| { | |
| "paper_id": "O15-1018", | |
| "header": { | |
| "generated_with": "S2ORC 1.0.0", | |
| "date_generated": "2023-01-19T08:09:57.836031Z" | |
| }, | |
| "title": "Points of Interest Extraction from Unstructured Web", | |
| "authors": [ | |
| { | |
| "first": "Ting-Yao", | |
| "middle": [], | |
| "last": "\u9ad8\u9706\u8000", | |
| "suffix": "", | |
| "affiliation": { | |
| "laboratory": "", | |
| "institution": "National Central University", | |
| "location": {} | |
| }, | |
| "email": "" | |
| }, | |
| { | |
| "first": "", | |
| "middle": [], | |
| "last": "Kao", | |
| "suffix": "", | |
| "affiliation": { | |
| "laboratory": "", | |
| "institution": "National Central University", | |
| "location": {} | |
| }, | |
| "email": "" | |
| }, | |
| { | |
| "first": "Hsiu-Min", | |
| "middle": [], | |
| "last": "\u838a\u79c0\u654f", | |
| "suffix": "", | |
| "affiliation": { | |
| "laboratory": "", | |
| "institution": "National Central University", | |
| "location": {} | |
| }, | |
| "email": "" | |
| }, | |
| { | |
| "first": "", | |
| "middle": [], | |
| "last": "Chuang", | |
| "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": "With the increased popularity of mobile devices, local search has become a new popular service. Therefore, we need a powerful POI (Points of Interest) database to support local search. In recent years, the web has become the largest data source of POIs. With the prevalence of Internet, people will share their travel experience and information of POIs that they had been visited on social network, their blogs, and even check-in post. Besides, many companies and organizations publish their business on their own websites, resulting a large number of POIs.", | |
| "pdf_parse": { | |
| "paper_id": "O15-1018", | |
| "_pdf_hash": "", | |
| "abstract": [ | |
| { | |
| "text": "With the increased popularity of mobile devices, local search has become a new popular service. Therefore, we need a powerful POI (Points of Interest) database to support local search. In recent years, the web has become the largest data source of POIs. With the prevalence of Internet, people will share their travel experience and information of POIs that they had been visited on social network, their blogs, and even check-in post. Besides, many companies and organizations publish their business on their own websites, resulting a large number of POIs.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Abstract", | |
| "sec_num": null | |
| } | |
| ], | |
| "body_text": [ | |
| { | |
| "text": "In this paper, we propose a POI database construction system from the immense data of the Web. Our system consists of two parts: the query-based crawler, and the POI extraction system. The goal of query-based crawler is to collect address-bearing pages (ABP) from the web as address is a good indicator of POIs. The second part is POI extraction system. We use CRF (Conditional Random Field) to train a Chinese postal address recognition model and a Chinese organization recognition model. After the extraction of addresses and POI names from ABP with these two CRF models, we then leant a model to pair an address and a POI name as a POI. Finally, we extract POI associated information for each POI to construct a complete POI data.", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "Keywords: electronic map, web crawler, information extraction, POI database. ", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "\u95dc\u9375\u8a5e\uff1a\u96fb\u5b50\u5730\u5716\u3001\u7db2\u8def\u722c\u87f2\u3001\u8cc7\u8a0a\u64f7\u53d6\u3001POI \u8cc7\u6599\u5eab", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "\u96fb\u5b50\u5730\u5716\u4e0d\u50c5\u662f\u6578\u4f4d\u5316\u5f8c\u7684\u5730\u5716\uff0c\u56e0\u70ba\u4e0d\u53d7\u9650\u65bc\u6709\u9650\u7684\u7a7a\u9593\uff0c\u53ef\u4ee5\u6839\u64da\u700f\u89bd\u8005\u7684\u9700\u6c42\uff0c", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "\u4e00\u3001 \u7dd2\u8ad6", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "http://openstreetmap.tw/ 2 http://wikimapia.org/", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "https://www.iyp.com.tw/ 4 http://www.ipeen.com.tw/", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "", | |
| "sec_num": null | |
| } | |
| ], | |
| "back_matter": [], | |
| "bib_entries": { | |
| "BIBREF0": { | |
| "ref_id": "b0", | |
| "title": "Location-based Web Search. The Geospatial Web", | |
| "authors": [ | |
| { | |
| "first": "D", | |
| "middle": [], | |
| "last": "Ahlers", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "S", | |
| "middle": [], | |
| "last": "Boll", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2007, | |
| "venue": "", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "55--66", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "D. Ahlers and S. Boll, Location-based Web Search. The Geospatial Web, pp. 55-66, Springer, 2007.", | |
| "links": null | |
| }, | |
| "BIBREF1": { | |
| "ref_id": "b1", | |
| "title": "Adaptive Geospatially Focused Crawling", | |
| "authors": [ | |
| { | |
| "first": "D", | |
| "middle": [], | |
| "last": "Ahlers", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "S", | |
| "middle": [], | |
| "last": "Boll", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2009, | |
| "venue": "China", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "D. Ahlers and S. Boll, Adaptive Geospatially Focused Crawling. CIKM, China, Nov. 2- 6, 2009.", | |
| "links": null | |
| }, | |
| "BIBREF2": { | |
| "ref_id": "b2", | |
| "title": "Latent Dirichlet allocation", | |
| "authors": [ | |
| { | |
| "first": "D", | |
| "middle": [ | |
| "M" | |
| ], | |
| "last": "Blei", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "A", | |
| "middle": [ | |
| "Y" | |
| ], | |
| "last": "Ng", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "M", | |
| "middle": [ | |
| "I" | |
| ], | |
| "last": "Jordan", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2003, | |
| "venue": "Journal of Machine Learning Research", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "993--1022", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "D. M. Blei, A. Y. Ng, and M. I. Jordan, Latent Dirichlet allocation. Journal of Machine Learning Research, 993-1022, 2003.", | |
| "links": null | |
| }, | |
| "BIBREF3": { | |
| "ref_id": "b3", | |
| "title": "LIBSVM: a library for support vector machines", | |
| "authors": [ | |
| { | |
| "first": "C.-C", | |
| "middle": [], | |
| "last": "Chang", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "C.-J", | |
| "middle": [], | |
| "last": "Lin", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2011, | |
| "venue": "ACM Transactions on Intelligent Systems and Technology", | |
| "volume": "2", | |
| "issue": "27", | |
| "pages": "1--27", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "C.-C. Chang, C.-J. Lin, LIBSVM: a library for support vector machines. ACM Transactions on Intelligent Systems and Technology, 2(27):1-27, 2011.", | |
| "links": null | |
| }, | |
| "BIBREF4": { | |
| "ref_id": "b4", | |
| "title": "Chinese Postal Address and Associated Information Extraction", | |
| "authors": [ | |
| { | |
| "first": "C.-H", | |
| "middle": [], | |
| "last": "Chang", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "C.-Y", | |
| "middle": [], | |
| "last": "Huang", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Y.-S", | |
| "middle": [], | |
| "last": "Su", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2012, | |
| "venue": "The 26th Annual Conference of the Japanese Society for Artificial Intelligence", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "C.-H. Chang, C.-Y. Huang and Y.-S. Su, Chinese Postal Address and Associated Information Extraction. The 26th Annual Conference of the Japanese Society for Artificial Intelligence, 2012.", | |
| "links": null | |
| }, | |
| "BIBREF5": { | |
| "ref_id": "b5", | |
| "title": "Effective Web Crawling for Chinese Addresses and Associated Information", | |
| "authors": [ | |
| { | |
| "first": "H.-M", | |
| "middle": [], | |
| "last": "Chuang", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "C.-H", | |
| "middle": [], | |
| "last": "Chang", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "T.-Y", | |
| "middle": [], | |
| "last": "Kao", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2014, | |
| "venue": "EC-Web", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "H.-M. Chuang, C.-H. Chang and T.-Y. Kao, Effective Web Crawling for Chinese Addresses and Associated Information. EC-Web, Munich, Germany, 2014.", | |
| "links": null | |
| }, | |
| "BIBREF6": { | |
| "ref_id": "b6", | |
| "title": "MapMarker: Extraction of Postal Addresses and Associated Information for General Web Pages", | |
| "authors": [ | |
| { | |
| "first": "Chia-Hui", | |
| "middle": [], | |
| "last": "Chang", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Shu-Ying", | |
| "middle": [], | |
| "last": "Li", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2010, | |
| "venue": "Web Intelligence", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Chia-Hui Chang, Shu-Ying Li, MapMarker: Extraction of Postal Addresses and Associated Information for General Web Pages. Web Intelligence, 2010.", | |
| "links": null | |
| }, | |
| "BIBREF7": { | |
| "ref_id": "b7", | |
| "title": "Web NER Model Generator Tool based on Google Snippets", | |
| "authors": [ | |
| { | |
| "first": "Y. -Y", | |
| "middle": [], | |
| "last": "Huang", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "C.-L", | |
| "middle": [], | |
| "last": "Chou", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "C.-H", | |
| "middle": [], | |
| "last": "Chang", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2015, | |
| "venue": "", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Y. -Y. Huang, C.-L. Chou, C.-H. Chang, Web NER Model Generator Tool based on Google Snippets, submitted for publication, 2015.", | |
| "links": null | |
| }, | |
| "BIBREF8": { | |
| "ref_id": "b8", | |
| "title": "ViDE: A Vision-based Approach for Deep Web Data Extraction", | |
| "authors": [ | |
| { | |
| "first": "Wei", | |
| "middle": [], | |
| "last": "Liu", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Xiaofeng", | |
| "middle": [], | |
| "last": "Meng", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Weiyi", | |
| "middle": [], | |
| "last": "Meng", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2007, | |
| "venue": "Transactions on Knowledge and Data Engineering", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Wei Liu, Xiaofeng Meng, Weiyi Meng, ViDE: A Vision-based Approach for Deep Web Data Extraction. Transactions on Knowledge and Data Engineering, IEEE, 2007", | |
| "links": null | |
| }, | |
| "BIBREF9": { | |
| "ref_id": "b9", | |
| "title": "\u7db2\u9801\u5546\u5bb6\u540d\u7a31\u64f7\u53d6\u8207\u5730\u5740\u914d\u5c0d\u4e4b\u7814\u7a76 (Store Name Extraction and Name-Address Matching on the Web). ROCLING", | |
| "authors": [ | |
| { | |
| "first": "Yu -Yang", | |
| "middle": [], | |
| "last": "Lin", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Chia-Hui", | |
| "middle": [], | |
| "last": "Chang", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2014, | |
| "venue": "", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Yu -Yang Lin, Chia-Hui Chang, \u7db2\u9801\u5546\u5bb6\u540d\u7a31\u64f7\u53d6\u8207\u5730\u5740\u914d\u5c0d\u4e4b\u7814\u7a76 (Store Name Extraction and Name-Address Matching on the Web). ROCLING, 2014.", | |
| "links": null | |
| }, | |
| "BIBREF10": { | |
| "ref_id": "b10", | |
| "title": "Study: 78 percent of local-mobile searches result in offline purchases", | |
| "authors": [ | |
| { | |
| "first": "G", | |
| "middle": [], | |
| "last": "Stirling", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2014, | |
| "venue": "", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "G. Stirling. \"Study: 78 percent of local-mobile searches result in offline purchases\", Search Engine Land. Apr. 9, 2014.", | |
| "links": null | |
| }, | |
| "BIBREF11": { | |
| "ref_id": "b11", | |
| "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 | |
| } | |
| }, | |
| "ref_entries": { | |
| "TABREF0": { | |
| "text": "Pages, ABP)\uff0c\u6211\u5011\u5f15\u5165 Chang \u7b49\u4eba\u5728 2012 \u5e74[5] \u548c Lin \u7b49\u4eba\u5728 2014 \u5e74[10]\u6240\u63d0\u51fa\u7684\u5169\u7a2e\u6a21\u578b\u4f86\u5f9e ABP \u4e2d\u64f7\u53d6\u51fa\u5730\u5740\uff1b\u7b2c\u4e8c\u500b\u6a21\u7d44\u5247\u662f POI \u64f7\u53d6\u6a21\u7d44\uff0c\u6211\u5011\u4f7f\u7528 Huang \u7b49\u4eba\u5728 2015 \u5e74[8]\u63d0\u51fa\u7684\u4e2d\u6587\u7d44\u7e54\u540d\u7a31\u8fa8\u8b58\u6a21\u578b\u4f86\u64f7\u53d6 ABP \u7576\u4e2d\u7684 POI \u540d\u7a31\u3002\u6700\u5f8c\u6211\u5011\u6703\u5c07\u8fa8\u8b58\u51fa\u7684 POI \u540d\u7a31\u4ee5\u53ca\u5730\u5740\u7d44\u6210\u8a31\u591a\u7b46 POI\uff0c\u4e26 \u900f\u904e POI \u914d\u5c0d\u9a57\u8b49\u6a21\u7d44\u5c07\u6b63\u78ba\u7684 POI \u8cc7\u6599\u653e\u5165\u8cc7\u6599\u5eab\u7576\u4e2d\u3002 NER)\u5f9e\u7db2\u9801\u4e2d\u64f7\u53d6\u5730\u5740\u4ee5\u53ca POI \u540d\u7a31\u4e26\u4e14\u5c07\u5176\u914d\u5c0d\uff0c\u5728\u5f97\u5230 POI\uff0c \u518d\u70ba\u5176\u627e\u51fa\u76f8\u95dc\u7684\u63cf\u8ff0\uff0c\u4f7f\u5f97\u8a72\u7b46 POI \u8cc7\u6599\u80fd\u5920\u5728\u8cc7\u8a0a\u6aa2\u7d22\u7cfb\u7d71\u4e2d\u88ab\u6aa2\u7d22\u3002\u56e0\u6b64\u672c\u7814 \u7a76\u7684\u4e3b\u8981\u6280\u8853\u53ef\u4ee5\u5206\u70ba\u5982\u4f55\u6709\u6548\u5730\u722c\u53d6\u76ee\u6a19\u7db2\u9801\u3001\u547d\u540d\u5be6\u9ad4\u8fa8\u8b58\u3001\u5730\u5740\u548c POI \u540d\u7a31\u7684 \u914d\u5c0d\u4ee5\u53ca\u76f8\u95dc\u8cc7\u8a0a\u64f7\u53d6\u3002 \u6211\u5011\u63a1\u7528\u7684\u4e2d\u6587\u5730\u5740\u64f7\u53d6\u65b9\u6cd5\u662f Chang \u7b49\u4eba\u65bc 2012 \u5e74\u6240\u63d0\u51fa\u7684\u6a21\u578b[5]\uff0c\u4f7f\u7528\u6a5f\u5668\u5b78\u7fd2 \u6cd5\u4e2d\u5e8f\u5217\u6a19\u8a18\u7684 CRF \u505a\u70ba\u5176\u8a13\u7df4\u53ca\u6e2c\u8a66\u6a21\u578b\uff0c\u914d\u5408\u53f0\u7063\u5730\u5740\u7684\u7279\u6027\u5efa\u7acb\u4e86 17 \u7a2e\u5730\u5740\u7279 \u5fb5\u4e26\u4e14\u4f7f\u7528 Start/End \u6a19\u8a18\u6cd5\uff0c\u63a5\u8457\u518d\u914d\u5408\u6975\u5927\u5206\u5b50\u5e8f\u5217\u6f14\u7b97\u6cd5 (Maximal Scoring Subsequences)\uff0c\u5176\u6e96\u78ba\u7387\u7d04\u5728 94%\u81f3 99%\u4e4b\u9593\u3002 \u800c\u4e2d\u6587\u7d44\u7e54\u540d\u7a31\u64f7\u53d6\u6a21\u7d44\u7684\u5efa\u7f6e\u5247\u662f\u4f7f\u7528 Huang \u7b49\u4eba\u65bc 2015 \u5e74\u63d0\u51fa\u7684\u65b9\u6cd5[8]\uff0c\u540c\u6a23\u662f \u4f7f\u7528 CRF \u505a\u70ba\u5176\u8a13\u7df4\u53ca\u6e2c\u8a66\u7684\u6a21\u578b\uff0c\u5229\u7528\u7d44\u7e54\u540d\u7a31\u4e2d\u5e38\u51fa\u73fe\u7684\u8a5e\u5f59(e.g., \u5e97\u3001\u516c\u53f8)\u4ee5 \u53ca\u7d44\u7e54\u540d\u7a31\u524d\u5f8c\u5e38\u51fa\u73fe\u7684\u8a5e\u5f59\u7b49\u7e3d\u5171\u5efa\u7acb\u4e86 18 \u7a2e\u7279\u5fb5\uff0c\u4e26\u4e14\u4f7f\u7528 Self-Testing \u4ee5\u53ca Tri-Training \u7b49\u65b9\u6cd5\u518d\u66f4\u9032\u4e00\u6b65\u5730\u63d0\u5347\u6e96\u78ba\u7387\uff0c\u6700\u7d42\u5176\u6e96\u78ba\u7387\u53ef\u4ee5\u5728\u975e\u7d50\u69cb\u5316\u7684\u7db2\u9801\u4e2d \u9054\u5230 86.13%\u3002 \u5728\u76f8\u95dc\u8cc7\u8a0a\u64f7\u53d6\u7684\u90e8\u5206\uff0c\u96d6\u7136 Li \u7b49\u4eba[7]\u8207 Chang \u7b49\u4eba[5]\u7684\u7814\u7a76\u4e2d\u90fd\u6709\u63d0\u5230\u6b64\u90e8\u5206\uff0c\u4f46 \u5176\u6548\u80fd\u4e26\u4e0d\u4f73\u3002\u5728 2012 \u5e74 Su[12]\u767c\u73fe\u4ed6\u5011\u7684\u505a\u6cd5\u904e\u5ea6\u7406\u60f3\u5316\u5404\u7b46\u7d00\u9304(Record)\u7684\u5132\u5b58\u6a19 \u7c64\u7686\u662f\u63a1\u7528\u540c\u4e00\u898f\u683c\u6a19\u6e96\uff0c\u82e5\u662f\u6a19\u7c64\u7684\u6a23\u5f0f\u7a0d\u6709\u4f8b\u5916\u51fa\u73fe\uff0c\u5c31\u6703\u767c\u751f\u9023\u9396\u932f\u8aa4\uff0c\u5c0e\u81f4\u64f7 \u53d6\u5931\u6557\u3002\u70ba\u4e86\u89e3\u6c7a\u6b64\u554f\u984c\uff0cSu[12]\u5c07 2010 \u5e74 Wei Liu \u6240\u63d0\u51fa\u57fa\u65bc\u8996\u89ba\u7684\u8cc7\u6599\u7d00\u9304\u64f7\u53d6\u6f14 \u7b97\u6cd5[9]\u5957\u7528\u5728\u5730\u5740\u76f8\u95dc\u8cc7\u8a0a\u64f7\u53d6\u7684\u7814\u7a76\u4e2d\uff0c\u4e26\u91cd\u4f5c Li[7]\u7684\u5be6\u9a57\uff0c\u5c07 F-measure \u7531 79.12% \u63d0\u6607\u81f3 95.04%\u3002 OR \u8857 OR \u6bb5 OR \u5df7 OR \u5f04)* \u865f\u505a\u70ba\u5730\u5740 pattern\u3002\u518d \u6839\u64da\u95dc\u9375\u5b57\u7684\u985e\u578b\uff0c\u5206\u4ee5\u4e0b\u5169\u7a2e\u4f5c\u6cd5\uff1a 1. \u985e\u5225\uff1a\u4ee5 26 \u500b\u7e23\u5e02\u52a0\u4e0a\u5730\u5740 pattern \u4ee5\u53ca\u985e\u5225\u95dc\u9375\u5b57(e.g., \u9910\u5ef3\u3001\u670d\u98fe\u3001\u4ea4\u901a)\u505a\u70ba \u67e5\u8a62\u8a5e\uff0c\u63a5\u8457\u53d6\u56de\u641c\u5c0b\u7d50\u679c\u7684\u524d 500 \u500b\u7db2\u9801\u3002 2. POI \u540d\u7a31\uff1a\u4ee5\u5730\u5740 pattern \u52a0\u4e0a POI \u540d\u7a31\u95dc\u9375\u5b57(e.g., \u6021\u5ba2\u5496\u5561\u3001\u661f\u5df4\u514b)\u505a\u70ba\u67e5\u8a62\u8a5e\uff0c \u63a5\u8457\u53d6\u56de\u641c\u5c0b\u7d50\u679c\u7684\u524d 20 \u500b\u7db2\u9801\u3002 \u5c0d\u65bc\u6bcf\u4e00\u500b\u6587\u4ef6 s\uff0c\u6211\u5011\u4f7f\u7528 LDA \u53d6\u5f97\u5176\u591a\u9805\u5f0f\u5206\u5e03\u03f4 \uff0c\u4e26\u5229\u7528\u6f5b\u85cf\u4e3b\u984c z \u8a08\u7b97 ( | ,\u03a6k)\u505a\u70ba LDA \u7522\u751f\u7684\u8a9e\u8a00\u6a21\u578b Plda(poi|s)\u7684\u4f30\u8a08\uff0c\u516c\u5f0f\u5982\u4e0b\uff1a Baseline\uff0c\u4f46\u662f\u5728 precision \u65b9\u9762\u537b\u9060\u9ad8\u65bc Baseline\u3002\u6211\u5011\u8a8d\u70ba\u539f\u56e0\u662f Baseline \u7684\u505a\u6cd5\u5c0d \u65bc\u6bcf\u500b\u5730\u5740\u90fd\u4e00\u5b9a\u80fd\u627e\u5230\u4e00\u500b POI \u540d\u7a31\u505a\u914d\u5c0d\uff0c\u56e0\u6b64\u5c0e\u81f4\u96d6\u7136\u6709\u8f03\u9ad8\u7684 recall \u4f46 precision \u537b\u76f8\u5c0d\u7684\u975e\u5e38\u4f4e\uff0c\u7136\u800c\u900f\u904e POI \u914d\u5c0d\u9a57\u8b49\u6a21\u7d44\u7684\u505a\u6cd5\u56e0\u70ba\u5224\u65b7\u8f03\u70ba\u56b4\u8b39\uff0c\u56e0\u6b64 \u9020\u6210 recall \u7684\u90e8\u5206\u7a0d\u4f4e\u4f46\u5728 precision \u7684\u90e8\u5206\u53ef\u4ee5\u6709\u975e\u5e38\u597d\u7684\u6548\u679c\u3002\u6211\u5011\u8a8d\u70ba\u5728 POI \u914d\u5c0d \u9019\u9805\u4efb\u52d9\u4e0a precision \u7684\u91cd\u8981\u6027\u9060\u9ad8\u65bc recall\uff0c\u56e0\u70ba\u6211\u5011\u4e0d\u80fd\u5920\u63d0\u4f9b\u932f\u8aa4\u7684 POI \u8cc7\u6599\u7d66\u4f7f \u7528\u8005\uff0c\u56e0\u6b64\u8981\u76e1\u53ef\u80fd\u7684\u78ba\u4fdd POI \u8cc7\u6599\u5eab\u4e2d\u7684\u8cc7\u6599\u7684\u6b63\u78ba\u6027\u3002 4.3 \u76f8\u95dc\u8cc7\u8a0a\u64f7\u53d6\u6548\u80fd\u8a55\u4f30 \u56e0\u70ba\u76f8\u95dc\u8cc7\u8a0a\u7684\u6b63\u78ba\u8207\u5426\u8f03\u96e3\u4ee5\u5224\u5b9a\uff0c\u56e0\u6b64\u70ba\u4e86\u6709\u6548\u8a55\u4f30\u76f8\u95dc\u8cc7\u8a0a\u7684\u6548\u80fd\u53ca\u5be6\u7528\u6027\uff0c\u6211 \u5011\u8a2d\u8a08\u4e86\u4e00\u500b IR \u5be6\u9a57\uff0c\u900f\u904e POI \u6aa2\u7d22\u7cfb\u7d71\u4f86\u6e2c\u91cf POI \u76f8\u95dc\u8cc7\u8a0a\u7684\u54c1\u8cea\u3002\u5728\u672c\u5be6\u9a57\u4e2d\uff0c \u6211\u5011\u900f\u904e\u4e0a\u500b\u5be6\u9a57\u4e2d\u6558\u8ff0\u7684 Baseline \u4f5c\u6cd5\u8a2d\u8a08\u4e86\u5169\u500b POI \u6aa2\u7d22\u7cfb\u7d71 POIDB_AI \u548c POIDB\uff0c \u5176\u4e2d POIDB_AI \u662f\u7531\u5305\u542b\u76f8\u95dc\u8cc7\u8a0a\u5728\u5167\u7684 POI \u8cc7\u6599\u5eab\u5efa\u7f6e\u7684 POI \u6aa2\u7d22\u7cfb\u7d71\uff0cPOIDB \u5247 \u662f\u7531\u9664\u4e86\u76f8\u95dc\u8cc7\u8a0a\u4ee5\u5916\u7684 POI \u8cc7\u6599\u6240\u5efa\u7f6e\u7684 POI \u6aa2\u7d22\u7cfb\u7d71\u3002\u6211\u5011\u900f\u904e\u5169\u500b POI \u6aa2\u7d22\u7cfb \u7d71\u56de\u50b3\u7684 POI \u7684\u6b63\u78ba\u6027(Average Precision@10)\u53ca\u6578\u91cf\u4f86\u6e2c\u91cf\u76f8\u95dc\u8cc7\u8a0a\u6b63\u78ba\u6027\u4ee5\u53ca\u5be6\u7528 \u6027\u3002\u6211\u5011\u9078\u64c7\u4e86 9 \u500b\u5730\u9ede\u505a\u70ba\u6aa2\u7d22\u4e2d\u5fc3\u9ede\uff0c\u5176\u4e2d\u5305\u542b\u4e86\u5e02\u4e2d\u5fc3\u3001\u5e02\u4e2d\u5fc3\u65c1\u7684\u5730\u5340\u4ee5\u53ca\u96e2 \u5e02\u4e2d\u5fc3\u8f03\u9060\u7684\u5730\u5340\uff0c\u7136\u5f8c\u6bcf\u500b\u4e2d\u5fc3\u9ede\u518d\u5404\u81ea\u914d\u5408 200m\u3001500m\u30011000m \u4e09\u7a2e\u4e0d\u540c\u7684\u6aa2\u7d22 \u534a\u5f91\uff0c\u5171\u5f62\u6210 27 \u7a2e\u7d44\u5408\u3002\u5c0d\u65bc\u6bcf\u4e00\u7a2e\u7d44\u5408\uff0c\u6211\u5011\u518d\u5206\u5225\u4f7f\u7528 18 \u500b\u8207\u65e5\u5e38\u751f\u6d3b\u8f03\u70ba\u76f8\u95dc \u7684\u67e5\u8a62\u8a5e\uff0c\u5982\uff1a\u9910\u5ef3\u3001\u670d\u98fe\u3001\u96fb\u5f71\u7b49\u4f86\u9032\u884c\u67e5\u8a62\uff0c\u5be6\u9a57\u7d50\u679c\u5982\u5716 \u5341\u4e8c\u6240\u793a\u3002\u5f9e\u5be6\u9a57\u7d50 \u679c\u4e2d\u53ef\u4ee5\u770b\u51fa POIDB_AI \u6240\u80fd\u67e5\u627e\u5230\u7684 POI \u6578\u91cf\u662f POIDB \u7684\u5169\u500d\u4ee5\u4e0a\uff0c\u800c\u4e14 POIDB_AI \u7684 AP@10 \u5728\u975e\u5e38\u5c0f\u7684\u6aa2\u7d22\u534a\u5f91\u4e2d\u4f9d\u7136\u53ef\u4ee5\u7dad\u6301\u5728 80%\u4ee5\u4e0a\u3002\u5f9e\u672c\u5be6\u9a57\u7684\u7d50\u679c\u8b93\u6211\u5011\u53ef \u4ee5\u78ba\u4fe1 POI \u76f8\u95dc\u8cc7\u8a0a\u64f7\u53d6\u6a21\u7d44\u6240\u64f7\u53d6\u7684\u76f8\u95dc\u8cc7\u8a0a\u53ef\u4ee5\u771f\u6b63\u7684\u5e6b\u52a9 POI \u7684\u6aa2\u7d22\u3002", | |
| "content": "<table><tr><td>\u6574\u5408\u5176\u80cc\u5f8c\u8cc7\u6599\u5229\u7528\u5716\u5c64\u758a\u52a0\u7684\u7279\u6027\u5c0d\u793e\u6703\u7d93\u6fdf\u8cc7\u6599\u9032\u884c\u6a19\u8a18\u8207\u5206\u6790\uff0c\u6240\u4ee5\u7522\u751f\u4e86\u76f8\u7576 \u591a\u65b0\u7a4e\u7684\u670d\u52d9\uff0c\u5982\u8cb7\u5c4b\u79df\u5c4b\u641c\u5c0b\u3001\u666f\u9ede\u641c\u5c0b\u7b49\u7b49\u3002\u53e6\u8005\uff0c\u7531\u65bc\u8fd1\u5e74\u884c\u52d5\u88dd\u7f6e\u7684\u9032\u6b65\u8207\u666e \u53ca\uff0c\u9023\u5e36\u4f7f\u5f97\u884c\u52d5\u5b9a\u4f4d\u8207\u76ee\u7684\u5730\u5c0e\u822a\u6210\u70ba\u4e00\u9805\u65b0\u8208\u7684\u71b1\u9580\u670d\u52d9\uff0c\u73fe\u4eca\u7684\u96fb\u5b50\u5730\u5716\u5927\u591a\u6574 \u5408\u4e86\u4ee5\u4e0a\u7684\u529f\u80fd\uff0c\u63d0\u4f9b\u4e86\u5b8c\u6574\u7684\u9069\u5730\u6027\u670d\u52d9(Location-based Service)\uff0c\u4f7f\u5f97\u5730\u5716\u641c\u5c0b\u6210\u70ba \u65e5\u5e38\u751f\u6d3b\u4e0d\u53ef\u6216\u7f3a\u7684\u529f\u80fd\u3002 \u96d6\u7136\u96fb\u5b50\u5730\u5716\u80fd\u5920\u63d0\u4f9b\u7d66\u6211\u5011\u8af8\u591a\u7684\u4fbf\u5229\uff0c\u4f46\u9664\u4e86\u57fa\u672c\u7684\u5730\u7406\u8cc7\u8a0a\u5916\uff0c\u96fb\u5b50\u5730\u5716\u9084\u5fc5\u9808 \u8981\u4ef0\u8cf4\u5176\u7cfb\u7d71\u5f8c\u65b9\u8c50\u5bcc\u4e14\u5145\u6c9b\u7684\u8cc7\u6599\u5eab\uff0c\u624d\u80fd\u66f4\u52a0\u7a81\u986f\u5176\u529f\u6548\u3002\u591a\u6578\u5730\u7406\u8cc7\u6599\u5eab\u90fd\u662f\u4f9d \u9760\u4eba\u5de5\u7de8\u8f2f\uff0c\u4f46\u662f\u8981\u5c07\u6240\u6709\u7684 POI \u90fd\u4f7f\u7528\u4eba\u5de5\u7684\u65b9\u5f0f\u52a0\u5165\u8cc7\u6599\u5eab\u662f\u4e00\u4ef6\u8017\u6642\u8cbb\u529b\u7684\u4e8b \u60c5\uff0c\u56e0\u6b64\u4e5f\u9650\u5236\u4e86\u73fe\u4eca\u5730\u9ede\u8cc7\u6599\u5eab\u7684\u5730\u9ede\u6578\u91cf\u8207\u5167\u5bb9\u3002\u7136\u800c\u5728\u7db2\u969b\u7db2\u8def\u76db\u884c\u7684\u73fe\u4eca\uff0c\u96d6 \u7136\u653f\u5e9c\u5de5\u696d\u5c40\u6216\u5546\u696d\u53f8\u6709\u4f01\u696d\u767b\u8a18\u8cc7\u6599\uff0c\u4f46\u4f01\u696d\u767b\u8a18\u540d\u7a31\u8207\u5e97\u5bb6\u540d\u7a31\u5f80\u5f80\u4e0d\u4e00\u81f4\uff0c\u4f8b\u5982 \u561f\u561f\u623f\u505c\u8eca\u5be6\u7531\u4e2d\u8208\u96fb\u5de5\u7d93\u71df\uff0c\u56e0\u6b64\u5373\u4f7f\u6709\u653f\u5e9c\u958b\u653e\u8cc7\u6599\uff0c\u5546\u5bb6 POI \u8cc7\u6599\u4ecd\u7136\u4e0d\u5920\u5b8c \u6574\uff0c\u4f46\u662f\u9664\u4e86\u653f\u5e9c\u6a5f\u69cb\u7684\u7db2\u7ad9\uff0cPOI \u4ea6\u7d93\u5e38\u4f34\u96a8\u5176\u63cf\u8ff0\u51fa\u73fe\u5728\u5176\u4ed6\u7db2\u9801\u4e2d\uff0c\u5982\u9023\u9396\u5546\u5e97 \u7684\u7db2\u9801\u3001\u90e8\u843d\u683c\u7684\u9910\u5ef3\u4ecb\u7d39\u53ca\u666f\u9ede\u4ecb\u7d39\uff0c\u751a\u81f3\u65bc\u793e\u7fa4\u7db2\u7ad9\u7684\u6253\u5361\u8cc7\u8a0a\u7b49\uff0c\u9019\u4e9b\u7db2\u9801\u4e2d\u6216 \u591a\u6216\u5c11\u90fd\u5305\u542b\u4e86 POI \u7684\u63cf\u8ff0\u8a0a\u606f\uff0c\u56e0\u6b64\u82e5\u80fd\u6709\u6548\u7387\u5730\u627e\u5230\u4e0a\u8ff0\u9019\u4e9b\u542b\u6709 POI \u8cc7\u6599\u7684\u7db2 \u9801\uff0c\u4e26\u7531\u7a0b\u5f0f\u81ea\u52d5\u5c07\u5176\u64f7\u53d6\u51fa\u53ef\u7528\u7684 POI \u8cc7\u6599\uff0c\u4fbf\u53ef\u6709\u6548\u5730\u64f4\u5c55\u8cc7\u6599\u5eab\u7684\u5730\u9ede\u6578\u91cf\u8207 \u5167\u5bb9\u3002 \u6839\u64da W3C \u7684\u5b9a\u7fa9\uff0c\u4e00\u500b POI \u6703\u5305\u542b\u8a31\u591a\u8cc7\u8a0a\uff0c\u50cf\u662f\u540d\u7a31\u3001\u4f4d\u7f6e\u3001\u96fb\u8a71\u4ee5\u53ca\u76f8\u95dc\u8cc7\u8a0a\u7b49 \u7b49\uff0c\u5176\u4e2d\u4f4d\u7f6e\u7528\u65bc\u5b9a\u4f4d\u6a19\u8a18\u5230\u5730\u5716\u4e0a\uff0c\u53ef\u7528\u5730\u5740\u6216\u7d93\u7def\u5ea7\u6a19\u8868\u793a\u3002\u7531\u65bc\u5730\u5740\u7684\u8b58\u5225\u7387\u8f03 \u9ad8\u76f8\u5c0d\u5176\u4ed6 POI \u8cc7\u8a0a\u66f4\u5bb9\u6613\u64f7\u53d6\uff0c\u56e0\u6b64\u672c\u7bc7\u8ad6\u6587\u4e2d\uff0c\u6211\u5011\u63d0\u51fa\u4e00\u500b POI \u8cc7\u6599\u5eab\u7684\u5efa\u7f6e \u7cfb\u7d71\uff0c\u4ee5\u5730\u5740\u64f7\u53d6\u505a\u70ba POI \u8fa8\u8b58\u7b56\u7565\uff0c\u4e26\u4e14\u5f9e\u5305\u542b\u5730\u5740\u7684\u7db2\u9801\u4e2d\u64f7\u53d6\u8207\u5730\u5740\u76f8\u5c0d\u61c9\u7684 POI \u540d\u7a31\u548c\u76f8\u95dc\u8cc7\u8a0a\uff0c\u7528\u4f86\u5efa\u7acb\u4e00\u500b POI \u8cc7\u6599\u5eab\uff0c\u63d0\u4f9b POI \u641c\u5c0b\u670d\u52d9\uff0cPOI \u8cc7\u6599\u7bc4\u4f8b\u5982 \u5716 \u4e00\u3002 \u5716 \u4e00\u3001POI \u7bc4\u4f8b \u5716 \u4e8c\u3001\u7db2\u9801\u4e2d\u7684 POI \u76f8\u95dc\u8cc7\u8a0a\u8207\u96dc\u8a0a \u672c\u7cfb\u7d71\u5305\u542b\u4e09\u500b\u6a21\u7d44\u3002\u7b2c\u4e00\u6a21\u7d44\u662f\u7db2\u9801\u7684\u722c\u53d6(Crawler)\uff0c\u6211\u5011\u9996\u5148\u4ee5\u5730\u5740\u70ba\u95dc\u9375\u5b57\u4e32\u4f86 \u8490\u96c6\u5305\u542b\u5730\u5740\u7684\u7db2\u9801(Address-bearing \u53e6\u5916\uff0c\u56e0\u70ba\u5927\u591a\u6578\u4f7f\u7528\u8005\u662f\u7531\u95dc\u9375\u5b57\u6216\u662f\u985e\u5225\u53cd\u67e5\u5546\u5e97\u5728\u5730\u5716\u4e0a\u7684\u4f4d\u7f6e\uff0c\u56e0\u6b64\u7528\u4ee5\u63cf\u8ff0 \u5730\u5740\u7684\u76f8\u95dc\u8cc7\u8a0a\u662f\u5426\u8db3\u5920\uff0c\u6703\u5927\u5e45\u5f71\u97ff\u67e5\u8a62\u7cfb\u7d71\u7684\u6aa2\u7d22\u6548\u80fd\uff0c\u70ba\u6b64\u6211\u5011\u63d0\u51fa\u4e86 POI \u76f8 \u95dc\u8cc7\u8a0a\u64f7\u53d6\u6a21\u7d44\uff0c\u70ba\u6bcf\u500b POI \u64f7\u53d6\u76f8\u95dc\u63cf\u8ff0\u4f86\u89e3\u6c7a\u6b64\u554f\u984c\u3002\u5982\u5716 \u4e8c\u6240\u793a\uff0c\u7db2\u9801\u4e2d\u5305\u542b \u8a31\u591a\u5730\u5740\u7684\u63cf\u8ff0\uff0c\u4f46\u540c\u6642\u4ecd\u6709\u8a31\u591a\u8207\u5730\u5740\u4e0d\u76f8\u95dc\u7684\u5167\u5bb9\u3002\u5728\u672c\u7bc7\u8ad6\u6587\u4e2d\u6211\u5011\u5c07\u61c9\u7528\u4e2d\u6587 \u7d44\u7e54\u540d\u7a31\u8fa8\u8b58\u6a21\u7d44\u4f86\u52a0\u5f37\u9019\u985e\u578b\u7db2\u9801\u7684\u5730\u5740\u76f8\u95dc\u8cc7\u8a0a\u64f7\u53d6\u3002 \u672c\u8ad6\u6587\u5171\u5206\u6210\u4e94\u500b\u7ae0\u7bc0\uff0c\u7b2c\u4e00\u7ae0\u70ba\u7dd2\u8ad6\uff0c\u8aaa\u660e\u7814\u7a76\u7684\u52d5\u6a5f\u8207\u76ee\u7684\u4e26\u7c21\u55ae\u7684\u4ecb\u7d39\u672c\u7bc7\u8ad6\u6587\uff1b \u7b2c\u4e8c\u7ae0\u70ba\u76f8\u95dc\u7814\u7a76\uff0c\u4ecb\u7d39\u548c\u672c\u8ad6\u6587\u76f8\u95dc\u7684\u7814\u7a76\uff1b\u7b2c\u4e09\u7ae0\u70ba\u7cfb\u7d71\u67b6\u69cb\u8207\u7814\u7a76\u65b9\u6cd5\uff0c\u8a73\u8ff0\u5982 \u4f55\u5f9e\u7db2\u8def\u4e2d\u627e\u5c0b ABP\uff0c\u4e26\u7531\u7a0b\u5f0f\u81ea\u52d5\u64f7\u53d6\u51fa POI\uff1b\u7b2c\u56db\u7ae0\u70ba\u5be6\u9a57\uff0c\u8a55\u4f30\u7cfb\u7d71\u7684\u6548\u80fd\u53ca POI \u8cc7\u6599\u7684\u6b63\u78ba\u6027\uff1b\u7b2c\u4e94\u7ae0\u70ba\u7d50\u8ad6\uff0c\u7e3d\u7d50\u672c\u8ad6\u6587\u7684\u8ca2\u737b\u3002 \u4e8c\u3001 \u76f8\u95dc\u7814\u7a76 \u8868 \u4e00\u3001POI \u914d\u5c0d\u9a57\u8b49\u6a21\u7d44\u7684\u7279\u5fb5 F Name Descriptions \u7684\u505a\u6cd5\uff0c\u900f\u904e\u52a0\u5165\u4e2d\u6587\u7d44\u7e54\u540d\u7a31\u4f86\u5e6b\u52a9\u6211\u5011\u5f9e\u5730\u5740\u7db2\u9801\u4e2d\u627e\u51fa\u5730\u5740\u7684\u76f8\u95dc\u8cc7\u8a0a\u7bc4\u570d\u3002 \u672c\u8ad6\u6587\u4e2d\u7684 POI \u76f8\u95dc\u8cc7\u8a0a\u64f7\u53d6\u6a21\u7d44\u7684\u8f38\u5165\u662f\u4e00\u5c0d\u7d44\u7e54\u540d\u7a31\u548c\u5730\u5740\u7684\u914d\u5c0d\u3002\u9996\u5148\u6211\u5011\u6703 ( | ) = \u03bb( * \u8868 \u4e09\u3001Baseline \u65b9\u6cd5\u5728\u4e0d\u540c\u7a97\u6846\u5927\u5c0f\u4e0b\u7684\u6548\u80fd\u6bd4\u8f03 ) + (1-\u03bb)P ( | ) Window Size 50 Window Size 100 Window Size 150 Items \u8fd1\u5e7e\u5e74\u4f86\uff0c\u7531\u65bc\u7db2\u8def\u4e0a\u5de8\u91cf\u8cc7\u6599\u7684\u7d2f\u7a4d\u8207\u884c\u52d5\u88dd\u7f6e\u7684\u666e\u53ca\uff0c\u5730\u7406\u8cc7\u8a0a\u6aa2\u7d22(Geographic 1 logC(a) # of search results for query a in log scale \u5148\u5c07\u7db2\u9801\u8f49\u63db\u6210\u6587\u4ef6\u7269\u4ef6\u6a21\u578b(Document Object Model)\u67b6\u69cb\uff0c\u4e26\u5c07\u7db2\u9801\u8996\u70ba\u6a39\u72c0\u7d50\u69cb\uff0c # Recognized POI names 10,773,585 20,539,371 30,144,909 Information Retrieval)\u4ee5\u53ca\u5340\u57df\u641c\u5c0b\u958b\u59cb\u53d7\u5230\u91cd\u8996\u3002\u570b\u969b\u9593\u5730\u7406\u8cc7\u8a0a\u6aa2\u7d22\u9818\u57df\u7684\u7814\u7a76\u4ee5 ACM SIGSpatial workshop on GIR \u8f03\u8ca0\u76db\u540d\uff0c\u81ea 2004 \u5e74\u8d77\u6536\u9304\u76f8\u95dc\u9818\u57df\u7684\u7814\u7a76\u5831\u544a\uff0c \u76f8\u95dc\u7814\u7a76\u4e3b\u984c\u5305\u62ec\u4e86\u5730\u7406\u8cc7\u8a0a\u7cfb\u7d71\u7684\u767c\u5c55\u6a21\u5f0f\u3001\u5730\u7406\u6578\u64da\u5eab\u7684\u5b58\u53d6\u8207\u7db2\u8def\u5167\u5bb9\u8207\u591a\u5a92\u9ad4 \u7684\u5206\u6790\u3001\u57fa\u65bc\u6587\u5b57\u8207\u5730\u7406\u8cc7\u8a0a\u7cfb\u7d71\u6574\u5408\u7684\u65b9\u6cd5(\u5982\u8cc7\u8a0a\u64f7\u53d6\u3001\u81ea\u7136\u8a9e\u8a00\u8655\u7406\u3001\u7a7a\u9593\u8cc7\u6599 \u7684\u7d22\u5f15\u8207\u641c\u5c0b\u7b49)\u3001\u4ee5\u53ca\u5730\u7406\u8853\u8a9e\u7684\u8b58\u5225\u8207\u6642\u7a7a(spatio-temporal)\u7684\u6982\u5ff5\u3002 \u53e6\u5916\u5247\u662f\u5f9e 2008 \u958b\u59cb\u8207 WWW \u540c\u6642\u8209\u8fa6\u7684 Workshop on Location and the Web(LocWeb)\uff0c \u5f8c\u7e8c\u4e5f\u5728 CHI\u3001IoT\u3001CIKM \u7b49\u6703\u8b70\u8209\u884c\uff0c\u67d0\u7a2e\u5c64\u6b21\u4f86\u770b\uff0cLocWeb \u8207 Web \u7684\u95dc\u4fc2\u66f4\u70ba \u7dca\u5bc6\u3002\u800c\u570b\u5167\u7814\u8a0e\u6703\u5247\u4ee5\u53f0\u7063\u5730\u7406\u8cc7\u8a0a\u5b78\u6703\u8209\u8fa6\u7684\u7814\u8a0e\u6703\u70ba\u4e3b\uff0c\u4e3b\u984c\u5305\u542b\u4e86\u5730\u7406\u7a7a\u9593\u6578 \u64da\u53ef\u8996\u5316\u3001\u5730\u7406\u8cc7\u8a0a\u7cfb\u7d71\u6280\u8853\u767c\u5c55\u8207\u6574\u5408\u61c9\u7528\u3001\u958b\u653e\u8cc7\u6599\u8207\u7fa4\u773e\u5916\u5305(Crowdsourcing)\u3001 \u9632\u6551\u707d\u8207\u8cc7\u901a\u6280\u8853\u6574\u5408\u3001\u4ee5\u53ca\u81ea\u7136\u74b0\u5883\u8cc7\u6e90\u7ba1\u7406\u8207\u74b0\u5883\u76e3\u6e2c\u76f8\u95dc\u7814\u7a76\u3002 \u76f8\u5c0d\u65bc\u5b78\u754c\u7684\u5c0f\u6578\u91cf\u8490\u96c6\u3001\u7279\u5b9a\u5c08\u696d\u6027\u7684\u554f\u984c\u63a2\u8a0e\uff0c\u696d\u754c\u5c0d\u65bc\u5730\u7406\u8cc7\u8a0a\u8207\u8de8\u9818\u57df\u6574\u5408\u7684 \u6f5b\u5728\u5546 \u6a5f\u8207\u7d93\u6fdf\u6548\u61c9\u4e0a\u66f4\u70ba\u7a4d\u6975\u3002\u4f8b\u5982 Google \u5728\u5730\u5716\u3001\u8857\u666f\u4e0a\u7684\u6295\u8cc7\uff0c\u540c\u6642\u6301\u7e8c\u300c\u514d \u8cbb\u300d\u958b\u653e\u4f7f\u7528\u5176\u670d\u52d9\uff0c\u5438\u5f15\u4e86\u5168\u7403\u4f7f\u7528\u8005\u7684\u529b\u91cf\u300c\u96f6\u6210\u672c\u300d\u8ca2\u737b\u5927\u91cf\u7684\u4f7f\u7528\u8005\u6a19\u8a18\uff0c\u7d2f \u7a4d\u4e86\u76ee\u524d\u4efb\u4e00\u500b\u570b\u5bb6\u7121\u80fd\u8207\u4e4b\u5339\u6575\u7684\u5927\u6578\u64da\u8cc7\u6599\u3002\u7121\u8ad6\u662f\u5728\u5730\u5716\u3001\u5730\u7406\u6578\u64da\u3001\u7db2\u9801\u6587\u5b57\u3001 \u5716\u7247\u53ca\u4f7f\u7528\u8005\u67e5\u8a62\u8a5e\u7d00\u9304\uff0c\u90fd\u8b93\u5176\u4ed6 LBS \u61c9\u7528\u670d\u52d9\u96e3\u671b\u5176\u9805\u80cc\u3002\u800c\u5176\u4ed6\u5546\u696d\u5de8\u64d8\u5982 Bing Maps\u3001Yahoo! Maps\u3001Apple Maps\u3001Facebook \u7684\u5730\u7406\u8cc7\u6599\u5eab\u6240\u64c1\u6709\u7684\u6578\u64da\u4ea6\u4e0d\u5bb9\u5c0f\u89b7\uff0c \u751a\u81f3\u662f\u5168\u7403\u6027\u975e\u71df\u5229\u7d44\u7e54\u7684\u5730\u7406\u4f4d\u7f6e\u8cc7\u8a0a\uff0c\u5982\uff1aOpenStreetMap 1 \u3001Wikimapia 2 \u7b49\u4e5f\u90fd\u5177 \u5099\u4e86\u6578\u5343\u842c\u7684 POI \u8cc7\u8a0a\u3002 Ahlers \u8207 Boll \u5728\u57fa\u65bc\u5730\u9ede\u7684\u7db2\u9801\u641c\u5c0b\u7814\u7a76\u4e2d[1]\uff0c\u63d0\u51fa\u4e86\u4e00\u500b\u5f9e\u7db2\u9801\u4e2d\u64f7\u53d6\u5730\u9ede\u7684\u7cfb\u7d71 \u67b6\u69cb\uff0c\u4e3b\u8981\u5206\u70ba crawling\u3001\u65b7\u8a5e\u8207\u7d22\u5f15\u7db2\u9801\uff0c\u4ee5\u53ca\u641c\u5c0b\u8207\u6392\u5e8f\u7b49\u4e09\u500b\u5b50\u7cfb\u7d71\uff0c\u5176\u4e2d\u4ed6\u5011 \u6240\u63a1\u7528\u7684 crawling \u7b56\u7565\u53c8\u53ef\u5206\u70ba\u5169\u7a2e\uff1a\u4ee5\u5730\u9ede\u5b57\u5178\u70ba\u4e3b\u548c\u4ee5\u95dc\u9375\u5b57\u70ba\u4e3b\u7684\u65b9\u6cd5[2]\uff0c\u900f \u904e\u81ea\u9069\u61c9\u5316(adaptive)\u7684\u5b78\u7fd2\u8207\u9810\u6e2c\u53ef\u80fd\u5305\u542b\u5730\u9ede\u7684\u7db2\u9801\u4f86\u6709\u6548\u63d0\u5347\u6574\u9ad4\u53ec\u56de\u7387(recall)\uff0c \u8a72\u7814\u7a76\u4e3b\u8981\u91dd\u5c0d\u5fb7\u570b\u591a\u500b\u57ce\u5e02\u9032\u884c\u7db2\u9801\u722c\u53d6\u8207\u7d22\u5f15\u3002 \u5728\u672c\u7bc7\u8ad6\u6587\u4e2d\uff0c\u6211\u5011\u5f9e\u7db2\u969b\u7db2\u8def\u4e2d\u627e\u51fa\u5305\u542b\u5730\u5740\u7db2\u9801\uff0c\u4e26\u4e14\u5229\u7528\u547d\u540d\u5be6\u9ad4\u8fa8\u8b58(Named Entity Recognition, \u4e09\u3001 \u7cfb\u7d71\u67b6\u69cb \u6211\u5011\u7684\u7cfb\u7d71\u67b6\u69cb\u5716\u5982\u5716 \u4e09\u6240\u793a\u3002\u672c\u7cfb\u7d71\u7684\u7b2c\u4e00\u90e8\u5206\u662f\u5229\u7528\u95dc\u9375\u5b57\u4ee5\u53ca\u5730\u5740 pattern \u7d44\u5408 \u800c\u6210\u7684\u67e5\u8a62\u8a5e\u900f\u904e Google \u641c\u5c0b\u5f15\u64ce\u4f86\u641c\u96c6\u5305\u542b\u5730\u5740\u7db2\u9801(ABP)\uff0c\u4e26\u4f7f\u7528\u4ee3\u7406\u4f3a\u670d\u5668\u63d0\u5347 \u641c\u5c0b\u6548\u7387\u3002\u672c\u7cfb\u7d71\u7684\u7b2c\u4e8c\u90e8\u5206\u5247\u5229\u7528\u5730\u5740\u8fa8\u8b58\u6a21\u7d44\u4ee5\u53ca\u4e2d\u6587\u7d44\u7e54\u540d\u7a31\u8fa8\u8b58\u6a21\u7d44\u627e\u51fa\u7db2\u9801 \u4e2d\u7684\u5730\u5740\u4ee5\u53ca POI \u540d\u7a31\uff0c\u63a5\u8457\u518d\u7528\u9019\u4e9b\u5730\u5740\u53ca\u7d44\u7e54\u540d\u7a31\u7d44\u6210 POI \u540d\u7a31\u8207\u4f4d\u7f6e\u7684\u57fa\u672c\u914d \u5c0d\u3002\u7b2c\u4e09\u90e8\u4efd\u5247\u70ba\u6bcf\u4e00\u500b POI \u914d\u5c0d\u64f7\u53d6\u5176\u76f8\u95dc\u8cc7\u8a0a\u3002\u6700\u5f8c\u5c07\u6bcf\u4e00\u7d44\u914d\u5c0d\u548c\u914d\u5c0d\u7684\u76f8\u95dc \u8cc7\u8a0a\u6574\u7406\u6210\u4e00\u7b46 POI\uff0c\u4e26\u653e\u5165 POI \u8cc7\u6599\u5eab\u4e2d\u4f9b\u4f7f\u7528\u8005\u67e5\u8a62\u3002 \u5716 \u4e09\u3001POI \u8cc7\u6599\u5eab\u5efa\u7f6e\u7cfb\u7d71\u67b6\u69cb\u5716 3.1 Query-based \u722c\u87f2 \u5728\u672c\u8ad6\u6587\u4e2d\uff0c\u6211\u5011\u8a2d\u8a08\u4e00\u500b Query-based \u722c\u87f2\u53d6\u5f97 ABP\u3002\u6211\u5011\u4e4b\u6240\u4ee5\u6536\u96c6 ABP \u7684\u539f\u56e0 \u662f\u56e0\u70ba\u5730\u5740\u76f8\u8f03\u65bc\u5176\u4ed6\u76f8\u95dc\u8cc7\u8a0a\u7279\u5fb5\u8f03\u70ba\u660e\u986f\uff0c\u56e0\u6b64\u4f7f\u5f97\u5730\u5740\u76f8\u5c0d\u5bb9\u6613\u8fa8\u8b58\uff0c\u6b64\u5916\u6bcf\u4e00 \u7b46 POI \u90fd\u9700\u8981\u6709\u7d93\u7def\u5ea6\u7684\u8cc7\u8a0a\u624d\u80fd\u5b9a\u4f4d\u5728\u96fb\u5b50\u5730\u5716\u4e0a\uff0c\u800c\u5730\u5740\u80fd\u5920\u900f\u904e\u8a31\u591a\u5de5\u5177\u8f49\u63db \u70ba\u7d93\u7def\u5ea6\u3002\u56e0\u6b64\u6211\u5011\u9078\u64c7\u4ee5\u5730\u5740\u70ba\u57fa\u790e\uff0c\u70ba\u6bcf\u4e00\u7b46\u5730\u5740\u64f7\u53d6\u5176\u540d\u7a31\u548c\u76f8\u95dc\u8cc7\u8a0a\u4e26\u5c07\u5176\u6574 \u7406\u6210 POI\u3002 \uf06c \u67e5\u8a62\u95dc\u9375\u5b57 \u70ba\u4e86\u4f7f\u641c\u5c0b\u5230\u7684\u7db2\u9801\u76e1\u53ef\u80fd\u5305\u542b\u5730\u5740\uff0c\u6211\u5011\u4f7f\u7528\"\u95dc\u9375\u5b57+\u5730\u5740 pattern\"\u505a\u70ba\u6211\u5011\u7684\u67e5 \u4f7f\u7528\u9023\u9396\u5546\u5e97\u540d\u7a31\u641c\u5c0b\u7684\u8a71\uff0cGoogle \u641c\u5c0b\u5f15\u64ce\u5927\u591a\u53ea\u6703\u56de\u50b3\u5b98\u65b9\u7db2\u7ad9\u7684\u9996\u9801\uff0c\u56e0\u6b64\u82e5 \u6211\u5011\u50c5\u4f7f\u7528\u4e0a\u9762\u5169\u7a2e\u67e5\u8a62\u95dc\u9375\u5b57\u7684\u8a71\uff0c\u5c07\u6703\u6f0f\u6389\u9019\u4e00\u985e\u578b\u7684\u7db2\u9801\u3002\u70ba\u6b64\u6211\u5011\u8a2d\u8a08\u4e86\u7b2c\u4e09 \u7a2e\u985e\u578b\u7684\u67e5\u8a62\u95dc\u9375\u5b57\u4f86\u89e3\u6c7a\u9019\u4e00\u554f\u984c\uff1a 3. \u9023\u9396\u5546\u5bb6\u540d\u7a31\uff1a\u4ee5\u9023\u9396\u5546\u5bb6\u540d\u7a31\u53ca\"\u9580\u5e02 or \u5206\u5e97\"\u505a\u70ba\u67e5\u8a62\u8a5e\uff0c\u63a5\u8457\u53d6\u56de\u641c\u5c0b\u7d50\u679c\u7684 \u524d 10 \u500b\u7db2\u9801\u3002 \u5716 \u56db\u3001Query-based \u722c\u87f2\u6240\u4f7f\u7528\u7684\u95dc\u9375\u5b57 \uf06c \u641c\u5c0b\u6548\u7387\u7684\u6539\u5584 2 logC(s) # of search results for query s in log scale 3 logC(a, s) # of search results for query a+s in log scale 4 R(a + s|a) the ratio of C(a+s) to C(a) 5 R(a + s|s) the ratio of C(a+s) to C(s) 6 P(a + s|Ta) the percentage of top 10 snippets from Qa that support POI pair (a,s) 7 P(a + s|Ts) the percentage of top 10 snippets from Qs that support POI pair (a,s) 8 P(a + s|Ta+s) the percentage of top 10 snippets from Qa+s that support POI pair (a,s) 9 Corr(a, s|Ta) Correlation of a and s in Ta 10 Corr(a, s|Ts) Correlation of a and s in Ts 11 Corr(a, s|Ta+s) Correlation of a and s in Ta+s 12 cos(Ta, Ts) the cosine similarity for snippet Ta and Ts 13 cos(Ta, Ta+s) the cosine similarity for snippet Ta and Ta+s 14 cos(Ts, Ta+s) the cosine similarity for snippet Ts and Ta+s 15 NDCG(s|Ta) the rank of s in top 10 snippets from Ta 16 NDCG(s|Ts) the rank of s in top 10 snippets from Ts 17 NDCG(s|Ta+s) the rank of s in top 10 snippets from Ta+s 18 Date(a, a + s) Da -Da+s in log scale 19 Date(s, a + s) Ds -Da+s in log scale \u63a5\u8457\u627e\u51fa\u5730\u5740\u6240\u5728\u6a39\u8449\u7bc0\u9ede\u7684\u4f4d\u7f6e\u4ee5\u53ca\u7d44\u7e54\u540d\u7a31\u6240\u5728\u6a39\u8449\u7bc0\u9ede\u7684\u4f4d\u7f6e(\u5982\u5716 \u4e03\u6240\u793a)\uff0c \u6700\u5f8c\u4ee5\u5730\u5740\u7bc0\u9ede\u548c\u96e2\u5730\u5740\u7bc0\u9ede\u6700\u8fd1\u7684\u7d44\u7e54\u540d\u7a31\u7bc0\u9ede\u4e4b\u6700\u5c0f\u5171\u540c\u7956\u5148\u7bc0\u9ede\u505a\u70ba\u65b0\u7684\u6839\u7bc0 \u9ede\uff0c\u4e26\u5c07\u8a72\u5b50\u6a39(\u5716 \u4e03\u4e2d\u865b\u7dda\u90e8\u5206)\u8996\u70ba\u6b64\u914d\u5c0d\u7684\u76f8\u95dc\u8cc7\u8a0a\u3002 2. \u641c\u5c0b\u7d50\u679c\u7247\u6bb5 POI \u7684\u76f8\u95dc\u8cc7\u8a0a\u4f86\u6e90\u9664\u4e86\u7db2\u9801\u672c\u8eab\u4e4b\u5916\uff0cGoogle Snippets \u4e5f\u662f\u6211\u5011\u8003\u616e\u7684\u9805\u76ee\uff0c\u56e0\u70ba\u7db2 \u9801\u4e2d\u51fa\u73fe\u7684\u5730\u5740\u672a\u5fc5\u8207\u8a72\u7db2\u9801\u7684\u4e3b\u984c\u6027\u76f8\u540c\uff0c\u56e0\u6b64\u82e5\u8cc7\u6599\u4f86\u6e90\u50c5\u4f7f\u7528\u7db2\u9801\u5167\u5bb9\uff0c\u53ef\u80fd\u6703 \u9020\u6210\u90e8\u5206\u914d\u5c0d\u7684\u76f8\u95dc\u8cc7\u8a0a\u5b8c\u5168\u932f\u8aa4\u7684\u554f\u984c\u3002\u70ba\u6b64\u6211\u5011\u4f7f\u7528\u914d\u5c0d\u4e2d\u7684\"\u540d\u7a31\"+\"\u5730\u5740\" \u505a\u70ba\u67e5\u8a62\u8a5e(\u5730\u5740\u8207\u540d\u7a31\u7686\u52a0\u4e0a\u96d9\u5f15\u865f)\uff0c\u53d6\u56de Google \u641c\u5c0b\u5f15\u64ce\u56de\u50b3\u524d\u5341\u7b46\u7db2\u9801\u7684 Snippets \u505a\u70ba\u8a72\u914d\u5c0d\u7684\u76f8\u95dc\u8cc7\u8a0a\u3002 (\u9019\u88e1\u7684\u03a6k\u662f\u4e3b\u984c k \u4e2d\u7684\u8a5e\u5206\u5e03\uff0c\u800c\u03f4 \u662f\u6587\u4ef6 s \u7684\u4e3b\u984c\u5206\u5e03) \u6211\u5011\u5229\u7528\u6b64\u516c\u5f0f\u7b97\u51fa\u7684 ( | )\uff0c\u70ba\u6bcf\u4e00\u7b46 POI \u76f8\u95dc\u8cc7\u8a0a\u4e2d\u7684\u6240\u6709\u53e5\u5b50\u7d66\u4e88\u4e00\u500b\u5206\u6578\u4e26 \u6392\u5e8f\uff0c\u6700\u5f8c\u9078\u64c7\u5206\u6578\u8f03\u9ad8\u7684\u53e5\u5b50\u505a\u70ba\u8a72 POI \u7684\u76f8\u95dc\u8cc7\u8a0a\u3002 \u56db\u3001 \u7d50\u679c\u5be6\u9a57 \u672c\u8ad6\u6587\u4e2d\u6211\u5011\u9032\u884c\u4e86\u4e09\u500b\u5be6\u9a57\uff0c\u5206\u5225\u91dd\u5c0d\u7cfb\u7d71\u7684\u591a\u500b\u6a21\u7d44\u9032\u884c\u6548\u80fd\u8207\u6548\u7387\u7684\u8a55\u4f30\u3002\u7b2c\u4e00 \u500b\u5be6\u9a57\u662f\u722c\u87f2\u7684\u641c\u5c0b\u6548\u7387\uff0c\u7b2c\u4e8c\u500b\u5be6\u9a57\u662f POI \u914d\u5c0d\u7684\u6e96\u78ba\u7387\u8a55\u4f30\uff0c\u6700\u5f8c\u7b2c\u4e09\u500b\u5be6\u9a57\u662f \u76f8\u95dc\u8cc7\u8a0a\u64f7\u53d6\u7684\u8a55\u4f30\u3002\u672c\u7814\u7a76\u7684\u5be6\u9a57\u4e2d\uff0c\u6211\u5011\u5b9a\u7fa9\u4e86\u4ee5\u4e0b\u5169\u500b\u6e2c\u91cf\u503c\uff1a \uf0d8 \u5730\u5740\u5305\u542b\u7387(ABR) ABR = \u5305\u542b\u5730\u5740\u7684\u7db2\u9801 / \u62dc\u8a2a\u7684\u7db2\u9801\u6578\u91cf \uf0d8 \u6295\u8cc7\u5831\u916c\u7387(ROI) P ( | ) = ( |\u03f4 ,\u03a6k) = \u2211 ( | ,\u03a6k) ( |\u03f4 ) \u5716 \u516b\u3001\u4e0d\u540c\u7684\u5730\u5740 pattern \u7684 ABR \u6bd4\u8f03\u5716 \uf06c Heroku \u4ee3\u7406\u4f3a\u670d\u5668\u641c\u5c0b\u6548\u7387 \u5728\u9019\u500b\u5be6\u9a57\u4e2d\uff0c\u6211\u5011\u4f7f\u7528\u50b3\u7d71\u641c\u5c0b\u722c\u87f2\u505a\u70ba\u57fa\u672c\u65b9\u6cd5\uff0c\u8207Heroku\u4ee3\u7406\u4f3a\u670d\u5668\u7684\u6539\u9032\u65b9\u6cd5 \u6bd4\u8f03ABP\u7684\u722c\u53d6\u6548\u7387\uff0c\u4e26\u4e14\u89c0\u5bdf\u641c\u96c6\u5230\u7684\u5730\u5740\u6578\u91cf\u6301\u7e8c3\u5468\uff0c\u5716\u4e5d\u70ba\u6548\u80fd\u6bd4\u8f03\u7d50\u679c\u3002\u5f9e \u5716\u4e5d\u4e2d\u53ef\u4ee5\u770b\u51fa\u4f7f\u7528Heroku\u4ee3\u7406\u4f3a\u670d\u5668\u7684\u65b9\u6cd5\u7684\u641c\u5c0b\u6548\u7387\u6bd4\u57fa\u672c\u65b9\u6cd5\u9084\u8981\u597d\u4e0a\u8a31\u591a\uff0c\u56e0 \u6b64\u5229\u7528\u4ee3\u7406\u4f3a\u670d\u5668\u7684\u65b9\u6cd5\u53ef\u4ee5\u660e\u986f\u7684\u589e\u9032\u6293\u53d6\u7684\u6548\u7387\u3002 # Distinct POI names 702,793 844,165 934,896 # Pairs 4,406,985 7,630,332 11,062,343 # POI 694,730 743,555 764,840 # Overlapping addresses with yellow pages 264,342 264,342 264,342 # Overlapping address containing correct POI pairs 107,257 121,932 129,913 # of addresses that are predicted correct 52,222 53,536 54,031 \u5716 \u5341\u4e00\u3001\u4f7f\u7528\u4e0d\u540c\u5927\u5c0f\u81ea\u52d5\u95dc\u9375\u8a5e\u5eab\u6bd4\u8f03\u6548\u80fd \u5f9e\u5716 \u5341\u4e00\u7684\u7d50\u679c\u4e2d\u6211\u5011\u53ef\u4ee5\u770b\u51fa\uff0c\u96d6\u7136\u5728 recall \u65b9\u9762 POI \u914d\u5c0d\u9a57\u8b49\u6a21\u7d44\u7684\u6548\u80fd\u7565\u4f4e\u65bc \u5716 \u5341\u4e8c\u3001POIDB_AI \u548c POIDB \u7684\u6548\u80fd\u6bd4\u8f03\u5716 \u4e94\u3001 \u7d50\u8ad6\u8207\u672a\u4f86\u7814\u7a76 \u96fb\u5b50\u5730\u5716\u7684\u767c\u5c55\u5927\u5927\u6539\u8b8a\u4e86\u73fe\u4ee3\u4eba\u7684\u751f\u6d3b\u7fd2\u6163\uff0c\u4e14\u5df2\u7d93\u5728\u6211\u5011\u65e5\u5e38\u4e2d\u626e\u6f14\u4e86\u4e0d\u53ef\u6216\u7f3a\u7684 \u91cd\u8981\u89d2\u8272\u3002\u5728\u667a\u6167\u578b\u624b\u6a5f\u666e\u53ca\u7684\u73fe\u4ee3\uff0c\u53ea\u8981\u5728\u80fd\u9023\u63a5\u7db2\u8def\u7684\u5730\u65b9\u96a8\u6642\u90fd\u80fd\u5920\u900f\u904e\u96fb\u5b50\u5730 \u5716\u7372\u53d6\u4efb\u4f55\u5730\u9ede\u7684\u8cc7\u6599\uff0c\u751a\u81f3\u9084\u6709\u8def\u7dda\u898f\u5283\u548c\u5373\u6642\u5c0e\u822a\u7b49\u9644\u52a0\u529f\u80fd\uff0c\u8b93\u4eba\u5011\u4e0d\u6703\u5728\u524d\u5f80 \u964c\u751f\u7684\u5730\u9ede\u6642\u611f\u5230\u4e0d\u4fbf\u3002\u96d6\u7136\u9019\u6a23\u83ab\u5927\u7684\u6539\u8b8a\u78ba\u5be6\u5f71\u97ff\u4e86\u6211\u5011\u7684\u65e5\u5e38\u751f\u6d3b\uff0c\u4f46\u662f\u8981\u501a\u9760 \u4eba\u5de5\u7684\u65b9\u5f0f\u4f86\u5efa\u7acb\u51fa\u542b\u6709\u6975\u5927\u91cfPOI\u7684\u96fb\u5b50\u5730\u5716\u662f\u4e00\u4ef6\u76f8\u7576\u8271\u96e3\u7684\u4efb\u52d9\u3002\u4e0d\u904e\u6b77\u53f2\u8f03\u96fb \u5b50\u5730\u5716\u60a0\u4e45\u7684\u7db2\u969b\u7db2\u8def\uff0c\u5176\u5be6\u65e9\u5df2\u7d2f\u7a4d\u4e86\u5927\u91cf\u7684\u5730\u7406\u8cc7\u8a0a\u53ef\u4f9b\u6211\u5011\u4f7f\u7528\u3002 \u672c\u8ad6\u6587\u900f\u904eQuery-based\u722c\u87f2\u5728\u7db2\u969b\u7db2\u8def\u4e2d\u627e\u51fa\u542b\u6709\u5730\u5740\u7684\u7db2\u9801(ABP)\uff0c\u4e26\u85c9\u7531\u547d\u540d\u5be6\u9ad4 \u8a62\u8a5e\u6a21\u578b\uff0c\u5176\u4e2d\u6211\u5011\u4f7f\u7528(\u8def \u53e6\u5916\uff0c\u7531\u65bc\u591a\u6578\u7684\u9023\u9396\u5546\u5e97\u90fd\u6703\u5728\u81ea\u5df1\u7684\u5b98\u65b9\u7db2\u7ad9\u4e0a\u4ecb\u7d39\u5206\u5e97\u8cc7\u8a0a\uff0c\u4f46\u662f\u5982\u679c\u6211\u5011\u76f4\u63a5 \u7531\u65bc Google \u641c\u5c0b\u5f15\u64ce\u5c0d\u65bc\u4e00\u822c\u4f7f\u7528\u8005\u7684\u67e5\u8a62\u4f7f\u7528\u91cf\u9650\u5236\uff0c\u5728\u514d\u8cbb\u7684\u60c5\u6cc1\u4e0b\u6211\u5011\u6c92\u6709\u8fa6 \u6cd5\u9023\u7e8c\u4f7f\u7528\u76f8\u540c\u7684 IP \u5c0d Google \u641c\u5c0b\u5f15\u64ce\u505a\u67e5\u8a62\uff0c\u6839\u64da\u6211\u5011\u7684\u89c0\u5bdf\uff0c\u82e5\u8981\u9577\u6642\u9593\u7a69\u5b9a\u7684 \u67e5\u8a62\uff0c\u5169\u6b21\u67e5\u8a62\u9593\u5927\u7d04\u9700\u8981\u9593\u9694 120 \u79d2\u9418\uff0c\u5426\u5247\u8a72 IP \u5c31\u6703\u88ab\u5c01\u9396\u3002\u5982\u6b64\u4e00\u4f86\u6703\u5927\u5e45\u589e \u52a0\u641c\u5c0b\u7684\u6210\u672c\uff0c\u56e0\u6b64\u6211\u5011\u4f7f\u7528 Heroku \u4ee3\u7406\u4f3a\u670d\u5668\u4f86\u89e3\u6c7a\u9019\u4e00\u554f\u984c\u3002 \u6211\u5011\u7684\u4f5c\u6cd5\u5982\u5716 \u4e94\u6240\u793a\uff0c\u9996\u5148\u6211\u5011\u900f\u904e Heroku \u4ee3\u7406\u4f3a\u670d\u5668\u53d6\u5f97 Google \u641c\u5c0b\u5f15\u64ce\u7684\u641c \u5c0b\u7d50\u679c\uff0c\u63a5\u8457\u767c\u51fa\u6307\u4ee4\u8b93 Heroku \u4ee3\u7406\u4f3a\u670d\u5668\u91cd\u65b0\u555f\u52d5\u4e26\u4e14\u9032\u5165\u7b49\u5f85\uff0c\u91cd\u65b0\u555f\u52d5\u5f8c\u7684 Heroku \u4ee3\u7406\u4f3a\u670d\u5668\u6703\u5f97\u5230\u4e00\u500b\u65b0\u7684 IP \u540c\u6642\u4e5f\u6703\u559a\u9192\u722c\u87f2\u7a0b\u5f0f\uff0c\u91cd\u8907\u4ee5\u4e0a\u7684\u6b65\u9a5f\u3002 \u6839\u64da\u6211\u5011\u7684\u89c0\u5bdf\uff0c\u82e5\u4f7f\u7528\u4e00\u822c\u7684\u65b9\u6cd5\u67e5\u8a62\uff0c\u4e00\u500b\u5c0f\u6642\u7d04\u7372\u5f97 20 \u6b21\u641c\u5c0b\u7d50\u679c\u3002\u82e5\u4f7f\u7528 Heroku \u4ee3\u7406\u4f3a\u670d\u5668\u7684\u4f5c\u6cd5\uff0c\u4e00\u500b\u5c0f\u6642\u80fd\u5920\u67e5\u8a62 70 \u6b21\uff0c\u76f8\u8f03\u4e4b\u4e0b\u6548\u7387\u63d0\u9ad8\u4e86 3.5 \u500d\u3002 \u5716 \u4e94\u3001\u4f7f\u7528\u4ee3\u7406\u4f3a\u670d\u5668\u7684\u722c\u87f2\u904b\u4f5c\u6d41\u7a0b\u5716 3.2 POI \u64f7\u53d6\u6a21\u7d44 F27)\u3002 20 Date(a + s) Today -Da+s in log scale 21 W(p, Ta+s) # of true words in snippet Ta+s 22 W(n, Ta+s) # of false words in snippets Ta+s 23 Lenmin(Ta+s) (NER)\u8fa8\u8b58\u627e\u51fa\u5176\u4e2d\u7684\u5730\u5740\u4ee5\u53ca\u7d44\u7e54\u540d\u7a31\uff0c\u63a5\u8457\u900f\u904e\u5730\u5740\u8207\u7d44\u7e54\u540d\u7a31\u914d\u5c0d\u7cfb\u7d71\u627e\u51fa\u6b63\u78ba ROI = \u4e0d\u91cd\u8907\u7684\u5730\u5740\u6578\u91cf / \u62dc\u8a2a\u7684\u7db2\u9801\u6578\u91cf \u7684\u914d\u5c0d\uff0c\u7136\u5f8c\u5f9e\u7db2\u9801\u4e2d\u6216\u662fGoogle Snippets\u4e2d\u6458\u8981\u6bcf\u4e00\u7d44\u914d\u5c0d\u7684\u76f8\u95dc\u8cc7\u8a0a\uff0c\u6700\u5f8c\u5c07\u9019\u4e9b \u5716 \u4e03\u3001ABP \u4e2d\u7684\u76f8\u95dc\u8cc7\u8a0a\u5b50\u6a39 \u8cc7\u6599\u505a\u70ba\u96fb\u5b50\u5730\u5716\u4e2d\u7684POI\u4f86\u4f7f\u7528\uff0c\u5982\u6b64\u5c31\u80fd\u5feb\u901f\u589e\u52a0\u96fb\u5b50\u5730\u5716\u4e2dPOI\u7684\u8cc7\u6599\u91cf\u3002 4.1 Query-based \u722c\u87f2\u641c\u5c0b\u6548\u7387 \u5716 \u5341\u3001Baseline \u5728\u4e0d\u540c\u7a97\u6846\u5927\u5c0f\u4e0b\u7684 POI \u914d\u5c0d\u6548\u80fd\u5716 \u6b64\u5916\uff0c\u70ba\u4e86\u77ad\u89e3\u85c9\u7531\u6211\u5011\u7684\u7cfb\u7d71\u81ea\u52d5\u5316\u6240\u7522\u751f\u7684\u5730\u5740\u76f8\u95dc\u8cc7\u8a0a\u662f\u5426\u80fd\u771f\u5be6\u61c9\u7528\u5230\u96fb\u5b50\u5730 the minimum word count of string between a and s in snippets Ta+s 24 Lenmax(Ta+s) the maximum word count of string between a and s in snippets Ta+s 25 AvgLen(Ta+s) \uf06c \u76f8\u95dc\u8cc7\u8a0a\u6458\u8981 \u6211\u5011\u5206\u5225\u4ee5\u5730\u5740 pattern \u7684\u6548\u80fd\u3001Heroku \u4ee3\u7406\u4f3a\u670d\u5668\u7684\u641c\u5c0b\u6548\u7387\u548c ABP \u7684\u641c\u5c0b\u6548\u7387\u9019\u4e09 \u500b\u5be6\u9a57\u4f86\u8a55\u4f30 Query-based \u722c\u87f2\u7684\u6548\u80fd\u3002 \u5716\u7684\u6aa2\u7d22\u4e0a\uff0c\u6211\u5011\u85c9\u7531\u8cc7\u8a0a\u6aa2\u7d22\u7684\u5be6\u9a57\u7d50\u679c\u986f\u793a\uff0cQuery-based\u722c\u87f2\u641c\u96c6\u800c\u4f86\u7684\u7db2\u9801\u4e2d \u5716\u4e5d\u3001\u4ee3\u7406\u4f3a\u670d\u5668\u65b9\u6cd5\u548c\u57fa\u672c\u65b9\u6cd5\u7684\u722c\u53d6\u6548\u7387\u6bd4\u8f03 \u5f9e\u5716 \u5341\u4e2d\u7684\u7d50\u679c\u53ef\u4ee5\u770b\u51fa\u5728\u7a97\u6846\u5927\u5c0f\u70ba100\u6642\u6709\u5e73\u8861\u7684\u6db5\u84cb\u7387\u8207\u6e96\u78ba\u7387\uff0c\u56e0\u6b64\u5f8c\u7e8c\u7684\u5be6 \u64f7\u53d6\u51fa\u7684POI\u8cc7\u6599\uff0c\u78ba\u5be6\u5177\u6709\u76f8\u7576\u7684\u5be6\u7528\u6027\u3002 the average word count of string between a and s in snippets Ta+s 26 W(Dict, Ta+s) the average count of connection words (e.g., address is, TEL is, located on) for each middle string in snippets Ta+s 27 MarkMap whether the pair is marked on Google Maps or not \u591a\u6578\u7684\u6642\u5019\u4f7f\u7528\u8005\u6703\u7528\u95dc\u9375\u5b57\u6216\u662f\u985e\u5225\u4f86\u67e5\u8a62\u5546\u5bb6\uff0c\u56e0\u6b64\u6211\u5011\u5fc5\u9808\u6709\u63cf\u8ff0 POI \u7684\u76f8\u95dc 1. \u5305\u542b\u5730\u5740\u7db2\u9801 \u5c0d\u65bc\u542b\u6709\u591a\u7b46\u5730\u5740\u7684\u7db2\u9801\uff0cSu \u63d0\u51fa\u7684\u76f8\u95dc\u8cc7\u8a0a\u64f7\u53d6\u65b9\u6cd5[12]\u53ef\u4ee5\u627e\u51fa\u6bcf\u4e00\u7b46\u5730\u5740\u7684\u76f8\u95dc \u8cc7\u8a0a\u7bc4\u570d\uff0c\u4f46\u662f\u9019\u4e00\u6f14\u7b97\u6cd5\u4e26\u4e0d\u9069\u7528\u65bc\u50c5\u5305\u542b\u55ae\u4e00\u5730\u5740\u7684\u7db2\u9801\uff0c\u540c\u6642\u56e0\u70ba HTML \u7684\u6b63 \u983b\u4f86\u4f30\u8a08 P(poi|s)\uff0c\u4e26\u85c9\u7531 Latent Dirichlet allocation (LDA)[3] \u7522\u751f\u7684\u8a9e\u8a00\u6a21\u578b\u4f86\u505a\u70ba \u5c0b\u6df1\u5ea6\u8a2d\u5b9a\u70ba\u524d500\u7b46\u7db2\u9801\u3002 \u70ba\u4e86\u6574\u5408\u53e5\u5b50\u7684\u76f8\u95dc\u5206\u6578\u8207\u4e3b\u984c\u6a21\u578b\u4ee5\u6709\u6548\u8403\u53d6\u51fa\u4ee3\u8868\u6027\u7684\u6458\u8981\uff0c\u6211\u5011\u4f7f\u7528\u8a5e\u983b\u8207\u5012\u8a5e \u5728\u904e\u4e86\u524d500\u7b46\u641c\u5c0b\u7d50\u679c\u5f8c\u5176ABR\u90fd\u5927\u5e45\u964d\u4f4e\uff0c\u56e0\u6b64\u70ba\u4e86\u6709\u6548\u7387\u5730\u8490\u96c6ABP\uff0c\u6211\u5011\u7684\u641c \u53ef\u4ee5\u7701\u7565\u4e0d\u8a08\u3002\u5047\u8a2d\u6bcf\u4e00\u500b\u6587\u4ef6\u7684\u6a5f\u7387 ( )\u90fd\u662f\u4e00\u81f4\u7684\uff0c\u56e0\u6b64 ( )\u4ea6\u53ef\u4ee5\u7701\u7565\u3002 \u5011\u6700\u5f8c\u9078\u64c7\u4f7f\u7528\u7b2c\u4e8c\u7a2e\u5730\u5740pattern\u505a\u70ba\u67e5\u8a62\u5b57\u4e32\u3002\u53e6\u5916\u503c\u5f97\u6ce8\u610f\u7684\u662f\uff0c\u9019\u5169\u7a2e\u5730\u5740pattern \u6211\u5011\u4f7f\u7528\u8c9d\u5f0f\u5b9a\u7406\u4f86\u4f30\u8a08 ( | )\uff0c\u5c0d\u540c\u4e00\u500b\u67e5\u8a62\u8a5e poi \u800c\u8a00\uff0c ( )\u662f\u56fa\u5b9a\u7684\uff0c\u6240\u4ee5 \u5bdf\u5230\u4e0d\u8ad6\u5c0d\u65bc\u54ea\u4e00\u7a2e\u6df1\u5ea6\uff0c\u7b2c\u4e8c\u7a2e\u5730\u5740pattern\u7684ABR\u90fd\u9ad8\u65bc\u7b2c\u4e00\u7a2e\u5730\u5740pattern\uff0c\u56e0\u6b64\u6211 \u8cc7\u8a0a\u4f86\u4f7f\u5f97\u9019\u985e\u578b\u7684\u67e5\u8a62\u80fd\u5920\u6210\u7acb\u3002\u672c\u8ad6\u6587\u4e2d\u7684 POI \u76f8\u95dc\u8cc7\u8a0a\u4f86\u6e90\u53ef\u4ee5\u5206\u70ba\u4ee5\u4e0b\u5169\u985e\uff1a ( | ) = ( | ) ( ) \u5206\u5225\u6bd4\u8f03\u524d100\u3001200\u3001300\u3001500\u3001750\u53ca1000\u7b46\u641c\u5c0b\u7d50\u679c\u7684ABR\u3002\u5f9e\u5716 \u516b\u4e2d\u6211\u5011\u53ef\u4ee5\u89c0 ( ) \u70ba\u4e86\u5206\u6790\u4e0d\u540c\u7684\u5730\u5740pattern\u5728\u4e0d\u540c\u6df1\u5ea6\u4e0b\u6240\u6293\u53d6\u56de\u4f86\u7684\u7db2\u9801\u7684\u5730\u5740\u5305\u542b\u7387(ABR)\uff0c\u6211\u5011 \u221d ( | ) 2. <\u57ce\u5e02\u540d\u7a31> * (\u8def OR \u8857 OR \u6bb5 OR \u5df7 OR \u5f04) * \u865f 3.3 POI \u76f8\u95dc\u8cc7\u8a0a\u64f7\u53d6 \u7d93\u904e POI \u76f8\u95dc\u8cc7\u8a0a\u64f7\u53d6\u5f8c\uff0c\u5c0d\u65bc\u6bcf\u4e00\u500b\u914d\u5c0d\u6211\u5011\u90fd\u6709\u8a31\u591a\u7684\u63cf\u8ff0\u8a0a\u606f\uff0c\u4f46\u6709\u4e9b\u63cf\u8ff0\u8a0a \u606f\u53ef\u80fd\u662f\u8207\u8a72\u914d\u5c0d\u7121\u95dc\u6216\u662f\u63cf\u8ff0\u4e0d\u5920\u7cbe\u78ba\u7684\u53e5\u5b50\uff0c\u56e0\u6b64\u6211\u5011\u4f7f\u7528\u8cc7\u8a0a\u6aa2\u7d22\u6a21\u578b\u4f86\u8a08\u7b97\u7d66 \u5b9a\u4e00\u500b POI \u540d\u7a31 poi \u505a\u70ba\u67e5\u8a62\u95dc\u9375\u5b57\u9019\u4e9b\u5019\u9078\u76f8\u95dc\u8cc7\u8a0a\u6240\u7522\u751f\u7684\u53e5\u5b50\u8207\u67e5\u8a62\u7684\u76f8\u95dc\u5206\u6578 ( | )\uff0c\u4ee5\u4fbf\u9078\u51fa\u6700\u76f8\u95dc\u7684\u53e5\u5b50 s \u505a\u70ba\u5730\u5740\u76f8\u95dc\u8cc7\u8a0a\u3002\u4e5f\u5c31\u662f\u8aaa\uff0c\u5c07\u76f8\u95dc\u8cc7\u8a0a\u53e5\u5b50\u8996 \u70ba\u6587\u4ef6\uff0c\u800c POI \u540d\u7a31\u8996\u70ba\u67e5\u8a62\u8a5e\uff1a \uf06c \u5730\u5740 pattern \u7684\u6548\u80fd\u8a55\u4f30 \u5728\u9019\u500b\u5be6\u9a57\u4e2d\u6211\u5011\u6bd4\u8f03\u4e86\u5169\u7a2e\u4e0d\u540c\u7684\u5730\u5740pattern\u7684\u6548\u80fd\uff1a 1. <\u57ce\u5e02\u540d\u7a31> * \u865f \uf06c ABP \u641c\u5c0b\u6548\u7387 \u5728\u9019\u500b\u5be6\u9a57\u4e2d\uff0c\u6211\u5011\u6bd4\u8f03\u4e86\u4e09\u7a2e\u4e0d\u540c\u722c\u87f2\u7684ABP\u641c\u5c0b\u6548\u7387\uff0cBaseline\u662f\u4e00\u822c\u7684\u5ee3\u5ea6\u512a\u5148 \u9a57\u6211\u5011\u90fd\u9078\u7528100\u500b\u5b57\u7684\u7bc4\u570d\u505a\u70ba\u6211\u5011\u9810\u8a2d\u7684\u7a97\u6846\u5927\u5c0f\u3002\u70ba\u4e86\u8a55\u4f30POI\u914d\u5c0d\u9a57\u8b49\u6a21\u7d44\u7684 \u5728\u672a\u4f86\u6211\u5011\u6703\u5c08\u6ce8\u5728\u5982\u4f55\u6574\u5408\u6211\u5011\u64f7\u53d6\u7684POI\u4ee5\u53ca\u73fe\u6709POI\u8cc7\u6599\u5eab\u4e2d\u7684POI\uff0c\u5982\u6b64\u4e00\u4f86\u4e0d \u6548\u80fd\uff0c\u6211\u5011\u96a8\u6a5f\u9078\u53d67500\u500b\u5730\u5740\u4e26\u7522\u751f\u4e8621,899\u500bPOI\u914d\u5c0d(\u5c0d\u65bc\u6bcf\u4e00\u500b\u5730\u5740\u6211\u5011\u90fd\u53d63 \u50c5\u50c5\u662f\u5229\u7528\u7db2\u8def\u4e0a\u7684\u8cc7\u6599\u5275\u5efa\u4e00\u500b\u5168\u65b0\u7684POI\u8cc7\u6599\u5eab\uff0c\u540c\u6642\u4e5f\u80fd\u5920\u5229\u7528\u73fe\u6709\u7684POI\u8cc7\u6599 \u500b\u5019\u9078POI\u914d\u5c0d)\u505a\u70ba\u6211\u5011\u7684\u8a13\u7df4\u8cc7\u6599\uff0c\u63a5\u8457\u6211\u5011\u505a3-folds cross-validation\u4f86\u8a55\u4f30\u6211\u5011\u7684 \u5eab\u518d\u9032\u4e00\u6b65\u7684\u8c50\u5bcc\u6211\u5011POI\u8cc7\u6599\u5eab\u4e2d\u7684\u8cc7\u6599\u3002\u6b64\u5916\u6211\u5011\u4e5f\u5e0c\u671b\u7cfb\u7d71\u9664\u4e86\u5b9a\u671f\u722c\u53d6\u65b0\u7684 POI\u914d\u5c0d\u9a57\u8b49\u6a21\u7d44\u7684\u6548\u80fd\uff0c\u5be6\u9a57\u7d50\u679c\u5982\u5716 \u5341\u4e00\u6240\u793a\u3002 POI\u4e4b\u5916\uff0c\u540c\u6642\u80fd\u5920\u5b9a\u671f\u6aa2\u67e5\u6211\u5011\u7684POI\u8cc7\u6599\u5eab\u4e2d\u73fe\u6709\u7684\u8cc7\u6599\uff0c\u4e26\u5c07\u904e\u671f\u7684POI\u904e\u6ffe\u6389\uff0c \u722c\u87f2\uff0c\u9ec3\u9801\u722c\u87f2\u662f\u5c08\u9580\u722c\u53d6\u5404\u7a2e\u9ec3\u9801\u8cc7\u6599\u7684\u722c\u87f2\uff0c\u4f8b\u5982\uff1a\u4e2d\u83ef\u9ec3\u9801 3 \u3001\u611b\u8a55\u7db2 4 \uff0cQuery-based \u722c\u87f2\u5247\u662f\u672c\u8ad6\u6587\u8a2d\u8a08\u7684\u722c\u87f2\uff0c\u6bd4\u8f03\u7d50\u679c\u5982\u8868 \u4e8c\u6240\u793a\u3002 \u4ee5\u78ba\u4fddPOI\u8cc7\u6599\u5eab\u80fd\u5920\u96a8\u6642\u63d0\u4f9b\u7d66\u4f7f\u7528\u8005\u6b63\u78ba\u7684POI\u3002</td></tr><tr><td>\u898f\u5316\u7684\u5931\u6557\uff0c\u9020\u6210\u4e0d\u5c11\u591a\u7b46\u5730\u5740\u7684\u7db2\u9801\u7121\u6cd5\u8655\u7406\u7684\u554f\u984c\u3002\u56e0\u6b64\u6211\u5011\u5728\u672c\u7bc7\u8ad6\u6587\u4e2d\u63d0\u51fa\u65b0 P(poi|s) smoothing \u7684\u65b9\u6cd5\uff0c\u518d\u52a0\u5165\u03bb\u4fc2\u6578\u8abf\u6574\u6b0a\u91cd\uff0c\u516c\u5f0f\u5982\u4e0b\uff1a</td></tr></table>", | |
| "type_str": "table", | |
| "num": null, | |
| "html": null | |
| } | |
| } | |
| } | |
| } |