| { | |
| "paper_id": "O15-1021", | |
| "header": { | |
| "generated_with": "S2ORC 1.0.0", | |
| "date_generated": "2023-01-19T08:10:14.752539Z" | |
| }, | |
| "title": "", | |
| "authors": [ | |
| { | |
| "first": "Po-Lin", | |
| "middle": [], | |
| "last": "\u9673\u67cf\u9716", | |
| "suffix": "", | |
| "affiliation": { | |
| "laboratory": "", | |
| "institution": "Chaoyang University of Technology", | |
| "location": {} | |
| }, | |
| "email": "" | |
| }, | |
| { | |
| "first": "", | |
| "middle": [], | |
| "last": "Chen", | |
| "suffix": "", | |
| "affiliation": { | |
| "laboratory": "", | |
| "institution": "Chaoyang University of Technology", | |
| "location": {} | |
| }, | |
| "email": "" | |
| }, | |
| { | |
| "first": "Shih-Hung", | |
| "middle": [], | |
| "last": "\u5433\u4e16\u5f18", | |
| "suffix": "", | |
| "affiliation": { | |
| "laboratory": "", | |
| "institution": "Chaoyang University of Technology", | |
| "location": {} | |
| }, | |
| "email": "" | |
| }, | |
| { | |
| "first": "", | |
| "middle": [], | |
| "last": "Wu", | |
| "suffix": "", | |
| "affiliation": { | |
| "laboratory": "", | |
| "institution": "Chaoyang University of Technology", | |
| "location": {} | |
| }, | |
| "email": "" | |
| } | |
| ], | |
| "year": "", | |
| "venue": null, | |
| "identifiers": {}, | |
| "abstract": "", | |
| "pdf_parse": { | |
| "paper_id": "O15-1021", | |
| "_pdf_hash": "", | |
| "abstract": [], | |
| "body_text": [ | |
| { | |
| "text": "EQUATION", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [ | |
| { | |
| "start": 0, | |
| "end": 8, | |
| "text": "EQUATION", | |
| "ref_id": "EQREF", | |
| "raw_str": "( | \u2212 +1 \u22121 ) = C( \u2212 +1 \u22121 ) C( \u2212 +1 \u22121 ) (1) \u5176\u4e2d C \u4ee3\u8868\u67d0\u500b\u5b57 W \u51fa\u73fe\u7684\u983b\uf961\u3002 \u4e00\u500b\uf906\u5b50\u662f\u7531 n \u500b\u5b57\u6240\u7d44\u6210\uff0c\u6240\u4ee5\u4e00\u6574\u500b\uf906\u5b50\u7684\u6a5f\uf961\u5c31\u53ef\u4ee5\u8a08\u7b97\u5982\u516c\u5f0f(2)\uff1a ( 1 ) = P( 1 , 2 , \u2026 , )", | |
| "eq_num": "(2)" | |
| } | |
| ], | |
| "section": "", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "EQUATION", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [ | |
| { | |
| "start": 0, | |
| "end": 8, | |
| "text": "EQUATION", | |
| "ref_id": "EQREF", | |
| "raw_str": "( 1 ) = ( 1 ) ( 2 | 1 ) ( 3 | 1 2 ) * \u2026 * ( 3 | 1 \u22121 ) = ( 1 )\u220f =2 \ufffd \ufffd 1 \u22121 \ufffd (3) \u516c\u5f0f(3)\u6539\u6210(4)\u662f\u7531\u65bc\u7121\u6cd5\u5f9e\u904e\u53bb\u7684\u8a9e\u6599\u4e2d\u4f86\u505a\u7121\u9650\u5b57\u7684\u9810\u6e2c\uff1a ( | 1 \u22121 ) \u2248 ( | \u2212 +1 \u22121 ) (4) \u4ee3\u8868\u4f9d\u64da\u524d(n-1)\u500b\u5b57\u51fa\u73fe\u7684\u6a5f\uf961\uf92d\u9810\u6e2c\u76ee\u524d\u7b2c n \u500b\u5b57\u6240\u51fa\u73fe\u7684\u6a5f\uf961\uff0c\u800c\u6240\u8b02\u7684 N-gram \u5c31\u662f\u7576 N=2 \u6642\uff0c\u7a31\u70ba bigram\uff0c\u5982\u516c\u5f0f(5)\uff1a ( | \u22121 )", | |
| "eq_num": "(5)" | |
| } | |
| ], | |
| "section": "", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "\u5728\u672c\u5be6\u9a57\u4e2d\u6240\u5efa\uf9f7\u7684\u8a9e\u8a00\u6a21\u578b\u63a1\u53d6 bigram \u4ee5\u53ca unigram \u7684\u6a21\u5f0f\u4ee5\u53ca\u662f\u5426\u5148\u7d93\u904e CKIP[12] \u65b7\u8a5e\uff0c\u7c21\u55ae\uf92d\uf96f bigram \u8a9e\u8a00\u6a21\u578b\u5c31\u662f\u7d71\u8a08\u5b8c\u8a9e\u6599\u4e4b\u5f8c\uff0c\u7d00\uf93f\u8a5e\u5f59\u4e2d\u6bcf\u4e00\u500b\u5b57\u51fa\u73fe\u7684\u689d \u4ef6\u4e0b\uff0c\u4e0b\u4e00\u500b\u5b57\u63a5\u5728\u6b64\u5b57\u5f8c\u9762\u7684\u6a5f\uf961\uff0c\u4e5f\u56e0\u70ba\u4e2d\u6587\u5b57\u4e2d\u5169\u5169\u5b57\u7684\u7d44\u5408\u6bd4\uf9b5\u8f03\u9ad8\uff0c\u56e0\u6b64\u6211 \u5011\u5be6\u9a57\u4f7f\u7528 bigram\u3002\u5982\u5716\u4e8c\u8868\u793a\uff0c\u6b64\u5716\u8209\uf9b5\uf96f\u660e\uff0c\u4ee5\"\u4eca\" \u8207\"\u5929\"\u70ba\uf9b5\uff0c\u7531\"\u4eca \" \u51fa\u73fe\u7684\u60c5\u6cc1\u4e0b\uff0c\u63a8\u6e2c\"\u5929\"\u51fa\u73fe\u7684\u6a5f\uf961\uff0c\u5c31\u7a31\u70ba bigram\uff0c\u540c\uf9e4\"\u5929\"\u8207\"\u7684\";\"\u5929\" \u8207\"\u6c23\"\u4e5f\ufa26\u662f bigram\u2026\u4f9d\u6b64\uf9d0\u63a8\uff0c\u800c\u5982\u679c\u53e5\u5b50\u5148\u7d93\u7531\u65b7\u8a5e\u6211\u5011\u5c31\u6703\u4ee5\u8a5e\u70ba\u55ae\u4f4d\uff0c bigram \u7684\u60c5\u6cc1\u5c31\u6703\u8b8a\u6210\u4ee5\"\u5929\u6c23\"\u8207\"\u771f\u597d\"\u70ba\uf9b5\uff0c\u7531\"\u5929\u6c23\"\u51fa\u73fe\u7684\u60c5\u6cc1\u4e0b\uff0c\u63a8\u6e2c \"\u771f\u597d\"\u51fa\u73fe\u7684\u6a5f\uf961\uff0c\u4e5f\u56e0\u6b64\u6211\u5011\u5c31\u80fd\u5f9e\u8a9e\u8a00\u6a21\u578b\u4e2d\u63a8\u7b97\u51fa\u67d0\u4e00\u500b\uf906\u5b50\u7684\u6a5f\uf961\u3002 \"Entropy\" \u662f\u5f88\u91cd\u8981\u7684\u8a55\u4f30\u6a19\u6e96\u4e4b\u4e00\uff0c\u5b83\u4e5f\u88ab\u5ee3\u6cdb\u7684\u4f7f\u7528\u5728\u6e2c\uf97e\u8cc7\u8a0a\u4e0a [13] \uff0c\"Entropy\" ", | |
| "cite_spans": [ | |
| { | |
| "start": 379, | |
| "end": 383, | |
| "text": "[13]", | |
| "ref_id": "BIBREF12" | |
| } | |
| ], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "EQUATION", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [ | |
| { | |
| "start": 0, | |
| "end": 8, | |
| "text": "EQUATION", | |
| "ref_id": "EQREF", | |
| "raw_str": "\u88ab\u5b9a\u7fa9\u70ba\u4e0b\uf99c\u7684\u5f0f\u5b50(6)\uff1a H(X) = \u2212 \u2211 ( ) log 2 ( ) \u2208 (6) \u5176\u4e2d\u96a8\u6a5f\u8b8a\uf969 X \u6db5\u84cb\u7684\u7bc4\u570d\u5305\u542b\u53ef\u9810\u6e2c\u7684 T \u96c6\u5408(\uf9b5\u5982\u5b57\u6bcd\uff0c\u5b57\u8a5e\u6216\u90e8\u5206\u7684\u8a9e\u97f3) \u3002P(x) \u3001 P'(x)\ufa26\u662f MLE \u6240\u8a08\u7b97\u51fa\uf92d\u7684\u6a5f\uf961\u503c\uff0c\u5be6\u969b\u4f7f\u7528\u6642\u5247\u662f\u5957\u7528\u4e0b\uf99c\u6539\u5beb\u904e\u7684\u516c\u5f0f(7)\uff1a H\u2032(X) = \u2212 \u2211 log 10 \u2032( ) \u2208", | |
| "eq_num": "(" | |
| } | |
| ], | |
| "section": "", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "\ufffd \ufffd \u2212 +1 \u22121 \ufffd = \ufffd \u2212 +1 \u22121 \ufffd\u2212 \ufffd \ufffd \u2212 +1 \u22121 \ufffd\ufffd \u2211 ( \u2212 +1 ) + \u03b3( \u2212 +1 \u22121 ) \ufffd \ufffd \u2212 +2 \u22121 \ufffd (12) \u5176\u4e2dD(c) = \ufffd 0 =0 1 =1 2 =2 3 \u22653 1 =1\u22122 1 1+2 2 * 2 1 2 =1\u22123 1 1+2 2 * 3 2 3 =1\u22124 1 1+2 2 * 4 3", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "Interpolated Kneser-Ney smoothing \u5176\u516c\u5f0f(13)\uff1a", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "3\u3001 Interpolated Kneser-Ney smoothing", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "EQUATION", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [ | |
| { | |
| "start": 0, | |
| "end": 8, | |
| "text": "EQUATION", | |
| "ref_id": "EQREF", | |
| "raw_str": "interpolated ( | \u22121 \u22122 ) =\u22cb trigram ( | \u22121 \u22122 ) + (1 \u2212\u22cb)[ ( | ) + (1 \u2212 ) ( )]", | |
| "eq_num": "(13)" | |
| } | |
| ], | |
| "section": "3\u3001 Interpolated Kneser-Ney smoothing", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "\u672c\u7bc7\u5be6\u9a57\u4e5f\u662f\u4f7f\u7528 Interpolated Kneser-Ney smoothing \u7684\u516c\u5f0f\uff0c\u4f46\u662f\u5be6\u9a57\u662f bigram \u8a9e\u8a00 \u6a21\u578b\uff0c\u56e0\u6b64\u6211\u5011\u5c07\u516c\u5f0f\u6539\u5beb\u6210 (14) : ", | |
| "cite_spans": [ | |
| { | |
| "start": 77, | |
| "end": 81, | |
| "text": "(14)", | |
| "ref_id": "BIBREF13" | |
| } | |
| ], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "3\u3001 Interpolated Kneser-Ney smoothing", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "EQUATION", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [ | |
| { | |
| "start": 0, | |
| "end": 8, | |
| "text": "EQUATION", | |
| "ref_id": "EQREF", | |
| "raw_str": "interpolated ( | \u22121 ) = (1 \u2212\u22cb)[ ( | ) + (1 \u2212 ) ( )", | |
| "eq_num": "(" | |
| } | |
| ], | |
| "section": "3\u3001 Interpolated Kneser-Ney smoothing", | |
| "sec_num": null | |
| } | |
| ], | |
| "back_matter": [], | |
| "bib_entries": { | |
| "BIBREF1": { | |
| "ref_id": "b1", | |
| "title": "Documents Ranking Based on a Hybrid Language Model for Information Retrieval", | |
| "authors": [ | |
| { | |
| "first": "Dequan", | |
| "middle": [], | |
| "last": "Zheng", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Feng", | |
| "middle": [], | |
| "last": "Yu", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Sheng", | |
| "middle": [], | |
| "last": "Tiejun", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "", | |
| "middle": [], | |
| "last": "Li", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2006, | |
| "venue": "IEEE International Conference on Information Acquisition", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "279--283", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Dequan Zheng, Feng Yu, Tiejun, Sheng Li, \"Documents Ranking Based on a Hybrid Language Model for Information Retrieval\" IEEE International Conference on Information Acquisition, Aug. 2006, pp: 279-283.", | |
| "links": null | |
| }, | |
| "BIBREF2": { | |
| "ref_id": "b2", | |
| "title": "A general Language Model for Information Retrieval", | |
| "authors": [ | |
| { | |
| "first": "Fei", | |
| "middle": [], | |
| "last": "Song", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "W", | |
| "middle": [ | |
| "Bruce" | |
| ], | |
| "last": "Croft", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 1999, | |
| "venue": "Proc. of Eighth International Conference on Information and Knowledge Management", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "316--321", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Fei Song, W. Bruce Croft , \"A general Language Model for Information Retrieval\", Proc. of Eighth International Conference on Information and Knowledge Management, 1999, pp: 316-321.", | |
| "links": null | |
| }, | |
| "BIBREF4": { | |
| "ref_id": "b4", | |
| "title": "A statistical approach to machine translation", | |
| "authors": [ | |
| { | |
| "first": "John; Della", | |
| "middle": [], | |
| "last": "Cocke", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Stephen", | |
| "middle": [ | |
| "A" | |
| ], | |
| "last": "Pietra", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Vincent", | |
| "middle": [ | |
| "J" | |
| ], | |
| "last": "Della Pietra", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Frederick", | |
| "middle": [ | |
| ";" | |
| ], | |
| "last": "Jelinek", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "John", | |
| "middle": [ | |
| "D" | |
| ], | |
| "last": "Lafferty", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Robert", | |
| "middle": [ | |
| "L" | |
| ], | |
| "last": "Mercer", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Paul", | |
| "middle": [ | |
| "S" | |
| ], | |
| "last": "Roossin", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 1990, | |
| "venue": "Computational Linguistics", | |
| "volume": "16", | |
| "issue": "2", | |
| "pages": "79--85", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Cocke, John; Della Pietra, Stephen A.; Della Pietra, Vincent J.; Jelinek, Frederick; Lafferty, John D.; Mercer, Robert L.; and Roossin, Paul S., \"A statistical approach to machine translation .\" Computational Linguistics, Volume 16 , Issue 2, 1990, pp: 79-85.", | |
| "links": null | |
| }, | |
| "BIBREF5": { | |
| "ref_id": "b5", | |
| "title": "Statistical Translation Model or Phrases", | |
| "authors": [ | |
| { | |
| "first": "S", | |
| "middle": [], | |
| "last": "Jason", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "David", | |
| "middle": [], | |
| "last": "Chang", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Chun-Jun", | |
| "middle": [], | |
| "last": "Yu", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "", | |
| "middle": [], | |
| "last": "Lee", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2001, | |
| "venue": "Processing of Computational Linguistics and Chinese Language", | |
| "volume": "6", | |
| "issue": "", | |
| "pages": "43--64", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Jason S Chang, David Yu, Chun-Jun Lee, \"Statistical Translation Model or Phrases\" In Processing of Computational Linguistics and Chinese Language, Vol. 6, No. 2, August 2001, pp: 43-64.", | |
| "links": null | |
| }, | |
| "BIBREF6": { | |
| "ref_id": "b6", | |
| "title": "Adaptive Statistical Language Modeling: a Maximum Entropy Approach", | |
| "authors": [ | |
| { | |
| "first": "Ronald", | |
| "middle": [], | |
| "last": "Rosenfeld", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 1992, | |
| "venue": "", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Ronald Rosenfeld, \"Adaptive Statistical Language Modeling: a Maximum Entropy Approach\" Ph.D. Thesis Proposal, Carnegie Mellon University, September 1992.", | |
| "links": null | |
| }, | |
| "BIBREF7": { | |
| "ref_id": "b7", | |
| "title": "ATree-Based Statistical Language Model for Natural Language Speech Recognition", | |
| "authors": [ | |
| { | |
| "first": "R", | |
| "middle": [], | |
| "last": "Lalit", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Peter", | |
| "middle": [ | |
| "F" | |
| ], | |
| "last": "Bahl", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "", | |
| "middle": [], | |
| "last": "Brown", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "V", | |
| "middle": [ | |
| "De" | |
| ], | |
| "last": "Peter", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Robert", | |
| "middle": [ | |
| "L" | |
| ], | |
| "last": "Souza", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "", | |
| "middle": [], | |
| "last": "Mercer", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 1989, | |
| "venue": "IEEE Transactions on Acoustics Speech and Signal Processing", | |
| "volume": "37", | |
| "issue": "7", | |
| "pages": "1001--1008", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Lalit R. Bahl, Peter F. Brown, Peter V. De Souza, Robert L. Mercer, \"ATree-Based Statistical Language Model for Natural Language Speech Recognition\", IEEE Transactions on Acoustics Speech and Signal Processing, Vol. 37, No. 7, July 1989, pp: 1001-1008.", | |
| "links": null | |
| }, | |
| "BIBREF8": { | |
| "ref_id": "b8", | |
| "title": "Pattern Recognition(Third Edition)", | |
| "authors": [ | |
| { | |
| "first": "Sergios", | |
| "middle": [], | |
| "last": "Theodoridis", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Konstantion", | |
| "middle": [], | |
| "last": "Koutroumbas", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": null, | |
| "venue": "", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "13--19", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Sergios Theodoridis and Konstantion Koutroumbas, \"Pattern Recognition(Third Edition) \", Academic Press. pp 13-19", | |
| "links": null | |
| }, | |
| "BIBREF9": { | |
| "ref_id": "b9", | |
| "title": "Word Segmentation in Sentence Analysis", | |
| "authors": [ | |
| { | |
| "first": "A.-D", | |
| "middle": [], | |
| "last": "Wu", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Z.-X", | |
| "middle": [], | |
| "last": "Jiang", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 1998, | |
| "venue": "International Conference on Chinese Information Processing", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "169--180", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Wu, A.-D., and Z.-X. Jiang, \"Word Segmentation in Sentence Analysis,\"International Conference on Chinese Information Processing, 1998, Beijing,China, pp: 169-180", | |
| "links": null | |
| }, | |
| "BIBREF10": { | |
| "ref_id": "b10", | |
| "title": "Estimation of Probabilities from Sparse Data for the Language Model Component of a Speech Recognizer", | |
| "authors": [ | |
| { | |
| "first": "", | |
| "middle": [], | |
| "last": "Slavam", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "", | |
| "middle": [], | |
| "last": "Katz", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 1987, | |
| "venue": "IEEE Transactions on ACOUSTICS, SPEECH, and SIGNAL PROCESSING", | |
| "volume": "35", | |
| "issue": "3", | |
| "pages": "400--401", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Slavam. Katz, \"Estimation of Probabilities from Sparse Data for the Language Model Component of a Speech Recognizer \", IEEE Transactions on ACOUSTICS, SPEECH, and SIGNAL PROCESSING, VOL. ASSP-35, NO.3, MARCH 1987, pp 400-401", | |
| "links": null | |
| }, | |
| "BIBREF11": { | |
| "ref_id": "b11", | |
| "title": "A Bit of Progress in Language Modeling, Extended Version", | |
| "authors": [ | |
| { | |
| "first": "J", | |
| "middle": [], | |
| "last": "Goodman", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2001, | |
| "venue": "", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "J. Goodman, \"A Bit of Progress in Language Modeling, Extended Version,\" Microsoft Research, Technical Report MSR-TR-2001-72, 2001.", | |
| "links": null | |
| }, | |
| "BIBREF12": { | |
| "ref_id": "b12", | |
| "title": "A maximum entropy approach to natural language processing", | |
| "authors": [ | |
| { | |
| "first": "Adam", | |
| "middle": [ | |
| "L" | |
| ], | |
| "last": "Berger", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Vincent", | |
| "middle": [ | |
| "J Della" | |
| ], | |
| "last": "Pietra", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Stephen", | |
| "middle": [ | |
| "A" | |
| ], | |
| "last": "Della Pietra", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 1996, | |
| "venue": "Computational Linguistics", | |
| "volume": "22", | |
| "issue": "1", | |
| "pages": "39--71", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Adam L. Berger, Vincent J. Della Pietra, Stephen A. Della Pietra, \"A maximum entropy approach to natural language processing\", Computational Linguistics, Volume 22, Issue 1, March 1996, pp: 39-71.", | |
| "links": null | |
| }, | |
| "BIBREF13": { | |
| "ref_id": "b13", | |
| "title": "An Empirical Study of Smoothing Techniques for Language Modeling", | |
| "authors": [ | |
| { | |
| "first": "S", | |
| "middle": [ | |
| "F" | |
| ], | |
| "last": "Chen", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Joshua", | |
| "middle": [], | |
| "last": "Goodman", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 1996, | |
| "venue": "Proc. of the 34th annual meeting on Association for Computational Linguistics", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "310--318", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "S. F. Chen, Joshua Goodman \"An Empirical Study of Smoothing Techniques for Language Modeling\", Proc. of the 34th annual meeting on Association for Computational Linguistics ,Santa Cruz, California, 1996, pp:310-318.", | |
| "links": null | |
| }, | |
| "BIBREF14": { | |
| "ref_id": "b14", | |
| "title": "A bit of Progress in Language Modeling", | |
| "authors": [ | |
| { | |
| "first": "J", | |
| "middle": [], | |
| "last": "Goodman", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2001, | |
| "venue": "", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "J. Goodman, \"A bit of Progress in Language Modeling\", Microsoft Research, Aug. 2001.", | |
| "links": null | |
| }, | |
| "BIBREF15": { | |
| "ref_id": "b15", | |
| "title": "A maximum entropy approach to natural language processing", | |
| "authors": [ | |
| { | |
| "first": "Adam", | |
| "middle": [ | |
| "L" | |
| ], | |
| "last": "Berger", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Vincent", | |
| "middle": [ | |
| "J Della" | |
| ], | |
| "last": "Pietra", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Stephen", | |
| "middle": [ | |
| "A" | |
| ], | |
| "last": "Della Pietra", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 1996, | |
| "venue": "Computational Linguistics", | |
| "volume": "22", | |
| "issue": "1", | |
| "pages": "39--71", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Adam L. Berger, Vincent J. Della Pietra, Stephen A. Della Pietra, \"A maximum entropy approach to natural language processing\", Computational Linguistics, Volume 22, Issue 1, March 1996, pp: 39-71.", | |
| "links": null | |
| }, | |
| "BIBREF16": { | |
| "ref_id": "b16", | |
| "title": "Error Diagnosis of Chinese Sentences Using Inductive Learning Algorithm and Decomposition-Based Testing Mechanism", | |
| "authors": [ | |
| { | |
| "first": "Ru-Yng", | |
| "middle": [], | |
| "last": "Chang", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Chung-Hsien", | |
| "middle": [], | |
| "last": "Wu", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Philips Kokoh", | |
| "middle": [], | |
| "last": "Prasetyo", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2012, | |
| "venue": "ACM Transactions on Asian Language Information Processing", | |
| "volume": "11", | |
| "issue": "1", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Ru-Yng Chang, Chung-Hsien Wu, and Philips Kokoh Prasetyo (2012). Error Diagnosis of Chinese Sentences Using Inductive Learning Algorithm and Decomposition-Based Testing Mechanism. ACM Transactions on Asian Language Information Processing, 11(1), article 3, March 2012.", | |
| "links": null | |
| } | |
| }, | |
| "ref_entries": { | |
| "TABREF0": { | |
| "type_str": "table", | |
| "html": null, | |
| "num": null, | |
| "text": "The 2015 Conference on Computational Linguistics and Speech Processing ROCLING 2015, pp. 218-232 \uf0d3 The Association for Computational Linguistics and Chinese Language Processing", | |
| "content": "<table><tr><td colspan=\"2\">\u7136\u5df2\u8003\u616e\u7acb\u610f\u53d6\u6750\u3001\u7d44\u7e54\u7d50\u69cb\u3001\u9063\u8a5e\u9020\u53e5\u3001\u932f\u5225\u5b57\u3001\u683c\u5f0f\u53ca\u6a19\u9ede\u7b26\u865f\u7b49\u56db\u9805\u6838\u5fc3\u6280\u5de7\u70ba (\u4e09)\u3001\u9063\u8a5e\u9020\u53e5 \u5c0f\u6642\u4e0d\u9593\u65b7\u5730\u63d0\u4f9b\u670d\u52d9\uff0c\u96a8\u6642\u63d0\u4f9b\u5b78\u7fd2\u7684\u6a5f\u6703\uff0c\u800c\u4e14\u5b78\u6821\u7684\u8001\u5e2b\u4e00\u6b21\u9762\u5c0d\u8a31\u591a\u7684\u5b78\u751f\uff0c \u9ad8\u65bc\u4e00\u5b9a\u7684\u9580\u6abb\u503c\u5c07\u6703\u63d0\u793a\u9019\u53ef\u80fd\u662f\u500b\u4e0d\u901a\u9806\u7684\u53e5\u5b50\uff0c\u8a55\u4f30\u7cfb\u7d71\u7684\u6548\u80fd\u7684\u90e8\u5206\uff0c\u6211\u5011\u628a</td></tr><tr><td colspan=\"2\">\u4e3b\u8ef8\"(\u9673\u6eff\u9298 2007\uff0c396)\u3002\u9019\u56db\u500b\u4f5c\u6587\u8a55\u91cf\u7bc4\u7587\u4e26\u4e0d\u662f\u4efb\u610f\u898f\u5b9a\u7684\uff0c\u800c\u662f\u4f9d\u7167\u4f5c\u6587\u7684 \u5b78\u751f\u96e3\u4ee5\u7372\u5f97\u5373\u6642\u7684\u8a55\u50f9\u56de\u994b\u3002\u5b78\u751f\u5beb\u7684\u4f5c\u6587\u7d93\u7531\u5206\u6563\u5f0f\u7684\u8a3a\u65b7\u6a21\u7d44(\u672c\u7cfb\u7d71\u70ba\u5176\u4e2d\u4e00 \u6e2c\u8a66\u7d50\u679c\u7d93\u7531\u4e2d\u6587\u6d41\u5229\u4eba\u662f\u4f86\u9032\u884c\u6aa2\u95b1\uff0c\u63a5\u8457\u4f7f\u7528 Recall \u8207 Precision \u8a55\u4f30\u7cfb\u7d71\u5075\u6e2c\u80fd</td></tr><tr><td colspan=\"2\">\u69cb\u6210\u904e\u7a0b\u4e2d\u6240\u9700\u8981\u7684\u5143\u7d20\u6c7a\u5b9a\u9019\u4e9b\u8a55\u91cf\u7bc4\u7587\u7684\u3002\u56e0\u6b64\u9019\u4e9b\u4f5c\u6587\u8a55\u91cf\u7bc4\u7587\u4e0d\u5bb9\u6613\u88ab\u8b8a\u66f4\u3002 \u6211\u5011\u521d\u6b65\u5206\u6790\u4e86\u4e00\u767e\u4efd\u570b\u4e2d\u4f5c\u6587\u7684\u7d50\u679c\u986f\u793a\uff0c\u5982\u679c\u4f5c\u6587\u5e73\u92ea\u76f4\u6558\u5f88\u5c11\u7528\u4fee\u98fe\u8a5e\u7684\u4f5c\u6587\u5927 \u500b\u8a3a\u65b7\u6a21\u7d44)\u5206\u5225\u8a3a\u65b7\u500b\u5225\u9762\u5411\u7684\u512a\u7f3a\u9ede\uff0c\u4e4b\u5f8c\u7522\u751f\u4e00\u4efd\u53ef\u64f4\u5c55\u7684\u4f5c\u6587\u8a3a\u65b7\u6e05\u55ae\u3002\u9019\u500b \uf98a\u3002\u6211\u5011\u5c07\u5206\u6790\u672c\u7cfb\u7d71\u7684\u5be6\u9a57\u7d50\u679c\uff0c\u4e26\u4e14\u6839\u64da\u7cfb\u7d71\u7f3a\u5931\uff0c\u4f86\u4e00\u6b65\u4e00\u6b65\u63d0\u51fa\u6539\u5584\u7684\u65b9\u6cd5\uff0c</td></tr><tr><td colspan=\"2\">\u4ee5\u4e0b\u8aaa\u660e\u5982\u4f55\u5c07\u4f5c\u6587\u8a55\u91cf\u70ba 6 \u7a2e\u4e0d\u540c\u7684\u7b49\u7d1a(\u5982\u8868\u4e00 [1])\uff0c\u800c\u672c\u7cfb\u7d71\u91dd\u5c0d\u56db\u500b\u9762\u5411\u4e2d-\u9063 \u8a5e\u9020\u53e5\u7684\u53e5\u5b50\u6d41\u66a2\u5ea6\u9032\u884c\u7814\u7a76\u3002 \u6e05\u55ae\u88e1\u6574\u5408\u5404\u9762\u5411\u6700\u5f8c\u7684\u8a3a\u65b7\u7d50\u679c\uff0c\u63d0\u4f9b\u5f8c\u9762\u8a55\u5206\u6a21\u7d44\u4ee5\u53ca\u96f7\u9054\u5716\u7684\u7522\u751f\u3002\u5728\u56db\u500b\u9762\u5411 \u5e0c\u671b\u672a\u4f86\uf901\u65b0\u7248\u672c\u7684\u7cfb\u7d71\uff0c\u80fd\u6539\u5584\u5be6\u9a57\u7d50\u679c\u3001\u63d0\u6607\u6548\u80fd\uff0c\u4e26\u4e14\u89c0\u5bdf\u7279\u6b8a\u6848\uf9b5\uff0c\u5305\u62ec\u7cfb\u7d71 \u6982\u662f\u4e09\u5230\u56db\u7d1a\u5206\u3002\u6d2a\u7f8e\u96c0(2013\uff0c 279-280)\u5efa\u8b70\u5b78\u751f\u4f7f\u7528\u300c\u6558\u4e8b\u52a0\u63cf\u5beb\u300d\u53d6\u4ee3\u300c\u55ae\u7d14\u6558 \u4e8b\u300d\u800c\u4e14\u5c07\u4fee\u98fe\u8a5e\u5206\u7d1a\uff0c\u5982\u4e0b\u8868\u4e8c\u3002\u6211\u5011\u4f7f\u7528\u570b\u4e2d\u4e09\u5e74\u7684\u570b\u6587\u8ab2\u672c\u88e1\u9762\u7684\u8a5e\u5f59\u4ee5\u53ca\u570b\u4e2d \u88e1\u9762\uff0c \u300c\u932f\u5225\u5b57\u3001\u683c\u5f0f\u8207\u6a19\u9ede\u7b26\u865f\u8a3a\u65b7\u6a21\u7d44\u300d\u6280\u8853\u4e0a\u662f\u76ee\u524d\u6700\u6210\u719f\u7684\uff0c\u5982\u6709\u660e\u986f\u7684\u932f\u8aa4 \u8aa4\u5224\u70ba\u4e0d\u6d41\u66a2\u7684\u53e5\u5b50\u4ee5\u53ca\u932f\u653e\u4e0d\u6d41\u66a2\u7684\u53e5\u5b50\uff0c\u9032\u4e00\u6b65\u5206\u6790\u932f\u8aa4\u7684\u539f\u56e0\uff0c\u8a2d\u6cd5\u6539\u5584\u7cfb\u7d71\u3002</td></tr><tr><td colspan=\"2\">\u4f5c\u6587\u8a9e\u6599\u5eab\u88e1\u9762\u7684\u8a5e\u5f59\uff0c\u9019\u4e9b\u8a5e\u5f59\u4e0d\u50c5\u5e9c\u548c\u570b\u4e2d\u751f\u7684\u4f7f\u7528\u7a0b\u5ea6\u4e5f\u4e0d\u6703\u6709\u8271\u6f80\u4ee5\u53ca\u5c11\u7528\u8a5e \u5c07\u5747\u6703\u88ab\u8a3a\u65b7\u51fa\u4f86\u4e26\u4e14\u7cfe\u6b63\u3002\u7576\u4f5c\u6587\u5404\u500b\u7684\u8a3a\u65b7\u7d50\u679c\u7522\u751f\u5f8c(\u7acb\u610f\u53d6\u6750\u8a3a\u65b7\u6a21\u7d44\u3001\u9063\u8a5e \u81ea\u7136\u8a9e\u8a00\u8655\uf9e4(Natural Language Processing, NLP)\u7684\uf9b4\u57df\u5305\u542b\u4e86\u8a9e\u97f3\u8fa8\uf9fc[7][8]\u3001\u8cc7\u8a0a\u6aa2</td></tr><tr><td colspan=\"2\">\u5f59\u7684\u767c\u751f\u3002 \u9020\u53e5\u8a3a\u65b7\u6a21\u7d44\u3001\u7d50\u69cb\u7d44\u7e54\u8a3a\u65b7\u6a21\u7d44\u7b49)\uff0c\u6211\u5011\u5c31\u53ef\u4ee5\u7d66\u4f5c\u6587\u8a55\u5b9a\u7b49\u7d1a\u3002\u4f9d\u7167\u4f5c\u6587\u5728\u56db\u500b \uf96a[2][3]\u3001\u6587\u4ef6\u5206\uf9d0\u3001\u624b\u5beb\u8fa8\uf9fc\u4ee5\u53ca\u6a5f\u5668\u7ffb\u8b6f[4] [5]\u2026\u7b49\u7b49\uff0c\u800c\u8a9e\u8a00\u6a21\u578b (Language Model, \u8868\u4e00\u3001\u570b\u4e2d\u751f\u57fa\u672c\u5b78\u529b\u6e2c\u9a57\u4f5c\u6587\u6e2c\u9a57\u8a55\u5206\u898f\u6e96[1] \u9762\u5411\u7684\u8868\u73fe\uff0c\u6a5f\u5668\u5b78\u7fd2\u7a0b\u5f0f\u53ef\u4ee5\u8a13\u7df4\u51fa\u7a69\u5b9a\u7684\u5206\u985e\u5668\uff0c\u5c07\u4f5c\u6587\u5206\u70ba\u96f6\u5230\u516d\u7d1a\u5206\uff0c\u7522\u751f\u5c0d LM)\u662f\u81ea\u7136\u8a9e\u8a00\u8655\uf9e4\u91cd\u8981\u7684\u6280\u8853\u4e4b\u4e00[6]\uff0c\u8a9e\u8a00\u6a21\u578b\u7d71\u8a08\u4e26\u4e14\u7d00\uf93f\uf9ba\u5927\uf97e\u8a9e\uf9be\u5eab\u7684\u8a5e\u983b</td></tr><tr><td colspan=\"2\">\u6458\u8981 \u56e0\u61c9\u81ea\u52d5\u5316\u4f5c\u6587\u6559\u5b78\u7cfb\u7d71\u4e4b\u9700\u6c42\uff0c\u6211\u5011\u5c07\u958b\u767c\u591a\u7a2e\u4e2d\u6587\u81ea\u7136\u8a9e\u8a00\u8655\u7406\u529f\u80fd\u3002\u672c\u6587\u5c07\u4ee5 \u4f5c\u6587\u53e5\u5b50\u7684\u901a\u9806\u7a0b\u5ea6\u5075\u6e2c\u70ba\u76ee\u6a19\uff0c\u6211\u5011\u63d0\u51fa\u57fa\u65bc\u8a9e\u8a00\u6a21\u578b(language model)\u7d50\u5408\u570b\u4e2d \u7d1a\u5206 \u570b\u6c11\u4e2d\u5b78\u5b78\u751f\u57fa\u672c\u5b78\u529b\u6e2c\u9a57\u5beb\u4f5c\u6e2c\u9a57\u8a55\u5206\u898f\u6e96\u4e00\u89bd\u8868 \u516d\u7d1a\u5206 \u61c9\u7684\u96f7\u9054\u5716\uff0c\u63a5\u8457\u8a55\u8a9e\u4f9d\u7167\u5404\u5225\u9762\u76f8\u8a3a\u65b7\u7684\u7d50\u679c\u7522\u751f\uff0c\u7136\u5f8c\u5408\u4f75\u4e00\u8d77\u5448\u73fe\u7279\u5fb5\u7d30\u7bc0\uff0c\u4f46 \u53ca\u6a5f\uf961\uff0c\u5b83\u7279\u6027\u5c31\u662f\u53ef\u4ee5\u4f9d\u64da\u904e\u53bb\u7684\u8a13\u7df4\u8cc7\u6599\uff0c\u4e5f\u5c31\u662f\u66fe\u7d93\u51fa\u73fe\u7684\u5b57\uff0c\u9810\u6e2c\u4e0b\u4e00\u500b\u5b57\u51fa \u8868\u4e8c\u3001\u4fee\u98fe\u8a5e\u5206\u7d1a\u8868(\u6d2a\u7f8e\u96c0 2013\uff0c 81) \u662f\u96fb\u8166\u53ef\u4ee5\u8a73\u7d30\u5730\u5c07\u5404\u7a2e\u7279\uff0c\u9019\u4e9b\u77e5\u8b58\u7684\u6aa2\u6e2c\u9700\u8981\u642d\u914d\u81ea\u7136\u8a9e\u8a00\u8655\u7406\u7684\u5de5\u5177\u7a0b\u5f0f\u4ee5\u53ca\u8a9e \u73fe\u7684\u6a5f\uf961\uff0c\u56e0\u6b64\u4e5f\u80fd\u85c9\u6b64\u8a08\u7b97\u51fa\u4e00\u500b\uf906\u5b50\u7684\u6a5f\uf961\uff0c\u6a5f\u7387\u8d8a\u5927\u4ee3\u8868\u9019\u53e5\u5b50\u8d8a\u5e38\u51fa\u73fe\uff0c\u4e5f\u5c31 \u516d\u7d1a\u5206\u7684\u6587\u7ae0\u662f\u512a\u79c0\u7684\uff0c\u9019\u7a2e\u6587\u7ae0\u660e\u986f\u5177\u6709\u4e0b\u5217\u7279\u5fb5\uff1a \u203b\u9063\u8a5e\u9020\u53e5\uff1a\u80fd\u7cbe\u78ba\u4f7f\u7528\u8a9e\u8a5e\uff0c\u4e26\u6709\u6548\u904b\u7528\u5404\u7a2e\u53e5\u578b\u4f7f \u8a00\u8cc7\u6e90\uff0c\u6700\u57fa\u790e\u7684\u524d\u8655\u7406\u5c31\u662f\u65b7\u8a5e\u4ee5\u53ca\u6a19\u8a3b\u8a5e\u6027(POS tagging)\uff0c\u7136\u5f8c\u4f9d\u7167\u5404\u6a21\u7d44\u9700\u6c42\u4f86 \u662f\u8d8a\u70ba\u901a\u9806\uff0c\u53cd\u4e4b\u5982\u679c\u6a5f\u7387\u8d8a\u4f4e\uff0c\u4ee3\u8868\u9019\u53e5\u5b50\u7684\u5beb\u6cd5\u5f88\u5c11\u51fa\u73fe\uff0c\u5982\u679c\u4e0d\u662f\u5275\u65b0\uff0c\u6975\u6709\u53ef \u56db\u7d1a\u5206\u7528\u8a9e \u4e94\u7d1a\u5206\u7528\u8a9e \u516d\u7d1a\u5206\u7528\u8a9e \u589e\u52a0\u8655\u7406\u7684\u77e5\u8b58\u3002\u6700\u5f8c\uff0c\uf96f\u660e\u5be6\u9a57\u7d50\u679c\u8207\u8a55\u4f30\u7684\u5206\u6790\u8a0e\uf941\uff0c\u85c9\u4ee5\u9a57\u8b49\u672c\uf941\u6587\u64f7\u53d6\u4e4b\u6548\u80fd\u3002 \u80fd\u662f\u5beb\u51fa\u4e86\u4e0d\u901a\u9806\u7684\u53e5\u5b50\uff0c\u6240\u4ee5\u8a9e\u8a00\u6a21\u578b\u4e5f\u80fd\u61c9\u7528\u5728\u4e2d\u6587\u53e5\u5b50\u6d41\u66a2\u5ea6\u5075\u6e2c\u7684\u65b9\u9762\u3002\u8a9e\u8a00 \u6587\u53e5\u6d41\u66a2\u3002 \u4e94\u7d1a\u5206 \u5f88\u7d2f \u75b2\u7d2f \u75b2\u618a \u6a21\u578b\u898f\u6a21\u76f8\u7576\u4f9d\u8cf4\u5927\u578b\u8a13\uf996\u8a9e\uf9be\uff0c\u8a13\uf996\u8a9e\uf9be\u7684\u6027\u8cea\u8d8a\u63a5\u8fd1\u6e2c\u8a66\u7684\u6587\u7ae0\uff0c\u6240\u5efa\uf9f7\u7684\u8a9e\u8a00\u6a21 \u4e94\u7d1a\u5206\u7684\u6587\u7ae0\u5728\u4e00\u822c\u6c34\u6e96\u4e4b\u4e0a\uff0c\u9019\u7a2e\u6587\u7ae0\u660e\u986f\u5177\u6709\u4e0b\u5217 \u7279\u5fb5\uff1a \u203b\u9063\u8a5e\u9020\u53e5\uff1a\u80fd\u6b63\u78ba\u4f7f\u7528\u8a9e\u8a5e\uff0c\u4e26\u904b\u7528\u5404\u7a2e\u53e5\u578b\u4f7f\u6587\u53e5 \u5f88\u5435 \u5435\u9b27 \u4e09\u3001\u7814\u7a76\u67b6\u69cb \u55a7\u56c2 \u578b\u6548\u679c\u8d8a\u597d\uff0c\u6240\u4ee5\u8a9e\uf9be\u5eab\u4e5f\u8981\u8ddf\u8457\u6539\u8b8a\u8207\u9069\u61c9\u3002</td></tr><tr><td colspan=\"2\">\u901a\u9806\u3002 \u751f\u4f5c\u6587\u8a9e\u6599\u77e5\uf9fc\u5eab\u7684\u65b9\u6cd5\uff0c\u4e26\u4e14\u4f7f\u7528\u8cc7\u8a0a\u6aa2\uf96a\u7684\u6280\u8853\u4f86\u6539\u5584\u7cfb\u7d71\u6548\u80fd\uff0c\u958b\u767c\u51fa\u7b2c\u4e00\u5957\u91dd \u5c0d\u53e5\u5b50\u901a\u9806\u7a0b\u5ea6\u7684\u5075\u6e2c\u7cfb\u7d71\uff0c\u80fd\uf901\u5feb\uf901\u6b63\u78ba\u5075\u67e5\u5b78\u751f\u6587\u7ae0\u5167\u5bb9\u4e0d\u901a\u9806\u7684\u5730\u65b9\u3002\u7cfb\u7d71\u5206\u70ba \u4e8c\u500b\u90e8\u4efd\uff1a\u8a9e\u8a00\u6a21\u578b\u8a13\u7df4\u6a21\u7d44\u548c\u4e2d\u6587\u8a9e\u6599\u64f7\u53d6\u6e2c\u8a66\u6a21\u7d44\u3002\u6211\u5011\u7684\u5be6\u9a57\u8b49\u660e\uf9ba\u4ee5\u8a9e\u8a00\u6a21\u578b \uf9e4\uf941\u70ba\u57fa\u790e\u7684\u53e5\u5b50\u901a\u9806\u5ea6\u81ea\u52d5\u5075\u6e2c\u7cfb\u7d71\u80fd\u5920\u6709\u6548\u5075\u6e2c\u4e0d\u901a\u9806\u7684\u53e5\u5b50\u3002\u63d0\u4f9b\u672c\u570b\u5b78\u751f\u6216\u5916 \u7c4d\u5b78\u751f\u5b78\u7fd2\u4f5c\u6587\u6642\u7684\u8f14\u52a9\u5de5\u5177\u3002 \u95dc\u9375\u8a5e\uff1a\u4e2d\u6587\uff0c\u4f5c\u6587\uff0c\u8a9e\u8a00\u6a21\u578b\uff0cN \u5143\u8a9e\u8a00\u6a21\u578b\uff0c\u53e5\u5b50\u6d41\u66a2\u5ea6 \u4e00\u3001\u7dd2\u8ad6 \u56db\u7d1a\u5206 \u56db\u7d1a\u5206\u7684\u6587\u7ae0\u5df2\u9054\u4e00\u822c\u6c34\u6e96\uff0c\u9019\u7a2e\u6587\u7ae0\u660e\u986f\u5177\u6709\u4e0b\u5217\u7279 \u5fb5\uff1a \u203b\u9063\u8a5e\u9020\u53e5\uff1a\u80fd\u6b63\u78ba\u4f7f\u7528\u8a9e\u8a5e\uff0c\u6587\u610f\u8868\u9054\u5c1a\u7a31\u6e05\u695a\uff0c\u4f46 \u6709\u6642\u6703\u51fa\u73fe\u5197\u8a5e\u8d05\u53e5\uff1b\u53e5\u578b\u8f03\u7121\u8b8a\u5316\u3002 \u4e09\u7d1a\u5206\u7684\u6587\u7ae0\u5728\u8868\u9054\u4e0a\u662f\u4e0d\u5145\u5206\u7684\uff0c\u9019\u7a2e\u6587\u7ae0\u660e\u986f\u5177\u6709 \u4e0b\u5217\u7279\u5fb5\uff1a \u4e0b\u5716\u4e00\u662f\u4e2d\u6587\u4f5c\u6587\u6d41\u66a2\u5ea6\u5075\u6e2c\u7cfb\u7d71\u904b\u4f5c\u7684\uf9ca\u7a0b\u5716\uff0c\u9996\u5148\u5c07\u6e2c\u8a66\u7684\u8cc7\uf9be\uff0c\u5305\u62ec\u624b\u5de5\u8a2d\u5b9a\u7684 \u8a9e\u8a00\u6a21\u578b\u6703\u57fa\u65bc\u4f7f\u7528\u65b9\u6cd5\u7684\u4e0d\u540c\u800c\u6709\u6240\u6539\u8b8a\uff0c\uf9b5\u5982\uff1a mixing \u8a9e\u8a00\u6a21\u578b\uff0c\u4f7f\u7528\u6df7\u5408\u591a\u7a2e \u7531\u65bc\u570b\u4e2d\u4e09\u5e74\u7684\u5404\u8272\u570b\u6587\u6559\u6750\u4ecd\u7136\u6709\u96e3\u6613\u5ea6\u7684\u5206\u5225\uff0c\u901a\u5e38\u4e09\u7d1a\u5206\u4ee5\u4e0b\u7684\u4f5c\u6587\u4f7f\u7528\u7684\u8a5e\u5f59 \u53e5\u5b50\u4ee5\u53ca\u4e2d\u6587\u4f5c\u6587\u8f38\u5165\u5230\u53e5\u5b50\u6d41\u66a2\u5ea6\u5075\u6e2c\u7cfb\u7d71\uff0c\u7cfb\u7d71\u6703\u81ea\u52d5\u5075\u6e2c\u8a08\u7b97\u5206\u6578\uff0c\u4e4b\u5f8c\u5206\u6578\u82e5 \uf967\u540c\u7684\u8a9e\u8a00\u6a21\u578b\u4f86\u6539\u5584\u4e2d\u6587\u65b7\u8a5e\u7684\u6548\u679c[9]\uff0c\u5716\u4e8c\u8209\uf9b5\uf96f\u660e\u975e\u65b7\u8a5e bigram \u4f7f\u7528\u55ae\u5b57\u5efa\u7acb \u90fd\u505c\u7559\u5728\u570b\u4e8c\u4ee5\u4e0b\uff0c\u6c92\u6709\u9054\u5230\u570b\u4e09\u7684\u7b49\u7d1a\u3002\u6240\u4ee5\u5206\u985e\u570b\u4e2d\u570b\u6587\u8ab2\u672c\u7684\u8a5e\u5f59\u7b49\u7d1a\u662f\u5177\u6709\u610f \u8a9e\u8a00\u6a21\u578b\uff0c\u5716\u4e09\u8209\uf9b5\uf96f\u660e\u65b7\u8a5e bigram \u5148\u65b7\u8a5e\u5f8c\u5efa\u7acb\u8a9e\u8a00\u6a21\u578b\u3002\u800c\u672c\u5be6\u9a57\u4e2d\u5206\u5225\u4f7f\u7528\u4e86 \u7fa9\u7684\u3002\u6587\u737b\u4e2d\u4f9d\u7167\u5404\u7d1a\u5206\u4f5c\u6587\u8a5e\u5f59\u7684\u7a0b\u5ea6\u4e0d\u540c\uff0c\u76f8\u540c\u610f\u601d\u4f46\u8868\u9054\u65b9\u5f0f\u4e0d\u540c\uff0c\u4f9d\u7167\u7b49\u7d1a\u7531 \u4e2d\u6587\u6587\u7ae0 \u65b0\u805e\u8a9e\u6599\u5eab\u4ee5\u53ca\u570b\u4e2d\u751f\u4f5c\u6587\u8a9e\u6599\u5eab\u4f86\u5efa\u7acb\u8a9e\u8a00\u6a21\u578b\u3002 \u4f4e\u5230\u9ad8\u5206\u985e\uff0c\u4f8b\u5982:3 \u7d1a\u5206:\u6211\u8096\u60f3\u80fd\u8003\u7b2c\u4e00\u540d\uff0c4 \u7d1a\u5206:\u8003\u7b2c\u4e00\u540d\u5c0d\u6211\u4f86\u8aaa\u771f\u662f\u975e\u5206\u4e4b\u60f3\uff0c \u4e2d\u6587\u6d41\u66a2\u5ea6\u5075\u6e2c\u7cfb\u7d71 5 \u7d1a\u5206:\u6211\u5984\u60f3\u80fd\u8003\u5230\u7b2c\u4e00\u540d\uff0c6 \u7d1a\u5206:\u89ac\u89a6\u8457\u7b2c\u4e00\u540d\u5bf6\u5ea7\u7684\u6211(\u6d2a\u7f8e\u96c0 2013\uff0c 70)\uff0c\u4ee5\u53ca Input 4 \u7d1a\u5206:\u9047\u5230\uff0c5 \u7d1a\u5206:\u76f8\u9022\uff0c6 \u7d1a\u5206:\u9082\u9005(\u6d2a\u7f8e\u96c0 2013\uff0c 81)\uff0c\u6211\u5011\u9019\u88e1\u4e3b\u8981\u91dd\u5c0d\u5197\u8d05\u8a5e \u4e09\u7d1a\u5206 \u203b\u9063\u8a5e\u9020\u53e5\uff1a\u7528\u5b57\u9063\u8a5e\u4e0d\u592a\u6070\u7576\uff0c\u6216\u51fa\u73fe\u932f\u8aa4\uff1b\u6216\u5197\u8a5e \u7684\u5075\u6e2c\u4f86\u8655\u7406\u3002 N-gram\u4e4b \u8d05\u53e5\u904e\u591a\u3002\u4e0a\u7684\u932f\u8aa4\uff0c\u4ee5\u81f4\u9020\u6210\u7406\u89e3\u4e0a\u7684\u56f0\u96e3\u3002 \u4e8c\u7d1a\u5206\u7684\u6587\u7ae0\u5728\u8868\u9054\u4e0a\u5448\u73fe\u56b4\u91cd\u7684\u554f\u984c\uff0c\u9019\u7a2e\u6587\u7ae0\u660e\u986f (\u56db)\u3001\u932f\u5225\u5b57\u3001\u683c\u5f0f\u8207\u6a19\u9ede\u7b26\u865f \u4e2d\u6587\u65b7\u8a5e \u5716\u4e8c\u3001\u8209\uf9b5\uf96f\u660e\u975e\u65b7\u8a5e bigram \u793a\u610f\u5716</td></tr><tr><td>\u4e8c\u7d1a\u5206</td><td>\u5177\u6709\u4e0b\u5217\u7279\u5fb5\uff1a</td></tr><tr><td colspan=\"2\">\u7531\u65bc\u73fe\u4ee3\u79d1\u6280\u4ee5\u53ca 3C \u7522\u54c1\u7684\u666e\u53ca\uff0c\u4f7f\u5f97\u5b69\u5b50\u983b\u7e41\u7684\u63a5\u89f8\u96fb\u8996\u3001\u7db2\u8def\u3001\u624b\u6a5f\u2026\u7b49\uff0c\u56e0\u6b64 \u203b\u9063\u8a5e\u9020\u53e5\uff1a\u9063\u8a5e\u9020\u53e5\u5e38\u6709\u932f\u8aa4\u3002 \u932f\u5225\u5b57\u90e8\u5206\u7cfb\u7d71\u53ef\u4ee5\u85c9\u7531\u6b63\u78ba\u4f5c\u6587\u7684\u8a9e\u6599\u5eab\u4f86\u5c0b\u627e\u3001\u6bd4\u5c0d\u65b0\u4f5c\u6587\u7684\u932f\u5225\u5b57\uff0c\u56e0\u6b64\u6211\u5011\u53ef \u8a9e\u8a00\u6a21\u578b 1</td></tr><tr><td colspan=\"2\">\u5bb9\u6613\u7f3a\u4e4f\u8207\u4eba\u4e4b\u9593\u4e92\u52d5\u3001\u6e9d\u901a\u4ee5\u53ca\u60c5\u611f\u7684\u8868\u9054\uff0c\u76f8\u5c0d\u7684\uff0c\u5b78\u751f\u5beb\u7684\u4f5c\u6587\u5e38\u5e38\u662f\u4ee5\u6d41\u6c34\u5e33 \u4ea4\u4ee3\u7d93\u904e\uff0c\u6709\u7684\u5b78\u6821\u751a\u81f3\u4e0d\u8003\u4f5c\u6587\uff0c\u4f46\u96a8\u8457\u6559\u80b2\u653f\u7b56\u7684\u8b8a\u52d5\uff0c\u570b\u4e2d\u6559\u80b2\u6703\u8003\u52a0\u5165\u4e86\u4f5c\u6587 \u8a55\u91cf\u7684\u9805\u76ee\uff0c\u4f7f\u7684\u4f5c\u6587\u518d\u5ea6\u53d7\u5230\u5b78\u751f\u53ca\u5bb6\u9577\u7684\u91cd\u8996\u3002\u53ef\u662f\u53d7\u9650\u65bc\u5b78\u6821\u6559\u5b78\u6642\u6578\uff0c\u4f5c\u6587\u8f03 \u4e00\u7d1a\u5206 \u4e00\u7d1a\u5206\u7684\u6587\u7ae0\u5728\u8868\u9054\u4e0a\u5448\u73fe\u6975\u56b4\u91cd\u7684\u554f\u984c\uff0c\u9019\u7a2e\u6587\u7ae0\u660e \u986f\u5177\u6709\u4e0b\u5217\u7279\u5fb5 \u203b\u9063\u8a5e\u9020\u53e5\uff1a\u7528\u5b57\u9063\u8a5e\u6975\u4e0d\u6070\u7576\uff0c\u9817\u591a\u932f\u8aa4\uff1b\u6216\u6587\u53e5\u652f \u4ee5\u5075\u6e2c\u932f\u5225\u5b57\u3002\u5c07\u4f5c\u6587\u683c\u5f0f\u8f49\u8b8a\u6210\u96fb\u8166\u53ef\u4ee5\u8fa8\u8b58\u7684\u683c\u5f0f\uff0c\u570b\u4e2d\u751f\u4f5c\u6587\u5728\u66f8\u5beb\u6642\u662f\u7531\u4e0a\u800c \u983b\u7387 \u4e0b\uff0c\u7531\u53f3\u81f3\u5de6\uff0c\u6b64\u5916\u6284\u5beb\u6a19\u984c\u6642\u8981\u7a7a 4 \u683c\uff0c\u6bcf\u6bb5\u524d\u9762\u7a7a\u5169\u683c\u3002\u4f5c\u6587\u9577\u5ea6\u5e38\u5e38\u6703\u5f71\u97ff\u4f5c\u6587 \u7684\u7b49\u7d1a\u5c31\u5982\u6d2a\u7f8e\u96c0(\u6703\u8003\u7684\u6838\u5fc3\u8001\u5e2b)\u63d0\u5230:\"\u4f5c\u6587\u8003\u984c\u7684\u5c0e\u6587\u5f8c\u9762\u5747\u51fa\u73fe\u300c\u6587\u9577\u4e0d\u9650\u300d \uff0c \u6aa2\u7d22 \u8cc7\uf9be\u5eab \u5716\u4e09\u3001\u8209\uf9b5\uf96f\u660e\u65b7\u8a5e bigram \u793a\u610f\u5716</td></tr><tr><td colspan=\"2\">\u5f31\u7684\u5b78\u751f\u5bb9\u6613\u7f3a\u5c11\u88dc\u6551\u7684\u6a5f\u6703\u3002\u6211\u5011\u8a8d\u70ba\u672a\u4f86\u81ea\u5b78\u4f5c\u6587\u4ee5\u53ca\u5728\u5bb6\u7df4\u7fd2\uff0c\u53ef\u4ee5\u85c9\u7531\u81ea\u52d5\u5316 \u8a3a\u65b7\u7d50\u679c\u53ef\u4ee5\u8b93\u5b78\u751f\u5c0d\u8a5e\u53e5\u7d44\u5408\u7684\u7406\u89e3\u529b\u6709\u6240\u63d0\u5347\uff0c\u5e6b\u52a9\u5b78\u751f\u5beb\u51fa\u8f03\u6d41\u66a2\u7684\u53e5\u5b50\uff0c\u85c9\u6b64 \u63d0\u9ad8\u4ed6\u5011\u7684\u4f5c\u6587\u5206\u6578\u3002\u7cfb\u7d71\u6240\u4f9d\u8cf4\u7684 N-gram \u8a9e\u8a00\u6a21\u578b\uff0c\u5b83\u7684\u7279\u6027\u662f\u8a08\u7b97\u5b57\u8a5e\u9593\u7d44\u5408\u7684 \u6a5f\uf961\uff0c\u6a5f\uf961\u8d8a\u9ad8\u7684\u8a71\u5b57\u8a5e\u7d44\u5408\u7684\u6b63\u78ba\u6027\u8d8a\u9ad8\u4e5f\u5c31\u662f\u8d8a\u6d41\u66a2\uff0c\u800c\u8a9e\u8a00\u6a21\u578b\u6548\u679c\u76f8\u7576\u4f9d\u8cf4\u5927 \u53ef\u4ee5\u4f7f\u7528\u5e73\u6ed1(smoothing)\u7684\u65b9\u6cd5\u89e3\u6c7a\uff1b\u4ee5\u53ca\u8de8\uf9b4\u57df\u7684\u554f\u984c\uff0c\u53ea\u8981\u8a13\uf996\u8a9e\uf9be\u7684\u6027\u8cea\u8d8a\uf967\u540c (\u4e8c)\u3001\u7d50\u69cb\u7d44\u7e54 \u6216\u8aa4\u7f6e\u6a19\u9ede\uff0c\u5169\u8005\u5360\u4e86\u5c07\u8fd1\u516d\u6210\u7684\u6bd4\u91cd\u81ea\u7136\u662f\u4e0d\u5bb9\u5ffd\u8996\u7684\" (\u6797\u7d20\u73cd 2007\uff0c159)\u3002 Output \u65bc\u6e2c\u8a66\u7684\u6587\u7ae0\uff0c\u6211\u5011\u6240\u5efa\uf9f7\u8a9e\u8a00\u6a21\u578b\u7684\u6548\u679c\u5c31\u8d8a\u5dee\uff0c\u56e0\u6b64\u8a9e\uf9be\u5eab\u4e5f\u8981\u8ddf\u8457\u6539\u8b8a\u3002 \u9019\u88e1\u4e3b\u8981\u8a55\u91cf\u6240\u5beb\u7684\u4f5c\u6587\u5167\u5bb9\u662f\u5426\u7b26\u5408\u4e3b\u984c\uff0c\u5c31\u5982\u8521\u82f1\u4fca(2006\uff0c1)\u63d0\u5230\"\u7acb\u610f\u53d6\u6750:\u4e3b \u8981\u5728\u8a55\u91cf\u5b78\u751f\u662f\u5426\u80fd\u5207\u5408\u984c\u65e8\u4e26\u9078\u64c7\u5408\u9069\u7684\u7d20\u6750\"\u3002 \u5230 6 \u6bb5\uff0c\u4e0d\u8981\u8d85\u904e 6 \u6bb5\u901a\u5e38\u662f\u56db\u7d1a\u5206\u4ee5\u4e0a\u3002\u5167\u6587\u7a7a\u767d\u7684\u8a71\u5c31\u662f\u96f6\u7d1a\u5206\u3002\u6839\u64da\u6797\u7d20\u73cd\u7684\u7814 \u7a76\u986f\u793a\u6a19\u9ede\u7b26\u865f\u932f\u8aa4\u7684\u6bd4\u4f8b\u5f88\u9ad8\"32.9%\u7684\u4f5c\u54c1\u6a19\u9ede\u4f7f\u7528\u4e0d\u7576\uff0c26.5%\u7684\u4f5c\u54c1\u65b7\u53e5\u4e0d\u7576 \u7d9c\u5408\u8a55\u5206 \u578b\u7684\u8a13\uf996\u8a9e\uf9be\uff0c\u9019\u662f\u8a9e\u8a00\u6a21\u578b\u7136\u80fd\u5f85\u514b\u670d\u7684\u7f3a\u9ede\uff0c\u4f8b\u5982\u8cc7\uf9be\u7a00\u758f(Data sparseness)\u7684\u554f\u984c\uff0c \u96f6\u7d1a\u5206 \u4f7f\u7528\u8a69\u6b4c\u9ad4\u3001\u5b8c\u5168\u96e2\u984c\u3001\u53ea\u6284\u5beb\u984c\u76ee\u6216\u8aaa\u660e\u3001\u7a7a\u767d\u5377 \u5169\u6bb5\u5927\u6982\u662f\u4e8c\u7d1a\u5206\u3002\u5982\u679c\u53ea\u6709\u4e09\u6bb5\u5927\u90e8\u5206\u6700\u9ad8\u662f\u4e09\u7d1a\u5206\uff0c\u901a\u5e38\u662f\u4e03\u5230\u5341\u4e8c\u884c\u3002\u81f3\u5c11\u5beb 4 (\u4e00)\u3001\u7acb\u610f\u53d6\u6750 \u6307\u8003\u81f3\u5c11\u5beb\u4e94\u767e\u4e8c\u5341\u5b57\uff0c\u6c92\u6709\u9019\u6a23\u7684\u5206\u91cf\uff0c\u90fd\u4e0d\u53ef\u80fd\u5f97\u5230\u9ad8\u5206! \"(\u6d2a\u7f8e\u96c0 2013\uff0c 102)\u3002 \u7684\u8001\u5e2b\u4e4b\u5c08\u5bb6\u5224\u65b7)\uff0c\u884c\u6578 4 \u884c\u4ee5\u4e0a\u7d04 6\u30017 \u884c\u4ee5\u4e0b\u4f46\u4e0d\u5305\u542b\u6284\u7da0\u984c\u76ee\u5f15\u5c0e\u7684\u53e5\u5b50\uff0c\u4e14\u6709 \u5206\u516c\u5f0f \u4f5c\u6587\u7684\u5206\u6bb5\u4e5f\u6703\u5f71\u97ff\u5230\u8a55\u5206\uff0c\u4e00\u7d1a\u5206\u5927\u90fd\u662f\u4e00\u5230\u4e09\u884c\u5beb\u6210\u6bb5\u6216\u5169\u6bb5(\u4f9d\u64da\u4fee\u6539\u570b\u4e2d\u4f5c\u6587 \u5957\u7528\u7b97 \u8a9e\u8a00\u6a21\u578b 2 \u7684\u4f5c\u6587\u6559\u5b78\u7cfb\u7d71\u8f14\u52a9\u3002\u800c\u672c\u7cfb\u7d71\u958b\u767c\u4f5c\u6587\u6559\u5b78\u7cfb\u7d71\u4e4b\u53e5\u5b50\u6d41\u66a2\u5ea6\u5075\u6e2c\uff0c\u7d93\u7531\u7cfb\u7d71\u56de\u994b\u7684 \u96e2\u7834\u788e\uff0c\u96e3\u4ee5\u7406\u89e3\u3002 \u300c\u6587\u9577\u4e0d\u9650\u300d \u662f\u70ba\u4e86\u6015\u4eba\u6279\u8a55\u4ee5\u5b57\u6578\u8ad6\u6587\u7ae0\u512a\u52a3\uff0c\u4e0d\u904e\u4e0d\u8981\u88ab\u9a19\u4e86\uff0c\u5b78\u6e2c\u81f3\u5c11\u5beb\u516d\u767e\u5b57\uff0c (\u4e00)\u3001N-gram \u8a9e\u8a00\u6a21\u578b</td></tr><tr><td>\u4e8c\u3001\u7814\u7a76\u52d5\u6a5f (\u4e94)\u3001\u7814\u7a76\u76ee\u7684</td><td>\u7cfb\u7d71\u5224\u65b7\u4e4b</td></tr><tr><td colspan=\"2\">\u76ee\u524d\u570b\u4e2d\u4f5c\u6587\u4fee\u6539\u7cfb\u7d71\u5c11\u4e86\u91dd\u5c0d\u9023\u63a5\u8a5e\u932f\u8aa4\u7684\u8655\u7406:\u6797\u7d20\u73cd\u5206\u6790\u570b\u4e2d\u4f5c\u6587\u7684\u932f\u8aa4\u4e2d\uff0c\u7d50 \u7d9c\u5408\u4e0a\u8ff0\u7684\uf96f\u660e\uff0c\u672c\uf941\u6587\u4e3b\u8981\u7684\u7814\u7a76\u5728\u65bc\u53e5\u5b50\u6d41\u66a2\u5ea6\u7684\u5075\u6e2c\uff0c\u6b63\u78ba\u5224\u65b7\u53e5\u5b50\u662f\u5426\u901a\u9806\uff0c \u7d50\u679c\u8207\u63d0\u793a</td></tr><tr><td colspan=\"2\">\u679c\u986f\u793a:\"\u5728\u884c\u6587\u4f48\u5c40\u65b9\u9762\u6240\u72af\u932f\u8aa4\u7684\u7d71\u8a08:\u6709 45.3%\u7684\u4f5c\u54c1\u5728\u6587\u610f\u7684\u627f\u63a5\u4e0a\u4e0d\u9023\u8cab\uff0c\u662f \u7cfb\u7d71\u8a2d\u8a08\u65bc\u53ef\u89e3\u6c7a\u4e00\u822c\u6027\u554f\u984c\uff0c\u53ef\u96a8\u8457\u8a13\u7df4\u96c6\u7684\u589e\u52a0\uff0c\u800c\u589e\u5f37\u5c0d\u53e5\u5b50\u7684\u5224\u65b7\u3002\u96d6\u7136\u9019\u53ea \u8981\u5e6b\u52a9\u5b78\u751f\u5beb\u597d\u7684\u4f5c\u6587\u9996\u5148\u8981\u8b93\u7cfb\u7d71\u77e5\u9053\u5982\u4f55\u5224\u65b7\u51fa\u4e00\u7bc7\u662f\u597d\u7684\u4f5c\u6587\uff0c\u570b\u4e2d\u57fa\u6e2c\u4f5c\u6587\u7684 \u8a55\u91cf\u4e3b\u8981\u4ee5\u56db\u500b\u7bc4\u7587\u70ba\u4e3b:\"\u57fa\u6e2c\u5beb\u4f5c\u6e2c\u9a57\u96d6\u7136\u63a1\u7528\u6574\u9ad4\u6027\u8a55\u5206\u65b9\u6cd5\uff0c\u4f46\u8a55\u5206\u7684\u6642\u5019\u4ecd \u6bd4\u8f03\u56b4\u91cd\u7684\u554f\u984c\"(\u6797\u7d20\u73cd 2007\uff0c158)\u3002\u8521\u82f1\u4fca(2006\uff0c2)\u63d0\u5230\"\u5728\u7d50\u69cb\u7d44\u7e54\u4e0a\u7684\u57fa\u672c\u8981 \u6c42\uff0c\u5247\u662f\u610f\u5ff5\u524d\u5f8c\u4e00\u81f4(\u9996\u5c3e\u9023\u8cab)\u548c\u7d50\u69cb\u52fb\u7a31\u3002 \u662f\u4e00\u5c0f\u8d77\u6b65\uff0c\u6b64\u7cfb\u7d71\u672a\uf92d\u5c07\u6703\u6574\u5408\u5230\u96fb\u8166\u4f5c\u6587\u81ea\u52d5\u8a55\u5206\u7cfb\u7d71\uff0c\u96fb\u8166\u81ea\u52d5\u8a55\u5206\u7cfb\u7d71\u80fd 24 \u5716\u4e00\u3001\u4e2d\u6587\u4f5c\u6587\u6d41\u66a2\u5ea6\u5075\u6e2c\u7cfb\u7d71\u904b\u4f5c\u7684\uf9ca\u7a0b\u5716</td></tr></table>" | |
| }, | |
| "TABREF2": { | |
| "type_str": "table", | |
| "html": null, | |
| "num": null, | |
| "text": "Turing(GT) [14]\u8207 modified Kneser-Ney (mKN) [15]\u7684\u6f14\u7b97\u6cd5\u6548\u679c\uf967\u932f\uff0c\u4ee5\u4e0b\u5c07\u6703 \u7c21\u55ae\u4ecb\u7d39 GT \u8207 KN \u7684\u6f14\u7b97\u6cd5\u3002", | |
| "content": "<table><tr><td colspan=\"2\">Good-1\u3001 Good-Turing Discounting(GT)</td><td/></tr><tr><td colspan=\"4\">Good-Turing \u7684\u6f14\u7b97\u6cd5\u662f\u8abf\u6574\u5f9e\"r\" (r: \u8868\u793a\u51fa\u73fe r \u6b21\u7684\u5b57\uf969)\u81f3\"r*\"\uff0c\u4f9d\u64da\u5b83\u662f\u4e8c\u9805\u5f0f\u5206</td></tr><tr><td>\u5e03\u7684\u5047\u8a2d \uff0c\u5982\u516c\u5f0f(10)\u3002</td><td/><td/></tr><tr><td colspan=\"3\">* = (r + 1) +1 r < M</td><td>(10)</td></tr><tr><td colspan=\"4\">\u5176\u4e2d r N \u662f N-gram \u4e2d\u51fa\u73fe r \u6b21\u7684\u5b57\uf969\uff0cM \u662f\u754c\u9650\u503c\u901a\u5e38\ufa26\u5c0f\u65bc 5\u3002\u7279\u5225\u9700\u8981\u6ce8\u610f\u7684</td></tr><tr><td colspan=\"3\">\u662f r =0 \u6642\uff0c\u4ee3\u8868 N-gram \u4e2d\u51fa\u73fe 0 \u6b21\u7684\u5b57\uf969\uff1a</td></tr><tr><td>* = 1</td><td/><td/></tr><tr><td>0</td><td/><td/></tr><tr><td colspan=\"3\">\u5176\u4e2d 0 N \u662f\u8868\u793a\u5f9e\u672a\u51fa\u73fe\u904e\uff0c \u56e0\u6b64\u6298\u6263\u904e\u5f8c\u6539\u5beb\u6210\u4e0b\u5f0f(11)\uff1a</td></tr><tr><td>( 1 \u2026 ) =</td><td>*</td><td/><td>(11)</td></tr><tr><td colspan=\"4\">Good-Turing \u50c5\u9069\u7528\u65bc r < 5\uff0c\u800c\u4e14\u5fc5\u9808\u91cd\u65b0\u6a19\u6e96\u5316\u4ee5\u78ba\u4fdd\u6a5f\uf961\u7e3d\u548c\u70ba 1\u3002\u5982\u6b64\u8abf\u6574\u904e\u5f8c\uff0c</td></tr><tr><td colspan=\"4\">\u539f\u672c\u51fa\u73fe\u983b\uf961\u70ba 0 \u7684\u5b57\uff0c\u5c07\u6703\u88ab\u8abf\u6574\u63d0\u6607\u6210\u70ba\u5c0f\uf969\u4f4d\uf969\uff0c\u6240\u4ee5\u907f\u514d\uf9ba\u6a5f\uf961\u70ba\uf9b2\u7684\u800c\u5c0e\u81f4</td></tr><tr><td>\u7121\u6cd5\u8a08\u7b97\u6574\u500b\uf906\u5b50\u6a5f\uf961\u7684\u932f\u8aa4\u60c5\u5f62\u3002</td><td/><td/></tr><tr><td colspan=\"3\">2\u3001 Modified Kneser-Ney discounting (mKN)</td></tr><tr><td/><td/><td/><td>7)</td></tr><tr><td>\u53e6\u5916\u518d\u5b9a\u7fa9 Perplexity\uff0c\u5982\u4e0b\u5f0f(8)\uff1a</td><td/><td/></tr><tr><td colspan=\"2\">Perplexity = 2</td><td/><td>(8)</td></tr><tr><td>\u5be6\u969b\u8a08\u7b97\u6642\u4ea6\u5957\u7528\u6539\u5beb\u904e\u7684\u516c\u5f0f\uff1a</td><td/><td/></tr><tr><td colspan=\"2\">Perplexity\u2032 = 10</td><td>\u2032 /</td><td>(9)</td></tr><tr><td colspan=\"4\">\u5176\u4e2d W \u662f\u4e00\u500b\uf906\u5b50\u7684\u55ae\u5b57\uf969\u9664\u4ee5 W \u7684\u76ee\u7684\u662f\u907f\u514d\u7576\u53e5\u5b50\u8d8a\u9577\u6642\u6a5f\u7387\u8d8a\u4f4e\u7684\u60c5\u6cc1\u767c\u751f\u3002</td></tr><tr><td colspan=\"4\">Perplexity \u8d8a\u4f4e\u4ee3\u8868\u53e5\u5b50\u4e2d\u5b57\u8a5e\u7684\u7d44\u5408\u6a5f\uf961\u5f88\u9ad8\uff0c\u4e5f\u5c31\u662f\uf96f\u9019\u500b\uf906\u5b50\u662f\u6bd4\u8f03\u591a\u4eba\u9019\u6a23\u5beb</td></tr><tr><td colspan=\"4\">\u7684\u7576\u7136\u4e5f\u6703\u8f03\u70ba\u901a\u9806\u3002\u4f46\u662f N-gram \u8a9e\u8a00\u6a21\u578b\u9084\u662f\u6709\u7f3a\u9ede\u5fc5\u9700\u8981\u514b\u670d\uff1a\u8a9e\u8a00\u6a21\u578b\u5728\u4e0d\u5920</td></tr><tr><td colspan=\"4\">\u9f90\u5927\u6642\u5c0c\u7121\u6cd5\u6db5\u84cb\u6240\u6709\u53ef\u80fd\u7684\u5b57\u8a5e\u7d44\u5408\uff0c\u4e5f\u5c31\u662f\u8cc7\uf9be\u7a00\u758f\u7684\u554f\u984c\uff0c\u5373\u6709\u4e9b\u5b57\u8a5e\u7684\u7d44\u5408\u6c92</td></tr><tr><td colspan=\"4\">\u6709\u88ab\u8a13\u7df4\u5230\uff0c\u4f7f\u7684\u5728\u67e5\u8a62\u983b\u7387\u6642\u6703\u6709\u96f6\u7684\u554f\u984c\u767c\u751f\uff0c\u5c0e\u81f4\u7121\u6cd5\u6b63\u78ba\u7b97\u5206\u7684\u932f\u8aa4\uf9fa\u6cc1\u3002\u56e0</td></tr><tr><td colspan=\"4\">\u6b64\u70ba\uf9ba\u89e3\u6c7a\u9019\u500b\u554f\u984c\u6211\u5011\u9084\u9700\u4f7f\u7528\u5e73\u6ed1(smoothing)\u7684\u65b9\u6cd5\u4f86\u6539\u5584\u6a5f\uf961\u70ba\uf9b2\u7684\u4f8b\u5916\u60c5</td></tr><tr><td>\u6cc1\u3002</td><td/><td/></tr><tr><td>(\u4e8c)\u3001Smoothing</td><td/><td/></tr><tr><td colspan=\"4\">Smoothing \u7684\u65b9\u6cd5\u53ef\u5206\u6210\u6a21\u5f0f\u7d50\u5408\u7684\u65b9\u6cd5[14]\u4ee5\u53ca\u6298\u6263\u7684\u65b9\u6cd5\uff0c\u6a21\u578b\u7d50\u5408\u7684\u65b9\u5f0f\u5c31\u662f\uf9dd</td></tr><tr><td colspan=\"4\">\u7528\u5167\u63d2\u6cd5\u548c\u88dc\u63d2\u6cd5\uff0cbigram \u7121\u6548\u6642\uff0c\u4f7f\u7528 unigram\uff1b\u800c\u6298\u6263\u7684\u65b9\u6cd5\u5c31\u662f\u8abf\u6574\u6a5f\uf961\uff0c\u5c07\u6a5f</td></tr><tr><td colspan=\"4\">\uf961\u8f03\u9ad8\u8005\u628a\u503c\u5206\u914d\u7d66\u6a5f\uf961\u70ba\uf9b2\u8005\u3002\u5be6\u9a57\u662f\u7528 Interpolated Kneser-Ney smoothing\u3002\u800c</td></tr></table>" | |
| } | |
| } | |
| } | |
| } |