{
    "paper_id": "O14-5001",
    "header": {
        "generated_with": "S2ORC 1.0.0",
        "date_generated": "2023-01-19T08:04:42.575546Z"
    },
    "title": "POI \u64f7\u53d6 \u64f7\u53d6 \u64f7\u53d6 \u64f7\u53d6:\u5546\u5bb6\u540d\u7a31 \u5546\u5bb6\u540d\u7a31 \u5546\u5bb6\u540d\u7a31 \u5546\u5bb6\u540d\u7a31\u8fa8\u8b58 \u8fa8\u8b58 \u8fa8\u8b58 \u8fa8\u8b58\u8207\u5730\u5740\u914d\u5c0d\u4e4b\u7814\u7a76 \u8207\u5730\u5740\u914d\u5c0d\u4e4b\u7814\u7a76 \u8207\u5730\u5740\u914d\u5c0d\u4e4b\u7814\u7a76 \u8207\u5730\u5740\u914d\u5c0d\u4e4b\u7814\u7a76 POI Extraction from the Web: Store Name Recognition and Address Matching \u6797\u80b2\u6698 \u6797\u80b2\u6698 \u6797\u80b2\u6698 \u6797\u80b2\u6698\u3001 \u3001 \u3001 \u3001\u5f35\u5609\u60e0 \u5f35\u5609\u60e0 \u5f35\u5609\u60e0 \u5f35\u5609\u60e0",
    "authors": [
        {
            "first": "Lin",
            "middle": [],
            "last": "Yu-Yang",
            "suffix": "",
            "affiliation": {
                "laboratory": "",
                "institution": "National Central University",
                "location": {}
            },
            "email": ""
        },
        {
            "first": "Chang",
            "middle": [],
            "last": "Chia-Hui",
            "suffix": "",
            "affiliation": {
                "laboratory": "",
                "institution": "National Central University",
                "location": {}
            },
            "email": ""
        },
        {
            "first": "",
            "middle": [],
            "last": "\u570b\u7acb\u4e2d\u592e\u5927\u5b78\u8cc7\u8a0a\u5de5\u7a0b\u5b78\u7cfb",
            "suffix": "",
            "affiliation": {
                "laboratory": "",
                "institution": "National Central University",
                "location": {}
            },
            "email": ""
        }
    ],
    "year": "",
    "venue": null,
    "identifiers": {},
    "abstract": "Mobility is one of the trends in 2014. According to the report of IDC (International Data Corporation), the worldwide shipments of tablets have exceeded PCs in 2013 Quarter 4, while smart phones has already exceeded other devices in unit shipments and market ratio. With this trend, many location-based services (LBS) have been proposed, for example, navigation, searching restaurants or gas stations. Therefore, how to construct a large POI (Point-of Interest) database is the key problem. In this paper, we solve three problems including Taiwan address normalization, store name extraction, and the matching of addresses and store names. To train a statistical model for store name extraction, we make use of existing store-address pair to prepare training data for sequence labeling. The model is trained using common characteristics from store names in addition to POS tags. When testing on search snippets, we obtain 0.791 F-measure for store name recognition.",
    "pdf_parse": {
        "paper_id": "O14-5001",
        "_pdf_hash": "",
        "abstract": [
            {
                "text": "Mobility is one of the trends in 2014. According to the report of IDC (International Data Corporation), the worldwide shipments of tablets have exceeded PCs in 2013 Quarter 4, while smart phones has already exceeded other devices in unit shipments and market ratio. With this trend, many location-based services (LBS) have been proposed, for example, navigation, searching restaurants or gas stations. Therefore, how to construct a large POI (Point-of Interest) database is the key problem. In this paper, we solve three problems including Taiwan address normalization, store name extraction, and the matching of addresses and store names. To train a statistical model for store name extraction, we make use of existing store-address pair to prepare training data for sequence labeling. The model is trained using common characteristics from store names in addition to POS tags. When testing on search snippets, we obtain 0.791 F-measure for store name recognition.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Abstract",
                "sec_num": null
            }
        ],
        "body_text": [],
        "back_matter": [],
        "bib_entries": {
            "BIBREF0": {
                "ref_id": "b0",
                "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": "Chang, C.-H., Huang, C.-Y., & Su, Y.-S. (2012). Chinese Postal Address and Associated Information Extraction. The 26th Annual Conference of the Japanese Society for Artificial Intelligence, 2012.",
                "links": null
            },
            "BIBREF1": {
                "ref_id": "b1",
                "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": "Chuang, H.-M., Chang, C.-H., & Kao, T.-Y. (2014). Effective Web Crawling for Chinese Addresses and Associated Information. in EC-Web, Munich, Germany, 2014.",
                "links": null
            },
            "BIBREF2": {
                "ref_id": "b2",
                "title": "Location-based Web search. Advanced Information and Knowledge Processing",
                "authors": [
                    {
                        "first": "A",
                        "middle": [],
                        "last": "Dirk",
                        "suffix": ""
                    },
                    {
                        "first": "B",
                        "middle": [],
                        "last": "Susanne",
                        "suffix": ""
                    }
                ],
                "year": 2007,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "55--66",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Dirk, A., & Susanne, B. (2007). Location-based Web search. Advanced Information and Knowledge Processing 2007, 55-66.",
                "links": null
            },
            "BIBREF4": {
                "ref_id": "b4",
                "title": "TEX: An efficient and effective unsupervised Web information extractor. Knowledge-Based Systems",
                "authors": [
                    {
                        "first": "R",
                        "middle": [
                            "C"
                        ],
                        "last": "Hassan",
                        "suffix": ""
                    },
                    {
                        "first": "A",
                        "middle": [],
                        "last": "Sleiman",
                        "suffix": ""
                    }
                ],
                "year": 2013,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "109--123",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Hassan, R. C., & Sleiman, A. (2013). TEX: An efficient and effective unsupervised Web information extractor. Knowledge-Based Systems, 2013, 109-123.",
                "links": null
            },
            "BIBREF5": {
                "ref_id": "b5",
                "title": "Conditional Random Fields: Probabilistic Models for Segmenting and Labeling Sequence Data. ICML Proceedings of the Eighteenth International Conference on Machine Learning",
                "authors": [
                    {
                        "first": "L",
                        "middle": [
                            "D"
                        ],
                        "last": "John",
                        "suffix": ""
                    },
                    {
                        "first": "M",
                        "middle": [],
                        "last": "Andrew",
                        "suffix": ""
                    },
                    {
                        "first": "N",
                        "middle": [
                            "C"
                        ],
                        "last": "Fernando",
                        "suffix": ""
                    }
                ],
                "year": 2001,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "282--289",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "John, L. D., Andrew, M., & Fernando, N.C. (2001). Conditional Random Fields: Probabilistic Models for Segmenting and Labeling Sequence Data. ICML Proceedings of the Eighteenth International Conference on Machine Learning, 282-289.",
                "links": null
            },
            "BIBREF6": {
                "ref_id": "b6",
                "title": "Application and Extraction of Postal Addresses and Related Information",
                "authors": [
                    {
                        "first": "S.-Y",
                        "middle": [],
                        "last": "Li",
                        "suffix": ""
                    }
                ],
                "year": 2009,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Li, S.-Y. (2009). Application and Extraction of Postal Addresses and Related Information, National Central University, 2009.",
                "links": null
            },
            "BIBREF7": {
                "ref_id": "b7",
                "title": "Chinese Organization Name Recognition Based on Multiple Features",
                "authors": [
                    {
                        "first": "Y",
                        "middle": [],
                        "last": "Ling",
                        "suffix": ""
                    },
                    {
                        "first": "J",
                        "middle": [],
                        "last": "Yang",
                        "suffix": ""
                    },
                    {
                        "first": "L",
                        "middle": [],
                        "last": "He",
                        "suffix": ""
                    }
                ],
                "year": 2012,
                "venue": "Pacific Asia conference on Intelligence and Security Informatics",
                "volume": "7299",
                "issue": "",
                "pages": "136--144",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Ling, Y., Yang, J., & L. He. (2012). Chinese Organization Name Recognition Based on Multiple Features. Pacific Asia conference on Intelligence and Security Informatics, 7299, 136-144.",
                "links": null
            },
            "BIBREF8": {
                "ref_id": "b8",
                "title": "ViDE: A Vision-Based Approach for Deep Web Data Extraction",
                "authors": [
                    {
                        "first": "W",
                        "middle": [],
                        "last": "Liu",
                        "suffix": ""
                    },
                    {
                        "first": "X",
                        "middle": [],
                        "last": "Meng",
                        "suffix": ""
                    },
                    {
                        "first": "W",
                        "middle": [],
                        "last": "Meng",
                        "suffix": ""
                    }
                ],
                "year": 2010,
                "venue": "Transactions on Knowledge and Data Engineering",
                "volume": "22",
                "issue": "3",
                "pages": "447--460",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Liu, W., Meng, X., & Meng, W. (2010). ViDE: A Vision-Based Approach for Deep Web Data Extraction. Transactions on Knowledge and Data Engineering, 22(3), 447-460.",
                "links": null
            },
            "BIBREF10": {
                "ref_id": "b10",
                "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": "Su, Y.-S. (2012). Associated Information Extraction for Enabling Entity Search on Electronic Map, National Central University, 2012.",
                "links": null
            },
            "BIBREF11": {
                "ref_id": "b11",
                "title": "Semi-joint labeling for Chinese named entity recognition",
                "authors": [
                    {
                        "first": "C.-W",
                        "middle": [],
                        "last": "Wu",
                        "suffix": ""
                    },
                    {
                        "first": "R",
                        "middle": [
                            "T"
                        ],
                        "last": "Tsai",
                        "suffix": ""
                    },
                    {
                        "first": ".-H",
                        "middle": [],
                        "last": "Hsu",
                        "suffix": ""
                    },
                    {
                        "first": "W.-L",
                        "middle": [],
                        "last": "",
                        "suffix": ""
                    }
                ],
                "year": 2008,
                "venue": "Proceedings of the 4th Asia information retrieval conference",
                "volume": "4993",
                "issue": "",
                "pages": "107--116",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Wu, C.-W., Tsai, R. T.-H., & Hsu, W.-L. (2008). Semi-joint labeling for Chinese named entity recognition. In Proceedings of the 4th Asia information retrieval conference, 4993, 107-116.",
                "links": null
            },
            "BIBREF12": {
                "ref_id": "b12",
                "title": "A Method of Chinese Organization Named Entity Recognition Based on Statistical Word Frequency",
                "authors": [
                    {
                        "first": "X",
                        "middle": [],
                        "last": "Yao",
                        "suffix": ""
                    }
                ],
                "year": 2011,
                "venue": "Part of Speech and Length. Broadband Network and Multimedia Technology (IC-BNMT)",
                "volume": "",
                "issue": "",
                "pages": "637--641",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "X. Yao. (2011). A Method of Chinese Organization Named Entity Recognition Based on Statistical Word Frequency, Part of Speech and Length. Broadband Network and Multimedia Technology (IC-BNMT), 637-641.",
                "links": null
            },
            "BIBREF13": {
                "ref_id": "b13",
                "title": "Automatic Recognition of Chinese Organization Name Based on Conditional Random Fields. Natural Language Processing and Knowledge Engineering",
                "authors": [
                    {
                        "first": "S",
                        "middle": [],
                        "last": "Zhang",
                        "suffix": ""
                    },
                    {
                        "first": "&",
                        "middle": [
                            "X"
                        ],
                        "last": "Wang",
                        "suffix": ""
                    }
                ],
                "year": 2007,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "229--233",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "S. Zhang, & X. Wang. (2007). Automatic Recognition of Chinese Organization Name Based on Conditional Random Fields. Natural Language Processing and Knowledge Engineering 2007, 229-233.",
                "links": null
            },
            "BIBREF15": {
                "ref_id": "b15",
                "title": "The 18th Conference on Artificial Intelligence (TAAI 2013",
                "authors": [
                    {
                        "first": "\u6539\u5584\u7db2\u9801\u5546\u5bb6\u591a\u6a19\u7c64",
                        "middle": [],
                        "last": "\u9673\u5b9c\u52e4\u3001\u8cf4\u90c1\u5a77\u3001\u838a\u79c0\u654f\u8207\u5f35\u5609\u60e0 ; \u3002 \u52a0\u5165 Google Snippets",
                        "suffix": ""
                    },
                    {
                        "first": "",
                        "middle": [],
                        "last": "\u5206\u985e",
                        "suffix": ""
                    }
                ],
                "year": 2013,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "6--8",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "\u9673\u5b9c\u52e4\u3001\u8cf4\u90c1\u5a77\u3001\u838a\u79c0\u654f\u8207\u5f35\u5609\u60e0(2013)\u3002 \u52a0\u5165 Google Snippets \u6539\u5584\u7db2\u9801\u5546\u5bb6\u591a\u6a19\u7c64 \u5206\u985e, The 18th Conference on Artificial Intelligence (TAAI 2013), 6-8.",
                "links": null
            }
        },
        "ref_entries": {
            "TABREF0": {
                "html": null,
                "num": null,
                "content": "<table><tr><td>\u8da8\u52e2\u4e2d\u5177\u6709\u81f3\u95dc\u91cd\u8981\u7684\u5730\u4f4d\uff0c\u56e0\u70ba\u88dd\u7f6e\u884c\u52d5\u5316\u7684\u56e0\u7d20\uff0c\u5927\u91cf\u67e5\u8a62\u9700\u6c42\u56e0\u6b64\u8a95 \u751f\uff0c\u4f8b\u5982\uff1a\u8def\u7dda\u5c0e\u822a\u3001\u67e5\u8a62\u9644\u8fd1\u9910\u5ef3\u3001\u52a0\u6cb9\u7ad9\u7b49\u3002\u9069\u5730\u6027\u670d\u52d9\u8981\u80fd\u5ee3\u6cdb\u7684\u63d0 \u4f9b\u670d\u52d9\uff0c\u901a\u5e38\u9700\u8981\u6709\u4e00\u500b\u5b8c\u6574\u7684 POI(Point of Interest)\u8cc7\u6599\u5eab\uff0c\u800c\u6574\u500b\u7db2\u8def \u5c31\u662f\u6700\u5927\u7684\u8cc7\u8a0a\u4f86\u6e90\u3002\u9019\u4e9b\u8cc7\u6599\u6e90\u81ea\u65bc\u7db2\u7ad9\u7ba1\u7406\u8005\u3001\u7fa4\u773e\u5916\u5305 (crowdsourcing) \u6216\u500b\u4eba\u4f7f\u7528\u8005\u6240\u5206\u4eab\u7684\u8cc7\u8a0a\uff0c\u5305\u62ec\u4e86\u5730\u5740\u3001\u540d\u7a31\u3001\u96fb\u8a71\u3001\u8a55\u8ad6\u7b49\u8cc7\u8a0a\u3002\u73fe\u5728 \u96d6\u7136\u6709\u5404\u7a2e\u64f7\u53d6\u5730\u5740\u76f8\u95dc\u8cc7\u8a0a\u7684\u65b9\u6cd5\uff0c\u4f46\u7d93\u5e38\u9762\u81e8\u7121\u6cd5\u53d6\u5f97\u660e\u78ba POI \u7684\u540d\u7a31\uff0c \u5728\u8cc7\u8a0a\u6aa2\u7d22\u4e0a\u53d7\u5230\u5f88\u5927\u7684\u9650\u5236\u3002 \u5728\u672c\u7bc7\u8ad6\u6587\u4e2d\uff0c\u6211\u5011\u63d0\u51fa\u4e00\u500b\u5546\u5bb6\u540d\u7a31\u8fa8\u8b58\u7684\u65b9\u6cd5\uff0c\u85c9\u7531\u6536\u96c6\u7db2\u8def\u4e0a\u5305\u542b\u5730 \u5740\u7684\u7db2\u9801\uff0c\u5efa\u7acb\u4e00\u500b\u5177\u6709\u5546\u5bb6\u540d\u7a31\u8207\u5730\u5740\u95dc\u806f\u6027\u7684\u8cc7\u6599\u5eab\uff0c\u4ee5\u63d0\u9ad8\u5730\u5740\u76f8\u95dc \u8cc7\u8a0a\u6aa2\u7d22\u7684\u6548\u679c\uff0c\u8b93\u4f7f\u7528\u8005\u5728\u4f7f\u7528\u884c\u52d5\u88dd\u7f6e\u67e5\u8a62\u6642\uff0c\u80fd\u76f4\u63a5\u8f38\u5165\u5e97\u5bb6\u540d\u7a31\u6216 \u95dc\u9375\u5b57\u67e5\u8a62\u5730\u5740\u4e4b\u670d\u52d9\uff0c\u63d0\u4f9b\u4fbf\u5229\u7684\u67e5\u8a62\u529f\u80fd\u3002\u5176\u4e2d\uff0c\u5728\u5546\u5bb6\u547d\u540d\u5be6\u9ad4\u8fa8\u8a8d \u4e0a\uff0c\u672c\u7bc7\u8ad6\u6587\u63d0\u51fa\u4e86\u5546\u5bb6\u8207\u7d44\u7e54\u540d\u7a31\u5728\u547d\u540d\u4e0a\u7684\u5171\u540c\u7279\u6027\uff0c\u5229\u7528\u6b64\u5171\u540c\u7279\u6027 \u7576\u4f5c\u7279\u5fb5\u52a0\u5165 CRF \u6a21\u578b\uff0c\u4ee5\u63d0\u4f9b N-Gram \u8207\u8a5e\u6027\u4e4b\u5916\u7684\u7279\u5fb5\u3002 \u95dc\u9375\u8a5e \u95dc\u9375\u8a5e \u95dc\u9375\u8a5e \u95dc\u9375\u8a5e\uff1a \uff1a \uff1a \uff1a\u5546\u5bb6\u5730\u7406\u8cc7\u8a0a\u6aa2\u7d22\u3001\u5546\u5bb6\u540d\u7a31\u64f7\u53d6\u3001\u5546\u5bb6\u540d\u7a31\u8207\u5730\u5740\u914d\u5c0d\u3001\u5e8f\u5217\u6a19 \u8a18\u3001\u689d\u4ef6\u96a8\u6a5f\u57df 1. \u7dd2\u8ad6 \u7dd2\u8ad6 \u7dd2\u8ad6 \u7dd2\u8ad6 \u6839\u64da\u570b\u969b\u6578\u64da\u8cc7\u8a0a IDC \u65bc 2013 \u5e74 9 \u6708\u8abf\u67e5\u5831\u544a\u986f\u793a\uff0c\u5e73\u677f\u96fb\u8166\u7684\u51fa\u8ca8\u91cf\u5728 2013 \u5e74\u7b2c\u56db \u5b63\u9996\u6b21\u8d85\u904e\u500b\u4eba\u96fb\u8166\uff0c\u800c\u667a\u6167\u578b\u624b\u6a5f\u4e0d\u8ad6\u5728\u51fa\u8ca8\u91cf\u6216\u5e02\u4f54\u7387\u65e9\u5c31\u9060\u9060\u8d85\u904e\u684c\u4e0a\u578b\u96fb\u8166\u548c \u53ef\u651c\u5f0f\u96fb\u8166\u7684\u7e3d\u548c\uff0cIDC \u751a\u81f3\u9810\u6e2c\u5e73\u677f\u96fb\u8166\u7684\u51fa\u8ca8\u91cf\u5c07\u5728 2015 \u5e74\u8d85\u904e\u684c\u4e0a\u578b\u96fb\u8166\u548c\u53ef\u651c \u5f0f\u96fb\u8166\u7684\u7e3d\u548c\u3002\u9019\u986f\u793a\u4e86\u884c\u52d5\u88dd\u7f6e\u7684\u666e\u53ca\u662f\u4e00\u7a2e\u4e0d\u53ef\u62b5\u64cb\u7684\u8da8\u52e2\u3002 \u884c\u52d5\u88dd\u7f6e\u7684\u666e\u53ca\u9020\u5c31\u5927\u91cf\u5730\u57df\u6027\u67e5\u8a62\u7684\u9700\u6c42\uff0c\u5176\u4e2d\u6700\u5e38\u898b\u7684\u4e00\u7a2e\u67e5\u8a62\uff0c\u5c31\u662f\u5c0b\u627e\u9644 \u8fd1\u7684\u9910\u5ef3\u6216\u52a0\u6cb9\u7ad9\uff0c\u6839\u64da Google \u65bc 2013 \u5e74\u7b2c\u4e00\u5b63\u53f0\u7063\u667a\u6167\u578b\u624b\u6a5f\u4f7f\u7528\u884c\u70ba\u8abf\u67e5(\u591a\u9078\u984c)\uff0c \u641c\u5c0b\u5167\u5bb9\u4f9d\u5e8f\u70ba\u7522\u54c1\u8cc7\u8a0a(60%)\u3001\u9910\u5ef3\u3001\u9152\u9928\u548c\u9152\u5427(51%)\u3001\u65c5\u904a(49%)\u3001\u5de5\u4f5c\u6a5f \u6703(29%)\u4ee5\u53ca\u8cfc\u5c4b\u3001\u79df\u5c4b\u8cc7\u8a0a(28%)\u3002\u7136\u800c\u7576\u4f7f\u7528\u8005\u5728\u96fb\u5b50\u5730\u5716\u4e0a\u641c\u5c0b\u9019\u4e9b\u5730\u9ede\u540d \u7a31 (POI\uff0cPoint of Interest)\u6642\uff0c\u7d93\u5e38\u7121\u6cd5\u627e\u5230\uff0c\u56e0\u70ba\u96fb\u5b50\u5730\u5716\u4e0a\u96d6\u6709\u5730\u9ede\u540d\u7a31\u6a19\u8a3b\uff0c\u4f46\u662f \u76f8\u95dc\u8cc7\u8a0a\u4e0d\u8db3\uff0c\u800c\u9019\u4e9b\u8cc7\u8a0a\u5176\u5be6\u5927\u591a\u53ef\u4ee5\u5728\u7db2\u9801\u4e2d\u627e\u5230\u3002\u56e0\u6b64\u4f7f\u7528\u8005\u5927\u591a\u5fc5\u9808\u958b\u555f\u700f\u89bd \u5668\u641c\u5c0b\u5546\u5bb6\u540d\u7a31\u627e\u51fa\u5730\u5740\uff0c\u4e26\u628a\u5730\u5740\u8f38\u5165\u81f3\u96fb\u5b50\u5730\u5716\u67e5\u8a62\u8def\u7dda\u3002\u4f46\u884c\u52d5\u88dd\u7f6e\u87a2\u5e55\u5c0f\uff0c\u4e14 \u8f38\u5165\u6587\u5b57\u4e0d\u4fbf\u5229\uff0c\u5982\u679c\u8981\u53cd\u8986\u67e5\u8a62\u5c07\u662f\u4e00\u4ef6\u8017\u6642\u8017\u529b\u7684\u5de5\u4f5c\u3002\u5982\u679c\u9019\u6642\u5019\u6709\u4e00\u500b\u5546\u5bb6\u5730 \u7406\u8cc7\u8a0a\u7cfb\u7d71\u80fd\u4e8b\u5148\u5c07\u7db2\u8def\u4e0a\u7684\u5546\u5bb6\u8cc7\u8a0a\u9032\u884c\u6574\u5408\uff0c\u6700\u5f8c\u63d0\u4f9b\u4e00\u500b APP \u76f4\u63a5\u8b93\u4f7f\u7528\u8005\u67e5\u8a62\uff0c \u5c07\u53ef\u4ee5\u5927\u5e45\u5ea6\u6e1b\u5c11\u4f7f\u7528\u8005\u8207\u88dd\u7f6e\u9593\u7684\u4e92\u52d5\u6b21\u6578\uff0c\u6709\u6548\u7684\u63d0\u4f9b\u641c\u5c0b\u7684\u4fbf\u5229\u6027\u3002 3 4 \u6797\u80b2\u6698\u8207\u5f35\u5609\u60e0 POI\u64f7\u53d6:\u5546\u5bb6\u540d\u7a31\u8fa8\u8b58\u8207\u5730\u5740\u914d\u5c0d\u4e4b\u7814\u7a76 5 6 \u6797\u80b2\u6698\u8207\u5f35\u5609\u60e0 POI\u64f7\u53d6 \u64f7\u53d6:\u5546\u5bb6\u540d\u7a31\u8fa8\u8b58\u8207\u5730\u5740\u914d\u5c0d\u4e4b\u7814\u7a76 7 8 \u6797\u80b2\u6698\u8207\u5f35\u5609\u60e0 POI\u64f7\u53d6:\u5546\u5bb6\u540d\u7a31\u8fa8\u8b58\u8207\u5730\u5740\u914d\u5c0d\u4e4b\u7814\u7a76 9 \u6797\u80b2\u6698\u8207\u5f35\u5609\u60e0 POI\u64f7\u53d6:\u5546\u5bb6\u540d\u7a31\u8fa8\u8b58\u8207\u5730\u5740\u914d\u5c0d\u4e4b\u7814\u7a76 11 12 \u6797\u80b2\u6698\u8207\u5f35\u5609\u60e0 POI\u64f7\u53d6 \u64f7\u53d6:\u5546\u5bb6\u540d\u7a31\u8fa8\u8b58\u8207\u5730\u5740\u914d\u5c0d\u4e4b\u7814\u7a76 13 14 \u6797\u80b2\u6698\u8207\u5f35\u5609\u60e0 POI\u64f7\u53d6:\u5546\u5bb6\u540d\u7a31\u8fa8\u8b58\u8207\u5730\u5740\u914d\u5c0d\u4e4b\u7814\u7a76 15 16 \u6797\u80b2\u6698\u8207\u5f35\u5609\u60e0 \u70ba\u5efa\u69cb\u5546\u5bb6\u5730\u7406\u8cc7\u6599\u5eab\uff0cChuang \u7b49\u4eba(POI\u64f7\u53d6:\u5546\u5bb6\u540d\u7a31\u8fa8\u8b58\u8207\u5730\u5740\u914d\u5c0d\u4e4b\u7814\u7a76 \u6240\u5f97\u8cc7\u8a0a\u6709\u9650\uff0c\u5c0d\u65bc\u55ae\u7b46\u5730\u5740\u7db2\u9801\u7684\u76f8\u95dc\u8cc7\u8a0a\u64f7\u53d6\u4ecd\u5c1a\u7121\u7814\u7a76\u3002 \u672c\u7814\u7a76\u5f9e\u5730\u5740\u64f7\u53d6\u7684\u89d2\u5ea6\u51fa\u767c\u505a\u70ba\u5546\u5bb6\u8fa8\u8b58\u7684\u6a19\u8a18\uff0c\u5229\u7528\u5df2\u6293\u53d6\u5927\u91cf\u5305\u542b\u5730\u5740\u7684\u7db2 \u9801\uff0c\u5148\u627e\u51fa\u7db2\u9801\u4e2d\u7684\u5730\u5740\uff0c\u518d\u85c9\u7531\u5730\u5740\u627e\u51fa\u5c0d\u61c9\u7684\u5546\u5bb6\u540d\u7a31\u9032\u884c\u914d\u5c0d\u3002\u63db\u8a00\u4e4b\uff0c\u7d66\u5b9a\u4e00 \u500b\u5df2\u77e5\u5730\u5740\uff0c\u6211\u5011\u5e0c\u671b\u80fd\u900f\u904e\u7db2\u8def\u8cc7\u6599\u64f7\u53d6\u51fa\u8a72\u5730\u9ede\u7684\u540d\u7a31(\u5982\uff1a\u5546\u5bb6\u540d\u7a31\u3001\u653f\u5e9c\u55ae\u4f4d\u2026 \u7b49)\u3002\u8209\u4f8b\u800c\u8a00\uff1a\u7576\u6211\u5011\u5df2\u6709\u5730\u5740\u300c\u65b0\u5317\u5e02\u677f\u6a4b\u5340\u4e2d\u5c71\u8def\u4e8c\u6bb5 88 \u865f 3F\u300d\uff0c\u6211\u5011\u5e0c\u671b\u80fd \u77e5\u9053\u9019\u500b\u5730\u5740\u5c0d\u61c9\u7684\u540d\u7a31\u300c\u5927\u921e\u91ab\u5b78\u7f8e\u5bb9\u8a3a\u6240\u300d\uff0c\u5982\u6b64\u5373\u53ef\u9032\u4e00\u6b65\u85c9\u7531\u5730\u5740\u3001\u540d\u7a31\u4e26\u5229 \u7528\u641c\u5c0b\u5f15\u64ce\u6536\u96c6\u66f4\u591a\u984d\u5916\u5546\u5bb6\u8cc7\u8a0a\u3002\u9019\u4e9b\u984d\u5916\u8cc7\u8a0a\u4e0d\u50c5\u53ef\u4ee5\u6709\u6548\u63d0\u6607\u5730\u5716\u4e0a\u641c\u5c0b\u4e5f\u5c31\u662f \u5730\u7406\u6aa2\u7d22\u7cfb\u7d71 GIS \u7684\u53ec\u56de\u7387\uff0c\u4e5f\u53ef\u63d0\u6607\u5546\u5bb6\u5206\u985e\u7684\u6e96\u78ba\u7387(\u9673\u5b9c\u52e4 \u7b49\uff0c2013)\u3002 \u5728\u8fa8\u8b58\u5546\u5bb6\u540d\u7a31\u7684\u90e8\u5206\uff0c\u672c\u7bc7\u8ad6\u6587\u4f7f\u7528\u4e86\u689d\u4ef6\u96a8\u6a5f\u57df (Conditional Random Field)\u7576 \u4f5c\u5b78\u7fd2\u6f14\u7b97\u6cd5\u3002\u76ee\u524d\u6709\u8a31\u591a\u95dc\u65bc\u4e2d\u6587\u7d44\u7e54\u540d\u7a31\u8fa8\u8a8d\u7684\u7814\u7a76 (Zhang et al., 2007) (Yao, 2o11) (Ling et al., 2012) (Wu et al., 2008)\uff0c\u53ef\u4ee5\u5f9e\u65b0\u805e\u6216\u4e00\u4e9b\u8f03\u6b63\u5f0f\u7684\u6587\u7ae0\u4e2d\u8403\u53d6\u51fa\u7d44\u7e54\u540d\u7a31\uff0c \u4f46\u662f\u4e26\u6c92\u6709\u5617\u8a66\u4ee5\u4e00\u500b CRF-Model \u76f4\u63a5\u5c0d\u5404\u7a2e\u7db2\u7ad9\u4e2d\u7684\u6574\u500b\u7db2\u9801\u5167\u5bb9\u9032\u884c\u4e2d\u6587\u7d44\u7e54\u540d\u7a31 \u8fa8\u8a8d\u3002\u9019\u5169\u8005\u4e4b\u9593\u4e0d\u540c\u8655\u5728\u65bc\u65b0\u805e\u985e\u6587\u7ae0\u5c6c\u65bc\u8f03\u6b63\u5f0f\u7684\u6587\u7ae0\u9ad4\u88c1\uff0c\u56e0\u6b64\u5bb9\u6613\u51fa\u73fe\u884c\u653f\u6a5f \u95dc\u8207\u6b63\u5f0f\u7684\u7d44\u7e54\u540d\u7a31\uff0c\u4f8b\u5982\uff1a\u884c\u653f\u9662\u548c\u7dad\u5fb7\u98df\u54c1\u6709\u9650\u516c\u53f8\uff0c\u4f46\u662f\u6574\u500b\u7db2\u8def\u4e0a\u5546\u5bb6\u7d44\u7e54\u540d \u7a31\u7684\u547d\u540d\u65b9\u5f0f\u50be\u5411\u5247\u4e0d\u540c\uff0c\u4f8b\u5982\uff1a\u543c\u725b\u6392\u3001\u52aa\u54c7\u514b\u5496\u5561\u3001\u963f\u5b24\u7956\u50b3\u83dc\u5305\u8089\u7cbd\u4ed9\u8349\u2026\u7b49\uff0c \u90fd\u662f\u5546\u5bb6\u7d44\u7e54\u540d\u7a31\u3002\u53e6\u5916\uff0c\u4e00\u500b\u5b8c\u6574\u7684\u7db2\u9801\u5167\u5bb9\u6709\u7d50\u69cb\u8207\u975e\u7d50\u69cb\u5316\u7684\u8cc7\u8a0a\u4ea4\u932f\u5448\u73fe\uff0c\u96d6 \u7136\u7d50\u69cb\u5316\u8cc7\u8a0a\u6703\u9020\u6210\u81ea\u7136\u8a9e\u8a00\u6587\u5b57\u5167\u5bb9\u7684\u7834\u788e\uff0c\u4f46\u9019\u4e9b\u7d50\u69cb\u4e5f\u96b1\u542b\u6709\u53ef\u5229\u7528\u7684\u8cc7\u8a0a\u3002 \u70ba\u4e86\u4f7f\u5546\u5bb6\u8fa8\u8b58\u80fd\u4ee5\u6700\u5c11\u4eba\u529b\u9032\u884c\u81ea\u52d5\u5316\u5b78\u7fd2\uff0c\u672c\u7814\u7a76\u4f7f\u7528\u81ea\u52d5\u6a19\u8a18\u65b9\u5f0f\u5efa\u7acb\u8a13\u7df4 \u8cc7\u6599\uff0c\u6211\u5011\u5148\u91dd\u5c0d\u90e8\u4efd\u7684\u9ec3\u9801\u7db2\u7ad9(\u5982 104 \u6c42\u8077\u7db2\u3001\u611b\u8a55\u7db2\u3001\u5de5\u5546\u540d\u9304\u7db2\u7ad9)\u64b0\u5beb Parser \u53d6\u5f97\u5927\u91cf\u5546\u5bb6\u540d\u7a31\u8207\u5730\u5740\u7684\u7d44\u5408\uff0c\u4e26\u4ee5\u9019\u4e9b\u5df2\u7d93\u53d6\u5f97\u7684\u5546\u5bb6\u540d\u7a31\u5c0d\u7db2\u9801\u8a9e\u6599\u9032\u884c\u81ea\u52d5\u6a19 \u8a18\uff0c\u518d\u5229\u7528\u81ea\u52d5\u6a19\u8a18\u5f8c\u7684\u8a9e\u6599\u8a13\u7df4 CRF \u5e8f\u5217\u6a19\u8a18\u6a21\u578b\u3002\u7136\u800c\u4e00\u500b\u5730\u5740\u53ef\u80fd\u51fa\u73fe\u5728\u591a\u500b\u7db2 \u9801\u4e4b\u4e2d\uff0c\u50c5\u53ea\u4ef0\u8cf4\u5176\u4e2d\u4e00\u500b\u7db2\u9801\u4e5f\u6709\u5931\u4e4b\u504f\u9817\u4e4b\u616e\uff0c\u56e0\u6b64\u6211\u5011\u4e5f\u6536\u96c6\u4e86 Google Snippets \u7576\u4f5c\u8a13\u7df4\u8cc7\u6599\u9032\u884c\u6bd4\u8f03\u3002\u672c\u7bc7\u8ad6\u6587\u7684\u7b2c\u4e8c\u500b\u4e3b\u984c\u5247\u662f\u5546\u5bb6\u5730\u5740\u7684\u914d\u5c0d\uff0c\u7531\u65bc\u4e00\u500b\u7db2\u9801\u53ef \u80fd\u5305\u542b\u591a\u500b\u5546\u5bb6\u540d\u7a31\uff0c\u6211\u5011\u5c0d\u7db2\u9801\u4ee5\u7c21\u55ae\u7684\u898f\u5247\u9032\u884c\u5206\u985e\u5f8c\uff0c\u4f7f\u7528\u4e86\u555f\u767c\u5f0f(heuristic) \u7684\u914d\u5c0d\u898f\u5247\uff0c\u5229\u7528\u5404\u985e\u578b\u7684\u7db2\u7ad9\u6240\u5177\u6709\u7684\u8868\u9054\u7279\u6027\uff0c\u5c0d\u5730\u5740\u8207\u5546\u5bb6\u540d\u7a31\u9032\u884c\u914d\u5c0d\u3002 \u672c\u7814\u7a76\u627f\u7e8c (Su, 2012) (Chuang et al., 2014)\u4e4b\u7814\u7a76\uff0c\u7d93\u7531\u722c\u53d6\u7db2\u9801\u4e0a\u5305\u542b\u5730\u5740\u7684\u5927 \u4e4b\u5730\u5740\u64f7\u53d6\u6a21\u578b\u64f7\u53d6\u51fa\u4e86\u53ef\u80fd\u542b\u6709\u53f0\u7063\u5730\u5740\u7684\u7db2\u9801\u8207\u5730\u5740\u6e05\u55ae\u3002\u672c\u7bc7\u8ad6\u6587\u4ee5\u5df2\u77e5\u53ef\u80fd\u542b \u6709\u53f0\u7063\u5730\u5740\u7684\u4e2d\u6587\u7db2\u9801\u3001\u6bcf\u7b46\u7db2\u9801\u7684\u5730\u5740\u6e05\u55ae\u3001\u5927\u91cf\u5546\u5bb6\u540d\u7a31\u6e05\u55ae\u4ee5\u53ca\u5df2\u77e5\u7684\u5730\u5740\u8207\u5546 \u5bb6\u540d\u7a31\u914d\u5c0d\u8cc7\u6599\u70ba\u57fa\u790e\uff0c\u63d0\u51fa\u4e86\u4e00\u500b\u5546\u5bb6\u540d\u7a31\u64f7\u53d6\u7cfb\u7d71\uff0c\u65b9\u6cd5\u5206\u70ba\u4e09\u5927\u6b65\u9a5f\uff1a\u5730\u5740\u7db2\u9801 \u7684\u524d\u8655\u7406\u3001\u5546\u5bb6\u540d\u7a31\u547d\u540d\u5be6\u9ad4\u8fa8\u8a8d\u3001\u53ca\u5730\u5740-\u5546\u5bb6\u540d\u7a31\u5339\u914d\u3002\u672c\u7814\u7a76\u5728\u4e09\u500b\u6a21\u578b\u806f\u5408\u6a19 \u8a18\u5546\u5bb6\u540d\u7a31\u7684\u65b9\u5f0f\u4e0b\uff0c\u5730\u5740\u8207\u5546\u5bb6\u540d\u7a31\u7684\u5e73\u5747\u914d\u5c0d\u6b63\u78ba\u7387\u70ba 0.57\u3002 \u672c\u8ad6\u6587\u5171\u6709\u4e94\u500b\u7ae0\u7bc0\uff0c\u7b2c\u4e00\u7bc0\u662f\u7dd2\u8ad6\uff0c\u8aaa\u660e\u7814\u7a76\u52d5\u6a5f\u8207\u80cc\u666f\uff1b\u7b2c\u4e8c\u7bc0\u662f\u76f8\u95dc\u7814\u7a76\uff0c \u4ecb\u7d39\u4e2d\u6587\u7d44\u7e54\u540d\u7a31\u8fa8\u8a8d\u548c\u5730\u5740\u76f8\u95dc\u8cc7\u8a0a\u64f7\u53d6\u7684\u76f8\u95dc\u7814\u7a76\u3002\u7b2c\u4e09\u7bc0\u662f\u65b9\u6cd5\uff0c\u6703\u8a73\u7d30\u4ecb\u7d39\u5982 \u4f55\u5c0d\u5730\u5740-\u7db2\u9801\u5206\u985e\u3001\u4e2d\u6587\u7d44\u7e54\u540d\u7a31\u8fa8\u8a8d\u4ee5\u53ca\u5730\u5740\u8207\u5546\u5bb6\u7d44\u7e54\u540d\u7a31\u7684\u914d\u5c0d\u3002\u7b2c\u56db\u7bc0\u662f\u6211\u5011 \u91dd\u5c0d\u73fe\u6709\u7684\u7db2\u9801\u4e2d\uff0c\u4f9d\u64da\u6211\u5011\u7684\u5206\u985e\uff0c\u6bcf\u985e\u96a8\u6a5f\u62bd\u53d6\u7db2\u9801\u9032\u884c\u7684\u5be6\u9a57\u8207\u7d50\u679c\u5206\u6790\u3002\u6700\u5f8c \u662f\u6211\u5011\u7684\u7d50\u8ad6\u4ee5\u53ca\u672a\u4f86\u7684\u5c55\u671b\u3002 2. \u76f8\u95dc\u7814\u7a76 \u76f8\u95dc\u7814\u7a76 \u76f8\u95dc\u7814\u7a76 \u76f8\u95dc\u7814\u7a76 \u64f7\u53d6\u5730\u5740\u76f8\u95dc\u8cc7\u8a0a\u727d\u6d89\u5230\u4e09\u500b\u9818\u57df\uff0c\u8cc7\u8a0a\u64f7\u53d6(Information Extraction)\u3001\u81ea\u7136\u8a9e\u8a00\u8655\u7406 (Natural Language Processing)\u8207\u8cc7\u8a0a\u6aa2\u7d22(Information Retrieval)\u3002\u9019\u4e09\u8005\u5f7c\u6b64\u9593\u4e92\u76f8 \u4ea4\u932f\uff0c\u5f88\u96e3\u7cbe\u78ba\u5207\u5272\u51fa\u5404\u81ea\u6240\u5c6c\u7684\u7bc4\u7587\u3002\u5927\u81f4\u4e0a\u4f86\u8aaa\uff0c\u8cc7\u8a0a\u64f7\u53d6\u4e3b\u8981\u662f\u5f9e\u5404\u7a2e\u7d50\u69cb\u5316\u8cc7 \u6599\u8207\u975e\u7d50\u69cb\u5316\u6587\u5b57\u8403\u53d6\u51fa\u7279\u5b9a\u8cc7\u8a0a\u7684\u65b9\u6cd5\uff0c\u800c\u81ea\u7136\u8a9e\u8a00\u8655\u7406\u5247\u5c6c\u65bc\u4eba\u5de5\u667a\u6167\u9818\u57df\u7684\u4e00\u500b \u5206\u652f\uff0c\u76ee\u7684\u5728\u65bc\u81ea\u52d5\u5316\u7684\u7406\u89e3\u4e26\u8655\u7406\u4eba\u985e\u6240\u4f7f\u7528\u7684\u8a9e\u8a00\u3002\u8cc7\u8a0a\u6aa2\u7d22\u5247\u662f\u5f9e\u5927\u91cf\u8cc7\u6599\u4e2d\u4ee5 \u6a5f\u7387\u7d71\u8a08\u6a21\u578b\u5c0d\u8cc7\u6599\u9032\u884c\u6392\u5e8f(rank)\u3001\u5efa\u7acb\u7d22\u5f15\uff0c\u5feb\u901f\u627e\u51fa\u4f7f\u7528\u8005\u76ee\u6a19\u6587\u4ef6\u7684\u65b9\u6cd5\u3002 \u672c\u7814\u7a76\u76f8\u95dc\u7684\u4e3b\u8981\u6280\u8853\uff0c\u5206\u5225\u70ba\u5982\u4f55\u6709\u6548\u722c\u53d6\u5305\u542b\u5730\u5740\u4e4b\u76ee\u6a19\u7db2\u9801\u3001\u5730\u5740\u76f8\u95dc\u8cc7\u8a0a \u64f7\u53d6\u8207\u547d\u540d\u5be6\u9ad4\u8fa8\u8a8d\u3002\u5730\u5740\u76f8\u95dc\u8cc7\u8a0a\u64f7\u53d6\u662f\u5728\u5f97\u77e5\u5730\u5740\u8cc7\u8a0a\u5f8c\uff0c\u5f9e\u542b\u6709\u5730\u5740\u7684\u7db2\u9801\u4e2d\u64f7 \u53d6\u51fa\u8207\u8a72\u5730\u5740\u76f8\u95dc\u7684\u8cc7\u8a0a\uff0c\u5982\uff1a\u96fb\u8a71\u3001\u7db2\u5740\u3001\u96fb\u5b50\u90f5\u4ef6\u3001\u8a55\u8ad6\u2026\u7b49\u8cc7\u8a0a\u3002\u547d\u540d\u5be6\u9ad4\u8fa8\u8a8d \u5247\u662f\u70ba\u4e86\u8fa8\u8a8d\u6587\u53e5\u6240\u63d0\u5230\u7684\u7279\u5b9a\u7a2e\u985e\u6982\u5ff5\uff0c\u5982\uff1a\u4eba\u540d\u3001\u5730\u540d\u3001\u7d44\u7e54\u540d\u7a31\u3002\u672c\u7ae0\u4e2d\u5c07\u4f9d\u5e8f \u4ecb\u7d39\u9019\u4e9b\u6280\u8853\u7684\u76f8\u95dc\u7814\u7a76\u3002 2.1 \u5305\u542b\u5730\u5740\u7684\u7db2\u9801\u6293\u53d6 \u5305\u542b\u5730\u5740\u7684\u7db2\u9801\u6293\u53d6 \u5305\u542b\u5730\u5740\u7684\u7db2\u9801\u6293\u53d6 \u5305\u542b\u5730\u5740\u7684\u7db2\u9801\u6293\u53d6\u8207\u5730\u7406\u8cc7\u8a0a\u6aa2\u7d22 \u8207\u5730\u7406\u8cc7\u8a0a\u6aa2\u7d22 \u8207\u5730\u7406\u8cc7\u8a0a\u6aa2\u7d22 \u8207\u5730\u7406\u8cc7\u8a0a\u6aa2\u7d22 \u9019\u88e1\u6240\u8b02\u7684\u5730\u7406\u8cc7\u8a0a\u6aa2\u7d22\uff0c\u662f\u5f9e\u7db2\u8def\u4e0a\u722c\u53d6\u5305\u542b\u5730\u9ede\u6216\u5730\u5740\u7684\u7db2\u9801\uff0c\u8403\u53d6\u5730\u7406\u8cc7\u8a0a\u4e26\u5229 \u7528\u6b64\u8cc7\u8a0a\u6392\u5e8f\u8207\u5efa\u7acb\u7d22\u5f15\uff0c\u63d0\u4f9b\u5feb\u901f\u6aa2\u7d22\u7684\u670d\u52d9\u3002\u76ee\u524d\u7684\u641c\u5c0b\u5f15\u64ce\uff0c\u50cf\u662f Google \u548c Yahoo \u4e5f\u5206\u5225\u5f9e 2005 \u8207 2002 \u5e74\u958b\u59cb\u63d0\u4f9b\u96fb\u5b50\u5730\u5716\u7684\u670d\u52d9\u3002\u800c\u9019\u4e9b\u670d\u52d9\u9700\u8981\u85c9\u7531\u4f7f\u7528\u8005\u7684\u6a19\u8a18 \u7b49\u7fa4\u773e\u5916\u5305\u7684\u65b9\u5f0f\u5efa\u7acb POI \u8cc7\u8a0a\u3002(Dirk &amp; Susanne , 2007)\u7b49\u4eba\u63d0\u51fa\u4e86\u4e00\u500b\u4ee5\u4f4d\u7f6e\u8cc7\u8a0a\u70ba \u57fa\u790e\u7684\u641c\u5c0b\u5f15\u64ce\uff0c\u53ef\u4ee5\u81ea\u52d5\u5f9e\u7db2\u8def\u8cc7\u6e90\u4e2d\u53d6\u5f97\u8207\u7a7a\u9593\u76f8\u95dc\u7684\u6587\u53e5\uff0c\u800c\u5728\u4ed6\u5011\u6700\u8fd1\u7684\u7814\u7a76 \u4e2d (Ahlers, 2013a; 2013b)\uff0c\u5247\u5c08\u6ce8\u5728\u5982\u4f55\u5f9e\u6df1\u5ea6\u7db2\u9801\u4f8b\u5982\u9ec3\u9801\u8207 Wikipedia \u64f7\u53d6\u51fa\u4f4d\u7f6e\u547d \u540d\u5be6\u9ad4\u3002\u7531\u65bc\u5730\u5740\u662f POI \u7684\u660e\u78ba\u6307\u6a19\uff0c\u56e0\u6b64 Chuang \u7b49\u4eba(Chuang et al., 2014)\u63d0\u51fa\u4ee5\u5ee3\u5ea6 \u512a\u5148\u641c\u5c0b\u3001\u9ec3\u9801\u722c\u87f2\u8207\u5730\u5740\u6a23\u7248\u67e5\u8a62\u4e09\u7a2e\u7b56\u7565\u722c\u53d6\u542b\u6709\u5730\u5740\u7684\u7db2\u9801\u3002\u5be6\u9a57\u7d50\u679c\u986f\u793a\u96d6\u7136 \u722c\u53d6\u9ec3\u9801\u7db2\u9801\u53ef\u4ee5\u8f03\u5feb\u53d6\u5f97\u5927\u91cf\u5730\u5740\uff0c\u7136\u800c\u5730\u5740\u6a23\u7248\u67e5\u8a62\u53ef\u4ee5\u88dc\u8db3\u9ec3\u9801\u6db5\u84cb\u5ea6\u4e0d\u8db3\u4e4b\u8655\uff0c \u4e5f\u662f\u5efa\u7acb\u5546\u5bb6\u67e5\u8a62\u670d\u52d9\u4e0d\u53ef\u6216\u7f3a\u7684\u65b9\u6cd5\u3002 2.2 \u5730\u5740\u8207\u76f8\u95dc\u8cc7\u8a0a\u64f7\u53d6 \u5730\u5740\u8207\u76f8\u95dc\u8cc7\u8a0a\u64f7\u53d6 \u5730\u5740\u8207\u76f8\u95dc\u8cc7\u8a0a\u64f7\u53d6 \u5730\u5740\u8207\u76f8\u95dc\u8cc7\u8a0a\u64f7\u53d6 \u5730\u5740\u64f7\u53d6\u662f\u56e0\u61c9\u5730\u5740\u8cc7\u8a0a\u6aa2\u7d22\u6240\u7522\u751f\u7684\u9700\u6c42\uff0c\u76ee\u7684\u662f\u5f9e\u7db2\u8def\u4e0a\u5927\u91cf\u7684\u7db2\u9801\u4e2d\uff0c\u64f7\u53d6\u53d6\u51fa \u5730\u5740\u8cc7\u8a0a\uff0c\u5728 2009 \u5e74 Li \u7684\u7814\u7a76\u4e2d (Li, 2009)\uff0cLi \u4ee5\u5e8f\u5217\u6a19\u8a18(Sequence Labeling)\u548c CRF \u6a21\u578b\u5c0d\u7f8e\u570b\u5730\u5340\u7684\u82f1\u6587\u5730\u5740\u9032\u884c\u8a13\u7df4\u8207\u6e2c\u8a66\u3002Li \u5229\u7528\u8a72\u5730\u5340\u5730\u5740\u7684\u7279\u6027\u5efa\u7acb\u4e86 14 \u7a2e\u7279\u5fb5\uff0c\u4e26\u4f7f\u7528 BIEO \u6a19\u8a18\u6cd5\uff0c\u5be6\u9a57\u7d50\u679c F-measure \u9054\u5230\u4e86 0.913 \u7684\u6e96\u78ba\u7387\u30022011 \u5e74 Huang \u5ef6\u7e8c\u4e86 Li \u7684\u7814\u7a76 (Chang et al., 2012)\uff0c\u5229\u7528 17 \u7a2e\u53f0\u7063\u5730\u5740\u7279\u5fb5\u548c BIEO \u53ca IO \u5169\u7a2e\u6a19\u8a18 \u6cd5\uff0c\u5176\u4e2d IO \u6a19\u8a18\u6cd5\u56e0\u70ba\u908a\u754c\u5075\u6e2c\u80fd\u529b\u8f03\u5f31\uff0c\u9700\u642d\u914d\u6975\u5927\u5206\u6578\u5b50\u5e8f\u5217(Maximal Scoring Subsequence)\u9032\u884c\u4fee\u6b63\u3002BIEO \u6a19\u8a18\u6cd5\u7684\u5be6\u9a57\u7d50\u679c F-measure \u7d04\u5728 0.96 \u81f3 0.99 \u4e4b\u9593\uff0cIO \u6a19\u8a18\u6cd5\u5247\u5728 0.94 \u81f3 0.96 \u4e4b\u9593\u3002 \u76f8\u95dc\u8cc7\u8a0a\u64f7\u53d6\u662f\u5730\u5740\u64f7\u53d6\u7684\u5ef6\u4f38\u7814\u7a76\uff0c\u76ee\u7684\u662f\u91dd\u5c0d\u5df2\u77e5\u7684\u5730\u5740\u64f7\u53d6\u51fa\u8207\u8a72\u5730\u5740\u6709\u95dc \u7684\u8a0a\u606f\uff0c\u5982\uff1a\u96fb\u8a71\u3001\u7db2\u5740\u3001\u96fb\u5b50\u90f5\u4ef6\u3001\u8a55\u8ad6\u2026\u7b49\u8cc7\u8a0a\u3002\u4e3b\u8981\u7684\u4f5c\u6cd5\u662f\u91dd\u5c0d\u5df2\u7d93\u6210\u529f\u64f7\u53d6 \u51fa\u7684\u5730\u5740\uff0c\u627e\u51fa\u53ef\u80fd\u7684\u4e0a\u4e0b\u908a\u754c\u3001\u5283\u51fa\u8cc7\u6599\u7bc4\u570d\u4f5c\u70ba\u8a72\u5730\u5740\u7684\u76f8\u95dc\u63cf\u8ff0\uff0c\u53ef\u4ee5\u8996\u70ba\u4e00\u7a2e \u6df1\u5ea6\u7db2\u9801\u8cc7\u6599\u64f7\u53d6(Deep Web Data Record Extraction)\u7684\u4e00\u7a2e\u7279\u4f8b\u3002\u5728 Li \u7684\u7814\u7a76\u4e2d\uff0c\u4e3b \u8981\u662f\u628a\u6240\u6709\u5730\u5740\u6240\u5728\u7684\u6587\u5b57\u8449\u7bc0\u9ede(Text Leaf Node)\u7576\u4f5c\u8d77\u9ede\uff0c\u5229\u7528\u9019\u4e9b\u7bc0\u9ede\u8d70\u8a2a\u81f3\u6839 \u7bc0\u9ede\u904e\u7a0b\u4e2d\uff0cHtml Tag \u7684\u8b8a\u5316\u7576\u4f5c\u908a\u754c\u9ede\u3002\u4f46\u662f Li \u7684\u65b9\u6cd5\u5c0d\u65bc\u7db2\u9801\u4e2d\u64c1\u6709\u5169\u7a2e\u4ee5\u4e0a\u7684\u5730 \u5740\u76f8\u95dc\u8cc7\u8a0a\u6392\u7248\u7121\u6cd5\u6709\u6548\u64f7\u53d6\uff0c\u70ba\u4e86\u89e3\u6c7a\u6b64\u554f\u984c\uff0cHuang \u6703\u5148\u91dd\u5c0d\u5404\u5730\u5740\u8def\u5f91\u7684\u76f8\u4f3c\u5ea6 \u4f5c\u51fa\u5206\u985e\uff0c\u518d\u91dd\u5c0d\u5404\u985e\u5be6\u884c Li \u7684\u65b9\u6cd5\u3002\u5728\u6700\u5f8c\u82f1\u6587\u5730\u5740\u76f8\u95dc\u8cc7\u8a0a\u64f7\u53d6\u7684\u5be6\u9a57\u4e2d\uff0cLi \u7684\u76f8 \u95dc\u8cc7\u8a0a\u64f7\u53d6\u7684 F-measure \u9054\u5230\u4e86 0.8689\uff0c\u800c\u52a0\u5165\u4e86 Huang \u7684\u6539\u9032\u5247\u63d0\u6607 0.0233\u3002 2012 \u5e74 Su (Su, 2012)\u767c\u73fe Li \u8207 Huang \u7684\u505a\u6cd5\u904e\u5ea6\u7c21\u5316\u5404\u7b46\u7d00\u9304(Record)\u7684\u7522\u751f\u6a21 \u7248 (Template) \uff0cLi \u8207 Huang \u7684\u505a\u6cd5\u4e2d\uff0c\u53ea\u8981\u6a21\u7248\u4e2d\u6709\u4efb\u4f55\u4e00\u7b46\u9078\u64c7\u6027\u8cc7\u6599 (Optional Data) \uff0c \u5c31\u6703\u767c\u751f\u9023\u9396\u932f\u8aa4\u3002\u70ba\u4e86\u89e3\u6c7a\u6b64\u554f\u984c\uff0cSu \u5c07 2010 \u5e74 Wei Liu \u6240\u63d0\u51fa\u57fa\u65bc\u8996\u89ba (Vision-Based) \u7684\u8cc7\u6599\u7d00\u9304(Data Record)\u64f7\u53d6\u6f14\u7b97\u6cd5\u5957\u7528\u5728\u5730\u5740\u76f8\u95dc\u8cc7\u8a0a\u64f7\u53d6\u7684\u7814\u7a76\u4e2d\uff0c\u4e26\u91cd\u4f5c Li \u7684\u5be6\u9a57\uff0c\u5c07 F-measure \u7531 0.7912 \u63d0\u6607\u81f3 0.9504\u3002 Li\u3001Huang \u548c Su \u7684\u7814\u7a76\u7686\u5c08\u6ce8\u65bc\u8cc7\u8a0a\u64f7\u53d6\u7684\u6548\u679c\u4e0a\uff0c\u4f46\u5176\u524d\u63d0\u662f\u7db2\u9801\u4e2d\u5b58\u5728\u591a\u500b\u5730 \u5740\u5b57\u4e32\u3002\u82e5\u63d0\u53ca\u5730\u5740\u76f8\u95dc\u8cc7\u8a0a\u7684\u7db2\u9801\u5167\u4e0d\u5b58\u5728\u5730\u5740\u5b57\u4e32\uff0c\u5247\u7121\u6cd5\u5f97\u77e5\u7db2\u9801\u5167\u542b\u8207 POI \u76f8 \u95dc\u7684\u8cc7\u8a0a\uff0c\u66f4\u4e0d\u53ef\u80fd\u6709\u5f8c\u7e8c\u8403\u53d6\u8cc7\u8a0a\u7684\u904e\u7a0b\u3002\u56e0\u6b64\uff0c\u672c\u7814\u7a76\u8a66\u5716\u64f7\u53d6\u51fa\u5730\u5740\u7684\u5546\u5bb6\u7d44\u7e54 \u540d\u7a31\uff0c\u4ee5\u5229\u5f8c\u7e8c\u7684\u76f8\u95dc\u8cc7\u8a0a\u8403\u53d6\u8207\u6aa2\u7d22\u3002 2.3 \u4e2d\u6587\u7d44\u7e54\u547d\u540d\u5be6 \u4e2d\u6587\u7d44\u7e54\u547d\u540d\u5be6 \u4e2d\u6587\u7d44\u7e54\u547d\u540d\u5be6 \u4e2d\u6587\u7d44\u7e54\u547d\u540d\u5be6\u9ad4\u8fa8\u8a8d \u9ad4\u8fa8\u8a8d \u9ad4\u8fa8\u8a8d \u9ad4\u8fa8\u8a8d \u547d\u540d\u5be6\u9ad4\u8fa8\u8a8d\u5c6c\u65bc\u8cc7\u8a0a\u8403\u53d6\u8207\u81ea\u7136\u8a9e\u8a00\u7684\u4e00\u500b\u5171\u540c\u5206\u652f\uff0c\u6b64\u7814\u7a76\u8d77\u56e0\u65bc\u4efb\u4f55\u7cfb\u7d71\u7686\u7121\u6cd5 \u7aae\u8209\u51fa\u6240\u6709\u7684\u8a5e\u5f59\u8207\u4ee3\u8868\u7684\u610f\u7fa9\uff0c\u56e0\u70ba\u518d\u5927\u7684\u8a5e\u5eab\u90fd\u6703\u6709\u6c92\u6536\u9304\u7684\u8a5e\u5f59(OOV word\uff0c Out-of-Vocabulary)\uff0c\u4e14\u540c\u6a23\u7684\u8a5e\u5f59\u5728\u4e0d\u540c\u7684\u5167\u5bb9\u4e2d\u5f88\u53ef\u80fd\u4ee3\u8868\u4e0d\u540c\u7684\u610f\u7fa9\u3002\u76ee\u524d\u7684\u4e3b \u8981\u65b9\u6cd5\u662f\u5229\u7528\u5e8f\u5217\u6a19\u8a18\u914d\u5408\u6a5f\u7387\u7d71\u8a08\u6a21\u578b\u8a08\u7b97\u51fa\u6700\u53ef\u80fd\u7684\u6a19\u8a18\u3002 -measure \u9054\u5230 0.9794\u3002 2011 \u5e74 Yao (2011)\u5247\u662f\u5c07\u4e2d\u6587\u7d44\u7e54\u540d\u7a31\u5206\u70ba\u4e09\u6bb5\uff1a\u524d\u7f6e\u8a5e(Prefix words)+\u4e2d\u9593\u8a5e (middle words)+\u8a18\u865f\u8a5e(mark words)(\u4f8b\u5982\uff1a\u4e2d\u570b+\u79fb\u52d5\u901a\u8a0a+\u516c\u53f8)\u4e14\u4e0d\u63a1\u7528\u73fe\u6709 \u7684\u6a21\u578b\uff0c\u4f7f\u7528\u81ea\u884c\u8a2d\u8a08\u7684\u7d71\u8a08\u65b9\u6cd5\uff0c\u8003\u616e\u7d44\u7e54\u540d\u7a31\u7684\u983b\u7387\u3001\u8a5e\u6027\u8207\u9577\u5ea6\uff0c\u914d\u5408\u4ee5\u4e0b\u5047\u8a2d \u9032\u884c\u8a08\u7b97\uff1a\u300c\u8a18\u865f\u8a5e\u80fd\u5b8c\u5168\u6536\u9304\u300d\u3001\u300c\u524d\u7f6e\u8a5e\u8207\u4e2d\u9593\u8a5e\u70ba\u540d\u8a5e\u3001\u5f62\u5bb9\u8a5e\u3001\u5e8f\u6578\u6216\u4f4d\u7f6e\u2026 \u7b49\u300d\u3001\u300c\u8a18\u865f\u8a5e\u5927\u90e8\u5206\u70ba\u540d\u8a5e\u300d\u548c\u300c\u7d44\u7e54\u540d\u7a31\u5c0f\u65bc\u7b49\u65bc 10 \u500b\u5b57\u300d\uff0c\u6700\u5f8c\u7684\u5be6\u9a57\u4f7f\u7528\u4e86\u4eba \u6c11\u7db2\u7684\u8a9e\u6599\u9032\u884c\u8a13\u7df4\uff0c\u4ee5\u4eba\u6c11\u7db2\u3001\u65b0\u83ef\u7db2 \u548c\u5317\u4eac\u90f5\u96fb\u5927\u5b78\u7db2\u7ad9\u9996\u9801\u7684\u65b0\u805e \u7576\u4f5c\u6e2c\u8a66\u8cc7 \u6599\u3002\u5e73\u5747\u6e96\u78ba\u7387\u6700\u9ad8\u9054\u5230 0.959\uff0c\u5e73\u5747\u53ec\u56de\u503c\u5247\u9054\u5230 0.8724\uff0c\u7686\u8d85\u904e\u96b1\u85cf\u99ac\u53ef\u592b\u6a21\u578b (HMM) \u8207\u6700\u5927\u71b5\u6a21\u578b(ME)\u3002 2012 \u5e74 Recognition)\u8fa8\u8b58\u4eba\u6c11\u65e5\u5831\u8207\u65b0\u6d6a\u7db2\u7684\u65b0\u805e\uff0cLing \u9996\u5148\u5c07\u8a9e\u6599\u7d93\u904e\u65b7\u8a5e\u4e26\u5c07\u4e2d\u6587\u7d44\u7e54\u540d \u7a31\u62c6\u89e3\u70ba\u591a\u500b\u4fee\u98fe\u8a5e(Modifiers)+\u6838\u5fc3\u7279\u5fb5\u8a5e(Core Feature Word)\u3002\u5728\u7d71\u8a08\u8a13\u7df4\u8cc7\u6599 \u5f8c\uff0c\u627e\u51fa\u5e38\u7528\u7684\u6838\u5fc3\u7279\u5fb5\u8a5e\uff0c\u5efa\u7acb\u6838\u5fc3\u7279\u5fb5\u8a5e\u5eab\u7576\u4f5c\u7d44\u7e54\u540d\u7a31\u7684\u7d50\u5c3e\uff0c\u4e26\u627e\u51fa 6 \u7a2e\u5de6\u908a \u754c\u7279\u5fb5(left-border features)\u5224\u65b7\u7d44\u7e54\u540d\u7a31\u7684\u8d77\u9ede\u3002\u5728\u53d6\u5f97\u7d44\u7e54\u540d\u7a31\u5019\u9078\u8005\u4e4b\u5f8c\uff0c\u5229\u7528 \u8a72\u7cfb\u7d71\u7684\u5e38\u898b\u932f\u8aa4\u6a21\u5f0f(Debugging Patterns)\u9032\u884c\u4fee\u6b63\u3002\u6700\u5f8c\u7684\u5be6\u9a57\u7d50\u679c\u986f\u793a\uff0cLing \u7684 \u65b9\u6cd5\u7684 F-measure \u6700\u9ad8\u9054\u5230\u4e86 0.8573\u3002 \u7136\u800c\u4e0a\u8ff0\u7814\u7a76\u7686\u8457\u91cd\u65b0\u805e\u8a9e\u6599\u4e4b\u547d\u540d\u5be6\u9ad4\u64f7\u53d6\uff0c\u5c0d\u65bc\u975e\u65b0\u805e\u6587\u4ef6\u7684\u4e00\u822c\u7db2\u9801\u64f7\u53d6\u4e26 \u672a\u8457\u58a8\u3002\u4e8b\u5be6\u4e0a\u7db2\u9801\u7684\u81ea\u7531\u5ea6\u4f7f\u5f97\u547d\u540d\u5be6\u9ad4\u64f7\u53d6\u76f8\u5c0d\u8f03\u70ba\u56f0\u96e3\uff0c\u9019\u4e5f\u662f\u672c\u7bc7\u8ad6\u6587\u7684\u6311\u6230 \u4e4b\u8655\u3002 3. \u5546\u5bb6\u540d\u7a31\u64f7\u53d6\u8207\u5730\u5740\u914d\u5c0d\u7cfb\u7d71 \u5546\u5bb6\u540d\u7a31\u64f7\u53d6\u8207\u5730\u5740\u914d\u5c0d\u7cfb\u7d71 \u5546\u5bb6\u540d\u7a31\u64f7\u53d6\u8207\u5730\u5740\u914d\u5c0d\u7cfb\u7d71 \u5546\u5bb6\u540d\u7a31\u64f7\u53d6\u8207\u5730\u5740\u914d\u5c0d\u7cfb\u7d71 \u5546\u5bb6\u540d\u7a31\u64f7\u53d6\u3002\u6211\u5011\u5f9e\u9019\u4e9b\u7db2\u9801\u4e2d\u904e\u6ffe\u51fa\u542b\u6709\u53f0\u7063\u5730\u5740\u7684\u53ef\u7528\u7db2\u9801\uff0c\u9032\u884c\u5546\u5bb6\u540d\u7a31\u64f7\u53d6\uff0c \u4e4b\u5f8c\u5229\u7528\u7db2\u7ad9\u7684\u7279\u6027\u5982\u6e05\u55ae\u7db2\u9801\u3001\u6df1\u5ea6\u8cc7\u8a0a\u7db2\u9801\u3001\u8a3b\u8173\u7db2\u9801\u3001\u53ca\u81ea\u7531\u6587\u5b57\u7db2\u9801\u7b49\u70ba\u6bcf\u4e00 \u500b\u5730\u5740\u914d\u5c0d\u5546\u5bb6\u540d\u7a31\u3002 3.1 \u5546\u5bb6\u540d\u7a31\u8fa8\u8a8d \u5546\u5bb6\u540d\u7a31\u8fa8\u8a8d \u5546\u5bb6\u540d\u7a31\u8fa8\u8a8d \u5546\u5bb6\u540d\u7a31\u8fa8\u8a8d \u672c\u7814\u7a76\u8a66\u5716\u5c0d\u7db2\u9801\u5167\u5bb9\u64f7\u53d6\u51fa\u6240\u6709\u7684\u5546\u5bb6\u540d\u7a31\uff0c\u9019\u88e1\u6240\u6307\u7684\u5546\u5bb6\u540d\u7a31\u6db5\u84cb\u4e86\u5404\u7a2e\u7bc4\u570d\uff1a \u660e\u78ba\u7684\u8208\u8da3\u9ede(POI\uff0cPoint of Interest)\u3001\u5be6\u969b\u7684\u7d44\u7e54\u540d\u7a31\u548c\u7522\u54c1\u7684\u5ee0\u5546\u540d\u7a31\u3002\u76ee\u524d\u5728\u547d \u540d\u5be6\u9ad4\u8fa8\u8a8d\u7684\u9818\u57df\uff0c\u901a\u5e38\u4f7f\u7528\u5e8f\u5217\u6a19\u8a18\u6cd5(Sequence Labeling)\u900f\u904e\u689d\u4ef6\u96a8\u6a5f\u57df(CRF) \u6a21\u578b\u9032\u884c\u8fa8\u8a8d\uff0c\u7136\u800c\u76e3\u7763\u5f0f\u5b78\u7fd2\u9700\u4ef0\u8cf4\u5927\u91cf\u7684\u8a13\u7df4\u8cc7\u6599\uff0c\u70ba\u6e1b\u5c11\u4eba\u5de5\u6a19\u8a18\u7684\u8ca0\u8377\uff0c\u672c\u6587 \u5229\u7528\u5df2\u77e5\u7684\u5546\u5bb6\u540d\u7a31\u5c0d\u7db2\u9801\u5167\u5bb9\u9032\u884c\u81ea\u52d5\u6a19\u8a18\uff0c\u4e26\u4ee5\u6a19\u8a18\u5f8c\u7684\u7db2\u9801\u6587\u5b57\u7576\u4f5c CRF \u7684\u8a13\u7df4 \u8cc7\u6599\u3002\u7576 CRF \u8a13\u7df4\u5b8c\u7562\u5f8c\uff0c\u5373\u53ef\u5c0d\u7db2\u9801\u5167\u5bb9\u9032\u884c\u5546\u5bb6\u540d\u7a31\u8fa8\u8b58\uff0c\u5efa\u7acb\u5546\u5bb6\u540d\u7a31\u6e05\u55ae\u3002\u4e0b \u9762\u5c07\u5206\u5225\u4ecb\u7d39\u672c\u7814\u7a76\u7684\u81ea\u52d5\u6a19\u8a18\u3001\u4ee5\u53ca\u8a13\u7df4\u8cc7\u6599\u7684\u6e96\u5099\u65b9\u5f0f\u3002 \u85c9\u7531 Web \u4e0a\u7684\u9ec3\u9801\u7db2\u7ad9\u6240\u63d0\u4f9b\u7684\u5546\u5bb6\u8cc7\u8a0a\uff0c\u6211\u5011\u53ef\u4ee5\u53d6\u5f97\u300c\u5730\u5740-\u5546\u5bb6\u540d\u7a31\u5c0d\u300d\u6e05 \u55ae\uff0c\u5c0d\u8a13\u7df4\u7db2\u9801\u9032\u884c\u81ea\u52d5\u6a19\u8a18\u3002\u7136\u800c\u7531\u65bc\u7db2\u9801\u7e3d\u6578\u9054 39.6 \u842c\u7b46\uff0c\u800c\u4e0d\u91cd\u8907\u7684\u5546\u5bb6\u540d\u7a31\u7e3d \u6578\u9ad8\u9054 68.8 \u842c\uff0c\u57fa\u65bc\u57f7\u884c\u6642\u9593\u7684\u8003\u91cf\uff0c\u7121\u6cd5\u5c0d\u6240\u6709\u7684\u7db2\u9801\u7684\u6bcf\u500b\u53e5\u5b50\u90fd\u6aa2\u67e5\u662f\u5426\u5b58\u5728\u5df2 \u77e5\u5546\u5bb6\u540d\u7a31\u3002\u56e0\u6b64\uff0c\u6211\u5011\u4ee5\u6bcf\u7b46\u7db2\u9801\u5df2\u77e5\u7684\u5730\u5740\u6e05\u55ae\u4f86\u52a0\u5feb\u6a19\u8a18\u901f\u5ea6\uff1a\u4e5f\u5c31\u662f\u8aaa\uff0c\u7cfb\u7d71 \u53ea\u6703\u4f9d\u64da\u7db2\u9801\u6240\u64c1\u6709\u7684\u5730\u5740\u67e5\u8a62\u5c0d\u61c9\u7684\u5546\u5bb6\u540d\u7a31\uff0c\u4e26\u5c0d\u7db2\u9801\u5167\u5bb9\u6383\u63cf\u9019\u4e9b\u5c0d\u61c9\u7684\u5546\u5bb6\u540d \u7a31\u662f\u5426\u5b58\u5728\uff0c\u82e5\u5b58\u5728\u5c31\u6703\u4ee5\u7279\u6b8a\u7684\u6a19\u7c64(Tag)\u4f86\u6a19\u8a3b\u9019\u4e9b\u5546\u5bb6\u540d\u7a31\u3002\u5716 1 \u5de6\u5716\u5373\u662f\u81ea\u52d5 \u6a19\u8a18\u81ea\u52d5\u7522\u751f\u8a13\u7df4\u8cc7\u6599\u7684\u6d41\u7a0b\u5716\u3002 \u5716 \u5716 \u5716 \u5716 1. \u500b\u5225\u5b8c\u6574\u7db2\u9801\u7684\u81ea\u52d5\u6a19\u8a18 \u500b\u5225\u5b8c\u6574\u7db2\u9801\u7684\u81ea\u52d5\u6a19\u8a18 \u500b\u5225\u5b8c\u6574\u7db2\u9801\u7684\u81ea\u52d5\u6a19\u8a18 \u500b\u5225\u5b8c\u6574\u7db2\u9801\u7684\u81ea\u52d5\u6a19\u8a18 \u672c\u7814\u7a76\u53e6\u5916\u4ee5\u5546\u5bb6\u540d\u7a31\u7576\u4f5c\u95dc\u9375\u5b57\u6536\u96c6 \u4ee5\u6240\u6709\u7684\u5df2\u77e5\u5546\u5bb6\u540d\u7a31\u5c0d\u9019\u4e9b Snippet \u96dc\u5ea6\u8207\u6a19\u8a18\u4e0d\u5b8c\u6574\u7684\u554f\u984c\u3002\u5982\u5716\u4e00\u53f3\u5716\u6240\u793a \u4e3b\u8981\u5dee\u7570\u9ede\u5728\u81ea\u52d5\u6a19\u8a18\u4e0d\u50c5\u53ea\u7528\u55ae\u4e00\u7684\u5546\u5bb6\u540d\u7a31\u4f86\u5354\u52a9\u6a19\u8a18 \u662f\u63a1\u7528\u6240\u7528\u5546\u5bb6\u540d\u7a31\u4f86\u9032\u884c\u6a19\u8a18 \u6e90\u7684\u8655\u7406\u65b9\u5f0f\u76f8\u540c\u3002\u8a13\u7df4\u8cc7\u6599\u8655\u7406\u6d41\u7a0b\u5982\u4e0b\u6240\u8ff0 \u2022 \u2022 \u2022 \u2022 \u524d\u8655\u7406 \u524d\u8655\u7406 \u524d\u8655\u7406 \u524d\u8655\u7406 \u91dd\u5c0d\u6bcf\u4e00\u500b\u539f\u59cb\u7db2\u9801\u5f8c\uff0c\u672c\u7cfb\u7d71\u9996\u5148\u4f7f\u7528 \u5bb9\u9023\u540c\u6a19\u984c\u64f7\u53d6\u6210\u6587\u5b57\u5167\u5bb9\u5f8c\u624d\u9032\u884c\u5f8c\u7e8c\u6b65\u9a5f \u589e\u5f37\uff0c\u7cfb\u7d71\u6703\u5148\u5c07\u6240\u6709\u5168\u5f62\u7b26\u865f\u8f49\u63db\u6210\u534a\u5f62\u7b26\u865f \u300c(\u300d\uff0c\u56e0\u70ba\u6b64\u7a2e\u62ec\u865f\u901a\u5e38\u542b\u6709\u88dc\u5145\u8aaa\u660e\u7684\u610f\u7fa9 \u7d71\u4e00\u8f49\u6210\u300c[\u300d\uff0c\u56e0\u70ba\u6b64\u7a2e\u62ec\u865f\u901a\u5e38\u5177\u6709\u5f37\u8abf\u7684\u610f\u601d \u6642\u9593\u2026\u7b49\u4ee5\u6b63\u898f\u8868\u793a\u6cd5\u53d6\u4ee3\u6210\u7279\u6b8a\u7684\u5e8f\u5217\u55ae\u5143 \u77ed\u5e8f\u5217\u7684\u9577\u5ea6\uff0c\u63d0\u6607\u8fa8\u8b58\u6548\u679c\u3002 \u2022 \u2022 \u2022 \u2022 \u6a23\u672c\u5e8f\u5217 \u6a23\u672c\u5e8f\u5217 \u6a23\u672c\u5e8f\u5217 \u6a23\u672c\u5e8f\u5217 \u5b8c\u6574\u7684\u7db2\u9801\u5167\u5bb9\u8207\u4e00\u822c\u7684\u6587\u7ae0\u76f8\u6bd4 \u9054\u65b9\u5f0f\u5c07\u6587\u5b57\u5167\u5bb9\u50b3\u9054\u7d66\u4f7f\u7528\u8005 \u6027\u2026\u7b49\u8cc7\u8a0a\u4ee5\u5217\u8868\u6216\u4f9d\u5e8f\u5217\u51fa\u7b49\u65b9\u5f0f\u5448\u73fe Testing Examples)\uff0c\u9032\u884c\u8a13\u7df4\u8207\u6e2c\u8a66 \u5f8c\uff0c\u79fb\u9664\u7a7a\u767d\u985e\u5b57\u5143\u3001\u4ee5\u9023\u7e8c\u4e09\u500b\u63db\u884c\u7b26\u865f\u7576\u4f5c\u5206\u9694\u7b26\u865f \u591a\u5340\u584a (Block) \uff0c\u4ee5\u542b\u6709\u5546\u5bb6\u540d\u7a31\u7684\u5340\u584a\u52a0\u4e0a\u524d\u5f8c\u5340\u584a \u9019\u6a23\u7684\u597d\u8655\u662f\u76e1\u53ef\u80fd\u8b93\u8a13\u7df4\u6a23\u672c\u6db5\u84cb\u5546\u5bb6\u540d\u7a31 \u5728\u6e2c\u8a66\u6642\uff0c\u4e5f\u63a1\u7528\u4e09\u884c\u6587\u5b57\u70ba\u4e00\u500b\u55ae\u4f4d\u7576\u4f5c\u6a23\u672c\u55ae\u5143\u9032\u884c\u6e2c\u8a66 \u500b\u5225\u5b8c\u6574\u7db2\u9801\u7684\u81ea\u52d5\u6a19\u8a18 \u500b\u5225\u5b8c\u6574\u7db2\u9801\u7684\u81ea\u52d5\u6a19\u8a18 \u500b\u5225\u5b8c\u6574\u7db2\u9801\u7684\u81ea\u52d5\u6a19\u8a18 \u500b\u5225\u5b8c\u6574\u7db2\u9801\u7684\u81ea\u52d5\u6a19\u8a18\u6d41\u7a0b \u6d41\u7a0b \u6d41\u7a0b \u6d41\u7a0b(\u5de6 \u5de6 \u5de6 \u5de6)\u8207 \u8207 \u8207 \u8207 Snippets \u81ea\u52d5\u6a19\u8a18\u6d41\u7a0b \u81ea\u52d5\u6a19\u8a18\u6d41\u7a0b \u81ea\u52d5\u6a19\u8a18\u6d41\u7a0b \u81ea\u52d5\u6a19\u8a18\u6d41\u7a0b(\u53f3 \u53f3 \u53f3 \u53f3 \u672c\u7814\u7a76\u53e6\u5916\u4ee5\u5546\u5bb6\u540d\u7a31\u7576\u4f5c\u95dc\u9375\u5b57\u6536\u96c6 Google \u641c\u5c0b\u5f15\u64ce\u63d0\u4f9b\u7684 20 \u7b46 Snippet Snippets \u4e2d\u7684\u53e5\u5b50\u9032\u884c\u6a19\u8a18\uff0c\u8a66\u5716\u964d\u4f4e\u500b\u5225\u7db2\u9801\u8cc7\u6599\u7684\u8907 \u5982\u5716\u4e00\u53f3\u5716\u6240\u793a\uff0c\u4ee5 Google Snippets \u70ba\u8cc7\u6599\u4f86\u6e90\u7684\u8655\u7406\u6d41\u7a0b \u4e0d\u50c5\u53ea\u7528\u55ae\u4e00\u7684\u5546\u5bb6\u540d\u7a31\u4f86\u5354\u52a9\u6a19\u8a18(\u7a31\u4e4b\u70ba UniLabeling \u662f\u63a1\u7528\u6240\u7528\u5546\u5bb6\u540d\u7a31\u4f86\u9032\u884c\u6a19\u8a18(\u7a31\u4e4b\u70ba FullLabeling)\uff0c\u5176\u9918\u7686\u8207\u4ee5\u6574\u500b\u7db2\u9801\u70ba\u8cc7\u6599\u4f86 \u8a13\u7df4\u8cc7\u6599\u8655\u7406\u6d41\u7a0b\u5982\u4e0b\u6240\u8ff0\uff1a \u672c\u7cfb\u7d71\u9996\u5148\u4f7f\u7528 Apache Tika\u2122 (Apache License, 2004)\u5c07\u7db2\u9801\u5167 \u5bb9\u9023\u540c\u6a19\u984c\u64f7\u53d6\u6210\u6587\u5b57\u5167\u5bb9\u5f8c\u624d\u9032\u884c\u5f8c\u7e8c\u6b65\u9a5f\u3002\u70ba\u4e86\u4f7f\u5e8f\u5217\u55ae\u5143(Tokens)\u7279\u5fb5\u7684\u5f37\u5ea6 \u7cfb\u7d71\u6703\u5148\u5c07\u6240\u6709\u5168\u5f62\u7b26\u865f\u8f49\u63db\u6210\u534a\u5f62\u7b26\u865f\uff0c\u5713\u5f27\u578b\u7684\u62ec\u865f\u300c(\u3001(\u3001\ufe59\u300d\u7d71\u4e00\u8f49\u6210 \u56e0\u70ba\u6b64\u7a2e\u62ec\u865f\u901a\u5e38\u542b\u6709\u88dc\u5145\u8aaa\u660e\u7684\u610f\u7fa9\u3002\u975e\u5713\u5f27\u578b\u7684\u62ec\u865f\u300c[\u3001{\u3001\u3014\u3001\uff5b\u3001\u3008\u3001 \u56e0\u70ba\u6b64\u7a2e\u62ec\u865f\u901a\u5e38\u5177\u6709\u5f37\u8abf\u7684\u610f\u601d\u3002\u7b2c\u4e8c\u6b65\u662f\u5c07\u63db\u884c\u7b26\u865f\u3001\u5730\u5740\u96fb\u8a71 \u4ee5\u6b63\u898f\u8868\u793a\u6cd5\u53d6\u4ee3\u6210\u7279\u6b8a\u7684\u5e8f\u5217\u55ae\u5143\uff0c\u9019\u4e9b\u53d6\u4ee3\u52d5\u4f5c\u80fd\u6709\u6548\u52a0\u5f37\u908a\u754c\u7279\u5fb5 \u3002 \u5b8c\u6574\u7684\u7db2\u9801\u5167\u5bb9\u8207\u4e00\u822c\u7684\u6587\u7ae0\u76f8\u6bd4\uff0c\u4e0d\u540c\u7684\u5730\u65b9\u5728\u65bc\u7db2\u9801\u6703\u5229\u7528\u7d50\u69cb\u5316\u8cc7\u8a0a\u3001\u6392\u7248\u7b49\u8868 \u9054\u65b9\u5f0f\u5c07\u6587\u5b57\u5167\u5bb9\u50b3\u9054\u7d66\u4f7f\u7528\u8005\uff0c\u56e0\u6b64\u5f88\u5c11\u6709\u5b8c\u6574\u7684\u53e5\u5b50\uff0c\u800c\u662f\u76f4\u63a5\u628a\u9805\u76ee\u3001\u540d\u7a31 \u7b49\u8cc7\u8a0a\u4ee5\u5217\u8868\u6216\u4f9d\u5e8f\u5217\u51fa\u7b49\u65b9\u5f0f\u5448\u73fe\u3002\u82e5\u6211\u5011\u63a1\u53d6\u50b3\u7d71\u7684\u53e5\u5b50\u6a23\u672c\u55ae\u5143(Trainin \u9032\u884c\u8a13\u7df4\u8207\u6e2c\u8a66\uff0c\u5f88\u96e3\u6709\u597d\u7684\u6210\u679c\u3002\u56e0\u6b64\u6211\u5011\u5c07\u7db2\u9801\u5167\u5bb9\u8f49\u6210\u6587\u5b57 \u4ee5\u9023\u7e8c\u4e09\u500b\u63db\u884c\u7b26\u865f\u7576\u4f5c\u5206\u9694\u7b26\u865f(Delimiter)\uff0c\u5c07\u6587\u5b57\u5207\u70ba\u8a31 \u4ee5\u542b\u6709\u5546\u5bb6\u540d\u7a31\u7684\u5340\u584a\u52a0\u4e0a\u524d\u5f8c\u5340\u584a\uff0c\u4ee5\u9023\u7e8c\u4e09\u5340\u584a\u70ba\u4e00\u500b\u8a13\u7df4\u6a23\u672c \u9019\u6a23\u7684\u597d\u8655\u662f\u76e1\u53ef\u80fd\u8b93\u8a13\u7df4\u6a23\u672c\u6db5\u84cb\u5546\u5bb6\u540d\u7a31\uff0c\u4e5f\u80fd\u6709\u8f03\u591a\u7684\u975e\u5546\u5bb6\u540d\u7a31\u7bc4\u4f8b\u3002 \u4e5f\u63a1\u7528\u4e09\u884c\u6587\u5b57\u70ba\u4e00\u500b\u55ae\u4f4d\u7576\u4f5c\u6a23\u672c\u55ae\u5143\u9032\u884c\u6e2c\u8a66\u3002 \u53f3 \u53f3 \u53f3 \u53f3) Snippets\uff0c\u4e26 \u8cc7\u6599\u7684\u8907 \u70ba\u8cc7\u6599\u4f86\u6e90\u7684\u8655\u7406\u6d41\u7a0b\uff0c UniLabeling)\uff0c\u800c \u5176\u9918\u7686\u8207\u4ee5\u6574\u500b\u7db2\u9801\u70ba\u8cc7\u6599\u4f86 \u5c07\u7db2\u9801\u5167 \u7279\u5fb5\u7684\u5f37\u5ea6 \u7d71\u4e00\u8f49\u6210 \u3001\u3014\u3001\uff5b\u3001\u3008\u3001\u2026\u300d \u5730\u5740\u96fb\u8a71\u3001 \u9019\u4e9b\u53d6\u4ee3\u52d5\u4f5c\u80fd\u6709\u6548\u52a0\u5f37\u908a\u754c\u7279\u5fb5\uff0c\u7e2e \u6392\u7248\u7b49\u8868 \u540d\u7a31\u3001\u5c6c Training or \u56e0\u6b64\u6211\u5011\u5c07\u7db2\u9801\u5167\u5bb9\u8f49\u6210\u6587\u5b57 \u5c07\u6587\u5b57\u5207\u70ba\u8a31 \u4ee5\u9023\u7e8c\u4e09\u5340\u584a\u70ba\u4e00\u500b\u8a13\u7df4\u6a23\u672c\uff0c \u3002\u540c\u6a23\u5730 \u2022 \u2022 \u2022 \u2022 \u5e8f\u5217\u55ae\u5143\u8207\u6a19\u8a18 \u5e8f\u5217\u55ae\u5143\u8207\u6a19\u8a18 \u5e8f\u5217\u55ae\u5143\u8207\u6a19\u8a18 \u5e8f\u5217\u55ae\u5143\u8207\u6a19\u8a18 \u4e00\u822c\u8aaa\u4f86\uff0c\u5728\u4eba\u540d\u8fa8\u8b58\u4e2d\uff0c\u96d6\u7136\u4eba\u540d\u6709\u5927\u91cf\u7684\u7d44\u5408\u8207\u53ef\u80fd\u6027\uff0c\u4f46\u662f\u4f9d\u7136\u6703\u6709\u6240\u8b02\u7684\u5e38\u7528 \u5b57\uff0c \u300c\u83dc\u5e02\u5834\u540d\u300d\u5c31\u662f\u4e00\u7a2e\u5f88\u597d\u7684\u4f8b\u5b50\u3002\u4f46\u662f\u5546\u5bb6\u7d44\u7e54\u540d\u7a31\u4e2d\u9664\u4e86\u7d50\u5c3e\u90e8\u4efd\u7684\u5e38\u7528\u8a5e\u5916\uff0c \u5728\u4e3b\u8981\u540d\u7a31\u4e0a\u5e7e\u4e4e\u6c92\u6709\u4efb\u4f55\u898f\u7bc4\uff0c\u4f8b\u5982\uff1a\u300c\u571f\u5730\u300d\u3001\u300c\u963f\u5b24\u7956\u50b3\u83dc\u5305\u8089\u7cbd\u4ed9\u8349\u300d\u4e2d\u6240\u6709 \u8a5e\u7686\u70ba\u5e38\u7528\u8a5e\u5f59\uff0c\u300c18 \u5ea6 c \u5de7\u514b\u529b\u5de5\u574a\u300d\u3001\u300c591 \u79df\u5c4b\u300d\u70ba\u4e2d\u82f1\u6578\u5b57\u5143\u4ea4\u932f\u51fa\u73fe\uff0c\u300c\u52aa \u54c7\u514b\u5496\u5561\u300d\u3001\u300c\u857e\u514b\u723e\u70d8\u57f9\u574a\u300d\u70ba\u97f3\u8b6f\u8a5e\uff0c\u300c\u8606\u8588\u82b1\u5712\u96f2\u5357\u98df\u5e9c\u300d\u3001\u300c\u4e09\u5cfd\u6b77\u53f2\u6587\u7269\u9928\u300d \u70ba\u5730\u540d\u3002\u50c5\u7ba1\u5982\u6b64\uff0c\u9019\u4e9b\u5546\u5bb6\u7d44\u7e54\u540d\u7a31\u7684\u8a5e\u6027\u537b\u6709\u5e38\u898b\u5e8f\u5217\uff0c\u5982\u540d\u8a5e+\u540d\u8a5e\u6216\u52d5\u8a5e\u3001\u5c08\u6709 \u540d\u8a5e+\u540d\u8a5e\u6216\u52d5\u8a5e\u3001\u6578\u5b57\u6216\u82f1\u6587+\u540d\u8a5e\u6216\u52d5\u8a5e \u2026\u7b49\uff0c\u6240\u4ee5\u8a5e\u6027\u662f\u4e00\u7a2e\u4e0d\u53ef\u5ffd\u7565\u7684\u91cd\u8981\u7279 \u5fb5\u3002\u56e0\u6b64\u5728\u5e8f\u5217\u55ae\u5143(Tokens)\u7684\u9078\u64c7\u4e0a\uff0c\u6211\u5011\u5229\u7528 Stanford Segmenter \u53ca POS Tagger \u5c07\u7db2\u9801\u7684\u6587\u5b57\u5167\u5bb9\u7d93\u904e\u65b7\u8a5e\u53ca\u8a5e\u6027(POS\uff0cPart of Speech)\u6a19\u8a18\uff0c\u4ee5\u8a5e\u70ba\u55ae\u4f4d\u9032\u884c\u8a13\u7df4 \u8207\u6e2c\u8a66\u3002\u7d93\u904e\u65b7\u8a5e\u7684\u5e8f\u5217\uff0c\u518d\u4ee5 B\u3001I\u3001E\u3001O \u56db\u7a2e\u6a19\u8a18\u4ee3\u8868\u5546\u5bb6\u540d\u7a31\u7684\u8d77\u59cb\u3001\u4e2d\u9593\u3001\u7d50 \u5c3e\u3001\u4ee5\u53ca\u975e\u5546\u5bb6\u540d\u7a31\u3002 \u2022 \u2022 \u2022 \u2022 \u7279\u5fb5 \u7279\u5fb5 \u7279\u5fb5 \u7279\u5fb5 \u4e00\u822c\u4eba\u5728\u5224\u65b7\u4e00\u6bb5\u6587\u5b57\u662f\u5426\u662f\u5546\u5bb6\u540d\u7a31\u6642\uff0c\u6703\u4f9d\u9760\u5169\u985e\u7279\u5fb5\uff0c\u7b2c\u4e00\u7a2e\u662f\u5916\u90e8\u7279\u5fb5 (Outside Feature)\uff0c\u9019\u7a2e\u7279\u5fb5\u843d\u5728\u5546\u5bb6\u540d\u7a31\u7684\u5de6\u53f3\uff0c\u4f46\u662f\u6b64\u7a2e\u7279\u5fb5\u7121\u6cd5\u6e96\u78ba\u5224\u65b7\u5546\u5bb6\u540d\u7a31\uff0c\u53ea\u80fd \u9032\u884c\u63a8\u6e2c\u4e0a\u7684\u8f14\u52a9\u3002\u7b2c\u4e8c\u7a2e\u5247\u662f\u5167\u90e8\u7279\u5fb5(Inside Feature)\uff0c\u5167\u90e8\u7279\u5fb5\u80fd\u63d0\u4f9b\u5f37\u70c8\u7684\u5224 \u65b7\u8cc7\u8a0a\uff0c\u56e0\u70ba\u7d55\u5927\u591a\u6578\u7684\u5546\u5bb6\u540d\u7a31\u90fd\u662f\u7531\u4e09\u500b\u90e8\u4efd\u6240\u7d44\u6210\uff1a\u771f\u540d(Real Name)\u3001\u7522\u54c1\u6216 \u670d\u52d9(Service or Product)\u3001\u5730\u6a19\u6027\u8a5e\u5f59(Landmark)\uff0c\u8209\u4f8b\u4f86\u8aaa\uff1a\u300c\u71e6\u5764 3C \u91cf\u8ca9\u5e97\u300d \u53ef\u4ee5\u62c6\u6210\u300c\u71e6\u5764/3C/\u91cf\u8ca9\u5e97\u300d\u6216\u300c\u71e6\u5764/3C \u91cf\u8ca9/\u5e97\u300d\u3002\u5373\u4f7f\u662f\u975e\u5e38\u77ed\u7684\u5546\u5bb6\u540d\u7a31\u90fd\u6703\u6709 \u9019\u7a2e\u7d50\u69cb\uff0c\u4f8b\u5982\uff1a\u300c\u9e97\u5b30\u623f\u300d\u662f\u300c\u9e97/\u5b30/\u623f\u300d\uff0c\u5176\u4e2d\u5b30\u662f\u6307\u63d0\u4f9b\u5152\u7ae5\u7528\u54c1\u3002 \u6211\u5011\u7d71\u8a08\u5df2\u77e5\u7684\u5546\u5bb6\u540d\u7a31\uff0c\u5c0d\u7d44\u7e54\u3001\u5efa\u7bc9\u3001\u623f\u9593\u3001\u5730\u6a19\u5efa\u7acb\u6e05\u55ae\uff0c\u4f8b\u5982\uff1a\u6703\u3001\u57ce\u3001 \u623f\u3001\u7ad9\u2026\u7b49\uff0c\u7576\u6bcf\u500b\u5e8f\u5217\u55ae\u5143(Tokens)\u662f\u4ee5\u6b64\u6e05\u55ae\u4e2d\u7684\u6587\u5b57\u70ba\u7d50\u5c3e\uff0c\u5c31\u8868\u793a\u5177\u6709\u5730\u6a19 \u6027\u8a5e\u5f59\u7684\u7279\u5fb5(Landmark Feature)\u3002\u53e6\u5916\uff0c\u6211\u5011\u4e5f\u6536\u96c6\u4e86\u9ec3\u9801\u7db2\u7ad9\u7684\u670d\u52d9\u3001\u7522\u54c1\u5efa\u7acb\u6e05 \u55ae \uff0c \u5982 \u679c \u5e8f \u5217 \u55ae \u5143 ( Tokens ) \u542b \u6709 \u6b64 \u6e05 \u55ae \u4e2d \u7684 \u8a5e \u5f59 \uff0c \u5c31 \u8868 \u793a \u5177 \u6709 \u7522 \u54c1 \u670d \u52d9 \u7279 \u5fb5 (Service/Product Feature)\u3002 \u7576\u6211\u5011\u6709\u4e86\u4e0a\u8ff0\u5169\u7a2e\u7279\u5fb5\uff0c\u554f\u984c\u5c31\u7c21\u5316\u6210\u5982\u4f55\u627e\u51fa\u771f\u540d(Real Name)\u7684\u90e8\u4efd\uff0c\u7db2\u9801 \u5167\u5bb9\u8207\u4e00\u822c\u6587\u7ae0\u4e0d\u540c\u7684\u5730\u65b9\u5728\u65bc\u540d\u7a31\u66f4\u50be\u5411\u65bc\u55ae\u7368\u51fa\u73fe\uff0c\u800c\u9bae\u5c11\u5b58\u5728\u65bc\u4e00\u6bb5\u5b8c\u6574\u7684\u53e5\u5b50 \u4e2d \uff0c\u6240\u4ee5\u4e00\u6bb5\u6587\u5b57\u610f\u601d\u7684\u8d77\u9ede\u5c31\u8b8a\u6210\u5f88\u91cd\u8981\u7684\u7279\u5fb5\uff1a\u5982\u679c\u4e00\u500b\u5e8f\u5217\u55ae\u5143(Tokens)\u662f\u6a23 \u672c\u55ae\u5143\u7684\u8d77\u9ede\u6216\u524d\u4e00\u500b\u5e8f\u5217\u55ae\u5143\u5c6c\u65bc\u7b26\u865f\u985e\uff0c\u5c31\u5177\u6709\u958b\u59cb\u7279\u5fb5(Start Feature)\uff0c\u53cd\u904e\u4f86 \u8aaa\uff0c\u7576\u5e8f\u5217\u55ae\u5143\u662f\u6a23\u672c\u55ae\u5143\u7684\u7d50\u5c3e\u6216\u4e0b\u4e00\u500b\u5e8f\u5217\u55ae\u5143\u5c6c\u65bc\u7b26\u865f\u985e\uff0c\u5c31\u5177\u6709\u7d50\u5c3e\u7279\u5fb5(End Feature)\u3002\u4f8b\u5982\uff1a\u7db2\u9801\u4e2d\u7684\u6a19\u8a9e\u300c[\u963f\u5b24\u7956\u50b3\u83dc\u5305\u8089\u7cbd\u4ed9\u8349]\u6709\u963f\u5b24\u7684\u7cbe\u795e\u50b3\u627f\u88fd\u4f5c\u51fa\u5ba2 \u5bb6\u50b3\u7d71\u7c73\u98df\u597d\u6ecb\u5473!\u300d\u8207\u7db2\u9801\u6a19\u984c\u300c\u963f\u5b24\u7956\u50b3\u83dc\u5305\u8089\u7cbd\u4ed9\u8349\u300d\u4e2d\uff0c\u524d\u8005\u7684\u300c\u963f\u5b24\u300d\u7684\u524d\u4e00 \u5e8f\u5217\u55ae\u5143\u70ba\u7b26\u865f\uff0c\u5f8c\u8005\u70ba\u5e8f\u5217\u55ae\u5143\u7684\u8d77\u9ede\u6240\u4ee5\u7686\u5177\u6709\u958b\u59cb\u7279\u5fb5\uff0c\u300c\u4ed9\u8349\u300d\u5247\u5177\u6709\u7d50\u5c3e\u7279 \u5fb5\u3002\u7cfb\u7d71\u6700\u5f8c\u9078\u64c7\u5c0d\u5546\u5bb6\u540d\u7a31\u5177\u6709\u5f37\u70c8\u5224\u65b7\u8cc7\u8a0a\u7684\u5167\u90e8\u7279\u5fb5\u52a0\u5165\u8a13\u7df4\u6a21\u578b\uff0c\u6240\u6709\u539f\u59cb\u7279 \u5fb5\u5217\u65bc\u8868 1\u3002 \u8868 \u8868 \u8868 \u8868 1. \u672c\u7814\u7a76\u6240\u4f7f\u7528\u7684\u539f\u59cb\u7279\u5fb5 \u672c\u7814\u7a76\u6240\u4f7f\u7528\u7684\u539f\u59cb\u7279\u5fb5 \u672c\u7814\u7a76\u6240\u4f7f\u7528\u7684\u539f\u59cb\u7279\u5fb5 \u672c\u7814\u7a76\u6240\u4f7f\u7528\u7684\u539f\u59cb\u7279\u5fb5 NO. Feature Explanation 1 Token \u500b\u5225\u8a5e Individual Word, e.g. 591, \u79df\u5c4b 2 isPOS \u8a5e\u6027 Part of Speech, e.g. NR, NN, CD 3 isStart \u6a23\u672c\u5e8f\u5217\u958b\u982d \u6216 \u77ed\u8a9e\u958b\u982d 4 isSymbol \u5c6c\u65bc\u7b26\u865f\u8a5e, e.g. (, [, breakline, !, : 5 isService/Product \u5c6c\u65bc\u670d\u52d9/\u7522\u54c1\u8a5e, e.g. 3C, \u58fd\u53f8, \u51fa\u79df, \u901a\u4fe1 6 isLandmark \u5c6c\u65bc\u5730\u6a19\u8a5e, e.g. \u5edf, \u838a, \u516c\u53f8, \u5e97 7 isEnd \u6a23\u672c\u5e8f\u5217\u7d50\u5c3e \u6216 \u77ed\u8a9e\u7d50\u5c3e 3.2 \u5730\u5740 \u5730\u5740 \u5730\u5740 \u5730\u5740-\u5546\u5bb6\u540d\u7a31\u5339\u914d \u5546\u5bb6\u540d\u7a31\u5339\u914d \u5546\u5bb6\u540d\u7a31\u5339\u914d \u5546\u5bb6\u540d\u7a31\u5339\u914d \u7576\u6211\u5011\u6709\u4e86\u5730\u5740\u8207\u5546\u5bb6\u540d\u7a31\u5f8c\uff0c\u4fbf\u53ef\u4ee5\u958b\u59cb\u9032\u884c\u914d\u5c0d\u3002\u7531\u65bc\u5404\u985e\u5225\u7684\u7db2\u9801\u7279\u6027\u5dee\u7570\u5f88\u5927\uff0c \u6240\u4ee5\u7cfb\u7d71\u6703\u91dd\u5c0d\u5404\u985e\u5225\u8a2d\u8a08\u5404\u81ea\u7684\u555f\u767c\u5f0f(heuristic)\u7684\u914d\u5c0d\u65b9\u5f0f\u3002\u9996\u5148\u6211\u5011\u4f9d\u7167\u7db2\u7ad9\u5c07 \u7db2\u9801\u5206\u6210\u4e0d\u540c\u7fa4\u7d44\uff0c\u63a5\u8457\u4f9d\u7db2\u7ad9\u4e2d\u7684\u5730\u5740\u8cc7\u8a0a\u5c07\u7db2\u9801\u5206\u6210\u56db\u985e\u3002 \u81ea\u7136\u8a9e\u8a00\u7db2\u9801\uff1a\u7576\u5730\u5740\u5b57\u4e32\u6240\u5728\u7684\u6587\u5b57\u7bc0\u9ede(Text Node)\u6709\u8d85\u904e 50 \u500b\u5b57\u5c31\u6703\u6b78\u985e \u81f3\u81ea\u7136\u8a9e\u8a00\u7db2\u9801 (\u8acb\u53c3\u8003\u5716 2) \uff0c\u56e0\u70ba\u6703\u9019\u500b\u9577\u5ea6\u76f8\u7576\u65bc\u4e00\u5c0f\u584a\u7247\u6bb5\u6587\u5b57 (Snippet) \u3002 \u4f4d\u65bc\u6b64\u7a2e\u7db2\u9801\u7684\u5546\u5bb6\u540d\u7a31\u5de6\u53f3\u5927\u591a\u63a5\u6709\u80fd\u610f\u6703\u5230\u8a72\u8655\u70ba\u5546\u5bb6\u540d\u7a31\u7684\u8a0a\u606f\uff0c\u4f8b\u5982\uff1a \u300c\u8d70 \u9032 edia cafa \u5e97\u88e1\u4e00\u773c\u671b\u53bb\u300d\u3001\u300c\u6211\u6628\u5929\u53bb\u4e86\u71e6\u5764 3C \u8cb7\u6771\u897f\u300d\u3002\u9019\u4e5f\u662f\u7db2\u9801\u4e2d\u552f\u4e00 \u63a5\u8fd1\u4e00\u822c\u6587\u7ae0\u7684\u985e\u5225\u3002\u901a\u5e38\u5177\u6709\u5916\u90e8\u7279\u5fb5 (Outside Feature)\u3002 \u8a3b\u8173\u8cc7\u8a0a\u7db2\u9801\uff1a\u7576\u4e00\u500b\u300c\u7db2\u7ad9\u300d\u5167\u8d85\u904e 80%\u7684\u7db2\u9801\u90fd\u6709\u76f8\u540c\u7684\u5730\u5740\u8207\u6587\u4ef6\u7269\u4ef6\u6a39\u8def \u5f91(DOM Tree Path)\uff0c\u9019\u4e9b\u5730\u5740\u5c31\u6703\u6b78\u985e\u81f3\u8a3b\u8173\u8cc7\u8a0a\u7db2\u9801\u3002\u6b64\u985e\u5225\u4e2d\u7684\u6240\u6709\u7db2\u7ad9\uff0c \u5546\u5bb6\u540d\u7a31\u5468\u570d\u7684\u6587\u5b57\u8cc7\u8a0a\u90fd\u6709\u5f88\u9ad8\u7684\u76f8\u4f3c\u5ea6\uff0c\u7d93\u5e38\u6703\u6709\uff1a\u300c\u672c\u7db2\u7ad9\u70ba\u2026\u300d\u300c\u2026\u7248\u6b0a \u6240\u6709\u300d\u3001\u00ae\u3001\u00a9\u3001\u5730\u5740\u3001\u96fb\u8a71\u2026\uff0c\u9019\u4e9b\u8cc7\u8a0a\u5728 N \u5143\u6587\u6cd5(N-Gram)\u7684\u7279\u5fb5\u4e0a\uff0c\u80fd\u63d0 \u4f9b\u6709\u7528\u7684\u8cc7\u8a0a\u3002 \u6e05\u55ae\u7db2\u9801\uff1a\u7576\u4e00\u500b\u7db2\u9801\u5167\u5305\u542b\u8d85\u904e 3 \u7b46\u5730\u5740\u6709\u76f8\u540c\u7684\u6587\u4ef6\u7269\u4ef6\u6a39\u8def\u5f91(DOM Tree Path) \uff0c\u9019\u4e9b\u5730\u5740\u5c31\u6b78\u985e\u70ba\u6e05\u55ae\u985e\u578b\u3002\u6e05\u55ae\u578b\u7684\u5546\u5bb6\u540d\u7a31\u96d6\u7136\u4e0d\u50cf\u81ea\u7136\u8a9e\u8a00\u7db2\u9801\u4e2d\uff0c \u5546\u5bb6\u540d\u7a31\u7684\u5de6\u53f3\u5177\u6709\u63cf\u8ff0\u6027\u7684\u6587\u5b57\uff0c\u4f46\u53d6\u800c\u4ee3\u4e4b\u7684\u662f\u5468\u570d\u5177\u6709\u63db\u884c\u7b26\u865f\u3001\u96fb\u8a71\u3001\u5730 \u5740\u3001\u6642\u9593\u7b49\u8cc7\u8a0a\uff0c\u85c9\u7531\u4e8b\u5148\u7528\u6b63\u898f\u8868\u793a\u6cd5\u53d6\u4ee3\u9019\u4e9b\u5b57\u4e32\u5f8c\uff0c\u4ea6\u80fd\u5229\u7528 N-Gram \u53d6\u5f97 \u6b64\u7279\u6027\u3002 \u6df1\u5ea6\u8cc7\u8a0a\u7db2\u9801(Detail Pages)\uff1a\u7576\u4e00\u500b\u7db2\u7ad9\u5167\u4e0d\u540c\u7db2\u9801\u7684\u5730\u5740\u6709\u76f8\u540c\u7684\u6587\u4ef6\u7269\u4ef6\u6a39\u8def \u5f91(DOM Tree Path)\uff0c\u4f46\u662f\u5730\u5740\u5b57\u4e32\u537b\u4e0d\u76f8\u540c\uff0c\u9019\u4e9b\u5730\u5740\u5c31\u6b78\u985e\u70ba\u6df1\u5ea6\u8cc7\u8a0a\u7db2\u9801\uff0c \u7576\u6211\u5011\u5f9e\u591a\u500b\u7db2\u9801\u4f86\u770b\u6642\uff0c\u5730\u5740\u548c\u5546\u5bb6\u540d\u7a31\u901a\u5e38\u64c1\u6709\u540c\u6a23\u7684\u6587\u4ef6\u7269\u4ef6\u6a39\u8def\u5f91(DOM Tree Path)\uff0c\u6211\u5011\u53ef\u4ee5\u900f\u904e\u6b64\u7279\u6027\u9032\u884c\u5546\u5bb6\u540d\u7a31\u7684\u4fee\u6b63\u3002 (a) (b) (c) (d) \u5716 \u5716 \u5716 \u5716 2. (a) \u81ea\u7136\u8a9e\u8a00\u7db2\u9801\u7bc4\u4f8b \u81ea\u7136\u8a9e\u8a00\u7db2\u9801\u7bc4\u4f8b \u81ea\u7136\u8a9e\u8a00\u7db2\u9801\u7bc4\u4f8b \u81ea\u7136\u8a9e\u8a00\u7db2\u9801\u7bc4\u4f8b (b) \u8a3b\u8173 \u8a3b\u8173 \u8a3b\u8173 \u8a3b\u8173\u7db2\u9801\u7bc4\u4f8b \u7db2\u9801\u7bc4\u4f8b \u7db2\u9801\u7bc4\u4f8b \u7db2\u9801\u7bc4\u4f8b (c) \u6e05\u55ae \u6e05\u55ae \u6e05\u55ae \u6e05\u55ae\u7db2\u9801\u7bc4\u4f8b \u7db2\u9801\u7bc4\u4f8b \u7db2\u9801\u7bc4\u4f8b \u7db2\u9801\u7bc4\u4f8b (d)\u6df1\u5ea6 \u6df1\u5ea6 \u6df1\u5ea6 \u6df1\u5ea6\u7db2\u9801 \u7db2\u9801 \u7db2\u9801 \u7db2\u9801 \u7bc4\u4f8b \u7bc4\u4f8b \u7bc4\u4f8b \u7bc4\u4f8b \u5c0d\u65bc\u7b2c\u4e00\u548c\u7b2c\u4e8c\u985e\u7db2\u9801\u800c\u8a00\uff0c\u5730\u5740\u6240\u5c0d\u61c9\u7684\u5546\u5bb6\u540d\u7a31\u901a\u5e38\u843d\u5728\uff1a\u7db2\u9801\u6a19\u984c\u3001\u5730\u5740\u524d\u3001 \u5730\u5740\u5f8c\u6216\u9ad8\u983b\u5546\u5bb6\u540d\u7a31\u3002\u82e5\u53ea\u6709\u4e00\u500b\u5730\u5740\uff0c\u5247\u7b2c\u4e00\u9806\u4f4d\u662f\u7db2\u9801\u6a19\u984c\u4e2d\u7684\u5546\u5bb6\u540d\u7a31\u3002\u5176\u6b21\uff0c \u4ee5\u9760\u8fd1\u5730\u5740\u7684\u5546\u5bb6\u540d\u7a31\u70ba\u512a\u5148\u914d\u5c0d\u5c0d\u8c61\uff0c\u300c\u5730\u5740\u524d\u300d\u7684\u914d\u5c0d\u65b9\u5f0f\u662f\u5c07\u5730\u5740\u8207\u6240\u5728\u4f4d\u7f6e\u7684 \u524d\u4e94\u884c\u5167\u7684\u5546\u5bb6\u540d\u7a31\u5217\u70ba\u914d\u5c0d\u5019\u9078\u8005\uff0c\u800c\u300c\u5730\u5740\u5f8c\u300d\u5247\u662f\u5c07\u5730\u5740\u8207\u6240\u5728\u4f4d\u7f6e\u7684\u5f8c\u5169\u884c\u5167 \u7684\u5546\u5bb6\u540d\u7a31\u5217\u70ba\u914d\u5c0d\u5019\u9078\u8005\uff0c\u7576\u591a\u500b\u5019\u9078\u8005\u8ddd\u96e2\u76f8\u540c\u6642\uff0c\u6703\u4ee5\u7db2\u9801\u4e2d\u51fa\u73fe\u8f03\u591a\u6b21\u7684\u5546\u5bb6 \u540d\u7a31\u70ba\u512a\u5148\uff0c\u82e5\u6b21\u6578\u5b8c\u5168\u76f8\u540c\u5247\u9078\u64c7\u4f4d\u65bc\u5730\u5740\u524d\u65b9\u7684\u5546\u5bb6\u540d\u7a31\u3002 \u81f3\u65bc\u7b2c\u4e09\u548c\u7b2c\u56db\u985e\u7db2\u9801\uff0c\u56e0\u70ba\u7db2\u9801\u901a\u5e38\u7531\u6a21\u677f(Template)\u548c\u7d00\u9304(Record)\u6240\u7d44 \u6210\uff0c\u800c\u76f8\u540c\u985e\u578b\u7684\u7d00\u9304\u6703\u653e\u7f6e\u5728\u985e\u4f3c\u8def\u5f91\u4e0b\uff0c\u6240\u4ee5\u5b58\u5728\u4e00\u500b\u5c08\u9580\u7684\u7814\u7a76\u9818\u57df\u7a31\u70ba Wrapper Induction\uff0c\u76ee\u7684\u662f\u900f\u904e\u53c3\u8003\u4e00\u500b\u6216\u591a\u500b\u7db2\u9801\u5167\u5bb9\u53cd\u5411\u63a8\u5c0e\u51fa\u6a21\u677f\u8207\u7d00\u9304\u3002\u672c\u7814\u7a76\u4e2d\u4f7f\u7528 \u53ef\u4ee5\u5c07\u591a\u500b\u7db2\u9801\u7684\u539f\u59cb\u6a94\u6587\u5b57\u5167\u5bb9\u7576\u4f5c\u8f38\u5165(\u4f5c\u8005\u7a31\u70ba TextSet)\uff0c\u900f\u904e\u5c0b\u627e\u5404\u6587\u4ef6\u6240\u64c1 \u6709\u7684\u5171\u4eab\u6a23\u5f0f(Shared Pattern)\u7576\u4f5c\u7d00\u9304\u7684\u5206\u9694\u9ede\uff0c\u7d93\u904e\u53cd\u8986\u5c0b\u627e\u5171\u4eab\u6a23\u5f0f\u8207\u5207\u5272\u5f8c\uff0c \u627e\u51fa\u6700\u5f8c\u7684\u8cc7\u6599\u7bc0\u9ede\u3002\u85c9\u7531 TEX (Hassan &amp; Sleiman, 2013) \u64f7\u53d6\u51fa\u7db2\u9801\u4e2d\u5177\u6709\u540c\u6027\u8cea\u7684 \u8cc7\u6599\u7bc0\u9ede\uff0c\u7576\u6709\u4e00\u5b9a\u6578\u91cf\u7684\u540c\u985e\u7bc0\u9ede\u88ab\u8a8d\u70ba\u662f\u5546\u5bb6\u540d\u7a31\u4e14\u5546\u5bb6\u540d\u7a31\u9577\u5ea6\u4f54\u7bc0\u9ede\u5167\u5bb9\u7684 20 \uff05\u4ee5\u4e0a\u6642\uff0c\u5247\u628a\u540c\u985e\u7684\u975e\u5546\u5bb6\u540d\u7a31\u7bc0\u9ede\u4e5f\u8996\u70ba\u5546\u5bb6\u540d\u7a31\u9032\u884c\u914d\u5c0d\u3002\u8209\u4f8b\u800c\u8a00\uff0c\u5716 2 \u4e2d\u7684 \u300c\u5929\u5929 100 \u526a\u9aee\u300d\u4e26\u6c92\u6709\u88ab CRF \u8fa8\u8b58\u51fa\u4f86\uff0c\u4f46\u662f\u5728\u540c\u7db2\u7ad9\u7684\u5176\u4ed6\u7db2\u9801\u4e2d\uff0c\u6b64\u7bc0\u9ede\u7684\u5167\u5bb9 \u904e\u6ffe\u51fa\u7d04 39 \u842c\u500b\u542b\u6709\u53f0\u7063\u5730\u5740\u7684\u7db2\u9801\uff0c\u7d93\u904e\u5730\u5740\u6b63\u898f\u5316\u5f8c\u542b 19 \u842c\u7b46\u53f0\u7063\u5730\u5740(\u8acb\u53c3\u8003 \u53d6 1 \u500b\u7db2\u9801\u9032\u884c\u5be6\u9a57\uff0c\u4f46 Detail Pages \u56e0\u70ba\u914d\u5c0d\u65b9\u6cd5\u9700\u53c3\u8003\u591a\u500b\u7db2\u9801\uff0c\u6240\u4ee5\u96a8\u6a5f\u6311\u9078\u4e86 \u7a31\u7576\u4f5c\u6e2c\u8a66\u8cc7\u6599\u3002\u800c\u8a13\u7df4\u8cc7\u6599\u5247\u96a8\u6a5f\u6311\u9078\u4e86 30,000 \u500b\u8a13\u7df4\u6a23\u672c\uff0c\u5305\u542b 51,775 \u500b\u4ee5\u81ea\u52d5\u6a19 \u500b\u554f\u984c\u3002 \u8b93\u4f7f\u7528\u8005\u5728\u5730\u5716\u4e0a\u63d0\u4f9b\u66f4\u70ba\u4fbf\u5229\u7684\u67e5\u8a62\u3002\u5730\u5740\u662f POI \u7684\u91cd\u8981\u6307\u6a19\u5982\u679c\u80fd\u627e\u51fa\u5730\u5740\u6240\u4ee3\u8868 \u8a18\u6cd5\u6a19\u8a18\u7684\u5546\u5bb6\u540d\u7a31\u3002 \u8cc7\u8a0a\u6aa2\u7d22\uff0c\u56e0\u6b64\u7cfb\u7d71\u6a19\u8a18\u7d50\u679c\u82e5\u80fd\u5305\u542b\u6b63\u78ba\u7b54\u6848(Gold)\uff0c\u6211\u5011\u5373\u8a8d\u5b9a\u6b63\u78ba\uff0c\u82e5\u662f\u50c5\u70ba\u6b63 \u6210 O)\u3002\u56e0\u6b64\u5728 Search Snippets Search Snippets \u5be6\u9a57\u4e2d\uff0c\u6211\u5011\u5617\u8a66\u63a2\u8a0e\u964d\u4f4e\u8a9e\u6599\u8907\u96dc\u5ea6\u8207\u6a19\u8a18\u4e0d\u5b8c\u5168\u5169 \u6211\u5011\u5617\u8a66\u63a2\u8a0e\u964d\u4f4e\u8a9e\u6599\u8907\u96dc\u5ea6\u8207\u6a19\u8a18\u4e0d\u5b8c\u5168\u5169 \u5316\u6027\u6975\u5927\u7684\u547d\u540d\u5be6\u9ad4\uff0c\u8f03\u96e3\u8fa8\u8b58\u51fa\u6b63\u78ba\u7684\u7b54\u6848\uff0c\u56e0\u6b64\u9700\u8981\u66f4\u591a\u7684\u7279\u5fb5\u8207\u63d0\u6607\u6a19\u8a18\u54c1\u8cea\u3002 \u6ce2\u4ee5\u884c\u52d5\u88dd\u7f6e\u70ba\u4e3b\u6d41\u7684\u8da8\u52e2\u4e2d\u5177\u6709\u81f3\u95dc\u91cd\u8981\u7684\u5730\u4f4d\uff0c\u5efa\u7acb\u4e00\u500b\u5b8c\u6574\u7684 POI \u8cc7\u6599\u5eab\uff0c\u53ef\u4ee5 \u78ba\u7b54\u6848\u7684\u90e8\u4efd\uff0c\u5247\u7d66 0~1 \u4e4b\u9593\u7684\u5206\u6578\uff0c\u4e26\u4f9d\u6b64\u5206\u6578\u8a08\u7b97 Precision\u3001Recall\u3001F-measure\uff1a \u6c92\u6709\u4f7f\u7528\u6240\u6709\u5df2\u77e5\u7684\u5546\u5bb6\u540d\u7a31\u9032\u884c\u6a19\u8a18 \u6c92\u6709\u4f7f\u7528\u6240\u6709\u5df2\u77e5\u7684\u5546\u5bb6\u540d\u7a31\u9032\u884c\u6a19\u8a18\uff0c\u6240\u4ee5\u9020\u6210\u4e86\u5927\u91cf\u6a19\u8a18\u932f\u8aa4(\u61c9\u70ba B/I/E/S /S\u3001\u537b\u6a19 \u6599\u4f86\u6e90\u518d\u4ee5\u81ea\u52d5\u6a19\u8a18\u9032\u884c\u8a13\u7df4\uff0c\u53ef\u80fd\u9020\u6210\u8a13\u7df4\u6a23\u672c\u7684\u54c1\u8cea\u4e0d\u4f73\uff0c\u56e0\u6b64\u5c0d\u5546\u5bb6\u540d\u7a31\u9019\u7a2e\u8b8a 2014 \u662f\u4e00\u500b\u884c\u52d5\u88dd\u7f6e\u7684\u6642\u4ee3\uff0c\u5927\u91cf\u7684\u9069\u5730\u6027\u670d\u52d9(LBS)\u56e0\u6b64\u8a95\u751f\uff0c\u800c POI \u8cc7\u6599\u5eab\u5728\u9019 11 \u500b\u7db2\u7ad9\uff0c\u6bcf\u500b\u7db2\u7ad9\u62bd\u53d6 10 \u500b\u7db2\u9801\u3002\u6700\u5f8c\u5c0d\u9019 410 \u500b\u7db2\u9801\u4eba\u5de5\u6a19\u8a18\u4e86 10,457 \u500b\u5546\u5bb6\u540d \u6642\u5019\u4f9d\u7136\u96e3\u4ee5\u6e96\u78ba\u5b9a\u51fa\u908a\u754c\u6a19\u6e96\uff0c\u4f8b\u5982\uff1a\u300c\u98ef\u5e97\u540d\u7a31\uff1a\u897f\u9580\u661f\u8fb0\u5927\u98ef\u5e97\u300d\u4e2d\uff0c\u300c\u897f\u9580\u300d \u4e8c\u5b57\u8a72\u4e0d\u8a72\u5217\u5165\u5546\u5bb6\u540d\u7a31\u4e2d\u6709\u8a31\u591a\u610f\u898b\u5206\u6b67\u7684\u60c5\u6cc1\uff0c\u7531\u65bc\u5546\u5bb6\u540d\u7a31\u4e3b\u8981\u63d0\u4f9b\u5f8c\u7e8c\u7684\u5730\u7406 \u6578\u91cf\u9054\u5230 30,000 \u6a23\u672c\u5e8f\u5217\u6642\uff0c Precision \u4e5f\u8f03\u5927\u5e45\u7684\u4e0b\u964d\u3002\u4e3b\u8981\u7684\u539f\u56e0\u53ef\u80fd\u5728\u65bc\u6211\u5011\u4f7f\u7528\u81ea\u52d5\u6a19\u8a18\u7522\u751f\u8a13\u7df4\u8cc7\u6599\u6642 \uff0c\u8fa8\u8b58\u6548\u679c\u4f9d\u7136\u53ea\u6709 0.328\uff0c\u96d6\u7136 Recall \u7372\u5f97\u63d0\u6607 \u4e3b\u8981\u7684\u539f\u56e0\u53ef\u80fd\u5728\u65bc\u6211\u5011\u4f7f\u7528\u81ea\u52d5\u6a19\u8a18\u7522\u751f\u8a13\u7df4\u8cc7\u6599\u6642 \u7372\u5f97\u63d0\u6607\uff0c\u4f46\u662f \u4e3b\u8981\u7684\u539f\u56e0\u53ef\u80fd\u5728\u65bc\u6211\u5011\u4f7f\u7528\u81ea\u52d5\u6a19\u8a18\u7522\u751f\u8a13\u7df4\u8cc7\u6599\u6642\uff0c\u4e26 \u7684\u6db5\u84cb\u5404\u985e\u5546\u5bb6\u7d44\u7e54\u540d\u7a31\u7684\u7279\u6027\u3002\u7b2c\u4e8c\uff0c\u7db2\u9801\u5c6c\u65bc\u4e00\u7a2e\u7d50\u69cb\u8907\u96dc\u7684\u8cc7\u6599\u4f86\u6e90\uff0c\u4ee5\u6b64\u7a2e\u8cc7 5. \u7d50\u8ad6 \u7d50\u8ad6 \u7d50\u8ad6 \u7d50\u8ad6 \u5546\u5bb6\u7d44\u7e54\u540d\u7a31\u5c6c\u65bc\u8b8a\u7570\u6027\u8f03\u5927\u7684\u4e00\u7a2e\u547d\u540d\u5be6\u9ad4\uff0c\u5728\u8a13\u7df4\u968e\u6bb5\u4e2d\uff0c\u8cc7\u6599\u7684\u6e96\u5099\u80fd\u5426\u76e1\u53ef\u80fd \u8868 2)\u3002\u5728\u7d93\u904e\u7db2\u9801\u5206\u985e\u5f8c\uff0c\u6211\u5011\u96a8\u6a5f\u6311\u9078\u5404\u985e\u4e2d\u7684 100 \u500b\u7db2\u7ad9\uff0c\u6bcf\u500b\u7db2\u7ad9\u4e2d\u5404\u96a8\u6a5f\u62bd \u5716 4 \u662f\u8a13\u7df4\u8cc7\u6599\u6578\u91cf\u5c0d Precision Precision\u3001Recall\u3001F1 \u5f71\u97ff\u7684\u8da8\u52e2\u5716\uff0c\u5716\u4e2d\u986f\u793a\u7576\u8a13\u7df4\u8cc7\u6599 \u5716\u4e2d\u986f\u793a\u7576\u8a13\u7df4\u8cc7\u6599 \u7d9c\u5408\u4ee5\u4e0a\u5be6\u9a57\u7d50\u679c\u4f86\u770b\uff0c\u6211\u5011\u8a8d\u70ba\u5f71\u97ff\u8fa8\u8b58\u6548\u80fd\u7684\u4e3b\u8981\u7684\u539f\u56e0\u6709\u4e09\u500b\uff1a\u7b2c\u4e00\u662f\u56e0\u70ba \u5716 \u5716 \u5716 \u5716 6. SnippetFullLabeling \u4e0d\u540c\u8a13\u7df4\u8cc7\u6599\u6578\u91cf\u7684\u6a21\u578b\u4e2d \u4e0d\u540c\u8a13\u7df4\u8cc7\u6599\u6578\u91cf\u7684\u6a21\u578b\u4e2d \u4e0d\u540c\u8a13\u7df4\u8cc7\u6599\u6578\u91cf\u7684\u6a21\u578b\u4e2d \u4e0d\u540c\u8a13\u7df4\u8cc7\u6599\u6578\u91cf\u7684\u6a21\u578b\u4e2d\uff0c \uff0c \uff0c \uff0cNER \u5c0d \u5c0d \u5c0d \u5c0d Match \u7684\u5f71\u97ff \u7684\u5f71\u97ff \u7684\u5f71\u97ff \u7684\u5f71\u97ff \u6a19\u8a18\u6bd4\u5c0d\u7684\u8a55\u4f30\u65b9\u5f0f\u5982\u4e0b\uff1a\u96d6\u7136\u6211\u5011\u6709\u660e\u78ba\u8a02\u51fa\u5546\u5bb6\u7d44\u7e54\u540d\u7a31\u7684\u5224\u5b9a\u898f\u5247\uff0c\u4f46\u5f88\u591a \u4e86 TEX \u4f5c\u70ba\u8f14\u52a9\u5de5\u5177(Hassan &amp; Sleiman, 2013)\uff0cTEX \u662f\u4e00\u500b Deep Web Crawling Tool\uff0c \u300cGM \u9020\u578b\u9928\u300d\u3001\u300c\u80af\u7279\u9020\u578b\u6c99\u9f8d\u300d\u2026\u7b49\u5df2\u88ab\u6210\u529f\u8fa8\u8b58\u70ba\u5546\u5bb6\u540d\u7a31\uff0c\u6240\u4ee5\u7cfb\u7d71\u4e5f\u6703\u5c07\u300c\u5929 \u5929 100 \u526a\u9aee\u300d\u8996\u70ba\u5546\u5bb6\u540d\u7a31\u3002\u672c\u7cfb\u7d71\u4e2d\uff0c\u9580\u6abb\u503c\u70ba 0.2\uff0c\u5373\u8a72\u7bc0\u9ede\u6709 20%\u4ee5\u4e0a\u7684\u5167\u5bb9\u88ab \u8a8d\u70ba\u662f\u5546\u5bb6\u540d\u7a31\uff0c\u5247\u5176\u9918\u7db2\u9801\u7684\u8a72\u7bc0\u9ede\u4e5f\u6703\u88ab\u8a8d\u70ba\u662f\u5546\u5bb6\u540d\u7a31\u3002 \u5716 \u5716 \u5716 \u5716 3. \u6df1\u5ea6\u8cc7\u6599\u7db2\u9801\u914d\u5c0d\u7bc4\u4f8b \u6df1\u5ea6\u8cc7\u6599\u7db2\u9801\u914d\u5c0d\u7bc4\u4f8b \u6df1\u5ea6\u8cc7\u6599\u7db2\u9801\u914d\u5c0d\u7bc4\u4f8b \u6df1\u5ea6\u8cc7\u6599\u7db2\u9801\u914d\u5c0d\u7bc4\u4f8b \u7576\u6211\u5011\u5229\u7528\u8def\u5f91\u627e\u51fa\u6240\u6709\u53ef\u80fd\u7684\u5546\u5bb6\u540d\u7a31\u5f8c\uff0c\u5c07\u958b\u59cb\u9032\u884c\u5be6\u969b\u914d\u5c0d\u3002\u6e05\u55ae\u578b\u7db2\u9801\u8207 \u6df1\u5ea6\u8cc7\u8a0a\u7db2\u9801\u7684\u914d\u5c0d\u65b9\u5f0f\u5927\u81f4\u76f8\u540c\uff1a\u4ee5\u6bcf\u7b46\u5730\u5740\u7684\u4e0a\u65b9\u5168\u90e8\u5167\u5bb9\u8207\u4e0b\u65b9\u5169\u884c\u5167\u7576\u4f5c\u914d\u5c0d \u5019\u9078\uff0c\u96e2\u5730\u5740\u8fd1\u7684\u512a\u5148\u914d\u5c0d\uff0c\u7576\u8ddd\u96e2\u76f8\u540c\u6642\uff0c\u4ee5\u5730\u5740\u524d\u65b9\u7684\u5546\u5bb6\u540d\u7a31\u70ba\u512a\u5148\u3002\u4f46\u6e05\u55ae\u578b \u7db2\u9801\u6703\u4ee5\u5730\u5740\u70ba\u754c\u7dda\uff0c\u5728\u6311\u9078\u914d\u5c0d\u5019\u9078\u8005\u6642\uff0c\u4e0d\u6703\u8d8a\u904e\u5730\u5740\u9032\u884c\u914d\u5c0d\u3002 \u53e6\u5916\u7576\u6211\u5011\u4ee5\u5730\u5740\u70ba\u95dc\u9375\u5b57\u6536\u96c6 Google Snippet \u5f8c\uff0c\u9019\u4e9b Snippets \u4e2d\u7684\u7db2\u9801\u7d50\u69cb\u8cc7 \u8a0a\u8f03\u5f31\uff0c\u4f46\u662f\u53ef\u4ee5\u540c\u6642\u53c3\u8003\u5927\u91cf\u8207\u8fd1\u671f\u76f8\u95dc\u7684\u7db2\u9801\u63d0\u9ad8\u53ef\u4fe1\u5ea6\uff0c\u6240\u4ee5\u7576\u6211\u5011\u4ee5\u5546\u5bb6\u540d\u7a31 \u7684 Snippet \u8a13\u7df4\u51fa CRF \u6a21\u578b\u5f8c\uff0c\u5c31\u76f4\u63a5\u4ee5\u67d0\u5730\u5740\u70ba\u95dc\u9375\u5b57\u6240\u5f97\u5230\u7684\u6240\u6709 Snippets \u4e2d\uff0c\u51fa \u73fe\u6700\u591a\u6b21\u7684\u5546\u5bb6\u540d\u7a31\u548c\u8a72\u5730\u5740\u9032\u884c\u914d\u5c0d\u3002 4. \u5be6\u9a57 \u5be6\u9a57 \u5be6\u9a57 \u5be6\u9a57 \u56e0\u70ba\u672c\u7814\u7a76\u662f\u5148\u9032\u884c\u5546\u5bb6\u540d\u7a31\u8fa8\u8a8d\uff0c\u518d\u5c07\u5730\u5740\u8207\u5df2\u77e5\u7684\u5546\u5bb6\u540d\u7a31\u9032\u884c\u914d\u5c0d\uff0c\u6240\u4ee5\u5be6\u9a57\u90e8 \u4efd\u4e5f\u4f9d\u7167\u9019\u5169\u500b\u968e\u6bb5\u4f86\u9032\u884c\u3002\u7b2c\u4e00\u968e\u6bb5\u7684\u5be6\u9a57\u70ba\u5546\u5bb6\u540d\u7a31\u8fa8\u8b58\u7387\uff0c\u8cc7\u6599\u4f86\u6e90\u6709\u5169\u7a2e\uff0c\u7b2c \u4e00\u7a2e\u662f\u4ee5[9]\u6240\u53d6\u5f97\u7684\u7d04 50 \u842c\u500b\u53ef\u80fd\u542b\u6709\u5730\u5740\u7684\u7db2\u9801\u7576\u4f5c\u539f\u59cb\u8cc7\u6599\uff0c\u5728\u7d93\u904e\u524d\u8655\u7406\u5f8c\uff0c \u2026 Node List by TEX Missed Entity Extracted Entity \u2026 \u8868 \u8868 \u8868 \u8868 2. \u4ee5\u500b\u5225\u5b8c\u6574\u7db2\u9801\u70ba\u8cc7\u6599\u4f86\u6e90\u7684\u8a13\u7df4\u8a9e\u6599\u8207\u6e2c\u8a66\u8cc7\u6599 \u4ee5\u500b\u5225\u5b8c\u6574\u7db2\u9801\u70ba\u8cc7\u6599\u4f86\u6e90\u7684\u8a13\u7df4\u8a9e\u6599\u8207\u6e2c\u8a66\u8cc7\u6599 \u4ee5\u500b\u5225\u5b8c\u6574\u7db2\u9801\u70ba\u8cc7\u6599\u4f86\u6e90\u7684\u8a13\u7df4\u8a9e\u6599\u8207\u6e2c\u8a66\u8cc7\u6599 \u4ee5\u500b\u5225\u5b8c\u6574\u7db2\u9801\u70ba\u8cc7\u6599\u4f86\u6e90\u7684\u8a13\u7df4\u8a9e\u6599\u8207\u6e2c\u8a66\u8cc7\u6599 Training Corpus Testing Data Raw Preprocessing FreeText Foot Detail List Sum # Sites -13,224 100 100 11 100 311 # Web Pages 508,038 396,093 100 100 110 100 410 # Addresses 272,987 190,180 219 156 467 807 1,649 # Stores --1,841 1,975 3,855 2,786 10,457 \u7b2c\u4e8c\u7a2e\u8cc7\u6599\u4f86\u6e90\u662f\u4f7f\u7528 Google \u641c\u5c0b\u5f15\u64ce\u6240\u53d6\u5f97\u7684\u7db2\u9801\u5167\u5bb9\u7247\u6bb5(Snippets\uff0c\u8acb\u53c3\u8003 \u8868 3)\uff0c\u5728\u8a13\u7df4\u8cc7\u6599\u7684\u90e8\u4efd\uff0c\u6211\u5011\u4ee5 11,138 \u7b46\u5546\u5bb6\u540d\u7a31\u9032\u884c\u67e5\u8a62\uff0c\u4ee5\u81ea\u52d5\u6a19\u8a18\u7684\u65b9\u5f0f\u7522 \u751f\u4e86\u5169\u7a2e\u8a13\u7df4\u8cc7\u6599\uff1aSnippetUniLabeling \u548c SnippetFullLabeling\uff0c\u5728 SnippetsUniLabeling \u4e2d\uff0c\u6211\u5011\u50c5\u4ee5\u95dc\u9375\u5b57\u7684\u5546\u5bb6\u540d\u7a31\u5c0d Snippets \u4e2d\u7684\u53e5\u5b50\u9032\u884c\u6a19\u8a18\uff0c\u5171\u6a19\u8a18\u4e86 222,121 \u500b\u5546 \u5bb6\u540d\u7a31\uff0c\u800c SnippetsFullLabeling \u4e2d\uff0c\u5247\u662f\u4ee5\u6240\u6709\u5df2\u77e5\u7684\u5546\u5bb6\u540d\u7a31\u5c0d Snippets \u4e2d\u6240\u6709\u53e5\u5b50 \u9032\u884c\u6a19\u8a18\uff0c\u5171\u6a19\u8a18\u4e86 390,113 \u500b\u5546\u5bb6\u540d\u7a31\uff0c\u85c9\u7531\u4e0d\u540c\u7684\u6a19\u8a18\u65b9\u5f0f\u7522\u751f\u4e0d\u540c\u7a0b\u5ea6\u7684\u96dc\u8a0a\uff0c \u4ee5\u4e86\u89e3\u96dc\u8a0a\u5c0d\u8fa8\u8b58\u7387\u7684\u5f71\u97ff\u3002\u5728\u6e2c\u8a66\u8cc7\u6599\u7684\u90e8\u4efd\u5247\u4ee5 6,963 \u7b46\u5730\u5740\u70ba\u95dc\u9375\u5b57\uff0c\u6536\u96c6\u6bcf\u7b46 \u5730\u5740\u6392\u540d\u524d 20 \u7684\u641c\u5c0b\u7d50\u679c(Snippets)\uff0c\u4ee5\u81ea\u52d5\u6a19\u8a18\u7684\u7b54\u6848\u9032\u884c\u6700\u5f8c NER \u6548\u80fd\u8a55\u4f30\u3002\u6700 \u5f8c\u518d\u5c0d\u5169\u985e\u8cc7\u6599\u9032\u884c\u4ea4\u53c9\u6e2c\u8a66\u3002 \u8868 \u8868 \u8868 \u8868 3. \u4ee5 \u4ee5 \u4ee5 \u4ee5 Search Snippets \u70ba\u8cc7\u6599\u4f86\u6e90\u7684\u8a13\u7df4\u8cc7\u6599\u8207\u6e2c\u8a66\u8cc7\u6599 \u70ba\u8cc7\u6599\u4f86\u6e90\u7684\u8a13\u7df4\u8cc7\u6599\u8207\u6e2c\u8a66\u8cc7\u6599 \u70ba\u8cc7\u6599\u4f86\u6e90\u7684\u8a13\u7df4\u8cc7\u6599\u8207\u6e2c\u8a66\u8cc7\u6599 \u70ba\u8cc7\u6599\u4f86\u6e90\u7684\u8a13\u7df4\u8cc7\u6599\u8207\u6e2c\u8a66\u8cc7\u6599 Training Data Testing Data # of Store Queries Tag Stores # of Address Queries Stores (Auto Labeling) Snippet Uni Labeling 11,138 222,121 6,963 70,449 Snippet Full Labeling 11,138 390,113 6,963 70,449 \u7b2c\u4e8c\u968e\u6bb5\u70ba\u5730\u5740\u8207\u5546\u5bb6\u540d\u7a31\u914d\u5c0d\u7684\u6b63\u78ba\u7387\uff0c\u91dd\u5c0d\u4e0d\u540c\u8cc7\u6599\u4f86\u6e90\u4ee5\u5404\u81ea\u7684\u65b9\u5f0f\u9032\u884c\u914d \u5c0d\uff0c\u7b2c\u4e00\u7a2e\u662f\u91dd\u5c0d\u4e0d\u540c\u7db2\u9801\u985e\u5225\u4ee5\u5404\u81ea\u7684\u555f\u767c\u5f0f(heuristic)\u898f\u5247\u9032\u884c\u914d\u5c0d\uff0c\u7b2c\u4e8c\u7a2e\u662f\u4ee5 Snippets \u4e2d\u5404\u5546\u5bb6\u540d\u7a31\u7684\u6700\u9ad8\u51fa\u73fe\u6b21\u6578\u9032\u884c\u914d\u5c0d\u3002 \u202b,\u0740\u0748\u0729\u202c \u0735\u202b\u0743\u073d\u0736\u074f\u0755\u202c\u6bd4\u5c0d\u5206\u6578 \u202b\u074a\u0745\u074f\u0745\u073f\u0741\u074e\u0732\u202c \u0d4c \u0735\u202b\u0743\u073d\u0736\u074f\u0755\u202c\u8fa8\u8b58\u51fa\u7684\u6240\u6709\u5546\u5bb6\u540d\u7a31\u8207 \u0734\u0741\u073f\u073d\u0748\u0748 \u0d4c \u0735\u202b\u0743\u073d\u0736\u074f\u0755\u202c\u8fa8\u8b58\u51fa\u7684\u6240\u6709\u5546\u5bb6\u540d\u7a31\u8207 \u6b64\u7a2e\u8a55\u4f30\u65b9\u5f0f\uff0c\u53ef\u4ee5\u89e3\u6c7a\u7576 CRF \u5b9a\u662f\u5426\u5c6c\u65bc\u5546\u5bb6\u540d\u7a31\u7684\u4e00\u90e8\u5206\u7684\u554f\u984c 4.1 \u5546\u5bb6\u540d\u7a31\u8fa8\u8b58\u7387 \u5546\u5bb6\u540d\u7a31\u8fa8\u8b58\u7387 \u5546\u5bb6\u540d\u7a31\u8fa8\u8b58\u7387 \u6bd4\u5c0d\u5206\u6578 \u0d4c \u1250 1, \u8868 \u8868 \u8868 \u8868 4. \u4ea4\u53c9\u6e2c\u8a66 \u4ea4\u53c9\u6e2c\u8a66 \u4ea4\u53c9\u6e2c\u8a66 \u4ea4\u53c9\u6e2c\u8a66 \u0745\u0742 \u202b\u0740\u0748\u0729\u202a\u0736\u073d\u0743\u5305\u542b\u202c\u074f\u0755\u0735\u202c \u202b\u0744\u0750\u0743\u074a\u0741\u072e\u0727\u0730\u0743\u073d\u0736\u202c \u202b\u0744\u0750\u0743\u074a\u0741\u072e\u0727\u0730\u0740\u0748\u0729\u202c , \u0745\u0742 \u202b\u0743\u073d\u0736\u074f\u0755\u0735\u202a\u0748\u0740\u5305\u542b\u202c\u0729\u202c \u8fa8\u8b58\u51fa\u7684\u6240\u6709\u5546\u5bb6\u540d\u7a31\u8207\u202b\u0740\u0748\u0729\u202c\u9032\u884c\u6bd4\u5c0d\u7684\u5206\u6578\u7e3d\u548c \u0735\u202b\u0743\u073d\u0736\u074f\u0755\u202c\u6240\u8fa8\u8b58\u51fa\u7684\u6240\u6709\u5546\u5bb6\u540d\u7a31\u6578\u91cf \u8fa8\u8b58\u51fa\u7684\u6240\u6709\u5546\u5bb6\u540d\u7a31\u8207\u202b\u0740\u0748\u0729\u202c\u9032\u884c\u6bd4\u5c0d\u7684\u5206\u6578\u7e3d\u548c \u4eba\u5de5\u6a19\u8a18\u7684\u6240\u6709\u5546\u5bb6\u540d\u7a31\u6578\u91cf CRF \u8fa8\u8b58\u51fa\u7684\u5546\u5bb6\u540d\u7a31\u908a\u754c\u5305\u542b\u5730\u540d\u3001\u767e\u5e74\u8001\u5e97\u2026\u7b49\u96e3\u4ee5\u5224 \u7b49\u96e3\u4ee5\u5224 TrainSet1 TrainSet2 TrainSet3 TrainSet4 TrainSet5 0 20,000 40,000 60,000 80,000 100,000 120,000 140,000 160,000 180,000 200,000 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 Snippet NER Whole Page Search Snippets Whole Page Model 0.305 0.473 Snippets Model (Full Labeling) 0.310 0.791 \u7b2c\u4e09\uff0c\u7576\u6211\u5011\u5229\u7528\u5df2\u77e5\u7684\u5546\u5bb6\u540d\u7a31\u9032\u884c\u6a19\u8a18\u6642\uff0c\u9019\u4e9b\u5df2\u77e5\u8cc7\u6599\u53ef\u80fd\u5b58\u5728\u4e0d\u6b63\u78ba\u3001\u4e0d\u9f4a\u5168 \u6216\u662f\u6b67\u7fa9\u6027\u7b49\u554f\u984c\uff0c\u9020\u6210\u81ea\u52d5\u6a19\u8a18\u7684\u7b2c\u4e00\u6b21\u932f\u8aa4\uff0c\u800c\u4e14\u64c1\u6709\u5927\u91cf\u7684\u5df2\u77e5\u540d\u7a31\u548c\u7db2\u9801\u6642\uff0c \u7121\u6cd5\u5c0d\u6240\u6709\u7db2\u9801\u4e2d\u7684\u6240\u6709\u5b57\u4e32\u90fd\u6aa2\u67e5\u662f\u5426\u5b58\u5728\u5546\u5bb6\u7d44\u7e54\u540d\u7a31\uff0c\u53ea\u80fd\u5229\u7528\u5730\u5740\u67e5\u8a62\u662f\u5426\u5b58 \u5728\u5c0d\u61c9\u7684\u5546\u5bb6\u540d\u7a31\uff0c\u9020\u6210\u7b2c\u4e8c\u6b21\u7684\u932f\u8aa4\uff0c\u82e5\u8981\u5728\u5408\u7406\u7684\u57f7\u884c\u6642\u9593\u5167\u89e3\u6c7a\u6b64\u554f\u984c\uff0c\u53ef\u80fd\u9700 F1 (Auto Labeling, Score Match) \u8981\u4f7f\u7528 Hadoop \u6216\u662f\u5176\u4ed6\u5206\u6563\u5f0f\u7cfb\u7d71\uff0c\u4ee5\u6240\u6709\u5df2\u77e5\u7684\u5546\u5bb6\u540d\u7a31\u9032\u884c\u6a19\u8a18\u4ee5\u63d0\u6607\u6a19\u8a18\u54c1\u8cea\u3002 \u5b9a\u662f\u5426\u5c6c\u65bc\u5546\u5bb6\u540d\u7a31\u7684\u4e00\u90e8\u5206\u7684\u554f\u984c\u3002 Sentences 7000 10000 30000 90000 190000 UniTagStores 10902 15669 45425 115014 222121 4.2 \u5730\u5740 \u5730\u5740 \u5730\u5740 \u5730\u5740-\u5546\u5bb6\u540d\u7a31 \u5546\u5bb6\u540d\u7a31 \u5546\u5bb6\u540d\u7a31 \u5546\u5bb6\u540d\u7a31 \u5339\u914d\u6b63\u78ba\u7387 \u5339\u914d\u6b63\u78ba\u7387 \u5339\u914d\u6b63\u78ba\u7387 \u5716 \u5716 \u5716 \u5716 7. \u4ee5\u5b8c\u6574\u7db2\u9801\u70ba\u8cc7\u6599\u4f86\u6e90\u7684\u914d\u5c0d\u6b63\u78ba\u7387 \u4ee5\u5b8c\u6574\u7db2\u9801\u70ba\u8cc7\u6599\u4f86\u6e90\u7684\u914d\u5c0d\u6b63\u78ba\u7387 \u4ee5\u5b8c\u6574\u7db2\u9801\u70ba\u8cc7\u6599\u4f86\u6e90\u7684\u914d\u5c0d\u6b63\u78ba\u7387 \u4ee5\u5b8c\u6574\u7db2\u9801\u70ba\u8cc7\u6599\u4f86\u6e90\u7684\u914d\u5c0d\u6b63\u78ba\u7387( ( ( (\u8a13\u7df4\u6a23\u672c\u6578 \u8a13\u7df4\u6a23\u672c\u6578 \u8a13\u7df4\u6a23\u672c\u6578 \u8a13\u7df4\u6a23\u672c\u6578\uff1a \uff1a \uff1a \uff1a4,398) ) ) ) \u5339\u914d\u6b63\u78ba\u7387 \u5546\u5bb6\u540d\u7a31\u8fa8\u8b58\u7387 \u6211\u5011\u9996\u5148\u4ee5\u500b\u5225\u5b8c\u6574\u7db2\u9801\u70ba\u8cc7\u6599\u4f86\u6e90 \u5011\u4ee5 Snippet \u70ba\u8cc7\u6599\u4f86\u6e90\uff0c\u5206\u5225\u5be6\u9a57\u4e86 \u52d5\u6a19\u8a18\u4e2d\uff0c\u96dc\u8a0a\u5c0d\u8fa8\u8b58\u6548\u80fd\u7684\u5f71\u97ff \u89c0\u5bdf\u4e0d\u540c\u4f86\u6e90\u7684\u8a13\u7df4\u8cc7\u6599\u6240\u8a13\u7df4\u51fa\u7684\u6a21\u578b \u7a76\u7684\u5728\u5546\u5bb6\u8fa8\u8b58\u90e8\u4efd\u7684\u6700\u5f8c\u8f38\u51fa \u5716 \u5716 \u5716 \u5716 4. \u5b8c\u6574\u7db2\u9801\u4e2d \u5b8c\u6574\u7db2\u9801\u4e2d \u5b8c\u6574\u7db2\u9801\u4e2d \u5b8c\u6574\u7db2\u9801\u4e2d \u5b8c\u6574\u7db2\u9801\u70ba\u8cc7\u6599\u4f86\u6e90\uff0c\u5be6\u9a57\u4e86\u8a13\u7df4\u8cc7\u6599\u6578\u91cf\u5c0d\u8fa8\u8b58\u6548\u80fd\u7684\u5f71\u97ff\u3002 \u5206\u5225\u5be6\u9a57\u4e86 Uni-Labeling \u548c Full-Labeling \u7684\u6548\u80fd\uff0c\u4ee5\u4e86\u89e3\u5728\u81ea \u96dc\u8a0a\u5c0d\u8fa8\u8b58\u6548\u80fd\u7684\u5f71\u97ff\uff0c\u7136\u5f8c\u5c0d\u5169\u7a2e\u8cc7\u6599\u4f86\u6e90\u6240\u8a13\u7df4\u51fa\u7684\u6a21\u578b\u9032\u884c\u4ea4\u53c9\u6e2c\u8a66 \u89c0\u5bdf\u4e0d\u540c\u4f86\u6e90\u7684\u8a13\u7df4\u8cc7\u6599\u6240\u8a13\u7df4\u51fa\u7684\u6a21\u578b\uff0c\u61c9\u7528\u5728\u4e0d\u540c\u6e2c\u8a66\u8cc7\u6599\u6642\u7684\u8868\u73fe\u3002\u6700\u5f8c\u662f\u672c\u7814 \u7a76\u7684\u5728\u5546\u5bb6\u8fa8\u8b58\u90e8\u4efd\u7684\u6700\u5f8c\u8f38\u51fa\u3002 \u5b8c\u6574\u7db2\u9801\u4e2d \u5b8c\u6574\u7db2\u9801\u4e2d \u5b8c\u6574\u7db2\u9801\u4e2d \u5b8c\u6574\u7db2\u9801\u4e2d\uff0c \uff0c \uff0c \uff0c\u8a13\u7df4\u8cc7\u6599\u6578\u91cf\u5c0d \u8a13\u7df4\u8cc7\u6599\u6578\u91cf\u5c0d \u8a13\u7df4\u8cc7\u6599\u6578\u91cf\u5c0d \u8a13\u7df4\u8cc7\u6599\u6578\u91cf\u5c0d F1 \u7684\u5f71\u97ff \u7684\u5f71\u97ff \u7684\u5f71\u97ff \u4e26\u4e0d\u662f\u4e00\u500b\u597d\u7684\u65b9\u5f0f\uff0c\u6703\u5927\u5e45\u5ea6\u53d7\u5230\u96dc\u8a0a\u8207\u6a23\u672c\u6578\u9650\u5236\u7684\u5f71\u97ff\u3002 \u7684\u5f71\u97ff FullTagStores 15289 21642 64379 189803 390113 \u7684\u5546\u5bb6\u540d\u7a31\uff0c\u518d\u4ee5\u5546\u5bb6\u540d\u7a31\u7576\u4f5c\u641c\u5c0b\u5f15\u64ce\u7684\u95dc\u9375\u5b57\u53d6\u5f97 POI \u7684\u76f8\u95dc\u8cc7\u8a0a\uff0c\u5c31\u53ef\u4ee5\u6210\u529f\u5efa \u5716 6 \u662f SnippetFullLabeling \u4ee5\u4e0d\u540c\u8a13\u7df4\u8cc7\u6599\u6240\u8a13\u7df4\u51fa\u7684\u6a21\u578b\u5c0d\u914d\u5c0d\u6b63\u78ba\u7387\u7684\u5f71\u97ff\uff0c\u5716\u4e2d \u3002\u63a5\u8457\u6211 UniLabel 0.134 0.243 0.176 0.175 0.086 \u7acb POI \u8cc7\u6599\u5eab\u3002 \u986f\u793a\u7576 NER \u7684\u6548\u80fd\u5927\u5e45\u63d0\u9ad8\u6642\uff0cMatch \u96d6\u7136\u8ddf\u8457\u4e0a\u5347\uff0c\u4f46\u50c5\u6709\u5fae\u5e45\u6210\u9577\u3002\u800c\u5728\u5b8c\u6574\u7db2\u9801 \u4ee5\u4e86\u89e3\u5728\u81ea FullLabel 0.564 0.624 0.679 0.740 0.791 \u70ba\u8cc7\u6599\u7684\u5be6\u9a57\u4e2d\uff0c\u96d6\u7136\u6211\u5011\u7121\u6cd5\u8fa8\u8a8d\u51fa\u6240\u6709\u7684\u5546\u5bb6\u540d\u7a31\uff0c\u4f46\u7d93\u7531\u555f\u767c\u5f0f(heuristic)\u7684\u914d \u904e\u53bb\u547d\u540d\u5be6\u9ad4\u4ee5\u65b0\u805e\u5831\u5c0e\u4e2d\u7684\u4eba\u540d\u3001\u5730\u540d\u3001\u7d44\u7e54\u540d\u64f7\u53d6\u70ba\u4e3b\u8ef8\uff0c\u76ee\u7684\u5728\u4e86\u89e3\u65b0\u805e\u4e2d \u7136\u5f8c\u5c0d\u5169\u7a2e\u8cc7\u6599\u4f86\u6e90\u6240\u8a13\u7df4\u51fa\u7684\u6a21\u578b\u9032\u884c\u4ea4\u53c9\u6e2c\u8a66\uff0c \u6700\u5f8c\u662f\u672c\u7814 \u5716 \u5716 \u5716 \u5716 5. \u4ee5 \u4ee5 \u4ee5 \u4ee5 Snippets \u70ba\u8cc7\u6599\u4f86\u6e90 \u70ba\u8cc7\u6599\u4f86\u6e90 \u70ba\u8cc7\u6599\u4f86\u6e90 \u70ba\u8cc7\u6599\u4f86\u6e90\uff0c \uff0c \uff0c \u96dc\u8a0a\u8207\u8a13\u7df4\u8cc7\u6599\u6578\u91cf\u5c0d\u6548\u80fd\u7684\u5f71\u97ff \u5c0d\u898f\u5247\uff0c\u53ef\u4ee5\u63d0\u6607\u914d\u5c0d\u7684\u6b63\u78ba\u7387\u3002\u5716 7 \u662f\u4ee5\u5b8c\u6574\u7db2\u9801\u70ba\u8cc7\u6599\u4f86\u6e90\uff0c\u5730\u5740-\u5546\u5bb6\u540d\u7a31\u914d\u5c0d\u6b63 \u7684\u4e8b\u4ef6\uff0c\u4f46\u5c0d\u65bc\u7db2\u8def\u4e0a\u7684\u8208\u8da3\u9ede POI \u7684\u6536\u96c6\u8f03\u5c11\u8457\u58a8\u3002\u672c\u7814\u7a76\u8a66\u5716\u76f4\u63a5\u5c0d\u6574\u500b\u7db2\u9801\u9032\u884c \uff0c\u96dc\u8a0a\u8207\u8a13\u7df4\u8cc7\u6599\u6578\u91cf\u5c0d\u6548\u80fd\u7684\u5f71\u97ff \u96dc\u8a0a\u8207\u8a13\u7df4\u8cc7\u6599\u6578\u91cf\u5c0d\u6548\u80fd\u7684\u5f71\u97ff \u96dc\u8a0a\u8207\u8a13\u7df4\u8cc7\u6599\u6578\u91cf\u5c0d\u6548\u80fd\u7684\u5f71\u97ff \u78ba\u7387\u7684\u5be6\u9a57\u7d50\u679c\u3002\u4ee5\u55ae\u4e00\u985e\u5225\u4f86\u770b\uff0c\u5728\u6df1\u5ea6\u8cc7\u8a0a\u7db2\u9801\u7684\u5be6\u9a57\u4e2d\uff0c\u5229\u7528\u6587\u4ef6\u7269\u4ef6\u6a39\u8def\u5f91\u7684 \u8fa8\u8a8d\uff0c\u96d6\u7136\u53d7\u9650\u65bc\u6a19\u8a18\u7684\u4e0d\u5168\uff0c\u5728\u547d\u540d\u5be6\u9ad4\u8fa8\u8a8d\u7684\u6548\u679c\u4e26\u4e0d\u597d\uff0c\u4f46\u662f\u5728\u6df1\u5ea6\u8cc7\u8a0a\u7db2\u9801(\u4e5f \u5728 Snippets \u65b9\u9762\u7684\u5be6\u9a57\uff0c\u6211\u5011\u6e2c\u8a66\u4e86\u8a13\u7df4\u8cc7\u6599\u6578\u91cf\u8207\u6a19\u8a18\u54c1\u8cea\u5c0d\u8fa8\u8b58\u6548\u80fd\u7684\u5f71\u97ff\uff0c \u76f8\u4f3c\u5ea6\u5f8c\uff0c\u53ef\u4ee5\u5c07\u914d\u5c0d\u6e96\u78ba\u7387\u63d0\u6607\u81f3 0.951\uff0c\u5e73\u5747\u6b63\u78ba\u7387\u5247\u70ba 0.573\u3002 \u662f\u542b\u6709\u6700\u591a\u5730\u5740\u7684\u7db2\u9801\u985e\u578b)\u7684\u5730\u5740-\u5546\u5bb6\u540d\u7a31\u914d\u5c0d\u4e2d\uff0c\u5229\u7528\u7db2\u9801\u9593\u7684\u76f8\u4f3c\u5ea6\u53ef\u4ee5\u53d6\u5f97 \u4ee5\u4e86\u89e3\u5728\u81ea\u52d5\u6a19\u8a18\u4e2d\uff0c\u96dc\u8a0a\u5c0d\u8fa8\u8b58\u6548\u80fd\u7684\u5f71\u97ff\u3002\u5982\u5716 5 \u6240\u793a\uff0c\u5728 UniLabeling \u6a21\u578b\u4e2d\uff0c\u7576 0.9514 \u7684\u6e96\u78ba\u7387\uff0c\u800c\u5e73\u5747\u6b63\u78ba\u7387\u5247\u70ba 0.5726\u3002\u800c Google Snippets \u7684\u65b9\u6cd5\u4e2d\uff0cNER \u6548\u80fd\u6700 \u8cc7\u6599\u589e\u52a0\u6642\uff0c\u8a13\u7df4\u8cc7\u6599\u542b\u6709\u7684\u96dc\u8a0a (\u6a19\u8a18\u4e0d\u5b8c\u5168) \u66f4\u70ba\u56b4\u91cd\uff0c\u4f7f\u5f97\u6548\u80fd\u4e0b\u964d\uff1b\u800c FullLabeling Sentences, Full Labeling Performance \u9ad8\u70ba 0.791\uff0c\u914d\u5c0d\u6b63\u78ba\u7387\u6700\u9ad8\u70ba 0.632\u3002 \u6a21\u578b\u56e0\u70ba\u4f7f\u7528\u6240\u6709\u7684\u5546\u5bb6\u540d\u7a31\u9032\u884c\u6a19\u8a18\uff0c\u6240\u4ee5\u96dc\u8a0a\u5927\u5e45\u6e1b\u5c11\uff0c\u5728\u8cc7\u6599\u589e\u52a0\u7684\u60c5\u6cc1\u4e0b\u53ef\u5927 \u5e45\u5ea6\u63d0\u6607\u6548\u80fd\uff0cFullLabeling \u6a21\u578b\u7684\u6548\u80fd\u6700\u9ad8\u70ba 0.791\u3002 \u4e0d\u904e\u5728 Search Snippets \u7684\u6e2c\u8a66\u8cc7\u6599\u4e2d\u4e26\u975e\u4f7f\u7528\u4eba\u5de5\u6a19\u8a18\u7684\u7b54\u6848\u9032\u884c\u9a57\u8b49\uff0c\u800c\u662f\u4f7f\u7528 \u81ea\u52d5\u6a19\u8a18\u7684\u7b54\u6848\u3002\u70ba\u4e86\u4e86\u89e3\u4f7f\u7528\u67d0\u4e00\u8a9e\u6599\u6240\u8a13\u7df4\u51fa\u7684\u6a21\u578b\u662f\u5426\u80fd\u61c9\u7528\u5728\u53e6\u4e00\u4e0d\u540c\u8a9e\u6599\u7684 \u6e2c\u8a66\u8cc7\u6599\uff0c\u6211\u5011\u5c0d\u500b\u5225\u7db2\u9801\u8207 Search Snippets \u9032\u884c\u4e86\u4ea4\u53c9\u6e2c\u8a66\uff0c\u6211\u5011\u4ee5\u5b8c\u6574\u7db2\u9801\u70ba\u8a13\u7df4 \u8cc7\u6599\u6240\u8a13\u7df4\u51fa\u7684\u6a21\u578b\u5c0d Snippet \u7684\u6e2c\u8a66\u8cc7\u6599\u9032\u884c\u6e2c\u8a66\uff0c\u540c\u6642\u4e5f\u4ee5 Snippet \u4e2d\u5169\u7a2e\u8a13\u7df4\u8cc7\u6599 \u6240\u8a13\u7df4\u51fa\u7684\u6a21\u578b\u5c0d 410 \u500b\u7db2\u9801\u9032\u884c\u6e2c\u8a66\u3002 \u5be6\u9a57\u7d50\u679c\u5982\u8868 4 \u6240\u793a\uff0c\u5716\u4e2d\u986f\u793a\u4e0d\u8ad6\u662f\u4f55\u7a2e\u6e2c\u8a66\u8cc7\u6599\u985e\u578b\uff0c\u7531 SnippetFullLabeling \u6240\u8a13\u7df4\u51fa\u7684\u6a21\u578b\u90fd\u5177\u6709\u6bd4\u8f03\u597d\u7684\u8fa8\u8b58\u6548\u679c\uff0c\u751a\u81f3\u6bd4\u500b\u5225\u5b8c\u6574\u7db2\u9801\u6240\u8a13\u7df4\u51fa\u7684\u6a21\u578b\u7528\u5728\u6e2c \u8a66\u540c\u985e\u8cc7\u6599\u9084\u8981\u9ad8\uff0c\u53ef\u898b\u5728\u81ea\u52d5\u6a19\u8a18\u4e2d\uff0c\u53ea\u4f7f\u7528\u90e8\u4efd\u5df2\u77e5\u7684\u5546\u5bb6\u540d\u7a31\u6240\u7522\u751f\u7684\u8a13\u7df4\u8cc7\u6599\uff0c 190000 NER, 0.791 Match, 0.632 0 20000 40000 60000 80000 100000 120000 140000 160000 180000 200000 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 TrainSet1 TrainSet2 TrainSet3 TrainSet4 TrainSet5 \u5728\u5be6\u9a57\u904e\u7a0b\u4e2d\uff0c\u6211\u5011\u767c\u73fe\u555f\u767c\u5f0f\u7684\u914d\u5c0d\u898f\u5247\u96d6\u7136\u53ef\u4ee5\u63d0\u6607 Detail Pages \u7684\u914d\u5c0d\u6b63\u78ba \u7387\uff0c\u4f46\u662f\u5176\u9918\u985e\u578b\u4f9d\u7136\u5f88\u4ef0\u8cf4\u547d\u540d\u5be6\u9ad4\u7684\u8fa8\u8a8d\u7d50\u679c\u3002\u82e5\u8981\u66f4\u9032\u4e00\u6b65\u63d0\u6607\u5546\u5bb6\u540d\u7a31\u7684\u8fa8\u8b58 \u7d50\u679c\uff0c\u6211\u5011\u89ba\u5f97\u53ef\u4ee5\u671d\u5169\u500b\u65b9\u5411\u9032\u884c\uff0c\u7b2c\u4e00\uff0c\u5fc5\u9808\u5c07\u5916\u90e8\u7279\u5fb5\u52a0\u5165\u7279\u5fb5\u77e9\u9663\u4e2d\uff0c\u56e0\u70ba\u5916 \u90e8\u7279\u5fb5\u96d6\u7136\u4e0d\u80fd\u660e\u78ba\u6307\u51fa\u5546\u5bb6\u540d\u7a31\uff0c\u4f46\u662f\u4f9d\u7136\u662f\u9032\u884c\u63a8\u6e2c\u7684\u91cd\u8981\u63d0\u793a\uff0c\u5728\u672a\u4f86\u6211\u5011\u5e0c\u671b \u80fd\u628a\u5916\u90e8\u7279\u5fb5\u548c\u8a5e\u983b\u52a0\u5165 CRF\uff0c\u63d0\u6607\u5546\u5bb6\u540d\u7a31\u7684\u8fa8\u8b58\u6548\u679c\u3002\u7b2c\u4e09\u662f\u5229\u7528\u5206\u6563\u5f0f\u7cfb\u7d71\u7684\u901f \u5ea6\uff0c\u5b8c\u6574\u6a19\u8a18\u5df2\u77e5(\u5927\u91cf)\u5df2\u77e5\u7684\u5546\u5bb6\u540d\u7a31\uff0c\u89e3\u6c7a\u81ea\u52d5\u6a19\u8a18\u7522\u751f\u7684\u8a13\u7df4\u8cc7\u6599\u54c1\u8cea\u4e0d\u4f73\u7684 \u554f\u984c\u3002 F1 (Auto Labeling, Score Match) \u53c3\u8003\u6587\u737b \u53c3\u8003\u6587\u737b \u53c3\u8003\u6587\u737b \u53c3\u8003\u6587\u737b</td></tr></table>",
                "text": "\u6458\u8981 \u6458\u8981 \u6458\u8981 \u6458\u8981\u884c\u52d5\u5316\u662f 2014 \u7684\u8da8\u52e2\u4e4b\u4e00\uff0c\u800c\u9069\u5730\u6027\u670d\u52d9(Location-based Service)\u5728\u9019\u6ce2 Chuang et al., 2014)\u5c0d\u65bc\u5305\u542b\u5730\u5740\u7db2\u9801\u63d0\u51fa\u4ee5 \u5ee3\u5ea6\u512a\u5148\u641c\u5c0b\u3001\u9ec3\u9801\u722c\u87f2\u3001\u8207\u5730\u5740\u6a23\u7248\u67e5\u8a62\u4e09\u7a2e\u6293\u53d6\u7a0b\u5f0f\uff0c\u4e26\u5229\u7528 Chang \u7b49\u4eba(Chang et al., 2012)\u7684\u5730\u5740\u64f7\u53d6\u7a0b\u5f0f\uff0c\u53d6\u5f97\u5927\u91cf\u4e2d\u6587\u5730\u5740\u3002Li \u8207 Chang(2009)\u4e26\u5b9a\u7fa9\u5730\u5740\u76f8\u95dc\u8cc7\u8a0a\u64f7\u53d6 \u554f\u984c\uff0c\u5e0c\u671b\u85c9\u6b64\u8c50\u5bcc\u6bcf\u500b POI \u7684\u76f8\u95dc\u8cc7\u8a0a\uff0c\u63d0\u9ad8\u5730\u7406\u8cc7\u8a0a\u6aa2\u7d22(Geographical Information Retrieval, GIR)\u7684\u53ec\u56de\u7387\u3002\u7136\u800c\u4e0d\u8ad6\u662f Li \u8207 Chang(2009)\u6216 Chang \u7b49\u4eba(Chang et al., 2012) \u6216 Chuang \u7b49\u4eba(Chuang et al., 2014)\u7684\u76f8\u95dc\u8cc7\u8a0a\u64f7\u53d6\u65b9\u6cd5\u90fd\u50c5\u80fd\u5f9e\u591a\u7b46\u5730\u5740\u7db2\u9801\u64f7\u53d6\u8cc7\u8a0a\uff0c Ling \u7b49\u4eba (Ling et al., 2012) \u4ee5\u898f\u5247\u5f0f\u7684\u8fa8\u8a8d\u65b9\u6cd5 (Rule-based Named-Entity Ahlers, D. (2013). Business entity retrieval and data provision for yellow pages by local search. Integrating IR technologies for professional search, ECIR, 2013. Ahlers, D. (2013). Lo major de dos idiomascross-lingual linkage of geotagged Wikipedia articles. Advances in Information Retrieval, 2013, 668-671. Apache Tika (2004). The Apache Software Foundation, [Online]. Available: http://tika.apache.org/.",
                "type_str": "table"
            }
        }
    }
}