e\u5f59\u76f8\u95dc\u8cc7\u8a0a fWord \u672c\u8a5e fPOS \u6210\u3001\u610f\u898b\u76f8\u95dc\u8cc7\u8a0a\u53ca\u6a19\u9ede\u7b26\u865f\u4e2d\u7684\u9a5a\u6b4e\u865f\u76f8\u95dc\u7279\u5fb5\u503c\u70ba\u672c\u7814\u7a76\u9996\u5148\u63d0\u51fa\u7684\u3002 \u6587\u53e5\u7d44\u6210\u76f8\u95dc\u7279\u5fb5\u503c\u5305\u62ec\u4f5c\u8005\u610f\u898b\u7684\u6587\u53e5\u5728\u6587\u53e5\u9577\u5ea6\u4e0a\u7684\u8cc7\u8a0a\u3002\u610f\u898b\u76f8\u95dc\u8cc7\u8a0a\u5247\u5305\u542b\u8868 \u8ff0\u95dc\u9375\u5b57 (operator) \uff0c\u5e0c\u671b\u4e86\u89e3\u4f5c\u8005\u767c\u8868\u7684\u610f\u898b\u4e2d\u662f\u5426\u8f03\u5e38\u4f7f\u7528\u7279\u5b9a\u7684\u8868\u8ff0\u95dc\u9375\u5b57\u3002\u8868 \u898b\u4e26\u6c92\u6709\u6a19\u8a18\u51fa\u610f\u898b\u6301\u6709\u8005\u7684\u4ee3\u8868\u8a5e\u70ba\u4f55\uff0c\u6240\u4ee5\u6211\u5011\u900f\u904e\u610f\u898b\u6301\u6709\u8005\u6a19\u8a18\u7a0b\u5e8f\u5f37\u5236\u6a19\u8a18 \u51fa\u4ee3\u8868\u8a5e\uff0c\u4e5f\u5c31\u662f\u5f9e\u6240\u6709\u8a5e\u4e2d\u627e\u51fa\u6700\u6709\u53ef\u80fd\u4ee3\u8868\u610f\u898b\u6301\u6709\u8005\u7684\u8a5e\u3002 \u900f\u904e\u524d\u7f6e\u8655\u7406\u3001\u4f5c\u8005\u610f\u898b\u8fa8\u8b58\u3001\u610f\u898b\u6301\u6709\u8005\u6a19\u8a18\u3001\u5f8c\u7f6e\u8655\u7406\u53ca\u7d50\u679c\u5408\u4f75\u4e94\u500b\u7a0b\u5e8f\uff0c\u6211\u5011 \u4f4d\u6a19\u8a18\u8005\u90fd\u5c07\u6b64\u53e5\u6a19\u8a18\u70ba\u610f\u898b\u53e5\uff0c\u5bec\u9b06\u610f\u898b\u53e5\u7684\u689d\u4ef6\u5247\u662f\u4e09\u4f4d\u6a19\u8a18\u8005\u4e2d\uff0c\u6709\u5169\u4f4d\u4ee5\u4e0a\u5c07 \u6b64\u53e5\u6a19\u8a18\u70ba\u610f\u898b\u53e5\u3002 \u672c\u5be6\u9a57\u5224\u65b7\u6b64\u53e5\u7684\u610f\u898b\u6301\u6709\u8005\u662f\u4e0d\u662f\u6587\u7ae0\u4f5c\u8005\uff0c\u4e5f\u5c31\u662f\u6b64\u53e5\u662f\u4e0d\u662f\u4f5c\u8005\u610f\u898b\uff0c\u6211\u5011\u4f7f\u7528 \u975e\u4f5c\u8005\u610f\u898b\u3001\u53ca\u6a19\u8a18\u6b67\u7570\u7684\u8a9e\u6599\u4e0d\u5217\u5165\u8a13\u7df4\u8a9e\u6599\u3002\u8868\u4e09\u986f\u793a\u5728\u9019\u4e09\u7a2e\u5be6\u9a57\u8a2d\u5b9a\u4e0b\uff0c\u6a19\u8a18 \u6b67\u7570\u7684\u8a9e\u6599\u5c0d\u4f5c\u8005\u610f\u898b\u8fa8\u8b58\u6548\u80fd\u7684\u5f71\u97ff\u3002 \u5be6\u9a57\u7d50\u679c\u986f\u793a\uff0c\u8996\u70ba\u4f5c\u8005\u610f\u898b\u7684\u8a2d\u5b9a\u6548\u80fd\u6700\u4f73\uff1a F \u503c\u70ba 79.98%\uff0c\u5176\u6b21\u70ba\u8996\u70ba\u975e\u4f5c\u8005\u610f \u7d50\u679c\u5408\u4f75\u7b56\u7565 \u6b63\u78ba\u6578 \u932f\u8aa4\u6578 set F \u503c \u5bec\u9b06\u610f\u898b\u53e5\u90e8\u4efd\u7684\u8a55\u4f30\u4e2d\uff0c\u672c\u7814\u7a76\u80fd\u9054\u5230\u7684\u6700\u4f73\u6548\u80fd\u662f set F \u503c 67.83%\u3002\u5728\u6a19\u8a18\u7d50\u679c \u932f\u8aa4\u7684\u5be6\u4f8b\u4e2d\uff0c\u6709 13% (254 \u53e5) \u662f\u6b63\u78ba\u7b54\u6848\u7684\u610f\u898b\u6301\u6709\u8005\u70ba\u55ae\u8a5e\u6216\u8a5e\u7d44\uff0c\u4f46\u7cfb\u7d71\u6a19\u8a18 \u51fa\u4e4b\u55ae\u8a5e\u6216\u8a5e\u7d44\u8207\u6b63\u78ba\u7b54\u6848\u4e0d\u7b26\uff0c\u4ee5\u4e0b\u5c07\u5206\u6790\u6a19\u8a18\u7d50\u679c\u932f\u8aa4\u7684\u5be6\u4f8b\uff0c\u4e26\u63d0\u51fa\u89e3\u6c7a\u7684\u65b9 \u7b56\u7565 A 829 340 \u56b4\u683c \u672c\u7814\u7a76\u63d0\u51fa\u4e00\u500b\u4ee5\u6a5f\u5668\u5b78\u7fd2\u65b9\u6cd5\u70ba\u57fa\u790e\u7684\u610f\u898b\u6301\u6709\u8005\u8fa8\u8b58\u65b9\u6cd5\uff0c\u4e26\u4e14\u4f9d\u7167\u6b64\u65b9\u6cd5\u5be6\u4f5c\u51fa 70.92% \u610f\u898b\u53e5 \u7b56\u7565 B 858 310 73.40% \u4e00\u5957\u610f\u898b\u6301\u6709\u8005\u8fa8\u8b58\u7cfb\u7d71\u3002 \u672c\u8a5e\u7684\u8a5e\u6027 fIsPronoun \u672c\u8a5e\u662f\u4e0d\u662f\u4ee3\u540d\u8a5e \u8a5e\u6027\u76f8\u95dc\u8cc7\u8a0a fIsNoun \u672c\u8a5e\u662f\u4e0d\u662f\u540d\u8a5e fIsPer \u672c\u8a5e\u662f\u4e0d\u662f\u4eba\u540d fIsLoc \u672c\u8a5e\u662f\u4e0d\u662f\u5730\u540d \u5177\u540d\u5be6\u9ad4\u8cc7\u8a0a fIsOrg \u672c\u8a5e\u662f\u4e0d\u662f\u7d44\u7e54\u540d fAfterParen \u672c\u8a5e\u662f\u5426\u5728\u4e0b\u5f15\u865f\u4e4b\u5f8c\u5169\u8a5e\uff0c\u4f8b\u5982\uff1a \u300e\u300d\u300f \u6216\u300c\u3011\u300d \u6a19\u9ede\u7b26\u865f\u8cc7\u8a0a fBeforeColon \u672c\u8a5e\u662f\u5426\u5728\u5192\u865f\u4e4b\u524d\u5169\u8a5e\uff0c\u4f8b\u5982\uff1a\u300c\uff1a\u300d\u6216\u300c:\u300d fNearSenStart \u672c\u8a5e\u662f\u5426\u9760\u8fd1\u53e5\u9996 fSenLen \u672c\u8a5e\u6240\u5728\u53e5\u4e2d\u7684\u8a5e\u6578 fWordOrder \u672c\u8a5e\u5728\u53e5\u4e2d\u7684\u8a5e\u5e8f \u6587\u53e5\u7d44\u6210\u8cc7\u8a0a fWordPerc \u672c\u8a5e\u5728\u53e5\u4e2d\u8a5e\u5e8f\u7684\u767e\u5206\u6bd4 fNearVerb \u540c\u53e5\u4e2d\u6700\u9760\u8fd1\u672c\u8a5e\u7684\u52d5\u8a5e fNearVerbPOS \u540c\u53e5\u4e2d\u6700\u9760\u8fd1\u672c\u8a5e\u7684\u52d5\u8a5e\u8a5e\u6027 \u524d\u5f8c\u6587 \u76f8\u95dc\u8cc7\u8a0a fDistNearVerb \u540c\u53e5\u4e2d\u672c\u8a5e\u5230\u52d5\u8a5e\u7684\u6700\u77ed\u8ddd\u96e2 fHasOpKW \u540c\u53e5\u4e2d\u6709\u6c92\u6709\u8868\u8ff0\u95dc\u9375\u5b57 fHasPosKW \u540c\u53e5\u4e2d\u6709\u6c92\u6709\u6b63\u9762\u610f\u898b\u8a5e fHasNegKW \u540c\u53e5\u4e2d\u6709\u6c92\u6709\u8ca0\u9762\u610f\u898b\u8a5e fHasNeuKW \u540c\u53e5\u4e2d\u6709\u6c92\u6709\u4e2d\u7acb\u610f\u898b\u8a5e fNearOpKW \u540c\u53e5\u4e2d\u6700\u9760\u8fd1\u672c\u8a5e\u7684\u8868\u8ff0\u95dc\u9375\u5b57 fNearPosKW \u540c\u53e5\u4e2d\u6700\u9760\u8fd1\u672c\u8a5e\u7684\u6b63\u9762\u610f\u898b\u8a5e fNearNegKW \u540c\u53e5\u4e2d\u6700\u9760\u8fd1\u672c\u8a5e\u7684\u8ca0\u9762\u610f\u898b\u8a5e fNearNeuKW \u540c\u53e5\u4e2d\u6700\u9760\u8fd1\u672c\u8a5e\u7684\u4e2d\u7acb\u610f\u898b\u8a5e fNearOpKWPOS \u540c\u53e5\u4e2d\u6700\u9760\u8fd1\u672c\u8a5e\u7684\u8868\u8ff0\u95dc\u9375\u5b57\u7684\u8a5e\u6027 fNearPosKWPOS \u540c\u53e5\u4e2d\u6700\u9760\u8fd1\u672c\u8a5e\u7684\u6b63\u9762\u610f\u898b\u8a5e\u7684\u8a5e\u6027 fNearNegKWPOS \u540c\u53e5\u4e2d\u6700\u9760\u8fd1\u672c\u8a5e\u7684\u8ca0\u9762\u610f\u898b\u8a5e\u7684\u8a5e\u6027 fNearNeuKWPOS \u540c\u53e5\u4e2d\u6700\u9760\u8fd1\u672c\u8a5e\u7684\u4e2d\u7acb\u610f\u898b\u8a5e\u7684\u8a5e\u6027 fDistOpKW \u540c\u53e5\u4e2d\u672c\u8a5e\u5230\u8868\u8ff0\u95dc\u9375\u5b57\u7684\u6700\u77ed\u8ddd\u96e2 fDistPosKW \u540c\u53e5\u4e2d\u672c\u8a5e\u5230\u6b63\u9762\u610f\u898b\u8a5e\u7684\u6700\u77ed\u8ddd\u96e2 fDistNegKW \u540c\u53e5\u4e2d\u672c\u8a5e\u5230\u8ca0\u9762\u610f\u898b\u8a5e\u7684\u6700\u77ed\u8ddd\u96e2 \u610f\u898b\u76f8\u95dc\u8cc7\u8a0a fDistNeuKW \u540c\u53e5\u4e2d\u672c\u8a5e\u5230\u4e2d\u7acb\u610f\u898b\u8a5e\u7684\u6700\u77ed\u8ddd\u96e2 \u4f5c\u8005\u610f\u898b\u8fa8\u8b58\u4f7f\u7528\u7684\u7279\u5fb5\u503c\u4e3b\u8981\u53ef\u5206\u70ba\u8a5e\u5f59\u3001\u8a5e\u6027\u3001\u5177\u540d\u5be6\u9ad4\u3001\u6a19\u9ede\u7b26\u865f\u3001\u6587\u53e5\u7d44\u6210\u53ca \u8ff0\u95dc\u9375\u5b57\u662f\u7528\u4f86\u8868\u9054\u610f\u898b\u7684\u8a5e\uff0c\u901a\u5e38\u70ba\u52d5\u8a5e\uff0c\u4f8b\u5982\uff1a \u300c\u8aaa\u300d \u3001 \u300c\u5831\u5c0e\u300d\u53ca\u300c\u4e3b\u5f35\u300d\u7b49\u3002\u6a19 \u9ede\u7b26\u865f\u4e2d\u7684\u9a5a\u5606\u865f\u5e38\u7528\u4f86\u8868\u9054\u500b\u4eba\u60c5\u7dd2\u7684\u53cd\u61c9\u3002 (\u4e09)\u3001\u610f\u898b\u6301\u6709\u8005\u6a19\u8a18 \u610f\u898b\u6301\u6709\u8005\u6a19\u8a18\u7684\u76ee\u7684\u662f\u8fa8\u8b58\u51fa\u610f\u898b\u6301\u6709\u8005\u7684\u4ee3\u8868\u8a5e\uff0c\u672c\u7814\u7a76\u5c07\u610f\u898b\u6301\u6709\u8005\u6a19\u8a18\u554f\u984c\u8996 \u70ba\u4e8c\u5143\u5206\u985e\u554f\u984c\uff0c\u8a66\u8457\u4f7f\u7528\u6c7a\u7b56\u6a39\u6f14\u7b97\u6cd5 (Decision Tree Algorithm) \u89e3\u6c7a\uff0c\u5be6\u4f5c\u4e0a\u4f7f\u7528 \u7684\u5957\u88dd\u8edf\u9ad4\u662f Mierswa \u7b49\u4eba[15] Lafferty \u7b49\u4eba[16] \u63d0\u51fa\u7684\u689d\u4ef6\u96a8\u6a5f\u57df\u6a21\u578b\u4f86\u6a19\u8a18\u51fa\u610f\u898b\u8a5e\u6709\u8005\u6240\u6db5\u84cb\u7684\u8a5e\u5f59\uff0c\u5be6\u4f5c\u4e0a\u4f7f \u7528\u7684\u5957\u88dd\u8edf\u9ad4\u662f Kudo[17] \u958b\u767c\u7684 CRF++ \u3002 \u610f\u898b\u6301\u6709\u8005\u6a19\u8a18\u4f7f\u7528\u7684\u7279\u5fb5\u503c\u4e3b\u8981\u53ef\u5206\u70ba\u4e3b\u8981\u53ef\u5206\u70ba\u8a5e\u5f59\u3001\u8a5e\u6027\u3001\u5177\u540d\u5be6\u9ad4\u3001\u6a19\u9ede\u7b26\u865f\u3001 \u6587\u53e5\u7d44\u6210\u3001\u524d\u5f8c\u6587\u53ca\u610f\u898b\u76f8\u95dc\u8cc7\u8a0a\u4e03\u7a2e\u985e\u5225\uff0c\u8868\u4e8c\u5217\u51fa\u610f\u898b\u6301\u6709\u8005\u6a19\u8a18\u6240\u6709\u4f7f\u7528\u7684\u7279\u5fb5 \u503c\uff0c\u5176\u4e2d\u524d\u5f8c\u6587\u3001\u610f\u898b\u76f8\u95dc\u8cc7\u8a0a\u53ca\u8a5e\u6027\u4e2d\u7684\u672c\u8a5e\u662f\u4e0d\u662f\u4ee3\u540d\u8a5e\u6216\u540d\u8a5e\u7279\u5fb5\u503c\u70ba\u672c\u7814\u7a76\u9996 \u5148\u63d0\u51fa\u7684\u3002 \u524d\u5f8c\u6587\u76f8\u95dc\u8cc7\u8a0a\u7684\u7279\u5fb5\u503c\u8003\u616e\u610f\u898b\u6301\u6709\u8005\u7684\u4ee3\u8868\u8a5e\u662f\u5426\u6703\u8f03\u5e38\u8207\u67d0\u4e9b\u52d5\u8a5e\u642d\u914d\u4f7f\u7528\u3002\u610f \u898b\u76f8\u95dc\u8cc7\u8a0a\u7684\u7279\u5fb5\u503c\u5247\u5305\u542b\u53e5\u4e2d\u8207\u610f\u898b\u95dc\u9375\u5b57\uff1a\u8868\u8ff0\u95dc\u9375\u5b57\u3001\u6b63\u9762\u610f\u898b\u8a5e\u3001\u8ca0\u9762\u610f\u898b\u8a5e \u53ca\u4e2d\u7acb\u610f\u898b\u8a5e\u76f8\u95dc\u7684\u7279\u5fb5\u503c\u3002\u6b63\u9762\u610f\u898b\u8a5e\u70ba\u8868\u9054\u6b63\u9762\u610f\u898b\u7acb\u5834\u7684\u8a5e\uff1a\u5982 \u300c\u540c\u610f\u300d \u3001 \u300c\u76f8\u4fe1\u300d \u3001 \u300c\u6210\u529f\u300d\u7b49\u3002\u8ca0\u9762\u610f\u898b\u8a5e\u70ba\u8868\u9054\u8ca0\u9762\u610f\u898b\u7acb\u5834\u7684\u8a5e\uff1a\u5982\u300c\u4e0d\u6703\u300d \u3001 \u300c\u53cd\u5c0d\u300d \u3001 \u300c\u6307\u63a7\u300d\u7b49\u3002 \u4e2d\u7acb\u610f\u898b\u8a5e\u70ba\u8868\u9054\u4e2d\u7acb\u610f\u898b\u7acb\u5834\u7684\u8a5e\uff1a\u5982\u300c\u672a\u7f6e\u8a55\u300d \u3001 \u300c\u5169\u96e3\u300d \u3001 \u300c\u53ef\u80fd\u300d\u7b49\u3002 NTCIR7 \u591a\u8a9e\u610f\u898b\u5206\u6790\u8a55\u4f30\u9805\u76ee\u7684\u8a13\u7df4\u96c6\u8f03\u5c0f\uff0c\u6211\u5011\u5f15\u5165 Blum \u548c Mitchell[18] \u63d0\u51fa\u7684 \u5354 \u540c \u8a13 \u7df4 (co-training) \u4f86 \u6539 \u5584 \u6548 \u80fd \u3002 \u5354 \u540c \u8a13 \u7df4 \u662f \u534a \u76e3 \u7763 \u5f0f \u6a5f \u5668 \u5b78 \u7fd2 \u65b9 \u6cd5 (semi-supervised learning method) \uff0c\u80fd\u7d50\u5408\u6a19\u8a18\u8cc7\u6599\u8207\u672a\u6a19\u8a18\u8cc7\u6599\u4e00\u8d77\u8a13\u7df4\u6a21\u578b\u3002\u672c\u7814 \u7a76\u6311\u9078 CRF \u9810\u6e2c\u4fe1\u5fc3\u503c\u8f03\u9ad8\u7684\u5be6\u4f8b\uff0c\u4ee5\u6587\u53e5\u70ba\u55ae\u4f4d\u56de\u994b\u5230\u8a13\u7df4\u8a9e\u6599\u4e2d\uff0c\u85c9\u6b64\u63d0\u5347\u7cfb\u7d71 \u6548\u80fd\u3002 (\u56db)\u3001\u5f8c\u7f6e\u8655\u7406 \u5916\u570b\u8b6f\u540d\u7684\u5177\u540d\u5be6\u9ad4\u5bb9\u6613\u5728\u65b7\u8a5e\u6642\u51fa\u73fe\u932f\u8aa4\uff0c\u9019\u6a23\u7684\u932f\u8aa4\u53ef\u80fd\u9020\u6210\u8fa8\u8b58\u51fa\u4e0d\u5b8c\u6574\u7684\u610f\u898b \u6301\u6709\u8005\uff0c\u56e0\u6b64\u6211\u5011\u53ef\u80fd\u9700\u8981\u4fee\u5fa9\u610f\u898b\u6301\u6709\u8005\u4e2d\u7684\u5177\u540d\u5be6\u9ad4\u3002\u672c\u7cfb\u7d71\u5047\u8a2d\u4e0d\u5b8c\u6574\u7684\u5177\u540d\u5be6 \u9ad4\u5728\u6587\u7ae0\u4e2d\u51fa\u73fe\u7684\u983b\u7387\u8207\u5b8c\u6574\u7684\u5177\u540d\u5be6\u9ad4\u76f8\u540c\uff0c\u6240\u4ee5\u672c\u7cfb\u7d71\u6703\u5c07\u610f\u898b\u6301\u6709\u8005\u6a19\u8a18\u7d50\u679c\u8ddf \u524d\u5f8c\u5b57\u4e32\u63a5\uff0c\u6e2c\u8a66\u7d44\u5408\u6210\u7684\u65b0\u8a5e\u8207\u539f\u8a5e\u5728\u6587\u7ae0\u4e2d\u51fa\u73fe\u7684\u983b\u7387\u662f\u5426\u76f8\u540c\uff0c\u76f8\u540c\u5247\u4ee5\u65b0\u8a5e\u53d6 \u4ee3\u539f\u8a5e\u3002\u4f8b\u53e5 4 \u70ba\u5177\u540d\u5be6\u9ad4\u4fee\u5fa9\u904e\u7a0b\u7684\u4e00\u4f8b\uff0c\u539f\u672c\u610f\u898b\u6301\u6709\u8005\u6a19\u8a18\u7684\u7d50\u679c\u662f\u300c\u8607\u54c8\u300d \uff0c \u62ec\u865f\u5167\u7684\u6578\u5b57\u4ee3\u8868\u8a72\u8a5e\u5728\u6587\u7ae0\u4e2d\u51fa\u73fe\u7684\u6b21\u6578\uff0c\u4fee\u5fa9\u5f8c\u53ef\u8f38\u51fa\u300c\u8607\u54c8\u6258\u300d \u3002\u900f\u904e\u9019\u6a23\u7684\u5177 \u540d\u5be6\u9ad4\u4fee\u5fa9\u65b9\u6cd5\uff0c\u672c\u7cfb\u7d71\u53ef\u4ee5\u5c07\u65b7\u8a5e\u932f\u8aa4\u7684\u5177\u540d\u5be6\u9ad4\u4fee\u5fa9\u70ba\u5b8c\u6574\u7684\u5177\u540d\u5be6\u9ad4\u3002 \u4f8b\u53e5 4\uff1a\u5370\u5c3c\u5f37\u4eba\u8607\u54c8\u6258\u7d71\u6cbb\u5370\u5c3c\u5345\u4e8c\u5e74 \u5c31\u53ef\u4ee5\u63d0\u5831\u6700\u5f8c\u8fa8\u8b58\u51fa\u4e4b\u610f\u898b\u6301\u6709\u8005\u3002 \u56db\u3001\u5be6\u9a57\u8207\u8a0e\u8ad6 \u672c\u7bc0\u5c07\u4ecb\u7d39\u672c\u5be6\u9a57\u4f7f\u7528\u7684\u8a9e\u6599\u8207\u8cc7\u6e90\uff0c\u4e26\u8a0e\u8ad6\u4f5c\u8005\u610f\u898b\u8fa8\u8b58\u5be6\u9a57\u3001\u610f\u898b\u6301\u6709\u8005\u6a19\u8a18\u5be6\u9a57 \u53ca\u610f\u898b\u6301\u6709\u8005\u8fa8\u8b58\u6574\u9ad4\u5be6\u9a57\u7684\u7d50\u679c\u3002 (\u4e00)\u3001NTCIR 7 \u591a\u8a9e\u610f\u898b\u5206\u6790\u8a55\u6bd4\u9805\u76ee\u4ecb\u7d39 \u7cbe\u78ba\u7387 (Precision) \u3001\u53ec\u56de\u7387 (Recall) \u3001F \u503c (F-score) \u53ca\u6b63\u78ba\u7387 (Accuracy) \u4f86\u8a55\u4f30\u7cfb \u7d71\u6548\u80fd\u3002\u6211\u5011\u7814\u7a76\u8a9e\u6599\u4e4b\u5f8c\u767c\u73fe\u90e8\u5206\u6587\u53e5\u6703\u6709\u4f5c\u8005\u610f\u898b\u6a19\u8a18\u6b67\u7570\u7684\u60c5\u5f62\uff0c\u4f8b\u5982\u4f8b\u53e5 5 \u7684 \u7b2c\u4e8c\u53e5\uff0c\u4e00\u7a2e\u8aaa\u6cd5\u8a8d\u70ba\u610f\u898b\u6301\u6709\u8005\u70ba\u300c\u81fa\u7063\u300d \uff0c\u53e6\u4e00\u7a2e\u8aaa\u6cd5\u8a8d\u70ba\u610f\u898b\u6301\u6709\u8005\u6a19\u8a18\u70ba\u6587\u7ae0 \u4f5c\u8005\u3002 \u898b\u7684\u8a2d\u5b9a\uff1a F \u503c\u70ba 77.40%\uff0c\u6548\u80fd\u6700\u5dee\u7684\u662f\u4e0d\u4f7f\u7528\u6a19\u8a18\u6b67\u7570\u7684\u8a9e\u6599\uff1a F \u503c\u70ba 65.10%\u3002 \u6cd5\u3002\u6211\u5011\u6839\u64da\u6b63\u78ba\u7b54\u6848\u8207\u7cfb\u7d71\u6a19\u8a18\u51fa\u4e4b\u7b54\u6848\u7684\u4f4d\u7f6e\u5206\u6790\uff0c\u5c07\u4e3b\u8981\u7684\u6a19\u8a18\u932f\u8aa4\u5206\u70ba 6 \u985e\uff1a \u7b56\u7565 A 1338 611 68.65% \u5bec\u9b06 \u672c\u7814\u7a76\u6839\u64da\u610f\u898b\u6301\u6709\u8005\u7684\u5206\u985e\u5c07\u610f\u898b\u6301\u6709\u8005\u8fa8\u8b58\u5206\u70ba\u4f5c\u8005\u610f\u898b\u8fa8\u8b58\u53ca\u610f\u898b\u6301\u6709\u8005\u6a19\u8a18 \u5f9e\u672c\u5be6\u9a57\u7d50\u679c\u53ef\u4ee5\u767c\u73fe\uff0c\u4f7f\u7528\u6a19\u8a18\u6b67\u7570\u7684\u8a9e\u6599\u80fd\u63d0\u5347\u7cfb\u7d71\u6548\u80fd\uff0c\u5c07\u6a19\u8a18\u6b67\u7570\u7684\u8a9e\u6599\u8996\u70ba 1.\u7b54\u6848\u7121\u95dc\u806f \u610f\u898b\u53e5 \u7b56\u7565 B 1372 576 70.40% \u5169\u90e8\u5206\u3002\u5728\u4f5c\u8005\u610f\u898b\u8fa8\u8b58\u4e2d\uff0c\u672c\u7814\u7a76\u63d0\u51fa\u8a5e\u5f59\u76f8\u95dc\u8cc7\u8a0a\u3001\u8a5e\u6027\u76f8\u95dc\u8cc7\u8a0a\u3001\u5177\u540d\u5be6\u9ad4\u8cc7\u8a0a\u3001 \u4f5c\u8005\u610f\u898b\u52a0\u5165\u8a13\u7df4\u6548\u80fd\u6700\u4f73\u3002 (\u56db)\u3001\u610f\u898b\u6301\u6709\u8005\u6a19\u8a18\u5be6\u9a57 \u95dc\u9375\u7b26\u865f\u8cc7\u8a0a\u3001\u6587\u53e5\u7d44\u6210\u8cc7\u8a0a\u53ca\u610f\u898b\u76f8\u95dc\u8cc7\u8a0a\u7b49\u7279\u5fb5\u503c\uff0c\u4e26\u4f7f\u7528\u652f\u63f4\u5411\u91cf\u6a5f\u4f86\u89e3\u6c7a\u6b64\u554f \u7121\u6cd5\u5224\u65b7\u6b63\u78ba\u7b54\u6848\u8207\u7cfb\u7d71\u6a19\u8a18\u51fa\u4e4b\u7b54\u6848\u4e4b\u9593\u7684\u95dc\u806f\uff0c\u6b64\u985e\u4f54\u6a19\u8a18\u932f\u8aa4\u7684 29.1%\u3002 2.\u591a\u64f7\u53d6\u524d\u5f8c\u4e00\u8a5e \u984c\u3002\u5728\u610f\u898b\u6301\u6709\u8005\u6a19\u8a18\u4e2d\uff0c\u672c\u7814\u7a76\u63d0\u51fa\u8a5e\u5f59\u76f8\u95dc\u8cc7\u8a0a\u3001\u8a5e\u6027\u76f8\u95dc\u8cc7\u8a0a\u3001\u5177\u540d\u5be6\u9ad4\u8cc7\u8a0a\u3001 (\u516d)\u3001\u8207 NTCIR 7 \u53c3\u8cfd\u968a\u4f0d\u6bd4\u8f03 \u95dc\u9375\u7b26\u865f\u8cc7\u8a0a\u3001\u6587\u53e5\u7d44\u6210\u8cc7\u8a0a\u3001\u524d\u5f8c\u6587\u76f8\u95dc\u8cc7\u8a0a\u53ca\u610f\u898b\u76f8\u95dc\u8cc7\u8a0a\u7b49\u7279\u5fb5\u503c\uff0c\u4e26\u4f7f\u7528\u689d\u4ef6 \u4f8b\u53e5 5\uff1a \u7b2c\u4e00\u53e5\uff1a\u81fa\u7063\u5b98\u65b9\u5df2\u6392\u9664\u6838\u6b66\u9078\u64c7\uff0c\u4e26\u5c07\u6700\u7d42\u5b89\u5168\u8a17\u4ed8\u65bc\u7f8e\u570b\uff0c \u7b2c\u4e8c\u53e5\uff1a\u56e0\u70ba\u81fa\u7063\u4e86\u89e3\uff0c\u6838\u6b66\u4e4b\u8def\u5c07\u6839\u672c\u7834\u58de\u8207\u7f8e\u570b\u4e4b\u9593\u7684\u95dc\u4fc2\u3002 \u6211\u5011\u5c07\u6a19\u8a18\u7684\u60c5\u5f62\u52a0\u4ee5\u5206\u6790\u3002\u5716\u4e09\u662f\u8a9e\u6599\u5eab\u4e2d\u4f5c\u8005\u610f\u898b\u6bd4\u4f8b\u793a\u610f\u5716\uff0c\u6211\u5011\u767c\u73fe\u5728 NTCIR7 \u7684\u8a9e\u6599\u5eab\u4e2d\uff0c\u88ab\u90e8\u5206\u6a19\u8a18\u8005\u6a19\u8a18\u70ba\u4f5c\u8005\u610f\u898b\uff0c\u5176\u4ed6\u6a19\u8a18\u8005\u6a19\u8a18\u70ba\u975e\u4f5c\u8005\u610f\u898b\u7684 \u6a19\u8a18\u6b67\u7570\u8a9e\u6599\u4f54 15%\uff0c\u8207\u88ab\u6240\u6709\u6a19\u8a18\u8005\u6a19\u8a18\u70ba\u4f5c\u8005\u610f\u898b\u7684\u6bd4\u4f8b (19%) \u76f8\u8ddd\u4e0d\u9060\u3002 NTCIR 7 NTCIR 6 \u672c\u5be6\u9a57\u7684\u76ee\u7684\u662f\u8fa8\u8b58\u51fa\u610f\u898b\u6301\u6709\u8005\u7684\u4ee3\u8868\u8a5e\uff0c\u4ee5\u4f9b\u5f8c\u7f6e\u8655\u7406\u7a0b\u5e8f\u7d44\u5408\u6210\u610f\u898b\u6301\u6709\u8005\u3002\u672c \u5be6\u9a57\u4f7f\u7528\u7684\u8a13\u7df4\u8a9e\u6599\u662f NTCIR7 \u8a13\u7df4\u96c6\uff0c\u6e2c\u8a66\u8a9e\u6599\u662f NTCIR7 \u6e2c\u8a66\u96c6\u4e2d\u7684\u56b4\u683c\u610f\u898b\u53e5 \u8207\u5bec\u9b06\u610f\u898b\u53e5\u3002\u672c\u5be6\u9a57\u5c07\u610f\u898b\u6301\u6709\u8005\u6a19\u8a18\u7684\u7d50\u679c\u7d44\u5408\u51fa\u6700\u5f8c\u7684\u610f\u898b\u6301\u6709\u8005\uff0c\u518d\u4f7f\u7528\u6b63\u78ba \u7cfb\u7d71\u6a19\u8a18\u51fa\u4e4b\u7b54\u6848\u5305\u542b\u6b63\u78ba\u7b54\u6848\uff0c\u4f46\u537b\u53c8\u591a\u5c07\u524d\u5f8c\u4e00\u8a5e\u5224\u65b7\u70ba\u610f\u898b\u6301\u6709\u8005\u7684\u4e00\u90e8 \u5206\uff0c\u4f8b\u5982\uff1a\u4e5f\u8a31\u8607\u54c8\u6258 \u8607\u54c8\u6258 \u8607\u54c8\u6258 \u8607\u54c8\u6258\u3001\u9b6f\u65af\u66fc \u9b6f\u65af\u66fc \u9b6f\u65af\u66fc \u9b6f\u65af\u66fc\u65e5\u524d\u3001\u4ed6\u5011 \u4ed6\u5011 \u4ed6\u5011 \u4ed6\u5011\u53ef\u4ee5\u3002\u7bc4\u4f8b\u4e2d\uff0c\u7c97\u9ad4\u5b57 \u7c97\u9ad4\u5b57 \u7c97\u9ad4\u5b57 \u7c97\u9ad4\u5b57\u4ee3\u8868\u6b63\u78ba\u7b54\u6848\uff0c\u6a19 \u96a8\u6a5f\u57df\u6a21\u578b\u4e26\u63d0\u51fa\u4e0d\u540c\u7684\u6a19\u8a18\u65b9\u5f0f\u4f86\u89e3\u6c7a\u6b64\u554f\u984c\u3002\u672c\u7814\u7a76\u63d0\u51fa\u5354\u540c\u8a13\u7df4\u4f86\u89e3\u6c7a\u8a13\u7df4\u8a9e\u6599 \u6211\u5011\u5c07\u672c\u7cfb\u7d71\u6548\u80fd\u8207 NTCIR7 \u53c3\u8cfd\u968a\u4f0d\u7684\u6548\u80fd\u6bd4\u8f03\uff0c NTCIR7 \u7684\u8a55\u4f30\u65b9\u5f0f\u5206\u70ba\u5169\u7a2e\uff0c \u904e\u5c11\u7684\u554f\u984c\uff0c\u4e26\u63d0\u51fa\u7d50\u679c\u5408\u4f75\u7b56\u7565\u4ee5\u63d0\u5347\u610f\u898b\u6301\u6709\u8005\u8fa8\u8b58\u6548\u80fd\u3002 \u4e00\u7a2e\u8a55\u4f30\u7cfb\u7d71\u63d0\u5831\u6b63\u78ba\u7684\u610f\u898b\u53e5\uff0c\u4e5f\u5c31\u662f\u6211\u5011\u5728\u610f\u898b\u6301\u6709\u8005\u6a19\u8a18\u4e2d\u4f7f\u7528\u7684\u8a55\u4f30\u65b9\u5f0f\uff0c\u53e6 \u5e95\u7dda\u7684\u5b57\u4ee3\u8868\u7cfb\u7d71\u6a19\u8a18\u51fa\u4e4b\u7b54\u6848\uff0c\u6b64\u985e\u4f54\u6a19\u8a18\u932f\u8aa4\u7684 18.1%\u3002 \u4e00\u7a2e\u5247\u8a55\u4f30 NTCIR7 \u8a9e\u6599\u4e2d\u6240\u6709\u7684\u610f\u898b\u53e5\u3002 \u672c\u7814\u7a76\u5be6\u4f5c\u51fa\u4e00\u5957\u610f\u898b\u6301\u6709\u8005\u8fa8\u8b58\u7cfb\u7d71\u3002\u672c\u7cfb\u7d71\u5728 NTCIR7 \u591a\u8a9e\u610f\u898b\u5206\u6790\u8a55\u6bd4\u9805\u76ee\u7e41 \u6578\u3001\u932f\u8aa4\u6578\u3001 set F \u503c (set F-score) \u4f86\u8a55\u4f30\u7cfb\u7d71\u6548\u80fd\u3002 NTCIR7 \u7684\u8a55\u4f30\u65b9\u5f0f\u6703\u5148\u8a55\u4f30 \u6bcf\u500b\u53c3\u8cfd\u968a\u4f0d\u6b63\u78ba\u63d0\u5831\u7684\u610f\u898b\u53e5\u6578\uff0c\u518d\u8a55\u4f30\u6bcf\u500b\u53c3\u8cfd\u968a\u4f0d\u6b63\u78ba\u63d0\u5831\u7684\u610f\u898b\u6301\u6709\u8005\uff0c\u53c3\u8cfd \u968a\u4f0d\u63d0\u5831\u7684\u610f\u898b\u53e5\u6578\u7b49\u65bc\u63d0\u5831\u7684\u610f\u898b\u6301\u6709\u8005\u6578\uff0c\u4e5f\u7b49\u65bc\u7b54\u6848\u7684\u610f\u898b\u6301\u6709\u8005\u6578\uff0c\u6b64\u6642\u7cbe\u78ba \u7387\u3001\u53ec\u56de\u7387\u3001 F \u503c\u8207\u6b63\u78ba\u7387\u6703\u76f8\u7b49\uff0c\u56e0\u6b64\u6211\u5011\u4e0d\u4f7f\u7528\u7cbe\u78ba\u7387\u3001\u53ec\u56de\u7387\u3001 F \u503c\uff0c\u800c\u4f7f \u7528 set F \u503c\u4f86\u8a55\u4f30\u6548\u80fd\uff0c set F \u503c\u7684\u5b9a\u7fa9\u5982\u4e0b\uff1a 3.\u64f7\u53d6\u51fa\u982d\u929c\u4f46\u672a\u64f7\u53d6\u51fa\u5176\u5f8c\u7684\u4eba\u540d \u7cfb\u7d71\u6a19\u8a18\u51fa\u4e4b\u7b54\u6848\u5305\u542b\u6b63\u78ba\u7b54\u6848\u4e2d\u4e4b\u982d\u929c\uff0c\u4f46\u537b\u6c92\u6709\u64f7\u53d6\u51fa\u982d\u929c\u5f8c\u7684\u4eba\u540d\uff0c\u4f8b\u5982\uff1a \u79d1\u7d22\u4f0f\u8457\u540d\u585e\u88d4\u9818\u8896 \u79d1\u7d22\u4f0f\u8457\u540d\u585e\u88d4\u9818\u8896 \u79d1\u7d22\u4f0f\u8457\u540d\u585e\u88d4\u9818\u8896 \u79d1\u7d22\u4f0f\u8457\u540d\u585e\u88d4\u9818\u8896\u7279\u62c9\u4f0a\u79d1\u7dad\u5951 \u7279\u62c9\u4f0a\u79d1\u7dad\u5951 \u7279\u62c9\u4f0a\u79d1\u7dad\u5951 \u7279\u62c9\u4f0a\u79d1\u7dad\u5951\uff0c\u6b64\u985e\u4f54\u6a19\u8a18\u932f\u8aa4\u7684 8.3%\u3002 4. \u64f7\u53d6\u51fa\u5f62\u5bb9\u8a5e\u4f46\u672a\u64f7\u53d6\u51fa\u5176\u5f8c\u7684\u666e\u901a\u540d\u8a5e \u9ad4\u4e2d\u6587\u8a9e\u6599\u4e2d\u53ef\u4ee5\u9054\u5230 F \u503c\u70ba 0.734 \u7684\u6548\u80fd\uff0c\u662f\u4f7f\u7528\u6a5f\u5668\u5b78\u7fd2\u65b9\u6cd5\u4e2d\u6548\u80fd\u6700\u4f73\u7684\uff0c\u4e5f \u8868\u516d\u986f\u793a\u672c\u7cfb\u7d71\u8207 NTCIR 7 \u53c3\u8cfd\u968a\u4f0d\u7684\u6548\u80fd\u6bd4\u8f03\u3002 NTCIR7 \u610f\u898b\u6301\u6709\u8005\u64f7\u53d6\u8a55\u6bd4\u9805 \u76f8\u7576\u63a5\u8fd1\u76ee\u524d\u6700\u4f73\u7cfb\u7d71\u7684\u6548\u80fd\u3002\u76ee\u524d\u6548\u80fd\u6700\u4f73\u7684\u65b9\u6cd5\u662f\u4f7f\u7528\u7d93\u9a57\u6cd5\u5247\u89e3\u6c7a\u672c\u554f\u984c\uff0c\u7d93\u9a57 \u76ee\u7684\u53c3\u8cfd\u968a\u4f0d\u5305\u542b\u9999\u6e2f\u4e2d\u6587\u5927\u5b78\u3001\u5317\u4eac\u5927\u5b78\u3001\u9f8d\u6372\u98a8\u79d1\u6280\u53ca\u53f0\u7063\u5927\u5b78\u56db\u968a\uff0c\u9999\u6e2f\u4e2d\u6587\u5927 \u6cd5\u5247\u8f03\u96e3\u91cd\u88fd\u8207\u9a57\u8b49\uff0c\u4f46\u7814\u7a76\u8005\u5f88\u5bb9\u6613\u5c31\u53ef\u4ee5\u91cd\u88fd\u8207\u9a57\u8b49\u672c\u7814\u7a76\u63d0\u51fa\u7684\u6a5f\u5668\u5b78\u7fd2\u65b9\u6cd5\u3002 \u5b78\u4f7f\u7528\u7d93\u9a57\u6cd5\u5247\u65b9\u6cd5\u3001\u5317\u4eac\u5927\u5b78\u4f7f\u7528\u689d\u4ef6\u96a8\u6a5f\u57df\u6a21\u578b\u3001\u9f8d\u6372\u98a8\u79d1\u6280\u4f7f\u7528\u652f\u63f4\u5411\u91cf\u6a5f\u3001\u53f0 \u672c\u7814\u7a76\u5206\u6790\u610f\u898b\u6301\u6709\u8005\u8fa8\u8b58\u8a13\u7df4\u8a9e\u6599\u4e2d\u6a19\u8a18\u6b67\u7570\u7684\u60c5\u5f62\uff0c\u4e26\u63d0\u51fa\u6700\u4f73\u7684\u61c9\u7528\u65b9\u5f0f\uff0c\u672c\u7814 \u7063\u5927\u5b78\u4f7f\u7528\u6c7a\u7b56\u6a39\u6f14\u7b97\u6cd5\u3002 \u7a76\u4e5f\u5206\u6790\u7cfb\u7d71\u8fa8\u8b58\u4e4b\u932f\u8aa4\u7d50\u679c\uff0c\u4e26\u4ee5\u5177\u540d\u5be6\u9ad4\u4fee\u5fa9\u53ca\u610f\u898b\u6301\u6709\u8005\u5c3e\u8a5e\u6a19\u8a18\u7684\u65b9\u6cd5\u4f86\u6539\u5584 Seki \u7b49\u4eba[19] \u5c0d\u6b64\u8a55\u6bd4\u9805\u76ee\u6709\u8a73\u7d30\u7684\u4ecb\u7d39\u3002 \u7cfb\u7d71\u6a19\u8a18\u51fa\u4e4b\u7b54\u6848\u5305\u542b\u6b63\u78ba\u7b54\u6848\u4e2d\u524d\u9762\u7684\u5f62\u5bb9\u8a5e\u4f46\u672a\u64f7\u53d6\u51fa\u5176\u5f8c\u7684\u666e\u901a\u540d\u8a5e\uff0c\u4f8b \u56b4\u683c\u610f\u898b\u53e5\u90e8\u4efd\uff0c\u4ee5\u7cfb\u7d71\u63d0\u5831\u6b63\u78ba\u7684\u610f\u898b\u53e5\u8a55\u4f30\uff0c\u672c\u7cfb\u7d71\u7684\u6700\u4f73\u6548\u80fd\u70ba F \u503c 73.40%\u3002 \u932f\u8aa4\u60c5\u6cc1\u3002 \u57f7\u884c\u904e\u7a0b\uff1a\u8607\u54c8(16) \u2192 \u8607\u54c8\u6258(16) \u983b\u7387\u76f8\u540c\u5f80\u5f8c\u7e7c\u7e8c \u2192 \u8607\u54c8\u6258\u7d71(1) \u983b\u7387\u4e0d\u540c\u6539\u5f80\u524d \u2192 \u4eba\u8607\u54c8\u6258(3) \u983b\u7387\u4e0d\u540c\u7d50\u675f (\u4e94)\u3001\u5408\u4f75\u4f5c\u8005\u610f\u898b\u8fa8\u8b58\u8207\u610f\u898b\u6301\u6709\u8005\u6a19\u8a18\u4e4b\u7d50\u679c \u4f5c\u8005\u610f\u898b\u8fa8\u8b58\u5c07\u610f\u898b\u53e5\u5340\u5206\u70ba\u4f5c\u8005\u610f\u898b\u53ca\u975e\u4f5c\u8005\u610f\u898b\u5169\u985e\uff0c\u610f\u898b\u6301\u6709\u8005\u6a19\u8a18\u5247\u53ef\u6a19\u8a18\u51fa \u610f\u898b\u6301\u6709\u8005\u7684\u4ee3\u8868\u8a5e\uff0c\u672c\u7814\u7a76\u5c07\u6b64\u5169\u90e8\u5206\u7684\u7d50\u679c\u5408\u4f75\uff0c\u7522\u751f\u6700\u5f8c\u63d0\u5831\u4e4b\u610f\u898b\u6301\u6709\u8005\u3002\u672c \u5716\u4e00\u3001\u7d50\u679c\u5408\u4f75\u7b56\u7565 A \u5716\u4e8c\u3001\u7d50\u679c\u5408\u4f75\u7b56\u7565 B \u7d50\u679c\u5408\u4f75\u7b56\u7565 A \u4e2d\uff0c\u6211\u5011\u76f8\u4fe1\u4f5c\u8005\u610f\u898b\u8fa8\u8b58\u5224\u65b7\u51fa\u7684\u4f5c\u8005\u610f\u898b\uff0c\u4f5c\u8005\u610f\u898b\u8fa8\u8b58\u5224\u65b7\u51fa \u975e\u4f5c\u8005\u610f\u898b\u7684\u6587\u53e5\u5247\u900f\u904e\u610f\u898b\u6301\u6709\u8005\u6a19\u8a18\u6a19\u51fa\u610f\u898b\u6301\u6709\u8005\u7684\u4f4d\u7f6e\u3002\u7d50\u679c\u5408\u4f75\u7b56\u7565 B \u4e2d\uff0c\u6211\u5011\u76f8\u4fe1\u4f5c\u8005\u610f\u898b\u8fa8\u8b58\u5224\u65b7\u51fa\u7684\u975e\u4f5c\u8005\u610f\u898b\uff0c\u4f5c\u8005\u610f\u898b\u8fa8\u8b58\u5224\u65b7\u51fa\u662f\u4f5c\u8005\u610f\u898b\u7684\u90e8 \u4efd\uff0c\u56e0\u70ba\u4e0d\u5920\u78ba\u5b9a\uff0c\u518d\u900f\u904e\u610f\u898b\u6301\u6709\u8005\u6a19\u8a18\u4e8c\u6b21\u6aa2\u67e5\u662f\u5426\u70ba\u4f5c\u8005\u610f\u898b\uff1a\u5982\u679c\u6c92\u6709\u6a19\u51fa\u610f \u6bd4\u9805\u76ee\u662f NTCIR 7 \u591a\u8a9e\u610f\u898b\u5206\u6790\u8a55\u4f30\u9805\u76ee\u7684\u524d\u8eab\uff0c\u8a9e\u6599\u5eab\u4e2d\u63d0\u4f9b\u76f8\u95dc\u53e5\u3001\u610f\u898b\u53e5\u3001\u610f \u898b\u50be\u5411\u3001\u610f\u898b\u6301\u6709\u8005\u7684\u6a19\u8a18\u3002 NTCIR6 \u6e2c\u8a66\u96c6\u5305\u62ec 29 \u500b\u4e3b\u984c\u30019,240 \u500b\u6587\u53e5\u30015,453 \u500b\u610f\u898b\u53e5\uff0c\u6211\u5011\u5229\u7528\u8a9e\u6599\u5eab\u4e2d\u95dc\u65bc\u610f\u898b\u53e5\u8207\u610f\u898b\u6301\u6709\u8005\u7684\u6a19\u8a18\u7576\u4f5c\u6211\u5011\u7684\u5be6\u9a57\u8a9e\u6599\u3002 (\u4e8c)\u3001\u5be6\u9a57\u8cc7\u6e90 \u672c\u7814\u7a76\u4f7f\u7528\u7684\u5be6\u9a57\u8cc7\u6e90\u5305\u62ec\u610f\u898b\u8a5e\u8a5e\u5178\u8207\u5177\u540d\u5be6\u9ad4\u8a5e\u5178\u3002\u610f\u898b\u8a5e\u8a5e\u5178\u7684\u90e8\u4efd\u5305\u542b\u6a19\u8a18\u8005 \u5f9e NTCIR 7 \u591a\u8a9e\u610f\u898b\u5206\u6790\u8a55\u6bd4\u9805\u76ee\u7684\u8a13\u7df4\u96c6\u4e2d\u6a19\u8a18\u51fa\u8868\u8ff0\u95dc\u9375\u5b57\u3001\u6b63\u9762\u610f\u898b\u8a5e\u3001\u8ca0\u9762 \u610f\u898b\u8a5e\u53ca\u4e2d\u7acb\u610f\u898b\u8a5e\u7b49\u610f\u898b\u8a5e\uff0c\u4e5f\u4f7f\u7528 Ku \u548c Chen[20] \u958b\u767c\u7684\u53f0\u5927\u610f\u898b\u8a5e\u8a5e\u5178 (NTUSD) \uff0c\u672c\u7814\u7a76\u5c07\u9019\u4e9b\u610f\u898b\u8a5e\u8a5e\u5178\u61c9\u7528\u65bc\u7279\u5fb5\u503c\u64f7\u53d6\u3002 \u672c\u7814\u7a76\u5f15\u5165\u4eba\u540d\u8a5e\u5178\u3001\u5730\u540d\u8a5e\u5178\u53ca\u7d44\u7e54\u540d\u8a5e\u5178\u4e09\u985e\u5177\u540d\u5be6\u9ad4\u8a5e\u5178\uff0c\u4f7f\u7528\u7684\u8a5e\u5178\u5305\u542b\u767e\u842c \u4eba\u540d\u5b57\u5178\u3001\u4e2d\u6587\u8a5e\u5eab\u3001\u4e2d\u592e\u793e\u8b6f\u540d\u6a94\u3001\u570b\u7acb\u7de8\u8b6f\u9928\u5c08\u696d\u5b57\u5178\u3001\u65e5\u672c\u5e38\u898b\u4e03\u5343\u500b\u59d3\u6c0f\u3001\u6559 \u80b2\u90e8\u5730\u540d\u8b6f\u540d\u8a5e\u5178\u3001\u5916\u570b\u5730\u540d\u8b6f\u540d\u53ca\u53f0\u7063\u516c\u71df\u4f01\u696d\u5217\u8868\u3002\u9019\u4e9b\u5177\u540d\u5be6\u9ad4\u8a5e\u5178\u5247\u61c9\u7528\u65bc\u5177 \u540d\u5be6\u9ad4\u8fa8\u8b58\u3002 (\u4e09)\u3001\u4f5c\u8005\u610f\u898b\u8fa8\u8b58\u5be6\u9a57 NTCIR7 \u8a13\u7df4\u96c6\u53ca NTCIR6 \u6e2c\u8a66\u96c6\uff0c\u6e2c\u8a66\u8a9e\u6599\u662f NTCIR7 \u6e2c\u8a66\u96c6\u4e2d\u7684\u5bec\u9b06\u610f\u898b\u53e5\u3002 \u4e0d\u5217\u5165\u8a13\u7df4\u8a9e\u6599 50.52% 91.53% 65.10% 77.28% \u53ef\u4ee5\u9054\u5230\u6700\u4f73\u6548\u80fd\u3002 \u672c\u5be6\u9a57\u7684\u76ee\u7684\u662f\u8fa8\u8b58\u610f\u898b\u53e5\u4e4b\u610f\u898b\u6301\u6709\u8005\u662f\u5426\u70ba\u6587\u7ae0\u4f5c\u8005\uff0c\u672c\u5be6\u9a57\u4f7f\u7528\u7684\u8a13\u7df4\u8a9e\u6599\u662f \u5728 NTCIR6 \u7684\u8a9e\u6599\u5eab\u4e2d\uff0c\u4e0d\u53ea\u662f\u610f\u898b\u53e5\uff0c\u6240\u6709\u7684\u6587\u53e5\u90fd\u6703\u88ab\u6a19\u4e0a\u767c\u8868\u8005\uff0c\u4ee3\u8868\u8868\u793a\u9019 \u500b\u8aaa\u6cd5\u7684\u4eba\u6216\u7d44\u7e54\uff0c NTCIR6 \u7684\u610f\u898b\u6301\u6709\u8005\u6a19\u8a18\u65b9\u6cd5\u662f\u5982\u679c\u53ef\u4ee5\u5f9e\u6587\u7ae0\u4e2d\u6a19\u8a18\u51fa\u610f\u898b \u6301\u6709\u8005\u5c31\u6a19\u8a18\uff0c\u5982\u679c\u4e0d\u80fd\uff0c\u5247\u610f\u898b\u6301\u6709\u8005\u70ba\u6587\u7ae0\u4f5c\u8005\uff0c\u6240\u4ee5\u7121\u6cd5\u5f97\u77e5 NTCIR6 \u7684\u8a9e\u6599 \u5eab\u4e2d\u6a19\u8a18\u51fa\u73fe\u6b67\u7570\u7684\u6587\u53e5\u5360\u5168\u90e8\u6587\u53e5\u7684\u6bd4\u4f8b\u3002 \u6211\u5011\u5f9e\u5be6\u9a57\u4e2d\u767c\u73fe\u4f7f\u7528\u610f\u898b\u53e5\u7576\u8a13\u7df4\u8a9e\u6599\u7684 F \u55ae\u7368\u4f7f\u7528\u5176\u4e2d\u4e00\u500b\u8a13\u7df4\u96c6\u70ba\u9ad8\uff0c\u6240\u4ee5\u672c\u5be6\u9a57\u4f7f\u7528 NTCIR 6 \u52a0 NTCIR 7 \u8a13\u7df4\u96c6\u4e2d\u7684\u610f \u898b\u53e5\u4f5c\u70ba\u8a13\u7df4\u8a9e\u6599\u3002 \u8868\u4e09\u3001\u6a19\u8a18\u6b67\u7570\u7684\u8a9e\u6599\u5c0d\u4f5c\u8005\u610f\u898b\u8fa8\u8b58\u6548\u80fd\u7684\u5f71\u97ff \u8a2d\u5b9a \u7cbe\u78ba\u7387 \u53ec\u56de\u7387 F \u503c \u6b63\u78ba\u7387 \u8996\u70ba\u4f5c\u8005\u610f\u898b 69.68% 93.85% 79.98% 83.49% \u8996\u70ba\u975e\u4f5c\u8005\u610f\u898b 64.87% 95.94% 77.40% 80.31% \u898b\u6301\u6709\u8005\u7684\u4e2d\u9593\u8a5e (I) \u53ca\u975e\u610f\u898b\u6301\u6709\u8005\u7d44\u6210\u8a5e (O) \u3002\u8868\u56db\u986f\u793a\u4f7f\u7528\u4e0d\u540c\u5206\u985e\u6f14\u7b97\u6cd5\u5c0d \u8868\u56db\u3001\u5206\u985e\u6f14\u7b97\u6cd5\u5c0d\u610f\u898b\u6301\u6709\u8005\u6a19\u8a18\u6548\u80fd\u7684\u5f71\u97ff \u5206\u985e\u6f14\u7b97\u6cd5 \u6b63\u78ba\u6578 \u932f\u8aa4\u6578 set F \u503c CHAID 564 605 48.16% CRF 817 351 69.89% \u56b4\u683c \u610f\u898b\u53e5 CRF+CHAID 825 344 70.57% CHAID 981 967 50.31% CRF 1317 631 67.57% \u5bec\u9b06 \u610f\u898b\u53e5 CRF+CHAID 1322 627 67.83% \u5de5\u5927\u5b78\u7684\u8907\u88fd\u5c08\u5bb6\u97cb\u65af \u5de5\u5927\u5b78\u7684\u8907\u88fd\u5c08\u5bb6\u97cb\u65af \u5de5\u5927\u5b78\u7684\u8907\u88fd\u5c08\u5bb6\u97cb\u65af \u5de5\u5927\u5b78\u7684\u8907\u88fd\u5c08\u5bb6\u97cb\u65af\u7279\u4f11\u751f \u7279\u4f11\u751f \u7279\u4f11\u751f \u7279\u4f11\u751f\u3001\u8eca\u71c8\u5ee0\u5824 \u8eca\u71c8\u5ee0\u5824 \u8eca\u71c8\u5ee0\u5824 \u8eca\u71c8\u5ee0\u5824\u7dad\u897f \u7dad\u897f \u7dad\u897f \u7dad\u897f\uff0c\u6b64\u985e\u4f54\u6a19\u8a18\u932f\u8aa4\u7d50\u679c\u7684 4.7%\u3002 \u6839\u64da\u9019\u4e9b\u6a19\u8a18\u7d50\u679c\u932f\u8aa4\u7684\u985e\u5225\uff0c\u6211\u5011\u63d0\u51fa\u5e7e\u7a2e\u65b9\u6cd5\u4f86\u6539\u5584\u7cfb\u7d71\u6548\u80fd\u3002\u5927\u90e8\u5206\u7684\u985e\u5225\u90fd\u6709 \u610f\u898b\u6301\u6709\u8005\u8a5e\u7d44\u4e2d\u9996\u8a5e\u3001\u5c3e\u8a5e\u4e0d\u660e\u78ba\u7684\u554f\u984c\uff0c\u6240\u4ee5\u6211\u5011\u63d0\u51fa\u589e\u52a0\u610f\u898b\u6301\u6709\u8005\u6a19\u7c64\u7684\u65b9 \u6cd5\u3002\u5f9e\u5be6\u9a57\u4e2d\u767c\u73fe\uff0c\u4f7f\u7528 HIO \u6a19\u7c64\u96c6\u5728\u56b4\u683c\u610f\u898b\u53e5\u90e8\u4efd\u7684\u8a55\u4f30\u4e2d\u53ef\u5f97\u5230\u6700\u4f73\u7684 set F \u503c 70.57%\u3002\u91dd\u5c0d\u7b2c 6 \u985e\u5177\u540d\u5be6\u9ad4\u64f7\u53d6\u4e0d\u5b8c\u6574\u7684\u554f\u984c\uff0c\u6211\u5011\u4e5f\u63d0\u51fa\u5177\u540d\u5be6\u9ad4\u4fee\u5fa9\u65b9\u6cd5\u4f86 \u89e3\u6c7a\u9019\u500b\u554f\u984c\uff0c\u5f9e\u5be6\u9a57\u4e2d\u767c\u73fe\uff0c\u52a0\u5165\u5354\u540c\u8a13\u7df4\u8207\u5177\u540d\u5be6\u9ad4\u4fee\u5fa9\u5728\u56b4\u683c\u610f\u898b\u53e5\u90e8\u4efd\u7684\u8a55\u4f30 \u4e2d\u53ef\u5f97\u5230\u6700\u4f73\u7684 set F \u503c 72.03%\uff0c\u6548\u80fd\u63d0\u5347\u4e86 1.46%\u3002 (\u4e94)\u3001\u610f\u898b\u6301\u6709\u8005\u8fa8\u8b58\u6574\u9ad4\u5be6\u9a57 \u672c\u5be6\u9a57\u7684\u76ee\u7684\u662f\u63a2\u8a0e\u4f7f\u7528\u4e0d\u540c\u7d50\u679c\u5408\u4f75\u7b56\u7565\u5c0d\u610f\u898b\u6301\u6709\u8005\u8fa8\u8b58\u6548\u80fd\u7684\u5f71\u97ff\uff0c\u5be6\u9a57\u8a2d\u5b9a\u70ba \u7d50\u679c\u5408\u4f75\u7b56\u7565 A \u8207\u7d50\u679c\u5408\u4f75\u7b56\u7565 B \uff0c\u8868\u4e94\u986f\u793a\u4f7f\u7528\u4e0d\u540c\u7d50\u679c\u5408\u4f75\u7b56\u7565\u7684\u7cfb\u7d71\u6548\u80fd\u3002 \u56b4\u683c\u610f\u898b\u53e5\u90e8\u4efd\u7684\u8a55\u4f30\u4e2d\u7b56\u7565 B \u7684 set F \u503c\u70ba 73.40%\uff0c\u6bd4\u7b56\u7565 A \u9ad8\u4e86 2.48%\uff0c\u5bec\u9b06 \u610f\u898b\u53e5\u90e8\u4efd\u7684\u8a55\u4f30\u4e5f\u53ef\u4ee5\u5f97\u5230\u985e\u4f3c\u7684\u7d50\u679c\u3002\u7d50\u679c\u986f\u793a\u4f5c\u8005\u610f\u898b\u8fa8\u8b58\u7a0b\u5e8f\u8f03\u64c5\u9577\u5224\u65b7\u975e\u4f5c \u8005\u610f\u898b\uff0c\u53ef\u80fd\u8207\u6211\u5011\u4f7f\u7528\u8f03\u591a\u8207\u975e\u4f5c\u8005\u610f\u898b\u76f8\u95dc\u7684\u7279\u5fb5\u503c\u6709\u95dc\uff0c\u5be6\u9a57\u7d50\u679c\u4e5f\u986f\u793a\u7b56\u7565 B \u4ee5 \u7cfb \u7d71 \u731c \u5c0d \u610f \u898b \u53e5 \u6578 \u8a55 \u4f30 \u4ee5 \u6240 \u6709 \u610f \u898b \u53e5 \u8a55 \u4f30 \u53c3 \u8cfd \u968a \u4f0d \u731c \u5c0d \u898b \u53e5 \u6578 \u7cbe \u78ba \u7387 \u53ec \u56de \u7387 F \u503c \u7cbe \u78ba \u7387 \u53ec \u56de \u7387 F \u9999 \u6e2f \u4e2d \u6587 \u5927 \u5b78 \u4eac \u5927 \u5b78 880 57.84% 57.84% 57.84% 13.03% 40.53% 19.72% \u9f8d \u6372 \u98a8 \u79d1 \u53f0 \u7063 \u5927 \u53e5 \u672c \u7cfb \u9999 \u6e2f \u4e2d \u6587 \u5927 \u5317 \u4eac \u5927 \u9f8d \u6372 \u98a8 \u79d1 \u53f0 \u7063 \u5927 \u672c \u7cfb \u7d71 \u53e5 \u5b78 1948 50.31% 50.31% 50.31% 14.43% 53.73% 22.75% \u898b \u6280 2070 56.47% 56.47% 56.47% 16.78% 40.02% 23.65% \u610f \u5b78 1364 58.72% 58.72% 58.72% 20.51% 36.84% 26.35% \u9b06 \u5b78 1134 82.54% 82.54% 82.54% 29.92% 43.05% 35.31% \u5bec \u7d71 1169 73.40% 73.40% 73.40% 12.38% 68.31% 20.97% \u5b78 1169 48.16% 48.16% 48.16% 8.14% 44.90% 13.78% \u898b \u6280 1213 54.91% 54.91% 54.91% 8.22% 52.95% 14.23% \u683c \u610f 757 82.\u5317 \u56b4 \u503c \u610f \u4f5c\u8005\u610f\u898b\u8fa8\u8b58\u6548\u80fd\u7684\u5f71\u97ff\u3002 \u6b63\u78ba\u7b54\u6848\u5305\u542b\u7cfb\u7d71\u6a19\u8a18\u51fa\u4e4b\u7b54\u6848\uff0c\u4f46\u5177\u540d\u5be6\u9ad4\u90e8\u4efd\u64f7\u53d6\u5f97\u4e0d\u5b8c\u6574\uff0c\u4f8b\u5982\uff1a\u5fb7\u5dde\u8fb2 \u5fb7\u5dde\u8fb2 \u5fb7\u5dde\u8fb2 \u5fb7\u5dde\u8fb2 \u8868\u516d\u3001\u610f\u898b\u6301\u6709\u8005\u8fa8\u8b58\u6574\u9ad4\u6548\u80fd\u2500\u8207 NTCIR 7 \u53c3\u8cfd\u968a\u4f0d\u6bd4\u8f03 \u7814\u7a76\u63d0\u51fa\u5169\u7a2e\u7d50\u679c\u5408\u4f75\u7b56\u7565\uff0c\u5716\u4e8c\u3001\u4e09\u70ba\u7d50\u679c\u5408\u4f75\u7b56\u7565 A \u8207 B \u7684\u793a\u610f\u5716\u3002 \u591a\u8a9e\u610f\u898b\u5206\u6790\u8a55\u6bd4\u9805\u76ee\u63d0\u4f9b\u82f1\u6587\u3001\u65e5\u6587\u3001\u7e41\u9ad4\u4e2d\u6587\u8207\u7c21\u9ad4\u4e2d\u6587\u7684\u8a9e\u6599\u5eab\uff0c\u8a9e\u6599\u5eab\u4e2d\u63d0\u4f9b \u76f8\u95dc\u53e5\u3001\u610f\u898b\u53e5\u3001\u610f\u898b\u50be\u5411\u3001\u610f\u898b\u5f37\u5ea6\u3001\u610f\u898b\u6301\u6709\u8005\u8207\u8a55\u8ad6\u76ee\u6a19\u7684\u6a19\u8a18\u3002\u8a9e\u6599\u5eab\u5206\u70ba\u8a13 \u7df4\u96c6\u8207\u6e2c\u8a66\u96c6\uff0c NTCIR7 \u8a13\u7df4\u96c6\u5305\u62ec 3 \u500b\u4e3b\u984c\u30011,509 \u500b\u6587\u53e5\u3001944 \u500b\u610f\u898b\u53e5\uff0cNTCIR7 \u6e2c\u8a66\u96c6\u5305\u62ec 14 \u500b\u4e3b\u984c\u30014,665 \u500b\u6587\u53e5\u30012,174 \u500b\u610f\u898b\u53e5\uff0c\u53c3\u8cfd\u8005\u5011\u6703\u4ee5\u6587\u53e5\u70ba\u55ae\u4f4d\u9032\u884c \u975e \u4f5c \u8005 \u610f \u898b \u4f5c \u8005 \u610f \u898b 66% 15% 19% \u4f5c \u8005 \u610f \u898b 65% 35% ?% \u975e \u4f5c \u8005 \u610f \u898b \u6578 \u7cfb\u7d71\u6b63\u78ba\u63d0\u5831\u7684\u610f\u898b\u53e5 \u8005\u7684\u610f\u898b\u53e5\u6578 \u7cfb\u7d71\u6b63\u78ba\u63d0\u5831\u610f\u898b\u6301\u6709 \u503c = setF (1) \u5be6\u9a57\u6bd4\u8f03\u5169\u7a2e\u5206\u985e\u6f14\u7b97\u6cd5\uff1a\u7528\u4f86\u89e3\u6c7a\u4e8c\u5143\u5206\u985e\u554f\u984c\u7684\u6c7a\u7b56\u6a39\u6f14\u7b97\u6cd5 CHAID \uff0c\u53ca\u7528\u4f86 \u89e3\u6c7a\u5e8f\u5217\u6a19\u8a18\u554f\u984c\u7684\u689d\u4ef6\u96a8\u6a5f\u57df\u6a21\u578b CRF \u3002 CHAID \u4f7f\u7528\u7684\u610f\u898b\u6301\u6709\u8005\u6a19\u8a18\u6a19\u7c64\u662f \u5982\uff1a\u8a72 \u8a72 \u8a72 \u8a72\u88c1\u6c7a \u88c1\u6c7a \u88c1\u6c7a \u88c1\u6c7a\u3001\u570b\u969b \u570b\u969b \u570b\u969b \u570b\u969b\u505c\u706b\u89c0\u5bdf\u5718 \u505c\u706b\u89c0\u5bdf\u5718 \u505c\u706b\u89c0\u5bdf\u5718 \u505c\u706b\u89c0\u5bdf\u5718\u3001\u7f8e\u570b\u5168\u570b \u7f8e\u570b\u5168\u570b \u7f8e\u570b\u5168\u570b \u7f8e\u570b\u5168\u570b\u516c\u5171\u5ee3\u64ad\u96fb\u53f0 \u516c\u5171\u5ee3\u64ad\u96fb\u53f0 \u516c\u5171\u5ee3\u64ad\u96fb\u53f0 \u516c\u5171\u5ee3\u64ad\u96fb\u53f0\uff0c\u6b64\u985e\u4f54\u6a19\u8a18\u932f\u8aa4\u7684 7.5%\u3002 5. \u984d\u5916\u64f7\u53d6\u51fa\u5176\u4ed6\u975e\u7b54\u6848\u8a5e \u7cfb\u7d71\u6a19\u8a18\u51fa\u4e4b\u7b54\u6848\u5305\u542b\u6b63\u78ba\u7b54\u6848\uff0c\u4f46\u537b\u53c8\u984d\u5916\u64f7\u53d6\u51fa\u5176\u4ed6\u7684\u8a5e\uff0c\u4f8b\u5982\uff1a\u72c4\u862d \u72c4\u862d \u72c4\u862d \u72c4\u862d\u5728\u8a18 \u8005\u6703\u3001\u4ed6 \u4ed6 \u4ed6 \u9999\u6e2f\u4e2d\u6587\u5927\u5b78\u7684\u6548\u80fd\u6700\u4f73\uff1a F \u503c\u70ba 82.30%\uff0c\u6bd4\u672c\u7cfb\u7d71\u9ad8\u4e86 8.90%\uff0c\u4f46\u672c\u7cfb\u7d71\u7684\u6548\u80fd\u4e5f \u6700\u7d42\u6211\u5011\u5e0c\u671b\u80fd\u5c07\u610f\u898b\u6301\u6709\u8005\u8fa8\u8b58\u7684\u7d50\u679c\u8207\u5176\u5b83\u610f\u898b\u63a2\u52d8\u7684\u7d50\u679c\u7d50\u5408\uff0c\u6574\u5408\u6210\u4e00\u5957\u80fd\u81ea \u6bd4\u5176\u4ed6\u4f7f\u7528\u6a5f\u5668\u5b78\u7fd2\u65b9\u6cd5\u7684\u968a\u4f0d\u9ad8\u4e86 15.09%\u4ee5\u4e0a\u3002\u4ee5\u6240\u6709\u610f\u898b\u53e5\u6578\u8a55\u4f30\u4e2d\uff0c\u672c\u7cfb\u7d71\u8207 \u52d5\u64f7\u53d6\u51fa\u610f\u898b\u53e5\u7684\u610f\u898b\u50be\u5411\u3001\u610f\u898b\u5f37\u5ea6\u3001\u610f\u898b\u6301\u6709\u8005\u53ca\u610f\u898b\u8a55\u8ad6\u76ee\u6a19\u7684\u610f\u898b\u63a2\u52d8\u7cfb\u7d71\uff0c \u9999\u6e2f\u4e2d\u6587\u5927\u5b78\u7684\u6548\u80fd\u5dee\u8ddd\u62c9\u8fd1\u5230 5.16%\u3002\u6bd4\u8f03\u56b4\u683c\u610f\u898b\u53e5\u8207\u5bec\u9b06\u610f\u898b\u53e5\u7684\u8a55\u4f30\uff0c\u53ef\u4ee5\u767c \u4ee5\u63d0\u4f9b\u4f7f\u7528\u8005\u66f4\u6709\u7528\u7684\u8cc7\u8a0a\u3002 \u73fe\u672c\u7cfb\u7d71\u8207\u5176\u4ed6\u7cfb\u7d71\u4e0d\u540c\uff0c\u8f03\u64c5\u9577\u65bc\u8fa8\u8b58\u56b4\u683c\u610f\u898b\u53e5\u7684\u610f\u898b\u6301\u6709\u8005\uff0c\u63db\u53e5\u8a71\u8aaa\uff0c\u672c\u7cfb\u7d71 \u4ed6\u795d\u8cc0\u5df4\u52d2\u65af\u5766\u7684\u79d1\u5b78\u5bb6\uff0c\u6b64\u985e\u4f54\u6a19\u8a18\u932f\u8aa4\u7d50\u679c\u7684 5.5%\u3002 \u64c5\u9577\u65bc\u8fa8\u8b58\u51fa\u8f03\u7121\u722d\u8b70\u7684\u610f\u898b\u6301\u6709\u8005\uff0c\u4e5f\u5c31\u662f\u8f03\u70ba\u53ef\u9760\u7684\u610f\u898b\u6301\u6709\u8005\u3002 \u610f\u898b\u5206\u6790\u3002\u56e0\u70ba NTCIR7 \u8a13\u7df4\u96c6\u8f03\u5c0f\uff0c\u672c\u5be6\u9a57\u7684\u8a13\u7df4\u8a9e\u6599\u52a0\u5165 NTCIR6 \u610f\u898b\u5206\u6790\u8a66 \u9a57\u8a55\u6bd4\u9805\u76ee (Opinion Analysis Pilot Task) \u7e41\u9ad4\u4e2d\u6587\u7684\u6e2c\u8a66\u96c6\uff0cNTCIR6 \u610f\u898b\u5206\u6790\u8a66\u9a57\u8a55 \u5716\u4e09\u3001\u8a9e\u6599\u5eab\u4e2d\u4f5c\u8005\u610f\u898b\u6bd4\u4f8b\u793a\u610f\u5716 YES \u8207 NO \u6a19\u7c64\uff0c CRF \u4f7f\u7528\u7684\u610f\u898b\u6301\u6709\u8005\u6a19\u8a18\u6a19\u7c64\u662f\u610f\u898b\u6301\u6709\u8005\u7684\u9996\u8a5e (H) \u3001\u610f 6. \u5177\u540d\u5be6\u9ad4\u64f7\u53d6\u4e0d\u5b8c\u6574 \u53c3\u8003\u6587\u737b |