| { | |
| "paper_id": "O16-1022", | |
| "header": { | |
| "generated_with": "S2ORC 1.0.0", | |
| "date_generated": "2023-01-19T08:04:54.365254Z" | |
| }, | |
| "title": "Facebook Activity Event Extraction System", | |
| "authors": [ | |
| { | |
| "first": "Yuan-Hao", | |
| "middle": [], | |
| "last": "\u6797\u5713\u7693", | |
| "suffix": "", | |
| "affiliation": { | |
| "laboratory": "", | |
| "institution": "National Central University", | |
| "location": {} | |
| }, | |
| "email": "" | |
| }, | |
| { | |
| "first": "", | |
| "middle": [], | |
| "last": "Lin", | |
| "suffix": "", | |
| "affiliation": { | |
| "laboratory": "", | |
| "institution": "National Central University", | |
| "location": {} | |
| }, | |
| "email": "" | |
| }, | |
| { | |
| "first": "Chia-Hui", | |
| "middle": [], | |
| "last": "\u5f35\u5609\u60e0", | |
| "suffix": "", | |
| "affiliation": { | |
| "laboratory": "", | |
| "institution": "National Central University", | |
| "location": {} | |
| }, | |
| "email": "" | |
| }, | |
| { | |
| "first": "", | |
| "middle": [], | |
| "last": "Chang", | |
| "suffix": "", | |
| "affiliation": { | |
| "laboratory": "", | |
| "institution": "National Central University", | |
| "location": {} | |
| }, | |
| "email": "" | |
| } | |
| ], | |
| "year": "", | |
| "venue": null, | |
| "identifiers": {}, | |
| "abstract": "The popularity of social networks has made them a perfect medium for activity or advertising campaign promotion. Therefore, many people use Facebook pages to announce their advertising campaign. The purpose of this study is to extract activity events by constructing two named entity recognition models, namely activity name and location, via a Web NER model generation tool [1]. We enhance the tool by improving the tokenizer and alignment technique. In addition, we also use a large database of FB checkin places for location name recognition improvement. For entity relation extraction, we apply sequential pattern mining to", | |
| "pdf_parse": { | |
| "paper_id": "O16-1022", | |
| "_pdf_hash": "", | |
| "abstract": [ | |
| { | |
| "text": "The popularity of social networks has made them a perfect medium for activity or advertising campaign promotion. Therefore, many people use Facebook pages to announce their advertising campaign. The purpose of this study is to extract activity events by constructing two named entity recognition models, namely activity name and location, via a Web NER model generation tool [1]. We enhance the tool by improving the tokenizer and alignment technique. In addition, we also use a large database of FB checkin places for location name recognition improvement. For entity relation extraction, we apply sequential pattern mining to", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Abstract", | |
| "sec_num": null | |
| } | |
| ], | |
| "body_text": [ | |
| { | |
| "text": "find rules for start date, end date, and location coupling. We use 1,300 posts from Facebook to test the activity event extraction performance. The experimental results show 0.727, 0.694 F 1score for activity name and location recognition; and 0.865, 0.72 F 1 -score for start and end date extraction. Overall, the extraction performance for activity event extraction is 0.708. Probability of s:", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "( ) = ( 0 ) \u00d7 ( 1 | 0 ) \u2026 \u00d7 ( \u22121 | \u22122 )", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "\u95dc\u9375\u8a5e\uff1a\u6d3b\u52d5\u4e8b\u4ef6\u64f7\u53d6\uff0c\u547d\u540d\u5be6\u9ad4\u8fa8\u8b58\uff0c\u793e\u7fa4\u5a92\u9ad4\u4e8b\u4ef6", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "Eq. ( ", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "\u95dc\u9375\u8a5e\uff1a\u6d3b\u52d5\u4e8b\u4ef6\u64f7\u53d6\uff0c\u547d\u540d\u5be6\u9ad4\u8fa8\u8b58\uff0c\u793e\u7fa4\u5a92\u9ad4\u4e8b\u4ef6", | |
| "sec_num": null | |
| } | |
| ], | |
| "back_matter": [], | |
| "bib_entries": { | |
| "BIBREF0": { | |
| "ref_id": "b0", | |
| "title": "A Tool for Web NER Model Generation Based on Google Snippets", | |
| "authors": [ | |
| { | |
| "first": "Y", | |
| "middle": [ | |
| "Y" | |
| ], | |
| "last": "Huang", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "C", | |
| "middle": [ | |
| "H" | |
| ], | |
| "last": "Chung", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2015, | |
| "venue": "Proceedings of the 27 th Conference on Computational Linguistics and Speech Processing", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "148--163", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Y. Y. Huang, C.H. Chung, \"A Tool for Web NER Model Generation Based on Google Snippets,\" Proceedings of the 27 th Conference on Computational Linguistics and Speech Processing, pp. 148-163, 2015.", | |
| "links": null | |
| }, | |
| "BIBREF1": { | |
| "ref_id": "b1", | |
| "title": "Foundations and Trends\u00ae in Databases", | |
| "authors": [ | |
| { | |
| "first": "Sunita", | |
| "middle": [], | |
| "last": "Sarawagi", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2008, | |
| "venue": "", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "261--377", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Sunita Sarawagi (2008), \"Information Extraction,\" Foundations and Trends\u00ae in Databases, pp. 261-377, 2008.", | |
| "links": null | |
| }, | |
| "BIBREF2": { | |
| "ref_id": "b2", | |
| "title": "Open domain event extraction from Twitter", | |
| "authors": [ | |
| { | |
| "first": "A", | |
| "middle": [], | |
| "last": "Ritter", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "O", | |
| "middle": [], | |
| "last": "Etzioni", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "S", | |
| "middle": [], | |
| "last": "Clark", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2012, | |
| "venue": "Proc. SIGKDD", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "1104--1112", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "A. Ritter, O. Etzioni, and S. Clark, \"Open domain event extraction from Twitter,\" Proc. SIGKDD, pp. 1104-1112, 2012.", | |
| "links": null | |
| }, | |
| "BIBREF3": { | |
| "ref_id": "b3", | |
| "title": "Chinese news event 5W1H semantic elements extraction for event ontology population", | |
| "authors": [ | |
| { | |
| "first": "Wang", | |
| "middle": [], | |
| "last": "Wei", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2012, | |
| "venue": "Proceedings of the 21 st International Conference Companion on World Wide Web", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "197--202", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Wei, Wang \"Chinese news event 5W1H semantic elements extraction for event ontology population,\" Proceedings of the 21 st International Conference Companion on World Wide Web, pp. 197-202, 2012.", | |
| "links": null | |
| }, | |
| "BIBREF4": { | |
| "ref_id": "b4", | |
| "title": "Identifying relevant temporal expressions for real-world events", | |
| "authors": [ | |
| { | |
| "first": "N", | |
| "middle": [], | |
| "last": "Kanhabua", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "S", | |
| "middle": [], | |
| "last": "Romano", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "A", | |
| "middle": [], | |
| "last": "Stewart", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2012, | |
| "venue": "Proceedings of the SIGIR 2012 Workshop on Time-aware Information Access", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "N. Kanhabua, S. Romano, and A. Stewart, \"Identifying relevant temporal expressions for real-world events,\" Proceedings of the SIGIR 2012 Workshop on Time-aware Information Access, 2012.", | |
| "links": null | |
| }, | |
| "BIBREF5": { | |
| "ref_id": "b5", | |
| "title": "Efficient Page-Level Data Extraction via Schema Induction and Verification", | |
| "authors": [ | |
| { | |
| "first": "T.-S", | |
| "middle": [], | |
| "last": "Chen", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "M.-C", | |
| "middle": [], | |
| "last": "", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Chen", | |
| "middle": [], | |
| "last": "", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "C.-H", | |
| "middle": [], | |
| "last": "", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Chang", | |
| "middle": [], | |
| "last": "", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2014, | |
| "venue": "Proceedings of the 15 th Conference on Technologies and Applications of Artificial Intelligencester", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "T.-S. Chen, M.-C, Chen, and C.-H, Chang, \"Efficient Page-Level Data Extraction via Schema Induction and Verification,\" Proceedings of the 15 th Conference on Technologies and Applications of Artificial Intelligencester, 2014.", | |
| "links": null | |
| }, | |
| "BIBREF6": { | |
| "ref_id": "b6", | |
| "title": "Deduplicating a places database", | |
| "authors": [ | |
| { | |
| "first": "N", | |
| "middle": [], | |
| "last": "Dalvi", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "M", | |
| "middle": [], | |
| "last": "Olteanu", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "M", | |
| "middle": [], | |
| "last": "Raghavan", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "P", | |
| "middle": [], | |
| "last": "Bohannon", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2014, | |
| "venue": "Proceedings of the 23rd international conference on World Wide Web", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "409--418", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "N. Dalvi, M. Olteanu, M. Raghavan, and P. Bohannon, \"Deduplicating a places database,\" Proceedings of the 23rd international conference on World Wide Web, pp. 409-418, 2014.", | |
| "links": null | |
| }, | |
| "BIBREF7": { | |
| "ref_id": "b7", | |
| "title": "On Chinese Postal Address and Associated Information Extraction", | |
| "authors": [ | |
| { | |
| "first": "C", | |
| "middle": [ | |
| "H" | |
| ], | |
| "last": "Chung", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "C.-Y", | |
| "middle": [], | |
| "last": "Huang", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Y.-Y.", | |
| "middle": [], | |
| "last": "Su", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2012, | |
| "venue": "Proceedings of the 26 th Annual Conference of the Japanese Society for Artificial Intelligence", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "C.H. Chung, C.-Y. Huang, and Y.-Y. Su, \"On Chinese Postal Address and Associated Information Extraction,\" Proceedings of the 26 th Annual Conference of the Japanese Society for Artificial Intelligence, 2012.", | |
| "links": null | |
| }, | |
| "BIBREF8": { | |
| "ref_id": "b8", | |
| "title": "Associated Information Extraction for Enabling Entity Search on Electronic Map", | |
| "authors": [ | |
| { | |
| "first": "Y. -S", | |
| "middle": [], | |
| "last": "Su", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2012, | |
| "venue": "", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Y. -S. Su, \"Associated Information Extraction for Enabling Entity Search on Electronic Map,\" National Central University, 2012.", | |
| "links": null | |
| }, | |
| "BIBREF9": { | |
| "ref_id": "b9", | |
| "title": "Heideltime: High quality rule-based extraction and normalization of temporal expressions", | |
| "authors": [ | |
| { | |
| "first": "J", | |
| "middle": [], | |
| "last": "Str\u00f6tgen", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "M", | |
| "middle": [], | |
| "last": "Gertz", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2010, | |
| "venue": "Proceedings of the 5 th International Workshop on Semantic Evaluation", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "J. Str\u00f6tgen, M. Gertz, \"Heideltime: High quality rule-based extraction and normalization of temporal expressions,\" Proceedings of the 5 th International Workshop on Semantic Evaluation, 2010.", | |
| "links": null | |
| } | |
| }, | |
| "ref_entries": { | |
| "FIGREF0": { | |
| "text": "Start Date rule (r1)Prefix word: \"\u65bc\" | \"\u6d3b\u52d5\u6642\u9593\" | \"\u5c55\u671f\" | \"\u6d3b\u52d5\u65e5\u671f\" \u2026 (r2)Suffix word: \"\u8d77\" | \"\u8209\u8fa6\" \u2026 (r5)Date Update: \"\u6d3b\u52d5\u6539\u5230\" \u2026\u2022 End Date rule (r3)Prefix word: \"\u622a\u6b62\u65e5\u671f\" | \"\u5373\u65e5\u8d77\u81f3\" \u2026 (r4)Suffix word: \"\u6b62\" | \"\u622a\u6b62\u5f8c\" \u2026 (r6)Date Update: \"\u6d3b\u52d5\u5ef6\u9577\u81f3\" \u2026 (r11)Negative rule: \"\u7d50\u6a19\u6642\u9593\" | \"\u539f\u8a02\u65bc\" \u2026\u2022 Special Date rule (r7)Confuse rule: \"\u8ca9\u8ce3\u6642\u9593\" | \"\u5fb5\u4ef6\u6642\u9593\" \u2026 \u2022 Location rule (r8)Prefix word: \"\u65bc\" | \"\u5047\" \u2026 (r9)Suffix word: \"\u5c55\u51fa\" | \"\u8209\u884c\" \u2026 (r10)LOC Update: \"\u65b0\u6d3b\u52d5\u5730\u9ede\" \u2026\u2022 Negative Activity rule: (r12)Cancel: \"\u6d3b\u52d5\u53d6\u6d88\" | \"\u66ab\u505c\u8209\u8fa6\" \u2026 \u8a3b: \u7070\u5e95\u662f\u4eba\u5de5\u5efa\u7acb\u7684\u898f\u5247 \u5e02\u300c\u4e3b\u59d4\u76c3\u300d\u5168\u570b\u9752\u5c11\u5e74 14 \u6b72\u7d1a\u7db2\u7403\u9326 1 \u6a19\u8cfd(C-2)\u2500\u57fa\u9686\u300d\u53ef\u80fd\u53ea\u8b58\u5225\u51fa\u90e8\u5206\u300c\u4e3b \u59d4\u76c3\u300d\u5168\u570b\u9752\u5c11\u5e74 14 \u6b72\u7d1a\u7db2\u7403\u9326\u6a19\u8cfd\u300d \uff0c\u53c3\u8003 Huang \u8a55\u4f30\u65b9\u6cd5\u5c0d\u65bc\u6bcf\u500b\u8fa8\u8b58\u5230\u7684\u547d\u540d \u5be6\u9ad4 e \u8207\u6b63\u78ba\u7b54\u6848\u7684\u547d\u540d\u5be6\u9ad4 a \u9032\u884c\u547d\u540d\u5be6\u9ad4\u8fa8\u8b58\u6548\u80fd\u7684\u8a55\u4f30\uff0c\u4e26\u5b9a\u7fa9 Eq. (3) P(e,a)\u3001 R(e,a)\u5206\u6578\uff0c\u52a0\u7e3d\u5206\u6578\u5f8c\u53d6\u5e73\u5747\u503c\u5f97\u5230\u6574\u9ad4\u7684 Eq. (4) Precision\u3001Recall \u8207 Eq. (5) F 1 -score\u3002", | |
| "type_str": "figure", | |
| "uris": null, | |
| "num": null | |
| }, | |
| "FIGREF1": { | |
| "text": "EventPrecision = ( P(ActSet, EAct) + I(Start, EStart) + I(End, EEnd) , + MP(Loc/Addr, ELoc/EAddr) ) / k Eq. (9) = ( R(ActSet, EAct) + I(Start, EStart) + I(End, EEnd) , + MR(Loc/Addr, ELoc/EAddr)", | |
| "type_str": "figure", | |
| "uris": null, | |
| "num": null | |
| }, | |
| "TABREF0": { | |
| "text": "Keywords: Activity Event Extraction, Named Entity Recognition, Social Media Event Places\u3002\u7531\u65bc FB \u6c92\u6709\u50cf Twitter \u63d0\u4f9b\u5c0d\u6574\u500b FB \u7db2\u7ad9\u767c\u6587\u7684\u641c\u5c0b API\uff0c\u56e0 \u6b64\u6211\u5011\u53ea\u80fd\u5229\u7528 FB Graph API \u5c0d 22 \u842c\u500b\u53f0\u7063\u516c\u958b\u7c89\u7d72\u5c08\u9801(\u900f\u904e\u89e3\u6790 1,400 \u842c\u7b46 FB \u65b9\u6cd5\uff0c\u52a0\u5165\u53ef\u81ea\u8a02\u7fa9\u7684 tokenizer \u6a21\u7d44\uff0c\u5141\u8a31\u5c0d token \u505a \u53bb\u8a5e\u5e79\u8a2d\u5b9a\u3002\u6392\u6bd4\u65b9\u5f0f\u662f\u53c3\u8003 T.-S. Chen[6]\u63d0\u51fa\u7684 Global alignment\uff0c\u91dd\u5c0d\u9577\u5ea6\u5927\u65bc k \u7684 \u6d3b\u52d5\u540d\u7a31\u6392\u6bd4\u6a19\u8a18\u689d\u4ef6\u5305\u62ec(1)\u4e0d\u5141\u8a31\u5169\u6392\u6bd4\u5e8f\u5217\u4e2d\u7684\u5b57\u5143 mismatch \u7684\u5c0d\u61c9\uff0c(2)\u5169\u76f8\u9130 matched token \u4e4b\u9593\u51fa\u73fe Gap \u6578\u81f3\u591a\u70ba MaxG\uff0c\u4e14(3)\u91cd\u8986\u7684 token \u6bd4\u4f8b\u5fc5\u9808\u5927\u65bc\u9580\u6abb\u503c r\uff0c\u6eff\u8db3\u4ee5\u4e0a\u4e09\u8005\u689d\u4ef6\u7684\u6bd4\u5c0d\u7cfb\u7d71\u5373\u5c07\u5176\u6a19\u8a18\u70ba\u51fa\u73fe\u7bc4\u4f8b\u3002 \u64f7\u53d6\u51fa\u7684\u5be6\u9ad4\u7cbe\u6e96\u5ea6\u3002 \u70ba\u4e86\u6574\u4f75 word-based \u548c character-based \u65b9\u6cd5\u4e26\u63d0\u4f9b\u66f4\u8c50\u5bcc\u524d\u8655\u7406\u529f\u80fd\u65b0\u589e Tokenizer \u6a21\u7d44\uff0c\u5be6\u4f5c\u900f\u904e\u79fb\u690d\u4e86 Lucene \u7684 Analyzer\uff0c\u53ef\u81ea\u884c\u66ff\u63db\u6839\u64da\u4efb\u52d9\u6240\u9700\u4e26\u81ea\u884c\u8a2d\u5b9a\u8a5e\u5e79 \u8a2d\u5b9a\u3002\u7576\u63a1\u7528 character based \u6642\uff0c\u6211\u5011\u4f7f\u7528 Jflex \u9019\u5957\u5de5\u5177\u5b9a\u7fa9\u5e38\u898b\u7684 token \u578b\u614b(\u5305\u62ec money\u3001alphanum\u3001Chinese or Japan\u3001URL\u3001E-mail \u7b49\u5171 20 \u500b)\uff0c\u800c\u7576\u63a1\u7528 word based \u6642 \u5247\u9810\u8a2d\u4f7f\u7528 IK Analyzer\uff0c\u4e0d\u904e\u7531\u65bc IK Analyzer \u6703\u6ffe\u9664\u672a\u5b9a\u7fa9\u7684 token\uff0c\u9020\u6210 token \u907a \u9ad8\u3002\u5b8c\u6574\u7684\u7d66\u5206\u65b9\u5f0f\u5b9a\u7fa9\u5982\u4e0b\u5716\u4e94\u3002 \u5176\u4e2d CoreFind \u76ee\u7684\u70ba\u627e\u51fa\u8f38\u5165\u5b57\u4e32\u4e2d\u53ef\u80fd\u51fa\u73fe\u7684\u6838\u5fc3\u5b57\u8a5e\u96c6\u5408(\u5229\u7528 MSRA NER \u6a21\u578b \u6a19\u8a18\u7684\u5be6\u9ad4\u52a0\u4e0a CoreDic \u6a19\u8a18\u7684\u5b57\u8a5e) \uff0cBgFind \u900f\u904e\u4e2d\u6587\u65b7\u8a5e\u4e26\u6392\u9664\u6838\u5fc3\u5b57\uff0c\u8207 LocBgDic \u4ea4\u96c6\u5f97\u5230\u7684\u80cc\u666f\u5b57\u8a5e\u96c6\u5408\uff0c\u5269\u9918\u4e0d\u5c6c\u65bc\u6838\u5fc3\u5b57\u548c\u80cc\u666f\u5b57\u7684\u96c6\u5408\u6211\u5011\u5c07\u5176\u5b9a\u7fa9\u70ba\u63cf\u8ff0\u5b57\u3002 CoreFind, BgFind \u53ca Descriptor \u4e09\u500b\u6a21\u7d44\u806f\u96c6\u6240\u5f97\u7684\u5b57\u8a5e\u6578\u5373\u662f Count \u51fd\u6578\u56de\u50b3\u503c\u3002\u6f14 \u7b97\u6cd5\u4e3b\u8981\u662f\u8a08\u7b97\u51fa n-gram \u53ca placei \u5171\u540c\u7684\u5be6\u9ad4\u6838\u5fc3\u8a5e CoreSet(\u53bb\u6389\u6c92\u6709\u51fa\u73fe\u5728 placei \u4e2d \u6838\u5fc3\u8a5e)\uff0c\u4ee5\u53ca\u5171\u540c\u80cc\u666f\u8a5e BgSet\uff0c\u4e26\u4f9d Eq.(1)\u8a08\u7b97 n-gram \u8207 placei \u7684\u76f8\u4f3c\u5ea6\uff0c\u82e5\u76f8\u4f3c\u5ea6 \u5927\u65bc\u9580\u6abb\u503c\uff0c\u5247\u7528\u8a72\u5730\u9ede placei \u9032\u884c Partial Alignment \u6a19\u8a18\u9019\u500b\u53e5\u5b50\uff0c\u9032\u884c\u6700\u5b8c\u6574\u6a19\u8a18\u3002 -score \u7d04\u5728 0.94 \u81f3 0.99 \u5340\u9593\u3002\u53e6\u5916\u672c\u6587\u7814\u7a76\u4e3b \u8981\u6587\u672c\u64f7\u53d6\u5c0d\u8c61\u70ba FB Post\uff0c\u4f46\u672a\u4f86\u4e8b\u4ef6\u64f7\u53d6\u7684\u4efb\u52d9\u6703\u64f4\u5145\u6210 Web Data Extraction\uff0c\u70ba\u4e86 \u76f8\u5bb9 HTML \u7db2\u9801(semi-structured)\u548c FB Post (free text)\u63a1\u7528 Su [9]\u7cfb\u7d71\u6a21\u7d44\u64f7\u53d6\u53f0\u7063\u5730\u5740\u3002 \u8868\u56db\u3001\u5730\u5740\u64f7\u53d6\u6a21\u578b\u8fa8\u8b58\u7279\u5fb5\u503c", | |
| "num": null, | |
| "html": null, | |
| "content": "<table><tr><td>\u73fe\u8c61\u5f80\u5f80\u6703\u4f7f\u5f97\u8cc7\u8a0a\u7f3a\u5c11\u6574\u5408\u8207\u548c\u4f7f\u7528\u8005\u7684\u4e92\u52d5\u3002\u5982\u679c\u80fd\u5c07\u4e0d\u540c\u7ba1\u9053\u7684\u8cc7\u8a0a\u5982 CityTalk\u3001 \u51fa\u4f86\u7684\u5143\u7d20\u586b\u5165\u5230 News Ontology Event model \u4f9b\u5f8c\u7e8c\u7684\u5229\u7528\u3002 \u6d3b\u52d5\u4e8b\u4ef6\u64f7\u53d6\u7684\u8f38\u51fa\u3002 \u7684\u5be6\u9ad4\u548c\u539f\u6587\u662f\u4e00\u81f4\u7684\u3002\u6211\u5011\u8981\u4fdd\u8b49\u6240\u6709\u5c0d\u6587\u672c\u505a\u7684\u524d\u8655\u7406\u90fd\u4e0d\u6703\u5f71\u97ff\u8b58\u5225\u51fa\u7684\u5be6\u9ad4\u3002 \u8f03\u9ad8\u7684\u512a\u5148\u6b0a\uff0c\u5982\u679c\u540c\u6a23\u9577\u5ea6\u5247\u770b Pattern \u7684\u89f8\u767c\u6a5f\u7387\uff0c\u8a08\u7b97\u65b9\u5f0f\u662f\u91dd\u5c0d Pattern s =</td></tr><tr><td>N. Kanhabua[5]\u63d0\u51fa\u7684\u7814\u7a76\u662f\u8ddf\u516c\u5171\u885b\u751f\u4e8b\u4ef6\u76f8\u95dc\u7684\u4e8b\u4ef6\u64f7\u53d6\uff0c\u4e3b\u984c\u662f\u75be\u75c5\u7206\u767c\u7684\u4e8b\u4ef6 \u8868\u4e8c\u3001\u6d3b\u52d5\u4e8b\u4ef6\u7684\u95dc\u4fc2 \u8868\u4e09\u3001\u4e0d\u540c\u65b9\u6cd5 tokenizer \u81ea\u52d5\u7522\u751f\u7684\u6d3b\u52d5\u540d\u7a31\u8fa8\u8b58\u7279\u5fb5\u503c\u7bc4\u4f8b \u6d3b\u52d5\u901a\u3001\u653f\u5e9c\u3001\u5b78\u6821\u7db2\u7ad9\u516c\u544a\u548c\u793e\u7fa4\u5a92\u9ad4\u505a\u7d50\u5408\uff0c\u4fbf\u53ef\u4ee5\u4f9d\u6d3b\u52d5\u7684\u53d7\u6b61\u8fce\u7a0b\u5ea6\u548c\u8a0e\u8ad6\u7a0b Activity Name Start Date End Date Location/Address ID \u8aaa\u660e \u9577 character based word based \u64f7\u53d6\u7cfb\u7d71\uff0c\u75be\u75c5\u7206\u767c\u7684\u4e8b\u4ef6\u5b9a\u7fa9\u5982\u8868\u4e00\uff0c\u5305\u62ec\u53d7\u5bb3\u4eba\u3001\u75be\u75c5\u3001\u6642\u9593\u53ca\u5730\u9ede\uff0c\u4e3b\u8981\u8981\u89e3\u6c7a \u300c\u4e16\u754c\u6587\u5316\u907a\u7522\u91cd\u6176\u5927\u8db3\u77f3 \u660e\u665a 3 \u6708 13 \u65e5 \u51ac\u5c71\u6cb3\u89aa\u6c34\u516c\u5712 \u2026 \u2026 \u2026 \u2026 \u2026 \u5ea6\uff0c\u63d0\u4f9b\u4e00\u500b\u6d3b\u52d5\u5730\u5716\u670d\u52d9(\u5982\u5716\u4e00\u6240\u793a) \uff0c\u50cf\u662f\u6d3b\u52d5\u7684\u8a55\u8ad6\u3001\u6d3b\u52d5\u526a\u5f71\u548c\u6d3b\u52d5\u7684\u5716\u7247/ \u5f71\u7247\uff0c\u9019\u4e9b\u5c0d\u65bc\u4e86\u89e3\u6d3b\u52d5\u9032\u884c\u548c\u53c3\u8207\u6709\u5f88\u5927\u7684\u5e6b\u52a9\u3002\u6240\u4ee5\u6574\u5408\u73fe\u6709\u7684\u6d3b\u52d5\u516c\u544a\u7db2\u7ad9\uff0c\u4e26 \u548c\u793e\u7fa4\u5a92\u9ad4\u505a\u7d50\u5408\u662f\u672c\u7814\u7a76\u7684\u76ee\u6a19\u3002 \u5728\u672c\u7814\u7a76\u7576\u4e2d\uff0c\u6211\u5011\u5c08\u6ce8\u65bc Facebook \u6d3b\u52d5\u4e8b\u4ef6\u7684\u64f7\u53d6\uff0c\u4e26\u63d0\u51fa\u65b9\u6cd5\u5f9e\u7c89\u7d72\u9801\u767c\u6587\u4e2d\u64f7 \u523b\u5f69\u71c8\u66a8\u727d\u624b\u5609\u5e74\u83ef\u300d (2016-02-06) (2016-03-13) 10 \u5e38\u898b\u65bc\u6d3b\u52d5\u540d\u7a31\u524d\u65b9\u7684 token 1 \u5230\u3001\u5728 \u8209\u8fa6\u3001\u53c3\u52a0 \u7684\u554f\u984c\u662f\u627e\u5230\u75be\u75c5\u548c\u75be\u75c5\u91cd\u8981\u7684\u6642\u7a0b\u8868\u9054\u5f0f\uff0c\u5b9a\u7fa9\u9019\u6a23\u7684\u554f\u984c\u70ba\u5206\u985e\u554f\u984c\uff0c\u4e26\u63d0\u51fa\u4e86\u76f8 \u95dc\u6392\u5e8f(relevance ranking) \u65b9\u6cd5\u53bb\u5224\u5225\u91cd\u8981\u7684\u6642\u7a0b\u8868\u9054\u5f0f\u3002\u5728\u4e0d\u540c\u5206\u985e\u65b9\u6cd5\u4e2d\u6700\u597d\u7684\u6548 \u679c\u662f\u63a1\u7528 J48\uff0c\u6e96\u78ba\u7387(accuracy)\u80fd\u9054\u5230 0.65\u3002 \u8868\u4e00\u3001\u65b0\u805e\u4e8b\u4ef6\u7684 5W1H \u5143\u7d20\u4e8b\u4ef6\u5c6c\u6027\u8aaa\u660e\u548c\u75be\u75c5\u7206\u767c\u7684\u4e8b\u4ef6\u5b9a\u7fa9 11 \u5e38\u898b\u65bc\u6d3b\u52d5\u540d\u7a31\u524d\u65b9\u7684 token 2 \u8209\u8fa6\u3001\u63a8\u51fa \u63a8\u51fa/\u300c\u3001\u4e00\u5e74\u4e00\u5ea6/\u7684 12 \u5e38\u898b\u65bc\u6d3b\u52d5\u540d\u7a31\u524d\u65b9\u7684 token 3 \u6d3b\u52d5\uff1a\u3001\u53c3\u52a0\u300c ID Feature ICCS ID Feature ICCS \u6d3b\u52d5/\u540d\u7a31/:\u3001\u7bc0\u76ee/\u540d\u7a31/: 13 \u5e38\u898b\u65bc\u6d3b\u52d5\u540d\u7a31\u5f8c\u65b9\u7684 token 1 \u3011\u3001\u300d 1 CountyCity \u7e23\u3001\u5e02 10 ChineseNo \u4e00\u3001\u4e8c\u3001\u4e09 \u5373\u65e5\u8d77\u3001\u958b\u5e55\u5f0f 14 \u5e38\u898b\u65bc\u6d3b\u52d5\u540d\u7a31\u5f8c\u65b9\u7684 token 2 \u8d77\u8dd1\u3001\u672c\u6b21 2 Township \u93ae\u3001\u9109\u3001\u5340, 11 AllDigits 42011\u30010937137659 \u8868\u6f14/\u6d3b\u52d5\u3001\u660e\u65e5/\u767b\u5834 15 \u5e38\u898b\u65bc\u6d3b\u52d5\u540d\u7a31\u5f8c\u65b9\u7684 token 3 \u4f86\u56c9~ 3 Village \u6751\u3001\u91cc\u3001\u9130 12 DigitLen1 5\u30016 \u71b1\u70c8/\u958b\u8dd1/\u56c9\u3001 /\u958b\u5e55/\u76db\u6cc1 \u2026 \u2026 \u2026 \u2026 4 StreetRoad \u9053\u3001\u8def\u3001\u8857 13 DigitLen2 11\u300132 \u2026 5 LaneAlley \u6bb5\u3001\u5df7\u3001\u5f04 14 DigitLen3 420\u3001260</td></tr><tr><td>\u4e00\u3001 \u7dd2\u8ad6 \u53d6\u6d3b\u52d5\u53ca\u5176\u91cd\u8981\u8cc7\u8a0a\uff0c\u7cfb\u7d71\u5c07\u64f7\u53d6\u51fa\u4f86\u7684\u6d3b\u52d5\u4e8b\u4ef6\u7d50\u5408\u96fb\u5b50\u5730\u5716\u8207\u6642\u9593\u8ef8\uff0c\u5e6b\u52a9\u4f7f\u7528\u8005 5W1H News 5W1H Event Disease Event e: (v, m, l, t) What \u62b5\u9054 disease m 6 HouseNo \u865f 15 DigitLen4 5566\u30011234 (\u4e94) \u5730\u9ede\u5be6\u9ad4\u7684\u8b58\u5225 7 Building \u6a13\u3001\u5ba4 16 DigitLen5 42011</td></tr><tr><td>\u7db2\u969b\u7db2\u8def\u767c\u5c55\u6539\u8b8a\u4e86\u4eba\u5011\u7372\u5f97\u6d3b\u52d5\u8a0a\u606f\u7684\u7fd2\u6163\uff0c\u5728\u904e\u53bb\u662f\u85c9\u7531\u96fb\u8996\u3001\u5e73\u9762\u5a92\u9ad4\u3001\u5ee3\u64ad\u5a92 \u9ad4\u7684\u5ba3\u50b3\u4f86\u5f97\u77e5\u6d3b\u52d5\u8cc7\u8a0a\uff0c\u4f46\u96a8\u8457\u7db2\u8def\u7684\u9032\u6b65\u548c\u793e\u7fa4\u5e73\u53f0\u7684\u84ec\u52c3\u767c\u5c55\uff0c\u5546\u5bb6\u6709\u4e86\u65b0\u7684\u50b3 \u64ad\u9014\u5f91\uff0c\u4e0d\u50c5\u589e\u52a0\u4e86\u8207\u5ba2\u6236\u9593\u7684\u4e92\u52d5\u6027\uff0c\u4e5f\u53ef\u4ee5\u66f4\u5feb\u56de\u61c9\u5ba2\u6236\u7684\u554f\u984c\u3002 \u5f97\u77e5\u6d3b\u52d5\u6216\u512a\u60e0\u7684\u8a0a\u606f\u662f\u65e5\u5e38\u751f\u6d3b\u53ca\u65c5\u904a\u4f11\u9592\u898f\u5283\u91cd\u8981\u7684\u4e00\u74b0\uff0c\u82e5\u80fd\u53d6\u5f97\u8f03\u591a\u7684\u6d3b\u52d5\u8cc7 \u4e86\u89e3\u7cfb\u7d71\u64f7\u53d6\u51fa\u4f86\u7684\u6d3b\u52d5\u4e8b\u4ef6\u3002 \u4e8c\u3001 \u76f8\u95dc\u7814\u7a76 When 8 \u65e5 time t Where \u6e25\u53f0\u83ef \u52a0\u62ff\u5927 8 ContactTag \u5730\u3001\u5740\u3001\u96fb\u3001\u8a71 17 DigitLong 327363\u30014227151 \u932f\u8aa4\u6a19\u8a18\u7684\u5730\u9ede\u6703\u5f71\u97ff\u5224\u5b9a\u6d3b\u52d5\u5730\u9ede\uff0c\u5c0d\u65bc\u5730\u9ede\u8fa8\u8b58\u6a21\u578b\u6211\u5011\u66f4\u770b\u91cd\u7cbe\u78ba\u7387(precision) 9 Punctuation \u3001\u3001\uff1a\u3001\uff1b location l Who \u4e2d\u570b\u570b\u5bb6\u4e3b\u5e2d\u80e1\u9326\u6fe4 victim v Whom \u52a0\u62ff\u5927\u9996\u90fd\u6e25\u53f0\u83ef How \u6d3b\u52d5\u6642\u9593\u5728\u767c\u6587\u4e2d\u6703\u4ee5\u6bd4\u8f03\u53e3\u8a9e\u7684\u65b9\u5f0f\u63d0\u53ca\u5982: \u672c\u5468\u516d\u3001\u660e\u5929\u4e0b\u5348\u7b49\u9019\u4e9b\u8868\u9054\u6d3b\u52d5\u6642\u9593 \u548c\u5be6\u9ad4\u908a\u754c\uff0c\u7cbe\u78ba\u7387\u662f\u4e0d\u8981\u932f\u8aa4\u5730\u8b58\u5225\u4e00\u4e9b\u5730\u9ede\uff0c\u63a1 CRF \u65b9\u6cd5\u53ec\u56de\u7387\u5f88\u9ad8\u4f46\u8b58\u5225\u932f\u8aa4 \u5716\u4e09\u3001\u4e8b\u4ef6\u64f7\u53d6\u7bc4\u4f8b\u8aaa\u660e \u7684\u9673\u8ff0\u4e5f\u662f\u7cfb\u7d71\u64f7\u53d6\u7684\u76ee\u6a19\uff0c\u6a19\u8a18\u7684\u6642\u9593\u91dd\u5c0d\u6642\u7a0b\u8868\u9054\u5f0f\u9032\u884c\u6a19\u8a18\uff0c\u8a72\u6a21\u7d44\u53c3\u8003 \u4e2d\u570b\u570b\u5bb6\u4e3b\u5e2d\u80e1\u9326\u6fe4\u62b5\u9054\u52a0\u62ff\u5927\u9996\u90fd\u6e25\u592a\u83ef\u9032\u884c\u8ecd\u4e8b\u8a2a\u554f (\u4e8c) \u8cc7\u6599\u6536\u96c6 \u4f8b\u5b50\u4e5f\u5f88\u591a\u4f8b\u5982:\"\u514d\u8cbb\"\u8b58\u5225\u6210\u5730\u9ede\uff0c\u53e6\u5916\u8b58\u5225\u51fa\u4f86\u7684\u5730\u9ede\u6709\u6642\u5019\u908a\u754c\u4e0d\u662f\u90a3\u9ebc\u5b8c\u6574 Heideltime[10]\uff0c\u52a0\u5165\u4e86\u53e3\u8a9e\u7684\u6642\u9593\u8868\u9054\u898f\u5247\u548c\u6b63\u898f\u5316\u6587\u5b57\u6b65\u9a5f(\u542b\u5168\u5f62\u5b57\u8f49\u534a\u5f62\u5b57) \uff0c \u4e09\u3001 \u7cfb\u7d71\u67b6\u69cb \u5728\u8cc7\u6599\u6536\u96c6\u65b9\u9762\uff0c\u7cfb\u7d71\u4e3b\u8981\u6703\u6536\u96c6 FB\u3001CityTalk \u53ca\u641c\u5c0b\u5f15\u64ce\u7684\u6d3b\u52d5\u95dc\u9375\u5b57\u67e5\u8a62\u7d50\u679c\u3002 \u5982: \"\u4e09\u7063\u9109\u4e94\u7a40\u5edf\u524d\u5ee3\u5834\uff02\u88ab\u8b58\u5225\u70ba\"\u4e09\u7063\uff02\u548c\"\u5edf\u524d\uff02\uff0c\u9019\u4e9b\u539f\u56e0\u90fd\u53ef\u80fd\u9020\u6210\u5730\u9ede \u89e3\u6c7a FB \u4e0a\u66f8\u5beb\u904e\u65bc\u81ea\u7531\u9020\u6210\u898f\u5247\u5339\u914d\u5931\u6557\u7684\u554f\u984c\uff0c\u4e26\u5229\u7528\u5339\u914d\u5230\u7684\u6642\u9593\u898f\u5247\u8f49\u63db\u6210\u660e \u672c\u7814\u7a76\u7684\u7cfb\u7d71\u67b6\u69cb\u5982\u5716\u4e8c\uff0c\u7cfb\u7d71\u9996\u5148\u9032\u884c\u8cc7\u6599\u7684\u8490\u96c6\uff0c\u5305\u62ec CityTalk \u548c FB \u7684\u767c\u6587\uff0c\u63a5 \u8a0a\uff0c\u5c31\u80fd\u65b9\u4fbf\u4eba\u5011\u66f4\u6709\u6548\u7684\u898f\u5283\u3002Facebook \u662f\u76ee\u524d\u53f0\u7063\u6700\u6d41\u884c\u7684\u793e\u7fa4\u5e73\u53f0\uff0c\u6bcf\u5929\u6709\u4e00 \u534a\u4ee5\u4e0a\u7684\u53f0\u7063\u4eba\u5728\u4f7f\u7528\u624b\u6a5f\uff0c\u9019\u6a23\u7684\u73fe\u8c61\u4f7f\u5f97\u4e0d\u5c11\u5546\u5bb6\u3001\u653f\u5e9c\u3001\u7d44\u7e54\u6703\u81ea\u5df1\u7d93\u71df\u7684\u7c89\u7d72 \u9801\uff0c\u4e26\u5728\u7c89\u7d72\u9801\u4e0a\u767c\u4f48\u6d3b\u52d5\u8a0a\u606f\u6216\u5546\u5bb6\u512a\u60e0\uff0c\u6211\u5011\u767c\u73fe\u9019\u4e9b\u8ddf\u6d3b\u52d5\u6709\u95dc\u7684\u8a0a\u606f\u6578\u91cf\u6703\u6bd4 \u4e00\u4e9b\u73fe\u6709\u7684\u6d3b\u52d5\u516c\u544a\u7db2\u7ad9(\u5982 CityTalk \u6216\u6d3b\u52d5\u901a)\u4f86\u7684\u66f4\u591a\u548c\u5373\u6642\uff0c\u56e0\u6b64\u672c\u6587\u7684\u76ee\u7684\u5373 \u70ba\u5f9e Facebook \u64f7\u53d6\u6d3b\u52d5\u8a0a\u606f\u3002 \u5716\u4e00\u3001\u6d3b\u52d5\u8a0a\u606f\u64f7\u53d6\u53ca\u6d3b\u52d5\u5730\u5716\u670d\u52d9\u7684\u5448\u73fe \u65e9\u671f\u793e\u7fa4\u5e73\u53f0\u5c1a\u672a\u84ec\u52c3\u767c\u5c55\u6642\uff0c\u60f3\u8981\u7372\u53d6\u6d3b\u52d5\u8a0a\u606f\u53ea\u80fd\u4ef0\u8cf4 CityTalk\u3001\u6d3b\u52d5\u901a\u9019\u985e\u7db2\u7ad9 \u67e5\u8a62\u6d3b\u52d5\u4e8b\u4ef6\u8cc7\u8a0a\uff0c\u5426\u5247\u5c31\u53ea\u80fd\u8f9b\u82e6\u5730\u700f\u89bd\u653f\u5e9c\u3001\u5b78\u6821\u7db2\u7ad9\u516c\u544a\u4f86\u5f97\u77e5\u4e8b\u4ef6\u3002\u800c\u9019\u6a23\u7684 Based)\u6216\u662f\u7d71\u8a08 \u5f0f(Statistical Model)\u3002 (\u4e00) \u793e\u7fa4\u5a92\u9ad4\u4e8b\u4ef6\u64f7\u53d6\u7cfb\u7d71 \u5e38\u898b\u7684\u4e8b\u4ef6\u64f7\u53d6\u505a\u6cd5\uff0c\u4e3b\u8981\u662f\u5229\u7528\u547d\u540d\u5be6\u9ad4\u8fa8\u8b58(NER)\u7684\u6280\u8853\u53bb\u8b58\u5225\u6587\u7ae0\u4e2d\u548c\u4e8b\u4ef6\u76f8\u95dc \u7684\u5be6\u9ad4\uff0c\u4e26\u9032\u884c association task \u8b58\u5225\u51fa\u5be6\u9ad4\u9593\u95dc\u4fc2\uff0c\u63a5\u8457\u900f\u904e\u4eba\u5de5\u5efa\u7acb\u7684\u898f\u5247\u53bb\u64f7\u53d6\u5b9a \u7fa9\u7684\u4e8b\u4ef6\u3002\u8209\u4f8b\u800c\u8a00\uff0cTweets calendar \u7cfb\u7d71[3]\u7684\u64f7\u53d6\u7684\u76ee\u6a19\u662f Twitter \u4e0a\u958b\u653e\u9818\u57df\u7684\u4e8b \u4ef6\uff0c\u5b9a\u7fa9\u4e8b\u4ef6\u70ba(Entity, Event Phrase, Date, Type)\uff0c\u5176\u76ee\u6a19\u662f\u64f7\u53d6\u4eba\u7269\u3001\u4e8b\u4ef6\u53ca\u65e5\u671f\u7b49\u4e09 \u7a2e\u8cc7\u8a0a\uff0c\u518d\u5c07\u4e8b\u4ef6\u5206\u985e\uff0c\u7d44\u5408\u6210\u4e8b\u4ef6\u7684 4-tuple \u5c6c\u6027\u5f97\u5230\u5982:(Steve Jobs, died, 10/6/11, DEATH)\u7684\u4e8b\u4ef6\u8cc7\u8a0a\u3002\u5176\u4f5c\u6cd5\u662f\u6a19\u8a18\u5b8c\u547d\u540d\u5be6\u9ad4\u4e4b\u5f8c\uff0c\u5229\u7528\u5361\u65b9\u6e2c\u5b9a(Chi square)\u4f86\u505a\u95dc \u4fc2\u7684\u9a57\u8b49\uff0c\u4ee5\u5f37\u5316\u5be6\u9ad4\u548c\u6642\u9593\u95dc\u4fc2\u4e26\u5f97\u51fa\u524d 100\u3001500\u30011,000 tuple\uff0c\u7d44\u5408\u6210\u5b8c\u6574\u7684\u4e8b\u4ef6 \u95dc\u4fc2\u3002\u63db\u8a00\u4e4b\uff0c\u4e8b\u4ef6\u5fc5\u9808\u88ab\u773e\u4eba\u591a\u6b21\u63d0\u53ca\uff0c\u624d\u6709\u8db3\u5920\u8cc7\u8a0a\u8b49\u660e\u4e8b\u4ef6\u70ba\u771f\u3002 (\u4e8c) \u65b0\u805e\u7684\u4e8b\u4ef6\u64f7\u53d6\u7cfb\u7d71 \u53e6\u5916\uff0cWang[4]\u5247\u63d0\u51fa\u4e00\u500b\u61c9\u7528\u65bc\u65b0\u805e\u5a92\u9ad4\u4e0a\u7684\u4e8b\u4ef6\u64f7\u53d6\u7cfb\u7d71\uff0c\u5176\u76ee\u7684\u662f 5W1H \u7684\u8a9e\u610f \u5c64\u7d1a\u7684\u5143\u7d20\u64f7\u53d6\u7cfb\u7d71\uff0c\u65b0\u805e\u4e8b\u4ef6\u7684 5W1H \u5143\u7d20\u4e8b\u4ef6\u5c6c\u6027\u5b9a\u7fa9\u5982\u8868\u4e00\u3002\u65b9\u6cd5\u662f\u8a2d\u8a08\u4e00\u4e9b \u7279\u5fb5(feature)\u5f9e\u65b0\u805e\u6a19\u984c\u4e2d\u627e\u51fa\u65b0\u805e\u4e2d\u7684\u4e3b\u984c\u53e5\uff0c\u63a5\u8457\u900f\u904e\u8a9e\u7fa9\u89d2\u8272\u6a19\u8a3b\uff0c\u6700\u5f8c\u5c07\u64f7\u53d6 FB \u7db2\u7ad9\u4e0a\u7684\u8cc7\u8a0a\uff0c\u5305\u62ec\u6253\u5361\u5730\u9ede\u3001\u53f0\u7063\u516c\u958b\u7c89\u7d72\u5c08\u9801\u3001\u548c FB \u4e8b\u4ef6\u7684\u767c\u6587\uff0c\u5176\u4e2d\u6253\u5361\u5730 \u78ba\u7684\u5e74\u6708\u65e5\u7684\u65e5\u671f\u683c\u5f0f\uff0c\u518d\u4fee\u5fa9\u8ddf\u9031\u76f8\u95dc\u898f\u5247\u6709\u95dc\u7684\u81ed\u87f2\u3002 \u8457\u900f\u904e\u6d3b\u52d5\u76f8\u95dc\u7684\u95dc\u9375\u5b57\u53d6\u5f97\u8ddf\u6d3b\u52d5\u8f03\u70ba\u76f8\u95dc\u7684\u767c\u6587\uff0c\u4e26\u5229\u7528\u6d3b\u52d5\u540d\u7a31\u8fa8\u8b58\u7684\u6a21\u578b\u9032\u884c \u6a19\u8a18\uff0c\u53ea\u6709\u5305\u542b\u6d3b\u52d5\u540d\u7a31\u7684\u8cbc\u6587\u624d\u6703\u5229\u7528\u6642\u7a0b\u8868\u9054\u5f0f\u3001\u5730\u9ede\u3001\u5730\u5740\u8fa8\u8b58\u7684\u6a21\u578b\u9032\u884c\u6a19\u8a18\u3002 \u6700\u5f8c\u5229\u7528\u4e8b\u4ef6\u95dc\u4fc2\u8026\u5408\u7684\u6a21\u7d44\u5c07\u767c\u6587\u7684\u6d3b\u52d5\u4e8b\u4ef6\u7684\u95dc\u4fc2\u627e\u51fa\u4f86\uff0c\u4e26\u653e\u5230\u4e8b\u4ef6\u7684\u8cc7\u6599\u5eab\uff0c \u4e26\u63d0\u4f9b\u4ecb\u9762\u7d66\u4f7f\u7528\u8005\u67e5\u770b\u64f7\u53d6\u51fa\u4f86\u7684\u6d3b\u52d5\u8cc7\u8a0a\u3002 \u5716\u4e8c\u3001\u7cfb\u7d71\u67b6\u69cb\u5716 (\u4e00) \u6d3b\u52d5\u4e8b\u4ef6\u5b9a\u7fa9 \u6d3b\u52d5\u4e8b\u4ef6\u7684\u5b9a\u7fa9\u53ef\u7531\u6d3b\u52d5\u540d\u7a31\u3001\u958b\u59cb\u3001\u7d50\u675f\u6642\u9593\u3001\u5730\u9ede(\u6216\u5730\u5740)\u56db\u500b\u57fa\u672c\u5143\u7d20\u7d44\u6210\uff0c \u7531\u65bc FB \u5927\u90e8\u5206\u63d0\u53ca\u4e8b\u4ef6\u7684\u767c\u6587\u90fd\u53ea\u63d0\u5230\u55ae\u4e00\u4e8b\u4ef6\uff0c\u56e0\u6b64\u672c\u7814\u7a76\u6d3b\u52d5\u4e8b\u4ef6\u64f7\u53d6\u4efb\u52d9\u5373\u5f9e \u6bcf\u7bc7\u8cbc\u6587\u4e2d\u5148\u884c\u8fa8\u8b58\u6d3b\u52d5\u540d\u7a31\uff0c\u518d\u64f7\u53d6\u6d3b\u52d5\u7684\u65e5\u671f\u53ca\u5730\u9ede\u3002\u4ee5\u5716\u4e09\u8cbc\u6587\u70ba\u4f8b\uff0c\u8868\u4e8c\u5373\u70ba \u9ede\u5305\u62ec 245 \u842c Object id \u6240\u5f97)\uff0c\u5206\u5225\u76e3\u807d\u95dc\u6ce8\u7684\u7c89\u7d72\u7db2\u9801\u53d6\u5f97\u767c\u6587\u8cc7\u6599\u3002\u5728 2015/9~2016/8 \u6708\u9593\u767c\u6587\u8490 \u96c6\u6a21\u7d44\u5171\u6536\u96c6\u7c89\u7d72\u9801\u7684\u767c\u6587 2,947 \u842c\u7bc7\u4ee5\u53ca\u548c\u8a72\u767c\u6587\u7684\u524d 100 \u7bc7\u56de\u61c9\u3002\u53e6\u5916\u7cfb\u7d71\u4e5f\u6536\u96c6 FB \u4e8b\u4ef6\u7684\u767c\u6587\uff0c\u4e26\u5229\u7528\u722c\u87f2\u7a0b\u5f0f\u6293\u53d6 CityTalk \u7db2\u7ad9\u4e0a\u7684\u6d3b\u52d5\u4e8b\u4ef6\u3002 (\u4e09) \u6d3b\u52d5\u540d\u7a31\u5be6\u9ad4\u7684\u8b58\u5225 \u6d3b\u52d5\u540d\u7a31\u7684\u6a21\u578b\u662f\u6839\u64da CityTalk \u7db2\u7ad9\u8490\u96c6\u56de\u4f86\u7684\u6d3b\u52d5\u540d\u7a31\u505a\u67e5\u8a62\u8a5e\uff0c\u5c0d\u641c\u5c0b\u5f15\u64ce\u8a62\u554f\u7d50 \u679c\uff0c\u4e26\u7d93\u7531\u81ea\u52d5\u6a19\u8a18\u5f97\u5230\u8a13\u7df4\u6587\u672c\uff0c\u6d3b\u52d5\u540d\u7a31\u5c6c\u65bc\u9577\u547d\u540d\u5be6\u9ad4\uff0c\u6240\u4ee5\u6b67\u7fa9\u6027\u7b49\u554f\u984c\u767c\u751f \u6a5f\u6703\u6bd4\u8f03\u5c11\uff0c\u56e0\u6b64\u6211\u5011\u53ef\u4ee5\u9032\u884c\u81ea\u52d5\u6a19\u8a18\u7372\u5f97\u5927\u91cf\u6a19\u8a18\u7684\u8a13\u7df4\u6587\u672c\u3002\u5b8c\u6574\u8a13\u7df4\u904e\u7a0b\u548c\u4f7f \u7528 Huang[1]\u7684\u6a21\u578b\u548c\u6539\u9032\u90e8\u5206\u53ef\u4ee5\u53c3\u8003\u4e0b\u5716\u56db\uff0c\u8a13\u7df4\u6a21\u578b\u7684\u90e8\u5206\u9996\u5148\u5c07\u6211\u5011\u722c\u87f2\u7a0b\u5f0f \u6488\u56de\u4f86\u7684\u6587\u672c\u900f\u904e tokenization \u6a21\u7d44\u5c07 token \u505a\u8f03\u597d\u7684\u5207\u5272\uff0c\u63a5\u8457\u900f\u904e\u81ea\u52d5\u6a19\u8a18\u6a21\u7d44\u81ea \u52d5\u6a19\u8a18\u5be6\u9ad4\uff0c\u6a19\u8a18\u51fa\u5be6\u9ad4\u7684\u6587\u672c\u6211\u5011\u4e0d\u6703\u6574\u6bb5\u4f7f\u7528\u800c\u662f\u900f\u904e String split \u6a21\u7d44\u53ea\u7559 entity \u524d\u5f8c\u56fa\u5b9a\u9577\u5ea6\u7684\u7bc4\u570d\uff0c\u63a5\u8457\u900f\u904e Feature Mining \u6a21\u7d44\u5f97\u5230\u5b57\u5178\u6a94\uff0c\u900f\u904e\u9019\u4e9b\u5b57\u5178\u6a94\u7d93\u7531 Generate Feature Matrix \u6a21\u7d44\u7522\u751f CRF ++\u8a13\u7df4\u683c\u5f0f\u4e26\u8a13\u7df4\u7522\u751f CRF Model\u3002\u5728\u4efb\u52d9\u4e2d\u6539 (\u56db) \u5de5\u5177\u7684\u64f4\u5145 \u5de5\u5177\u7684\u64f4\u5145\u662f\u70ba\u4e86\u6539\u9032\u539f\u5148 Huang[1]\u5de5\u5177\u5728 FB \u6587\u672c\u6548\u80fd\u4e0d\u597d\u7684\u554f\u984c\uff0c\u9664\u4e86\u4e0a\u8ff0\u6539\u5584 \u9577\u5be6\u9ad4 Uni-Labeling \u6a21\u7d44\u6a19\u8a18\u7684\u6e96\u78ba\u6027\u4e4b\u5916\uff0c\u4e26\u65b0\u589e\u9577\u5be6\u9ad4\u6392\u6bd4 Full-Labeling \u6a21\u7d44\u3002\u8f03 \u5927 \u7684 \u6539 \u8b8a \u662f \u64f4 \u5145 \u6846 \u67b6 \u7684 \u53ef \u64f4 \u5c55 \u6027 (scalability) \uff0c\u65b0\u589e(1) \u4e0d\u540c\u8cc7\u6599\u4f86\u6e90\u7684\u591a\u7dda\u7a0b \u5716\u4e94\u3001FBPlaceDB \u5730\u9ede\u8fa8\u8b58\u6a19\u8a18\u6f14\u7b97\u6cd5 ), respectively 5. (n-gram, placei) = \u00d7 | | + (1 \u2212 ) \u00d7 | | \u00d7 (| (n-gram)|, | (placei)|) Eq. (1) (\u4e03) \u95dc\u4fc2\u7684\u8026\u5408 \u9019\u500b\u6b65\u9a5f\u662f\u5c07\u64f7\u53d6\u51fa\u5be6\u9ad4\u76f8\u95dc\u8cc7\u8a0a\uff0c\u914d\u5c0d\u6210\u5b8c\u6574\u7684\u6d3b\u52d5\u4e8b\u4ef6\uff0c\u6211\u5011\u63a1\u7528\u898f\u5247\u5f0f\u7684\u65b9\u5f0f\uff0c \u900f\u904e\u5faa\u5e8f\u6a23\u5f0f\u63a2\u52d8\uff0c\u5feb\u901f\u627e\u5230\u548c\u6d3b\u52d5\u76f8\u95dc\u7684\u6587\u5b57\u7279\u5fb5\u3002\u7531\u65bc\u900f\u904e\u7cfb\u7d71\u6a19\u8a18\u6d3b\u52d5\u540d\u7a31\u6a21\u7d44 \u53ef\u80fd\u6703\u5f97\u5230\u591a\u500b\u6d3b\u52d5\u540d\u7a31\uff0c\u6211\u5011\u5c07\u5f9e\u4e2d\u9078\u64c7\u7cfb\u7d71\u8a8d\u70ba\u6700\u5408\u9069\u7684\u6d3b\u52d5\u540d\u7a31\u3002\u6311\u9078\u65b9\u6cd5\u662f\u91dd \u5584 Huang[1]\u7684\u6392\u6bd4(Alignment)\u5716\u56db\u3001\u6d3b\u52d5\u540d\u7a31\u8fa8\u8b58\u6a21\u578b\u5efa\u7acb\u548c\u66f4\u65b0 Huang \u5de5\u5177\u7684\u8aaa\u660e (multithreading)\u6a19\u8a18\u548c\u6a19\u8a18\u7d50\u679c\u7684\u5009\u5132\u3001(2)\u652f\u63f4 word-based \u65b9\u6cd5\u7684\u6a19\u8a18(\u5982\u8868\u4e09) \u3001(3) \u5c0d\u6bcf\u4e00\u500b\u6a19\u8a18\u6d3b\u52d5\u540d\u7a31 t \u548c\u7cfb\u7d71\u6a19\u8a18\u7684\u6240\u6709\u6d3b\u52d5\u540d\u7a31 aSet \u53bb\u6392\u6bd4\uff0c\u4e26\u5f97\u5230\u5171\u540c\u4ea4\u96c6\u6bd4\u4f8b \u8209\u4f8b\u800c\u8a00\uff0c\u53e5\u5b50 s1=\u300c\u4e3b\u8fa6\u55ae\u4f4d\uff1a\u9ad8\u96c4\u5e02\u653f\u5e9c\u52de\u5de5\u5c40\u8a13\u7df4\u5c31\u696d\u4e2d\u5fc3\u300d \uff0c\u5229\u7528\u4e0d\u540c\u7684 n-gram \u662f\u5229\u7528 MSRA \u8a13\u7df4\u51fa\u7684\u5be6\u9ad4(\u5305\u542b\u4eba\u540d\u3001\u5730\u540d\u3001\u7d44\u7e54\u540d)\u8fa8\u8b58\u6a21\u578b\uff0c\u6a19\u8a18 92 \u842c\u500b\u9ec3\u9801\u5546 \u5bb6\u540d\u7a31\u4e2d\u51fa\u73fe\u7684\u5be6\u9ad4\uff0c\u4e26\u522a\u6389\u51fa\u73fe\u983b\u7387\u5c11\u65bc 6 \u6b21\u7684\u5be6\u9ad4\u540d\u7a31\uff0c\u505a\u70ba\u6838\u5fc3\u5b57\u5178\u6a94(7,993 \u8a5e) \uff1b \u800c\u5730\u9ede\u5546\u5bb6\u80cc\u666f\u5b57\u5178\u6a94 LocBgDic \u6536\u96c6\u65b9\u5f0f\u5247\u662f\u5c07\u5546\u5bb6\u540d\u7a31\u7d93\u904e\u4e2d\u6587\u65b7\u8a5e\u5f8c\uff0c\u53d6\u8a5e\u983b\u5927 \u65bc 500 \u4e14\u5b58\u5728\u5e96\u4e01\u65b7\u8a5e\u548c MMSeg \u9810\u8a2d\u8a5e\u5eab\u4e2d\u4f46\u4e0d\u5c6c\u65bc CoreDic \u7684\u5b57\u8a5e(1,361 \u8a5e)\u3002 \u5229\u7528\u9019\u5169\u500b\u5b57\u5178\u6a94\uff0c\u6211\u5011\u53ef\u4ee5\u5c0d\u6bcf\u4e00\u500b\u53e5\u5b50\u4e2d\u7684 n-gram \u53ca\u5176\u67e5\u8a62\u5230\u7684 place \u8a55\u5206\u3002\u7d66\u5206 \u539f\u5247\u70ba(1)\u907f\u514d\u8ddf Location \u6216\u5546\u5bb6\u7121\u95dc\u7684\u5b57\u6703\u5f97\u5230\u5206\u6578\u3002(2)\u7576\u8981\u6a19\u8a18\u7684\u5c0d\u8c61\u548c\u8cc7\u6599\u5eab\u7684 \u5730\u9ede\u76f8\u95dc\u5ea6\u5224\u5b9a\u5206\u6578\u8d85\u904e\u9580\u6abb\u503c\uff0c\u6211\u5011\u8a8d\u70ba\u5176\u5be6\u53ef\u8996\u70ba\u76f8\u95dc\uff0c\u5373\u53ef\u4ee5\u5730\u9ede\u6a19\u8a18\u53e5\u5b50\u3002(3) \u53ef\u67e5\u8a62\u5230\"\u9ad8\u96c4\u5e02\u653f\u5e9c\"\u3001\"\u65b0\u5317\u5e02\u653f\u5e9c\u52de\u5de5\u5c40\"\u3001\"\u52de\u5de5\u5c40\u8a13\u7df4\u5c31\u696d\u4e2d\u5fc3\"\u3001\"\u6597\u516d\u5c31 \u696d\u4e2d\u5fc3\"\u3001\"\u9ad8\u96c4\u5e02\u653f\u5e9c\u52de\u5de5\u5c40\u52de\u5de5\u6559\u80b2\u751f\u6d3b\u4e2d\u5fc3\"\u3001\"\u9ad8\u96c4\u5e02\u653f\u5e9c\u52de\u5de5\u5c40\u8a13\u7df4\u5c31\u696d\u4e2d\u5fc3 \u5927\u5bee\u8077\u8a13\u5834\u57df\"\u7b49\u76f8\u95dc\u7684 FB \u6253\u5361\u9ede\uff0c\u900f\u904e\u6f14\u7b97\u6cd5\u6700\u5f8c\u53e5\u5b50\u4e2d\u6703\u6a19\u8a18\u5230\u7684\u5730\u9ede\u70ba\u300c\u9ad8\u96c4 \u5e02\u653f\u5e9c\u52de\u5de5\u5c40\u8a13\u7df4\u5c31\u696d\u4e2d\u5fc3\u300d \uff0c\u96d6\u7136\u9019\u7b46\u5730\u9ede\u540d\u7a31\u4e0d\u5728 FBPlaceDB \u4e2d\uff0c\u4f46\u662f\u56e0\u70ba\u8207\"\u9ad8 \u96c4\u5e02\u653f\u5e9c\u52de\u5de5\u5c40\u8a13\u7df4\u5c31\u696d\u4e2d\u5fc3\u5927\u5bee\u8077\u8a13\u5834\u57df\"\u7d93\u904e\u6392\u6bd4\uff0c\u537b\u80fd\u5b8c\u6574\u7684\u9032\u884c\u6a19\u8a18\u3002 (\u516d) \u5730\u5740\u5be6\u9ad4\u7684\u8b58\u5225\u548c\u6642\u7a0b\u8868\u9054\u5f0f\u6a19\u8a18\u6a21\u7d44 \u6700\u9ad8\u7684\u6d3b\u52d5\u540d\u7a31\u505a\u70ba\u6a19\u8a18\u7684\u8f38\u51fa\u3002 \u96d6\u7136\u6bcf\u500b\u4eba\u4ecb\u7d39\u6d3b\u52d5\u8cc7\u8a0a\u7684\u65b9\u5f0f\u6709\u5f88\u591a\u7a2e\uff0c\u4f46\u6703\u6709\u4e00\u4e9b\u5e38\u898b\u4ecb\u7d39\u6d3b\u52d5\u8cc7\u8a0a\u7684\u8a9e\u53e5\uff0c\u6211\u5011 \u60f3\u6cd5\u662f\u627e\u51fa\u9019\u4e9b\u8a9e\u53e5\u51fa\u73fe\u7684\u7279\u5b9a\u5b57\uff0c\u7528\u5176\u5efa\u7acb\u64f7\u53d6\u6d3b\u52d5\u6642\u9593\u7684\u898f\u5247\uff0c\u6211\u5011\u7684\u6a23\u672c\u6578\u662f\u5305 \u542b\u6642\u7a0b\u8868\u9054\u5f0f\u4e14\u662f\u8ddf\u6d3b\u52d5\u8cc7\u8a0a\u6709\u95dc\u7684 40 \u842c\u53e5\u53e5\u5b50\uff0c\u63a5\u8457\u900f\u904e\u5faa\u5e8f\u6a23\u5f0f\u63a2\u52d8(sequential pattern mining)\uff0c\u627e\u51fa\u524d 800 \u6a23\u7248\uff0c\u7d93\u7531\u4eba\u5de5\u5224\u5b9a\u7559\u4e0b 79 Pattern\uff0c\u5176\u4e2d\u5305\u62ec\u8d77\u59cb 51 \u500b \u65e5\u671f\u898f\u5247\u300114 \u500b\u7d50\u675f\u65e5\u671f\u898f\u5247\u3001\u4ee5\u53ca 14 \u689d Date Confuse \u898f\u5247\uff0c\u5efa\u7acb\u6d3b\u52d5\u6642\u9593 Pattern \u64f7 \u53d6\u898f\u5247\u3002\u53e6\u5916\u7cfb\u7d71\u4e5f\u624b\u52d5\u5efa\u7acb\u4e00\u4e9b\u8f14\u52a9\u7684\u898f\u5247\u4ee5\u8b58\u5225\u7279\u6b8a\u7684\u6d3b\u52d5\u767c\u6587\u6848\u4f8b\u548c\u8b58\u5225\u6d3b\u52d5\u5730 \u65b7\u53e5\u6a21\u7d44\uff0c\u548c(4)\u5931\uff0c\u56e0\u6b64\u6211\u5011\u6539\u5beb\u90e8\u5206\u7a0b\u5f0f\u79fb\u9664\u9019\u6a23\u7684\u8a2d\u5b9a\uff0c\u4e26\u6dfb\u52a0\u61c9\u7528\u5e96\u4e01\u65b7\u8a5e\u548c MMSeg \u81ea\u5b9a\u7fa9\u7684 \u8a5e\u5eab\uff0c\u5982\u679c\u8981\u4f7f\u7528\u5176\u5b83\u4e2d\u6587\u65b7\u8a5e\uff0c\u53ef\u81ea\u884c\u5c01\u88dd\u65b7\u8a5e\u4f86\u53d6\u4ee3\u9810\u8a2d\u3002\u6a19\u8a18\u7cbe\u6e96\u5ea6\u662f\u6307\u8b58\u5225\u51fa \u6bd4\u8f03\u597d\u7684\u5be6\u9ad4\u908a\u754c\u61c9\u53d6\u5f97\u8f03\u9ad8\u7684\u5206\u6578\uff0c\u5982 n-gram1=\"\u6e05\u83ef/\u5927\u5b78/\u65fa\u5b8f/\u9928\uff02\u8207 n-gram2= \u9ede\uff0c\u5b8c\u6574\u7684\u5236\u5b9a\u898f\u5247\u5728\u5716\u516d\u3002\u9019\u4e9b\u898f\u5247\u6703\u6709\u512a\u5148\u9806\u5e8f\uff0c\u5c0d\u65bc\u6bd4\u8f03\u5b8c\u6574\u7684 Pattern \u6703\u6709\u6bd4</td></tr></table>", | |
| "type_str": "table" | |
| }, | |
| "TABREF2": { | |
| "text": "Precision \u8f03\u4f73\u7684\u512a\u9ede\uff0c\u4f46\u5728 Recall \u8f03\u4f4e\u3002 (\u56db) \u5730\u9ede\u6a21\u578b\u7684\u8a55\u4f30 \u6211\u5011\u4f7f\u7528\u6253\u5361\u6b21\u6578\u8d85\u904e 1,000 \u6b21\u7684 10.3 \u842c FB \u5730\u9ede\u505a\u70ba\u7a2e\u5b50\uff0c\u67e5\u8a62 Google \u524d 10 \u7b46\u641c\u5c0b \u7d50\u679c\uff0c\u4ee5\u53ca FB \u8cbc\u6587\u5206\u5225\u5f97\u5230 17 \u842c\uff0c1,317 \u842c\u53e5\u5b50(FB \u8cbc\u6587\u641c\u5c0b\u63a1 Partial match \u67e5\u8a62\uff0c Recall \u4ee5\u53ca\u6700\u597d\u7684 0.452 \u7684F 1 -score\u3002\u53e6\u5916\u6211\u5011 \u5c07\u5169\u8005\u65b9\u6cd5\u505a\u7d50\u5408\u7684 LNER(FB posts)+FBLocDic \u80fd\u5c07 Recall \u63d0\u5347\u5230 0.656 \u4e14 Precision \uff0c\u6839\u64da\u7b54\u6848\u542b\u6709 k \u6d3b\u52d5\u4e8b\u4ef6\u8cc7\u8a0a\uff0c\u5b9a\u7fa9\u6d3b\u52d5\u70ba k-tuple (k=1 to 4)\uff0c \u82e5\u7cfb\u7d71\u64f7\u53d6\u51fa\u6d3b\u52d5\u540d\u7a31 EAct \u3001\u8d77\u59cb\u65e5\u671f EStart \u3001\u7d50\u675f\u65e5\u671f EEnd\u3001\u4ee5\u53ca\u5730\u9ede ELoc \u6216 EAddr \uff0c \u6211\u5011\u5206\u5225\u5b9a\u7fa9 Eq. (6)\u5f0f\u500b\u5225\u5be6\u9ad4\u64f7\u53d6\u7684 P, R \u5206\u6578 2 \u3002Eq. (7) \u5f0f\u70ba\u6d3b\u52d5\u6642\u9593\u8cc7\u8a0a\u5224\u5b9a\u662f\u5426", | |
| "num": null, | |
| "html": null, | |
| "content": "<table><tr><td>Start, End, Loc/Addr) 1 \u6b63\u78ba\u7684\u5206\u6578\uff0c\u6211\u5011\u5c07\u958b\u59cb\u3001\u7d50\u675f\u6642\u9593\u5206\u5225\u4f30\u91cf\u3002\u53e6\u5916 Eq. (8) \u70ba\u6d3b\u52d5\u5730\u9ede\u8cc7\u8a0a\u6b63\u78ba\u7684\u5206 \u500b\u65b9\u6cd5\uff0cword-based \u6709 \u6240\u4ee5\u53ef\u80fd\u6703\u5f97\u5230\u67e5\u8a62\u8a5e\u7121\u95dc\u7684\u6587\u672c\uff0c\u53e6\u5916\u53e5\u5b50\u7d71\u8a08\u662f\u67e5\u8a62\u7d50\u679c\u5f97\u5230\u6587\u672c\u6240\u542b\u53e5\u5b50\u6578\u91cf) \uff0c \u6578\u3002\u518d\u5c07\u9019 k \u500b\u5be6\u9ad4\u5206\u6578\u505a\u5e73\u5747\u4fbf\u80fd\u5206\u5225\u5f97\u5230 Eq. (9) EventPrecision \u53ca Eq. (10)</td></tr><tr><td>\u5229\u7528 Uni-Labeling \u6a19\u8a18\u5730\u9ede\u540d\u7a31\uff0c\u5206\u5225\u6a19\u8a18\u4e86 62,723 \u53ca 148,801 \u500b\u5730\u9ede\u5be6\u9ad4\uff0c\u4f9b\u7d66 CRF++ EventRecall\u3002\u6d3b\u52d5\u4e8b\u4ef6\u95dc\u4fc2\u8a55\u4f30\u5b9a\u7fa9\u5982\u4e0b:</td></tr><tr><td>F 1 -score = Web NER Model Generation \u5de5\u5177\u914d\u5408 Google Snippets \u548c FB \u8cbc\u6587\u7684\u65b9\u6cd5\uff0cFBLocDic \u5247 2 \u00d7 \u00d7 + Eq. (5) \u8a13\u7df4\u5730\u9ede\u540d\u7a31\u8fa8\u8b58\u6a21\u578b\uff0c\u5176 \u4e2d LNER(Google snippets)\u53ca LNER(FB posts)\u662f\u5229\u7528\u6539\u5584\u5f8c\u7684 P(ActSet, EAct)= max \u2208 |EAct \u2229 | | | , |EAct \u2229 | R(ActSet, EAct)= max \u2208 | |</td></tr><tr><td>(\u4e09) \u6d3b\u52d5\u540d\u7a31\u6a21\u578b\u7684\u8a55\u4f30 \u70ba\u672c\u6587\u914d\u5408 FBPlaceDB \u6a19\u8a18\u65b9\u6cd5\u6240\u63d0\u7684\u65b9\u6cd5\u3002\u5229\u7528\u5730\u9ede\u8fa8\u8b58\u7684\u8cc7\u6599\u96c6\u8a55\u4f30\u7d50\u679c\u5982\u5716\u516b</td></tr><tr><td>\u6240\u793a\uff0c\u7531\u65bc\u6e2c\u8a66\u7b54\u6848\u5c0d\u5730\u9ede\u548c\u5730\u5740\u548c\u7d44\u7e54\u7684\u5340\u5206\uff0c\u4e14\u8a13\u7df4\u7a2e\u5b50\u6a94\u6709\u9109\u93ae\u5340\u548c\u8857\u9053\u540d\u7d44\u6210</td></tr><tr><td>\u7684\u5730\u9ede\u540d\u7a31\uff0c\u9020\u6210\u6703\u6a19\u8a18\u5230\u90e8\u5206\u5730\u5740\uff0c\u666e\u904d Precision \u90fd\u7dad\u6301\u5728 6 \u6210\u4ee5\u4e0b\uff0cFBLocDic \u7cbe</td></tr><tr><td>\u78ba\u7387\u8f03\u9ad8\u6a19\u8a18\u5be6\u9ad4\u908a\u754c\u4e5f\u8f03\u6e96\uff0c\u4f46\u53ec\u56de\u7387\u904e\u4f4e\u662f\u8a72\u65b9\u6cd5\u7684\u7f3a\u9ede\u3002\u6539\u9032\u5f8c\u5de5\u5177\u4ee5\u53ca FB \u8cbc</td></tr><tr><td>\u5716\u4e03\u3001Activity Name NER Model \u6bd4\u8f03 \u5206\u6790\u767c\u73fe\u7d66\u5b9a\u76f8\u540c\u7684 Google snippets \u6587\u672c\u9032\u884c\u81ea\u52d5\u6a19\u8a18\uff0cHuang \u7684\u5de5\u5177\u5728\u6d3b\u52d5\u540d\u7a31(\u9577 \u5be6\u9ad4\u63a1\u6392\u6bd4\u65b9\u5f0f Uni-Labeling)\u53ea\u80fd\u6a19\u8a18 2,221 \u500b\u5be6\u9ad4\uff0c\u800c\u6539\u9032\u7684\u6392\u6bd4\u65b9\u5f0f Uni-Labeling \u5247\u53ef\u6a19\u8a18 77,766 \u500b\u5be6\u9ad4\uff0c\u986f\u793a\u900f\u904e\u65b0\u589e tokenizer \u548c\u91cd\u65b0\u5be6\u505a\u7684\u6392\u6bd4\u65b9\u5f0f\uff0c\u80fd\u81ea\u52d5\u6a19\u8a18 \u5716\u516b\u3001Location NER Model \u6bd4\u8f03 (\u4e94) \u6d3b\u52d5\u4e8b\u4ef6\u64f7\u53d6\u7684\u6548\u80fd \u6d3b\u52d5\u4e8b\u4ef6\u95dc\u4fc2\u7684\u4f30\u91cf\u65b9\u5f0f\u5148\u4ee5\u6bcf\u7bc7\u6587\u7ae0\u8b58\u5225\u51fa\u500b\u5225\u5be6\u9ad4\u9032\u884c\u8a55\u4f30\uff0c\u518d\u5b9a\u7fa9\u6574\u500b\u6d3b\u52d5\u4e8b\u4ef6 0.013 0.025 0 Precision Recall F1-Score \u5728\u62fc\u97f3\u548c\u975e\u62fc\u97f3\u7684\u6587\u672c\u3002\u4e26\u9054\u5230 0.5 \u7684F 0.361 0.02 0.038 0.494 0.215 0.299 0.563 0.455 0.5 0.8 Activity Name Model Comparison Huang baseline Stanford baseline ANER-U ANER-F ANER-FW \u6587\u8a13\u7df4\u6240\u5f97\u7684\u6a21\u578b\u53ef\u4ee5\u9054\u5230\u6700\u597d 0.618 \u7684 \u50c5\u6709 0.02 \u4e0b\u964d\u3002 0.604 0.529 0.397 0.454 0.282 0.385 0 0.2 0.4 0.6 Precision Recall F1-Score 0.165 0.139 0.17 0.153 0.168 0.254 0.441 0.262 0.329 0.356 0.452 0.334 0.442 0.2 0.4 0.6 Performance 0.536 Location NER Model Comparison Stanford Huang FBLocDic LNER(Google snippets) LNER(FB posts) LNER(FB posts)+ Performance 0.618 0.656 0.8 FBLocDic</td></tr><tr><td>\u7684\u64f7\u53d6\u6548\u80fd\u3002\u5982\u524d\u6240\u8ff0\uff0c\u7531\u65bc FB \u4e0a\u7684\u6d3b\u52d5\u8cbc\u6587\u57fa\u672c\u4e0a\u662f\u4e3b\u8ff0\u4e00\u500b\u4e3b\u8981\u6d3b\u52d5\u4e8b\u4ef6\uff0c\u56e0\u6b64</td></tr><tr><td>\u7d66\u5b9a\u4e00\u7bc7\u6d3b\u52d5\u8cbc\u6587\u53ca\u4eba\u5de5\u6a19\u8a18\u7684\u6d3b\u52d5\u3001\u8d77\u59cb\u65e5\u671f\u3001\u7d50\u675f\u65e5\u671f\u4ee5\u53ca\u5730\u9ede\u56db\u500b\u7b54\u6848(ActSet,</td></tr></table>", | |
| "type_str": "table" | |
| }, | |
| "TABREF3": { | |
| "text": "\u5de6\u53f3\u7684F 1 -score\uff0c\u5c0d\u65bc\u8d77\u59cb\u65e5\u671f\u66f4\u53ef\u9054 \u5230 0.865 \u7684F 1 -score\u3002 \u6700\u5f8c\u6211\u5011\u8a55\u4f30\u7cfb\u7d71\u5c07\u6d3b\u52d5\u5730\u9ede\u6295\u5c04\u5230\u96fb\u5b50\u5730\u5716\u4e0a\u7684\u4f4d\u7f6e\uff0c\u8ddf\u4eba\u5de5\u5224\u5b9a\u7684\u771f\u5be6\u5ea7\u6a19\u7684\u8aa4\u5dee\u3002 \u5f9e 1,300 \u7bc7\u6587\u7ae0\u4e2d\uff0c\u5229\u7528\u4eba\u5de5\u5224\u5b9a\u7684\u6d3b\u52d5\u4f4d\u7f6e GPS \u548c\u7cfb\u7d71\u81ea\u52d5\u5316\u5b9a\u4f4d\u7684 713 \u7b46 GPS \u9032 \u884c\u5be6\u9a57(\u76ee\u524d\u7cfb\u7d71\u6bcf\u7bc7\u6587\u7ae0\u53ea\u6703\u81ea\u52d5\u5b9a\u4f4d 1 \u500b GPS\uff0c\u81ea\u52d5\u5b9a\u4f4d 713 \u4ee3\u8868\u5c0d 713 \u7bc7\u6587\u7ae0\u505a \u81ea\u52d5\u5b9a\u4f4d)\u3002\u6263\u9664 75 \u7b46\u7cfb\u7d71\u9032\u884c\u81ea\u52d5\u5b9a\u4f4d\u3001\u4f46\u7121\u4eba\u5de5\u5224\u5b9a\u7b54\u6848\u7684 GPS\uff0c\u91dd\u5c0d\u6709\u7b54\u6848\u7684 638 \u7b46 GPS \u6211\u5011\u7d71\u8a08\u5be6\u969b\u81ea\u52d5\u5316\u5b9a\u4f4d\u548c\u7b54\u6848\u6d3b\u52d5\u4e8b\u4ef6\u4f4d\u7f6e GPS \u7684\u5dee\u8ddd\uff0c\u5c07\u7d50\u679c\u986f\u793a\u5728\u5716\u4e5d\u3002 \u5176\u4e2d\u516b\u6210\u56db\u7684\u9810\u6e2c\u5c11\u65bc 4 \u516c\u91cc(\u5e73\u5747 0.15 \u516c\u91cc)\uff0c\u8207\u8868\u516d\u4e2d\u7684 0.849 \u7684 Precision \u76f8\u8fd1\u3002", | |
| "num": null, | |
| "html": null, | |
| "content": "<table><tr><td colspan=\"7\">535 0~4 km 0.15 \u76ee\u524d\u7cfb\u7d71\u5c0d\u65bc\u6bcf\u7bc7\u6587\u7ae0\u76ee\u524d\u91dd\u5c0d\u5167\u6587\u7684\u55ae\u4e00\u4e8b\u4ef6\u53bb\u505a\u64f7\u53d6\uff0c\u4f46\u662f\u5982\u679c\u6587\u7ae0\u63d0\u53ca\u591a\u500b\u4e8b\u4ef6\uff0c 50 15 22 5 11 10.15 43.19 120.81 198.89 284.89 0 10 1,000 0 300 600 5~24 km 25~74 km 75~169 km 170~229 km 230~404 km \u7d2f \u7a4d \u8aa4 \u5dee \u8ddd \u96e2 ( \u516c \u91cc ) \u8aa4\u5dee\u7bc4\u570d \u7cfb\u7d71\u81ea\u52d5\u5316\u5b9a\u4f4d\u548c\u5be6\u969b\u4f4d\u7f6e\u7684\u4f30\u91cf \u5c6c\u65bc\u9019\u7bc4\u570d\u7684\u6578\u91cf \u5e73\u5747\u8aa4\u5dee \u4e5f\u9700\u64f7\u53d6\u6587\u7ae0\u4e2d\u63d0\u5230\u7684\u591a\u500b\u4e8b\u4ef6\u3002\u6b64\u5916\u6587\u7ae0\u4e2d\u63d0\u5230\u7684\u591a\u500b\u5b50\u6d3b\u52d5\u4e5f\u662f\u64f7\u53d6\u611f\u8208\u8da3\u7684\u76ee\u6a19\uff0c \u56e0\u70ba\u9019\u4e9b\u4e8b\u4ef6\u7684\u5b50\u6d3b\u52d5\u548c\u5b50\u6d3b\u52d5\u63d0\u53ca\u7684\u63cf\u8ff0\u548c\u6642\u9593\u66f4\u80fd\u5920\u5e6b\u52a9\u4eba\u5011\u5feb\u901f\u4e86\u89e3\u8a72\u6d3b\u52d5\u7684 \u8a73\u60c5\uff0c\u5982\u679c\u80fd\u64f7\u53d6\u9019\u6a23\u7684\u8cc7\u8a0a\uff0c\u5c31\u80fd\u63d0\u4f9b\u4eba\u5011\u6d3b\u52d5\u6392\u7a0b\u529f\u80fd\u548c\u6d3b\u52d5\u7684\u63a8\u85a6\u3002\u6b64\u5916\u50cf\u4e00\u4e9b \u7279\u6b8a\u60c5\u6cc1\u767c\u751f\u4f8b\u5982\u98b1\u98a8\u9020\u6210\u6d3b\u52d5\u53d6\u6d88\uff0c\u7cfb\u7d71\u61c9\u8a72\u8a3b\u8a18\u6d3b\u52d5\u56e0\u70ba\u4f55\u7a2e\u539f\u56e0\u53d6\u6d88\uff0c\u907f\u514d\u63d0\u4f9b \u932f\u8aa4\u6d3b\u52d5\u8a0a\u606f\u3002\u6700\u5f8c\u5e0c\u671b\u6211\u5011\u7684\u4efb\u52d9\u80fd\u63a8\u5ee3\u5230\u6574\u500b\u4e0d\u53ea\u6709\u793e\u7fa4\u5a92\u9ad4\u7684 Web \u4e0a\uff0c\u5f9e\u53f0\u7063 \u7684\u7db2\u7ad9\uff0c\u653f\u5e9c\u3001\u5b78\u6821\u3001\u552e\u7968\u7db2\u7ad9\u516c\u544a\u81ea\u52d5\u5316\u64f7\u53d6\u6d3b\u52d5\u516c\u544a\uff0c\u63d0\u4f9b\u66f4\u5b8c\u6574\u8c50\u5bcc\u7684\u6d3b\u52d5\u8a0a\u606f\u3002 \u64f7\u53d6\u898f\u5247\uff0c\u56e0\u6b64\u6211\u5011\u6574\u9ad4\u7684\u64f7\u53d6\u6548\u80fd\u53ef\u4ee5\u9039\u5230 0.7 \u5716\u4e5d\u3001\u6d3b\u52d5\u4e8b\u4ef6\u9810\u6e2c\u6d3b\u52d5\u4f4d\u7f6e(GPS)\u8a55\u4f30\u5be6\u9a57 \u53c3\u8003\u6587\u737b</td></tr><tr><td>\u4e94\u3001 \u7d50\u8ad6</td><td/><td/><td/><td/><td/><td>Eq. (12)</td></tr><tr><td colspan=\"7\">\u8868\u4e94\u3001\u4e0d\u540c k \u8cc7\u6599\u96c6\u500b\u5225\u7684\u4e8b\u4ef6\u5c6c\u6027\u548c\u6d3b\u52d5\u4e8b\u4ef6\u64f7\u53d6\u6548\u80fd \u672c\u7814\u7a76\u5efa\u69cb\u4e86\u4e00\u500b FB \u4e8b\u4ef6\u64f7\u53d6\u7cfb\u7d71\uff0c\u4e26\u4e3b\u52d5\u8490\u96c6\u793e\u7fa4\u5a92\u9ad4\u8cc7\u6599\uff0c\u6574\u5408\u5206\u6563\u7684\u7c89\u7d72\u9801\u767c</td></tr><tr><td colspan=\"7\">F 1 -score \u5e03\u7684\u6d3b\u52d5\u767c\u6587\uff0c\u64f7\u53d6\u6d3b\u52d5\u91cd\u8981\u8cc7\u8a0a\uff0c\u63d0\u4f9b\u641c\u5c0b FB \u6d3b\u52d5\u767c\u6587\u7684\u529f\u80fd\uff0c\u4e26\u5c07\u64f7\u53d6\u5230\u7684\u6d3b\u52d5 Event attribute #posts Activity Name Start Date End Date Loc/Add Precision Recall F 1 -score F 1 -score 1-tuple 45 0.766 NA NA NA 0.776 0.757 0.766 \u4e8b\u4ef6\u5728\u96fb\u5b50\u5730\u5716\u4e0a\u986f\u793a\u65b9\u4fbf\u4f7f\u7528\u8005\u67e5\u770b\u3002\u5728\u7cfb\u7d71\u767c\u5c55\u904e\u7a0b\u4e2d\uff0c\u767c\u73fe\u4e2d\u6587\u547d\u540d\u5be6\u9ad4\u8fa8\u8b58\u6a21</td></tr><tr><td>2-tuple</td><td>124</td><td>0.732</td><td>0.8587</td><td>NA</td><td>0.39</td><td>0.651 0.648 0.650</td></tr><tr><td colspan=\"7\">3-tuple \u7d44\u4ecd\u6709\u5f88\u5927\u7684\u9032\u6b65\u7a7a\u9593\uff0c\u5c24\u5176\u662f\u5c0d\u65bc\u66f8\u5beb\u8f03\u81ea\u7531\u7684 FB \u767c\u6587\uff0c\u900f\u904e\u6539\u5584\u81ea\u52d5\u6a19\u8a18\u7684\u65b9\u6cd5\uff0c 547 0.727 0.881 0.704 0.719 0.742 0.724 0.732</td></tr><tr><td>4-tuple</td><td>584</td><td>0.731</td><td>0.853</td><td>0.744</td><td>0.687</td><td>0.705 0.685 0.694</td></tr><tr><td colspan=\"7\">Total/Avg \u53ef\u4ee5\u5927\u5e45\u6539\u5584\u6a19\u8a18\u7684\u6e96\u78ba\u7387\u548c\u6a19\u8a18\u91cf\uff0c\u89e3\u6c7a Huang \u5728\u975e\u62fc\u97f3(\u4e2d\u6587)\u6392\u6bd4\u65b9\u6cd5\u6548\u80fd\u4e0d\u597d 1300 0.727 0.865 0.720 0.694 0.718 0.700 0.708</td></tr><tr><td/><td colspan=\"5\">\u8868\u516d\u3001\u500b\u5225\u7684\u4e8b\u4ef6\u5c6c\u6027\u548c\u6d3b\u52d5\u4e8b\u4ef6\u64f7\u53d6\u6548\u80fd</td><td/></tr><tr><td colspan=\"7\">Item \u7684\u554f\u984c\u3002\u540c\u6642\u6211\u5011\u4e5f\u52a0\u5f37\u6587\u672c\u524d\u8655\u7406\u5f48\u6027\u4ee5\u9069\u61c9\u64f4\u5145\u548c\u5ba2\u88fd\u5316\u3002\u900f\u904e\u8490\u96c6 FB \u4e0a\u7684\u8cc7\u6599\uff0c Performance Activity Name Start Date End Date Loc/Add Event</td></tr><tr><td colspan=\"7\">Precision Recall \u53ef\u4ee5\u6709\u6548\u7684\u8a13\u7df4\u6211\u5011\u7684\u4e2d\u6587\u547d\u540d\u5be6\u9ad4\u8fa8\u8b58\u6a21\u7d44\u3002\u53e6\u5916\u6211\u5011\u900f\u904e\u5e8f\u5217\u6a23\u5f0f\u63a2\u52d8\u627e\u51fa\u6709\u7528\u7684 0.729 0.884 0.942 0.849 0.718 0.726 0.848 0.583 0.587 0.700</td></tr><tr><td colspan=\"7\">F 1 -score \u7279\u5fb5\uff0c\u8f14\u52a9\u6d3b\u52d5\u64f7\u53d6\u7684\u5224\u65b7\uff0c\u5c0d\u63d0\u5347\u6e96\u78ba\u7387\u6709\u5f88\u5927\u7684\u5e6b\u52a9\u3002\u6700\u5f8c\u5be6\u9a57\u900f\u904e\u4eba\u5de5\u6a19\u8a18\u7684 0.727 0.865 0.720 0.694 0.708</td></tr><tr><td colspan=\"7\">\u6d3b\u52d5\u4e8b\u4ef6\u95dc\u4fc2\u5be6\u9a57\u7d50\u679c\u8868\u4e94\u3001\u8868\u516d\u96d6\u7136\u5728\u524d\u4e00\u7bc0\u4e2d\uff0c\u500b\u5225\u5be6\u9ad4\u540d\u7a31\u8fa8\u8b58\u6548\u679c\u53ea\u6709 0.5 \u5de6</td></tr><tr><td colspan=\"7\">1,300 \u7bc7\u767c\u6587\u8a55\u4f30\u547d\u540d\u5be6\u9ad4\u8fa8\u8b58\u548c\u4e8b\u4ef6\u64f7\u53d6\u7684\u6548\u80fd\u3002\u7d71\u8a08\u5f9e 2015/6/23 \u622a\u81f3 2016/8/8\uff0c\u91dd</td></tr><tr><td colspan=\"7\">\u53f3\uff0c\u4f46\u662f\u7531\u65bc\u6d3b\u52d5\u8cbc\u6587\u4e2d\u53ef\u80fd\u4ee5\u4e0d\u540c\u65b9\u5f0f\u63d0\u53ca\u6d3b\u52d5\u76f8\u95dc\u8cc7\u8a0a\uff0c\u52a0\u4ee5\u5faa\u5e8f\u6a23\u5f0f\u63a2\u52d8\u6240\u5f97\u7684</td></tr><tr><td colspan=\"7\">\u5c0d FB Post \u9032\u884c\u6d3b\u52d5\u4e8b\u4ef6\u64f7\u53d6\uff0c\u7cfb\u7d71\u5728\u9019\u6bb5\u6642\u9593\u5171\u622a\u53d6\u4e86 11 \u842c 1,931 \u500b\u6d3b\u52d5\u4e8b\u4ef6(\u6709\u540c</td></tr><tr><td colspan=\"7\">1 \u82e5\u6587\u7ae0\u7121\u8cc7\u6599\u5247\u6a19\u8a18\u70ba NULL\uff0c\u65bc\u6d3b\u52d5\u4e8b\u4ef6\u64f7\u53d6\u6548\u80fd\u6642\uff0c\u4e0d\u4e88\u8a08\u7b97\u3002 \u6642\u63d0\u53ca\u6d3b\u52d5\u540d\u7a31\u548c\u6d3b\u52d5\u6642\u9593\u6291\u6216\u662f\u540c\u6642\u63d0\u53ca\u6d3b\u52d5\u540d\u7a31\u548c\u6d3b\u52d5\u5730\u9ede/\u5730\u5740)\u3002</td></tr><tr><td colspan=\"7\">2 \u76ee\u524d\u7cfb\u7d71\u9810\u6e2c\u4e8b\u4ef6\u5c6c\u6027\u90fd\u53ea\u9810\u6e2c\u4e00\u500b\uff0c\u53d6 max \u70ba\u6211\u5011\u7cfb\u7d71\u9810\u6e2c\u8ddf\u6240\u6709\u7684\u7b54\u6848\u6a19\u8a18\u53bb\u770b\u80fd\u62ff\u5230\u7684\u5206\u6578</td></tr></table>", | |
| "type_str": "table" | |
| } | |
| } | |
| } | |
| } |