| { | |
| "paper_id": "O14-1006", | |
| "header": { | |
| "generated_with": "S2ORC 1.0.0", | |
| "date_generated": "2023-01-19T08:04:29.822175Z" | |
| }, | |
| "title": "Research on Hakka Word Segmentation Processes in Chinese-to-Hakka Text-to-Speech System", | |
| "authors": [ | |
| { | |
| "first": "", | |
| "middle": [], | |
| "last": "\u9ec3\u8c50\u9686", | |
| "suffix": "", | |
| "affiliation": {}, | |
| "email": "" | |
| }, | |
| { | |
| "first": "", | |
| "middle": [], | |
| "last": "\u570b\u7acb\u806f\u5408\u5927\u5b78", | |
| "suffix": "", | |
| "affiliation": {}, | |
| "email": "" | |
| } | |
| ], | |
| "year": "", | |
| "venue": null, | |
| "identifiers": {}, | |
| "abstract": "Language is a major tool for cultural inheritance especially for the minority nationality, for example Hakka and aborigine language in Taiwan. As second ethnic besides Minnan dialect, the population of Hakka in Taiwan is one seventh. According to the recently reports of Hakka usage survey in Taiwan, the difficulties to inherit the culture of Hakka is missed in spoken Hakka language, the reason is the environments for learning and has led to the results of descending population for communicating by Hakka. It will become crucial for the cultural inheritance of Hakka. Therefore, we has developed the Text-to-Speech method and system for Hakka language, and our goal is building environments for leaning the Hakka language, our some applied system such as: 58", | |
| "pdf_parse": { | |
| "paper_id": "O14-1006", | |
| "_pdf_hash": "", | |
| "abstract": [ | |
| { | |
| "text": "Language is a major tool for cultural inheritance especially for the minority nationality, for example Hakka and aborigine language in Taiwan. As second ethnic besides Minnan dialect, the population of Hakka in Taiwan is one seventh. According to the recently reports of Hakka usage survey in Taiwan, the difficulties to inherit the culture of Hakka is missed in spoken Hakka language, the reason is the environments for learning and has led to the results of descending population for communicating by Hakka. It will become crucial for the cultural inheritance of Hakka. Therefore, we has developed the Text-to-Speech method and system for Hakka language, and our goal is building environments for leaning the Hakka language, our some applied system such as: 58", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "Abstract", | |
| "sec_num": null | |
| } | |
| ], | |
| "body_text": [ | |
| { | |
| "text": "(2) ", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "\u5176\u4e2d P(W i |W i-1 )\u53ef\u5229\u7528 maximum likelihood estimation(MLE)\u4f86\u8a08\u7b97\uff1a C(W i-1 ,W i ) P(W i |W i-1 ) \u8d77 C(W i-1 (3) ) \u82e5\u9047\u5230\u8a13\u7df4\u8cc7\u6599 C(W i-1 , W i ) = 0 \u6642\uff0c\u6211\u5011\u5c07 Bi-gram \u6a5f\u7387\u4ee5\u70ba \u03b1 \u503c\u4ee3\u66ff\uff0c\u8f49\u63db\u5982\u4e0b\uff1a C(W i-1 ,W i ) P(W i |W i-1 ) = \u2211 W \u2208 v C(W i-1 ,W) , if C(W i-1 , W i ) > 0 (4) \u5176\u4e2d P(W i )\uff1a \u03b1 , if C(W i-1 , W i ) = 0 P(W i ) = 1+C(W i ) (5) v+[\u2211 v C(W j )] v j=1 C(W j ) = 47079\uff0c\u70ba\u5ba2\u8a9e Uni-gram \u8a9e\u8a00\u6a21\u578b\u4e2d\u7684\u7e3d\u8a5e\u983b\u6578\u3002 V = 8931\uff0c\u70ba Uni-gram \u8a9e\u8a00\u6a21\u578b\u7684\u7e3d type \u6578(\u8a5e\u6578)\u3002 \u03b1 = 10 -3 \uff0c\u70ba\u4ee3\u66ff\u7576 C(W , W ) = 0 \u6642\uff0c\u7528\u4ee5\u66ff\u4ee3 C(W i-1 ,W i ) \u7684\u503c\uff0c\u9019\u662f\u7d93\u7531\u5be6\u9a57\u5f97\u5230\uff0c\u6211 \u5011\u6e2c\u8a66\u4e86 10 -1 , 10 -2 , 10 -3 , \u2026 , 10 -8 \u7b49\u503c\u3002 \u2211 W \u2208 v C(W i-1 ,W) \u5716\u516d\u3001\u4e2d\u6587\u65b7\u8a5e\u642d\u914d\u5ba2\u8a9e Uni-gram \u53ca Bi-", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "", | |
| "sec_num": null | |
| }, | |
| { | |
| "text": "\u5177\u6709\u4e2d\u6587\u6587\u53e5\u548c\u5ba2\u6587\u6587\u53e5 1 \u5c0d 1 \u5c0d\u61c9\u7684\u5e73\u884c\u8cc7\u8a0a\u7684\u8a9e\u6599\u3002", | |
| "cite_spans": [], | |
| "ref_spans": [], | |
| "eq_spans": [], | |
| "section": "", | |
| "sec_num": null | |
| } | |
| ], | |
| "back_matter": [], | |
| "bib_entries": { | |
| "BIBREF1": { | |
| "ref_id": "b1", | |
| "title": "\u6797\u5343\u7fd4\uff0cChinese Word Segmentation using Specialized HMM\uff0c\u570b\u7acb\u4e2d\u592e\u5927\u5b78\u8cc7\u8a0a\u5de5\u7a0b\u6240\u78a9\u58eb \u8ad6\u6587\uff0c2005\u3002", | |
| "authors": [], | |
| "year": null, | |
| "venue": "", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "\u6797\u5343\u7fd4\uff0cChinese Word Segmentation using Specialized HMM\uff0c\u570b\u7acb\u4e2d\u592e\u5927\u5b78\u8cc7\u8a0a\u5de5\u7a0b\u6240\u78a9\u58eb \u8ad6\u6587\uff0c2005\u3002", | |
| "links": null | |
| }, | |
| "BIBREF4": { | |
| "ref_id": "b4", | |
| "title": "Chinese Word Segmentation as Character Tagging", | |
| "authors": [ | |
| { | |
| "first": "Nianwen", | |
| "middle": [], | |
| "last": "Xue", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2003, | |
| "venue": "Computational Linguistics", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Nianwen Xue, Chinese Word Segmentation as Character Tagging, Computational Linguistics 2003", | |
| "links": null | |
| }, | |
| "BIBREF6": { | |
| "ref_id": "b6", | |
| "title": "A Two-Stage Statistical Word Segmentation System for Chinese, Proceeding of The Second SIGHAN Workshop on Chinese Language Processing", | |
| "authors": [ | |
| { | |
| "first": "Guhong", | |
| "middle": [], | |
| "last": "Fu", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "K", | |
| "middle": [ | |
| "K" | |
| ], | |
| "last": "Luke", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 2003, | |
| "venue": "", | |
| "volume": "17", | |
| "issue": "", | |
| "pages": "156--159", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Guhong Fu and K.K. Luke., A Two-Stage Statistical Word Segmentation System for Chinese, Proceeding of The Second SIGHAN Workshop on Chinese Language Processing 2003, Vol. 17, pp.156- 159.", | |
| "links": null | |
| }, | |
| "BIBREF7": { | |
| "ref_id": "b7", | |
| "title": "Word Identification For Mandarin Chinese Sentences", | |
| "authors": [ | |
| { | |
| "first": "Shing-Huan", | |
| "middle": [], | |
| "last": "Keh-Jiann Chen", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "", | |
| "middle": [], | |
| "last": "Liu", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": null, | |
| "venue": "Proceedings of COLING 1992", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "101--107", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Keh-Jiann Chen and Shing-Huan Liu, Word Identification For Mandarin Chinese Sentences, Proceedings of COLING 1992, pp.101-107.", | |
| "links": null | |
| }, | |
| "BIBREF8": { | |
| "ref_id": "b8", | |
| "title": "Unknown Word Detection and Segmentation of Chinese Using Statistical and Heuristic Knowledge", | |
| "authors": [ | |
| { | |
| "first": "Jian-Yun", | |
| "middle": [], | |
| "last": "Nie", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Marie-Louise", | |
| "middle": [], | |
| "last": "Hannan", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Wanying", | |
| "middle": [], | |
| "last": "Jin", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 1995, | |
| "venue": "Communications of COLIPS", | |
| "volume": "5", | |
| "issue": "", | |
| "pages": "47--57", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Jian-Yun Nie, Marie-Louise Hannan, and Wanying Jin, Unknown Word Detection and Segmentation of Chinese Using Statistical and Heuristic Knowledge, Communications of COLIPS 1995, Vol. 5, pp.47-57.", | |
| "links": null | |
| }, | |
| "BIBREF9": { | |
| "ref_id": "b9", | |
| "title": "Word Segmentation", | |
| "authors": [ | |
| { | |
| "first": "Andi", | |
| "middle": [], | |
| "last": "Wu", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Zixin", | |
| "middle": [], | |
| "last": "Jiang", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": 1998, | |
| "venue": "Sentence Analysis, International Conference on Chinese Information Processing in Beijing China", | |
| "volume": "", | |
| "issue": "", | |
| "pages": "169--180", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Andi Wu, Zixin Jiang, Word Segmentation In Sentence Analysis, International Conference on Chinese Information Processing in Beijing China 1998, pp.169-180.", | |
| "links": null | |
| }, | |
| "BIBREF10": { | |
| "ref_id": "b10", | |
| "title": "Improved Source-Channel Models for Chinese Word Segmentation, the 41st Annual Meeting on Association for Computational Linguistics 2003", | |
| "authors": [ | |
| { | |
| "first": "Mu", | |
| "middle": [], | |
| "last": "Jian-Feng Gao", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "Chang-Ning", | |
| "middle": [], | |
| "last": "Li", | |
| "suffix": "" | |
| }, | |
| { | |
| "first": "", | |
| "middle": [], | |
| "last": "Huang", | |
| "suffix": "" | |
| } | |
| ], | |
| "year": null, | |
| "venue": "", | |
| "volume": "1", | |
| "issue": "", | |
| "pages": "272--279", | |
| "other_ids": {}, | |
| "num": null, | |
| "urls": [], | |
| "raw_text": "Jian-feng Gao, Mu Li, and Chang-Ning Huang, Improved Source-Channel Models for Chinese Word Segmentation, the 41st Annual Meeting on Association for Computational Linguistics 2003, Vol. 1, pp.272-279.", | |
| "links": null | |
| } | |
| }, | |
| "ref_entries": { | |
| "FIGREF0": { | |
| "type_str": "figure", | |
| "text": "gram \u8a9e\u8a00\u6a21\u578b\u7684\u6df7\u5408\u5f0f\u5206\u6578\u7b97\u6cd5\uff0c\u7bc4\u4f8b\u3002 \u4ee5\u4e0a\u5716\u70ba\u4f8b\uff0c\u5982\u8981\u8a08\u7b97\u4e0a\u689d\u65b7\u8a5e\u5e8f\u5217 Seo\u03b3e(S, \u9019, \u96bb, \u9593\u623f, \u4e2a, \u5149\u7dda, \u6bcb, \u597d)\uff0c\u5176 Mix-gram \u7684 \u5206\u6578\u8a08\u7b97\u516c\u5f0f\u5982\u4e0b\uff1a Seo\u03b3e(S, \u9019, \u96bb, \u9593\u623f, \u4e2a, \u5149\u7dda, \u6bcb, \u597d) = [P(\u9019-S) * P(\u9019)\uff3d * [P(\u96bb|\u9019) * P(\u96bb)\uff3d * [P(\u9593\u623f-\u96bb) * P(\u9593\u623f)\uff3d * [P(\u4e2a|\u9593\u623f) * P (\u4e2a)\uff3d * [P(\u5149\u7dda-\u4e2a) * P(\u5149\u7dda)\uff3d * [P(\u6bcb|\u5149\u7dda) * P(\u6bcb)\uff3d * [P(\u597d-\u6bcb) * P(\u597d)\uff3d )\u3001\u53ec\u56de\u7387(Recall)\u3001\u4ee5\u53ca F-\u5206\u6578(F-score)\u4f86\u8a55\u4f30\u7cfb\u7d71\u7684\u6548\u80fd\uff0c\u9019\u4e09\u7a2e\u65b9\u6cd5\u7684\u5b9a\u7fa9\u5982\u4e0b\u6240\u793a\uff1a \u9593\uff0c\u7531\u5b57\u4e32 A \u8f49\u63db\u6210\u5b57\u4e32 B \u7684\u6700\u5c0f\u7de8\u8f2f\u8ddd\u96e2(Insertions, Deletions \u6216 Substitutions)\uff0c\u8a08\u7b97\u65b9\u5f0f\u5982\u4e0b\uff1a D(i \u4e00 1, j) + Inse\u03b3tCost(t\u03b1\u03b3get i ) D(i, j) = min D(i \u4e00 1, j \u4e00 1) + Su \u5fc5 stituteCost(sou\u03b3ee j , t\u03b1\u03b3get i ) D(i, j \u4e00 1) + DeleteCost(sou\u03b3ee j ) (9) \u5176\u4e2d\uff1a Su \u5fc5 stituteCost = 0 if t\u03b1\u03b3get\ufe5di\ufe5e = sou\u03b3ee\ufe5dj\ufe5e 1 othe\u03b3wise Inse\u03b3tCost = \u4e4b\u9593\u7684\u503c\uff0c\u5373\u70ba A\u3001B \u5b57\u4e32\u9593\u7684\u76f8\u4f3c\u5ea6\uff1a Simil\u03b1\u03b3ity(A, \u7684) = 1 \u4e00 D(A,B) M\u03b1\u03c7 Length(A,B)", | |
| "num": null, | |
| "uris": null | |
| }, | |
| "TABREF0": { | |
| "content": "<table><tr><td>\u5408\uff0c\u518d\u4f7f\u7528\u7d93\u9a57\u6cd5\u5247\u5254\u9664\u4e0d\u53ef\u80fd\u7684\u8a5e\u5f59\u7d44\u5408\uff0c\u7136\u5f8c\u518d\u5229\u7528\u5256\u6790\u5668\u89e3\u6c7a\u5269\u4e0b\u7684\u6b67\u7fa9\u554f\u984c\uff0c\u6700\u5f8c\u5f97 \u6587\u53e5\u65b7\u8a5e\u7b54\u6848\u7684\u540c\u6642\uff0c\u627e\u51fa\u4e00\u4e9b\u5c1a\u672a\u88ab\u6536\u9304\u5728\u8fad\u5178\u4e2d\u7684\u570b\u5ba2\u8a9e\u5c0d\u7167\u4e4b\u8a5e\u76ee\u3002\u6700\u5f8c\uff0c\u4e5f\u91dd\u5c0d\u6bcf\u500b 3.3.1 \u8a9e\u6599\u4f86\u6e90 \u6211\u5011\u5c07\u5ba2\u59d4\u6703\u56db\u7e23\u8154\u8a8d\u8b49\u6559\u6750\u7684\u5ba2\u8a9e\u8a9e\u6599 6640 \u53e5\u53e5\u5b50\uff0c\u4f9d\u7167\u53e5\u5b50\u7684\u7de8\u865f\u9806\u5e8f\uff0c\u7531\u5c0f\u5230\u5927 \u539f\u5247 2\uff1a\u5982\u679c\u6709\u5f88\u660e\u986f\u4e14\u96e2\u8b5c\u7684\u65b7\u8a5e\u932f\u8aa4\uff0c\u8981\u4eba\u5de5\u4ecb\u5165\u4fee\u6b63\u3002 \u5728\u6b64\u4f8b\u5b50\u4e2d\uff0c\u5ba2\u8a9e\u7684\u300c\u9ebc\u4e2a\u300d\u7ffb\u6210\u4e2d\u6587\u300c\u4e00\u4e9b\u300d \uff0c\u4f9d\u5ba2\u8a9e\u6587\u610f\u4f86\u770b\u904e\u65bc\u6a21\u7cca\u3002\u56e0\u6b64\uff0c\u53ef\u900f 4.1 \u7cfb\u7d71\u67b6\u69cb \u4ed4\u300d\u9019\u7b46\u500b\u570b\u5ba2\u8a9e\u5c0d\u7167\u8cc7\u6599\uff0c\u4f46\u7919\u65bc\u4e2d\u6587\u65b7\u8a5e\u908a\u754c\u4e4b\u9650\u5236\uff0c\u53ea\u80fd\u7531\u4e00\u500b\u8a5e\u300c\u6ed1\u6e9c\u6e9c\u4ed4\u300d\u8f49\u6210 \u4f9d\u7167\u4e0a\u8ff0\u65b9\u6cd5\uff0c\u7522\u751f\u4ee5\u4e0b\u65b0\u589e\u5019\u9078\u5217\u8868\uff1a 4.3 \u5ba2\u8a9e\u65b7\u8a5e\u65b9\u6cd5 4.3.2 \u4e2d\u6587\u65b7\u8a5e\u642d\u914d\u5ba2\u8a9e\u8a5e\u983b\u7684\u76f4\u63a5\u7ffb\u8b6f\u6cd5 \u672c\u65b7\u8a5e\u65b9\u6cd5\u7684\u5206\u6578\u8a08\u7b97\u65b9\u5f0f\uff0c\u662f\u6df7\u5408\u5f0f Mix-gram(Uni-gram+Bi-gram)\u5206\u6578\u7b97\u6cd5\uff0c\u5982\u4e0b\u5217\u5f0f</td></tr><tr><td>\u5217\u5206\u6578\u7684\u7b97\u6cd5\uff0c\u627e\u51fa\u5206\u6578\u6700\u9ad8\u7684\u5ba2\u8a9e\u8a5e\u8f49\u63db\u5e8f\u5217\uff0c\u4f86\u5f97\u5230\u7b2c\u4e8c\u968e\u6bb5\u8f49\u63db\u5f8c\u7684\u7d50\u679c\u3002 \u5230\u65b7\u8a5e\u5e8f\u5217\u3002\u5728\u5be6\u9a57\u4e0a F \u5206\u6578\u9ad8\u9054 99%\u3002 \u800c\u6bd4\u8f03\u4e0d\u540c\u7684\u505a\u6cd5\u662f Gao \u7b49\u4eba [20]\u63d0\u51fa\u7684\u5c08\u6709\u540d\u8a5e\u6cd5\u5247\uff0c\u9019\u4e9b\u6cd5\u5247\u5e6b\u52a9\u7cfb\u7d71\u6293\u53d6\u672a\u77e5\u7684\u5c08 \u6709\u540d\u8a5e\u5982:\u4eba\u540d\u3001\u5730\u540d\u3001\u7d44\u7e54\u540d\u3001\u5916\u4f86\u8a9e\u97f3\u8b6f\u4eba\u540d\u3002\u5728\u6700\u5f8c\u5be6\u9a57\u4e0a F \u5206\u6578\u7d04\u70ba 96%\u5de6\u53f3\u3002 2.2 \u5ba2\u8a9e\u65b7\u8a5e \u8a5e\u76ee\uff0c\u9032\u884c\u4eba\u5de5\u6821\u6b63\u5de5\u4f5c\uff0c\u53bb\u9664\u91cd\u8907\u6216\u4e0d\u5408\u7406\u7684\u8a5e\u76ee\u4ee5\u53ca\u6a19\u8a18\u62fc\u97f3\u3002 \u8868\u4e00\u3001\u570b\u5ba2\u8a9e\u5c0d\u7167\u8fad\u5178\u8cc7\u6599\u6a23\u8c8c \u6b04\u4f4d \u5167\u5bb9 \u8aaa\u660e ID 13267 \u8cc7\u6599\u5eab\u4e2d ID \u6709\u4e2d\u6587\u3001\u5ba2\u8a9e\u7684\u8a5e\u76ee\u3001\u62fc\u97f3\u53ca\u4f8b\u53e5\uff0c\u5982\u8868\uff1a \u8868\u4e09\u3001\u5ba2\u59d4\u6703\u8a9e\u6599\u7684\u8cc7\u6599\u6a23\u8c8c \u8868\u5341\u3001\u6a19\u8a18\u539f\u5247 2 \u7bc4\u4f8b \u539f\u65b7\u8a5e \u9019(Nep) \u7fa4(Nf) \u5c0f\u5b69\u5b50(Na) \uff0c(COMMACATEGORY) \u6bcf(Nes) \u5929\u90fd(Na) \u4eba\u5de5\u4fee\u6b63 \u9019(Nep) \u7fa4(Nf) \u5c0f\u5b69\u5b50(Na) \uff0c(COMMACATEGORY) \u6bcf(Nes) \u5929(Nf) \u90fd \u6a19\u8a18\u5de5\u5177\u4e2d\u7684\u8a5e\u5178\u67e5\u8a62\u529f\u80fd\uff0c\u627e\u5230\u5ba2\u8a9e\u8a5e\u300c\u9ebc\u4e2a\u300d\u80fd\u7ffb\u6210\u7684\u570b\u8a9e\u8a5e\u6709\u54ea\u4e9b\uff0c\u767c\u73fe\u5230\u300c\u4ec0\u9ebc\u300d\u9019 \u500b\u4e2d\u6587\u8a5e\u6700\u8cbc\u8fd1\u6587\u610f\u3002\u56e0\u6b64\uff0c\u624b\u52d5\u5c07\u4e2d\u6587\u53e5\u5b50\u7684\u300c\u4e00\u4e9b\u300d\u6539\u70ba\u300c\u4ec0\u9ebc\u300d \uff0c\u91cd\u65b0\u8207\u300c\u9ebc\u4e2a\u300d\u914d\u5c0d\u3002 \u539f\u5247 5\uff1a\u82e5\u6c92\u8fa6\u6cd5\u6a19\u8a18\u5b8c\u4e00\u6574\u53e5\uff0c\u4f46\u90e8\u5206\u8a5e\u80fd\u5c0d\u61c9\uff0c\u5247\u9078\u64c7\u300c\u653e\u68c4\uff0c\u5132\u5b58\u8a5e\u914d\u5c0d\u7d50\u679c\u300d \u3002 \u5728\u672c\u8ad6\u6587\u7684\u5ba2\u8a9e\u65b7\u8a5e\u7cfb\u7d71\u4e2d\uff0c\u8f38\u5165\u662f\u300c\u4e2d\u6587\u6587\u53e5\u300d \uff0c\u8f38\u51fa\u662f\u300c\u5ba2\u8a9e\u6587\u53e5\u300d\u7684\u65b7\u8a5e\u53ca\u8a5e\u6027\u6a19\u8a18 \u7d50\u679c\u3002\u7c21\u800c\u8a00\u4e4b\uff0c\u4ee5\u4e2d\u6587\u6587\u53e5\u8f38\u5165\u5f8c\uff0c\u7d93\u6b64\u5ba2\u8a9e\u65b7\u8a5e\u6a21\u7d44\u8655\u7406\uff0c\u7522\u751f\u5177\u6709\u65b7\u8a5e\u8207\u8a5e\u6027\u7684\u5ba2\u8a9e\u6587 \u505a\u66f4\u9032\u4e00\u6b65\u7684\u6587\u53e5\u5206\u6790\u8655\u7406\u3002\u5982\uff1a\u7528\u65bc Hakka Text to Speech (HTTS)\u4e2d\u7684\u505c\u9813\u9810\u4f30\u6a21\u578b\u4e2d\uff0c\u9810\u6e2c \u300c\u6ed1/\u6e9c\u6e9c\u300d\u5169\u500b\u8a5e\u3002\u6b64\u60c5\u6cc1\u6703\u76f4\u63a5\u7684\u5f71\u97ff\u5230\u8f49\u63db\u7684\u6b63\u78ba\u7387\u3002 \u8868\u5341\u4e03\u3001\u4e2d\u6587\u65b7\u8a5e\u908a\u754c\u9650\u5236\uff0c\u7bc4\u4f8b 2 \u4e2d\u6587\u65b7\u8a5e \u5927\u5bb6/\u4f86\u53bb/\u5ba2\u5bb6\u5e84/\u8d70/\u4e00/\u8d70/\u3002 \u8868\u4e8c\u5341\u3001\u5ba2\u8a9e\u8a5e\u65b0\u8a5e\u983b\u5019\u9078\u8868 \u4e2d\u6587\u8a5e \u8a5e\u983b \u96d5\u523b 266 4.3.1 \u4e2d\u6587\u65b7\u8a5e\u642d\u914d\u570b\u5ba2\u8a9e\u5c0d\u7167\u8fad\u5178\u7684\u76f4\u7ffb\u6cd5 \u672c\u65b9\u6cd5\u7684\u6d41\u7a0b\uff1a 2. \u67e5\u627e\u570b\u5ba2\u8a9e\u5c0d\u7167\u8fad\u5178\uff0c\u627e\u51fa\u5c0d\u61c9\u7684\u5ba2\u8a9e\u8a5e\uff0c\u4e26\u9078\u64c7\u8cc7\u6599 ID \u6392\u5e8f\u7b2c\u4e00\u4f4d\u8005\uff0c\u4e26\u5c07\u4e2d\u6587\u8a5e\u8f49\u63db \u4ee5\u4e0b\u662f\u672c\u65b9\u6cd5\u7684\u7a0b\u5f0f\u6d41\u7a0b\uff1a 1. \u900f\u904e\u4e2d\u6587\u65b7\u8a5e\u7cfb\u7d71\u5f97\u5230\u65b7\u8a5e\u53ca\u6a19\u8a5e\u6027\u7684\u7d50\u679c\u3002 3. \u6311\u9078\u8a5e\u6027\u8207\u8a5e\u6027\u6a19\u8a18\u7d50\u679c\u76f8\u540c\u8005\u4e14\u8a5e\u983b\u6700\u9ad8\u3001\u5b57\u9577\u6700\u77ed\u8005\u70ba\u7d50\u679c\uff0c\u82e5\u90fd\u627e\u4e0d\u5230\u5247\u4ee5\u4e2d\u6587\u8a5e \u5b50\uff1a Seo\u03b3e(< S >, W 1 , W 2 , W 3 , \u2026 , W n ) = \u5206\u70ba\u8a9e\u6599 A \u767c\u7684\u5ba2\u8a9e\u65b7\u8a5e\u7b54\u6848\u6a19\u8a18\u5de5\u5177\uff0c\u4ee5\u534a\u81ea\u52d5\u4eba\u5de5\u5224\u65b7\u65b9\u5f0f\uff0c\u6a19\u8a18\u51fa\u5ba2\u8a9e\u65b7\u8a5e\u7684\u7b54\u6848\u3002\u5176\u4e2d\u6bcf\u4eba\u7684\u6a19 \u5728(P) \u6c99\u6d32(Na) \u4e0a(Nes) \u73a9(VC) \u6454\u8de4(VA) \u3002(P) \u53e5\u7684\u8f38\u51fa\u3002\u6a19\u8a18\u65b7\u8a5e\u53ca\u8a5e\u6027\u5169\u7a2e\u7279\u5fb5\u5f8c\u7684\u7d50\u679c\uff0c\u53ef\u518d\u4f7f\u7528\u6587\u6cd5\u5256\u6790\u5668\u5206\u6790\u3001\u5f97\u5230\u6587\u6cd5\u7d50\u69cb\u6a39\uff0c \u4e2d\u6587\u53e5\u5b50 \u5927\u5bb6\u4f86\u53bb\u5ba2\u5bb6\u5e84\u8d70\u4e00\u8d70\u3002 \u690d\u7269 532 1. \u900f\u904e\u4e2d\u6587\u65b7\u8a5e\u7cfb\u7d71\u5f97\u5230\u4e2d\u6587\u65b7\u8a5e\u53ca\u6a19\u8a5e\u6027\u7d50\u679c\u3002 2. \u67e5\u627e\u570b\u5ba2\u8a9e\u5c0d\u7167\u8a5e\u5178\uff0c\u627e\u51fa\u5c0d\u61c9\u7684\u5ba2\u8a9e\u8a5e\u3002 i=2 \u03b1\u03b3gmin \u4e00 \ufe5blog e</td></tr><tr><td>\u56e0\u7919\u65bc\u4eba\u529b\u6709\u9650\u3001\u8a9e\u6599\u6536\u96c6\u7684\u56f0\u96e3\uff0c\u4ecd\u6709\u8a31\u591a\u7121\u6cd5\u7a81\u7834\u4e4b\u8655\uff0c\u5982\uff1a\u8a9e\u6599\u7684\u898f\u6a21\u3001\u4eba\u5de5\u6a19\u8a18 \u56e0\u5ba2\u8a9e\u8a9e\u6599\u7a00\u758f\u7684\u7de3\u6545\uff0c\u76ee\u524d\u91dd\u5c0d\u4e2d\u6587\u8f49\u5ba2\u6587\u7684\u76f8\u95dc\u7814\u7a76\u975e\u5e38\u5c11\uff0c\u5ba2\u8a9e\u65b7\u8a5e\u7cfb\u7d71\u7684\u5be6\u505a\uff0c Chinese \u5e74\u8f15\u4eba \u4e2d\u6587\u8a5e\u7528\u8a5e \u5ba2\u8a9e\u8a5e \u5149\u7dda \u8a18\u901f\u5ea6\u5982\u8868\u56db\uff1a (Da) \u5728(P) \u6c99\u6d32(Na) \u4e0a(Nes) \u73a9(VC) \u6454\u8de4(VA) \u3002(P) \u8868\u5341\u56db\u3001\u6a19\u8a18\u539f\u5247 5 \u7bc4\u4f8b \u51fa\u53e5\u5b50\u4e2d\u7684 no break\u3001minor break \u53ca major break \u4e09\u7a2e\u505c\u9813\u985e\u578b\uff0c\u8b93\u5408\u51fa\u7684\u8a9e\u97f3\u53ef\u8fa8\u5ea6\u66f4\u9ad8\u3001\u8b93 \u4e2d\u6587\u7ffb\u5ba2\u8a9e \u5927\u5bb6/\u4f86\u53bb/\u5ba2\u5bb6\u5e84/\u884c/\u4e00/\u884c/\u3002 \u4fe1\u4ef0 443 \u6210\u8a72\u8a5e\u3002 \u70ba\u7d50\u679c\u3002</td></tr><tr><td>\u8cc7\u6599\u7684\u6b63\u78ba\u6027\u3002\u4f46\u4f7f\u7528\u672c\u8ad6\u6587\u65b9\u6cd5\u7684\u5ba2\u8a9e\u65b7\u8a5e\u6cd5\uff0c\u4ee5\u5167\u90e8\u6e2c\u8a66\u7d50\u679c\u53ef\u77e5\uff0c\u82e5\u5728\u8a13\u7df4\u8a9e\u6599\u5145\u8db3\u7684 \u5148\u4ee5(\u4e00)\u8f38\u5165\u70ba\u5ba2\u8a9e\u3001(\u4e8c)\u8f38\u5165\u70ba\u4e2d\u6587\uff0c\u5206\u70ba\u5169\u5927\u7a2e\u985e\u3002\u7b2c\u4e00\u985e\u662f\u76f4\u63a5\u91dd\u5c0d\u5ba2\u8a9e\u6587\u53e5\u505a\u65b7\u8a5e\uff0c\u7b2c Hakka \u5f8c\u751f\u4eba \u5ba2\u8a9e\u8a5e\u7528\u8a5e \u4e2d\u6587\u8a5e \u5149\u7dda \u8868\u56db\u3001\u5ba2\u8a9e\u65b7\u8a5e\u6a19\u8a18\u5de5\u5177\u7684\u6a19\u8a18\u6642\u9593\u7d71\u8a08\uff0c\u6642\u9593\u55ae\u4f4d\uff1a\u79d2\u3002 \u4e2d\u6587 \u4e00\u8eab\u4e7e\u5f46\u7c21\u55ae\u7c97\u964b\u7684\u8863\u670d\u90fd\u6c92\u6709\u80fd\u529b\u8cb7\u3002 \u4f7f\u7528\u8005\u80fd\u66f4\u8f15\u6613\u7684\u807d\u61c2\u53e5\u5b50\u5167\u5bb9\u3002 \u6b63\u78ba\u7b54\u6848 \u5927\u5bb6/\u4f86\u53bb/\u5ba2\u5bb6\u5e84/\u9076\u9076\u554a/\u3002 \u516b\u4ed9 355 3. \u82e5\u6b65\u9a5f 2 \u6642\u67e5\u4e0d\u5230\u5c0d\u61c9\u7684\u5ba2\u8a9e\u8a5e\uff0c\u5247\u6cbf\u7528\u4e2d\u6587\u65b7\u8a5e\u7684\u7d50\u679c\u3002 \u5716\u4e94\u70ba\u672c\u65b9\u6cd5\u6d41\u7a0b\uff1a</td></tr><tr><td>1. \u7dd2\u8ad6 \u4e00\u500b\u65b7\u8a5e\u7cfb\u7d71\u7684\u6548\u679c\uff0c\u901a\u5e38\u8ddf\u8a9e\u6599\u7684\u5927\u5c0f\u6709\u95dc\u3002\u4f46\u76ee\u524d\u5ba2\u8a9e\u8a9e\u6599\u7684\u6536\u96c6\u975e\u5e38\u56f0\u96e3\uff0c\u73fe\u6709\u7684 \u96fb\u5b50\u8cc7\u6599\uff0c\u5982\uff1a\u5ba2\u59d4\u6703\u521d\u7d1a\u3001\u4e2d\u9ad8\u7d1a\u7684\u8a8d\u8b49\u6559\u6750\u3001\u6559\u80b2\u90e8\u7de8\u8457\u7684\u570b\u5c0f\u5ba2\u8a9e\u6559\u6750\u2026\u7b49\uff0c\u5c0d\u65bc\u81ea\u7136 \u8a9e\u8a00\u8655\u7406\u4f86\u8aaa\uff0c\u8cc7\u6599\u898f\u6a21\u4ecd\u7136\u5c6c\u6975\u5c0f\u91cf\u8a9e\u6599\u3002\u56e0\u6b64\uff0c\u60f3\u8981\u5efa\u7f6e\u51fa\u66f4\u591a\u7684\u5ba2\u8a9e\u8a9e\u6599\uff0c\u5e7e\u4e4e\u90fd\u9700\u8981 \u5f9e\u5ba2\u8a9e\u66f8\u7c4d\u3001\u6587\u7ae0\u4e2d\uff0c\u900f\u904e\u4eba\u5de5\u8f38\u5165\u3001\u5efa\u7f6e\u6210\u96fb\u5b50\u6a94\u7684\u65b9\u5f0f\u4f86\u53d6\u5f97\u3002\u4f46\u6709\u4e86\u9019\u4e9b\u6587\u672c\u8cc7\u6599\u53ea\u662f \u7b2c\u4e00\u6b65\uff0c\u5f8c\u7e8c\u4ecd\u6709\u8a31\u591a\u7684\u8655\u7406\u5de5\u4f5c\uff0c\u5982\uff1a\u65b7\u8a5e\u3001\u8a5e\u6027\u6a19\u8a18\u7684\u8655\u7406\uff0c\u64f7\u53d6\u51fa\u9019\u4e9b\u8a9e\u8a00\u7279\u5fb5\u5f8c\uff0c\u624d \u80fd\u505a\u66f4\u9032\u4e00\u6b65\u7684\u5206\u6790\u8207\u61c9\u7528\u3002 \u5ba2\u8a9e\u8a5e\u7684\u5224\u5b9a\u662f\u4e00\u4ef6\u56b4\u8b39\u7684\u4e8b\u60c5\uff0c\u7406\u8ad6\u4e0a\u6211\u5011\u5fc5\u9808\u9075\u7167\u8a5e\u7684\u5b9a\u7fa9 1 \u4f86\u6a19\u8a18\uff0c\u4f46\u6709\u6975\u5c11\u6578\u7684\u60c5 \u6cc1\u4e0b\uff0c\u6211\u5011\u4ecd\u6703\u5c07\u8a5e\u7d44\u6a19\u8a18\u6210\u4e00\u500b\u5ba2\u8a9e\u8a5e\uff0c\u5982\uff1a\u6ed1\u6e9c\u6e9c\uff0c\u5728\u4e2d\u6587\u65b7\u8a5e\u88ab\u65b7\u70ba\uff1a\u6ed1/\u6e9c\u6e9c\uff0c\u6211\u5011\u8996 \u5b83\u662f\u4e00\u500b\u8a5e\u3002\u800c\u5c0d\u65bc\u975e\u5ba2\u8a9e\u8a9e\u8a00\u5c08\u5bb6\u7684\u6a19\u8a18\u4eba\u54e1\u4f86\u8aaa\uff0c\u5176\u6700\u6709\u6548\u7387\u7684\u65b9\u6cd5\uff0c\u662f\u900f\u904e\u5177\u6709\u5e73\u884c\u8cc7 \u8a0a 2 \u7684\u8a9e\u6599\uff0c\u5148\u5c07\u4e2d\u6587\u8a9e\u6599\u8f38\u5165\u81f3\u4e2d\u6587\u7684\u6587\u53e5\u8655\u7406\u7cfb\u7d71\uff0c\u53d6\u5f97\u4e2d\u6587\u7684\u65b7\u8a5e\u3001\u8a5e\u6027\u6a19\u8a18\u7684\u7279\u5fb5\u5f8c\uff0c \u518d\u5c0d\u5176\u5c0d\u61c9\u7684\u5ba2\u8a9e\u6587\u7ae0\uff0c\u4ee5\u4eba\u5de5\u65b9\u5f0f\u53bb\u5224\u65b7\u5ba2\u8a9e\u8a5e\u7684\u908a\u754c\u8207\u8a5e\u6027\u7684\u6a19\u8a18\u3002\u9019\u500b\u65b9\u6cd5\u666e\u904d\u88ab\u4f7f\u7528 \u65bc\u540c\u985e\u578b 3 \u7d71\u4e2d\uff0c\u5728\u6587\u53e5\u65b7\u8a5e\u8cc7\u8a0a\u6a19\u8a18\u7684\u6280\u8853\u65b9\u9762\u5df2\u7d93\u76f8\u7576\u6210\u719f\uff0c\u800c\u5ba2\u8a9e\u8207\u4e2d\u6587\u7684\u6587\u6cd5\u7d50\u69cb\u4e5f\u76f8\u8fd1\uff0c\u5be6\u969b \u4e0a\u4e2d\u6587\u7684\u65b7\u8a5e\u3001\u8a5e\u6027\u7279\u5fb5\uff0c\u5e7e\u4e4e\u90fd\u80fd\u76f4\u63a5\u5c0d\u61c9\u65bc\u5ba2\u8a9e\u8a5e\u3002 \u76ee\u524d\u5ba2\u8a9e\u8a9e\u6599\u7684\u6536\u96c6\u8207\u5efa\u7f6e\uff0c\u5728\u5b78\u754c\u6709\u8a31\u591a\u5b78\u8005\u5c08\u5bb6\u5df2\u7a4d\u6975\u7684\u5728\u505a\u52aa\u529b\uff0c\u5efa\u7f6e\u51fa\u5ba2\u8a9e\u7814\u7a76 \u7684\u76f8\u95dc\u57fa\u790e\u8cc7\u6599\u5eab\u3002\u5982\u5c4f\u6771\u6559\u80b2\u5927\u5b78\u7684\u300c\u5b78\u8853\u7814\u7a76\u57fa\u790e\u5efa\u7f6e\u66a8\u5ba2\u5bb6\u6587\u5316\u7814\u7a76\u8a08\u756b [2]\u300d \uff0c\u4ed6\u5011\u6b77 \u6642\u4e86\u81f3\u5c11\u4e09\u5e74\u7684\u6642\u9593\uff0c\u5728\u6536\u96c6\u3001\u5efa\u7f6e\u5ba2\u8a9e\u8a9e\u6599\u53ca\u8a5e\u983b\u5eab\u3002\u9019\u9805\u5275\u8209\u80fd\u6709\u52a9\u65bc\u5ba2\u8a9e\u6587\u53e5\u8655\u7406\u7684\u767c \u5c55\uff0c\u5982\uff1a\u5ba2\u8a9e\u65b7\u8a5e\u7cfb\u7d71\u3001\u5ba2\u8a9e\u6587\u53e5\u5206\u6790\u7cfb\u7d71\u3001\u5ba2\u8a9e\u6587\u53e5\u5256\u6790\u7cfb\u7d71\u3001\u5ba2\u8a9e\u8a9e\u97f3\u5408\u6210\u7cfb\u7d71\u3001\u667a\u6167\u578b \u7684\u5ba2\u8a9e\u8f38\u5165\u6cd5\u2026\u7b49\uff0c\u90fd\u975e\u5e38\u9700\u8981\u8db3\u5920\u7684\u5ba2\u8a9e\u8a9e\u6599\u4f86\u652f\u6301\u5176\u767c\u5c55\u3002 \u60c5\u6cc1\u4e0b\uff0c\u80fd\u5f97\u5230\u4e00\u500b\u4e0d\u932f\u7684\u65b7\u8a5e\u6548\u80fd\uff0c\u5176\u5167\u90e8\u6e2c\u8a66\u7684\u7cbe\u78ba\u5ea6\u9054 94.46%\u3002\u76f8\u4fe1\u5728\u672a\u4f86\u6301\u7e8c\u589e\u52a0\u5ba2 \u8a9e\u8a9e\u6599\u7684\u898f\u6a21\u5f8c\uff0c\u672c\u8ad6\u6587\u6240\u63d0\u51fa\u7684\u65b9\u6cd5\uff0c\u6548\u80fd\u6703\u6709\u66f4\u986f\u8457\u7684\u63d0\u5347\u3002 2. \u6587\u737b\u63a2\u8a0e 2.1 \u4e2d\u6587\u65b7\u8a5e \u4e2d\u6587\u65b7\u8a5e\u6cd5\u6bcf\u5e74\u90fd\u6709\u65b0\u7684\u7814\u7a76\u8207\u6280\u8853\uff0c\u751a\u81f3\u6bcf\u5e74\u90fd\u6703\u6709\u8209\u8fa6\u65b7\u8a5e\u6bd4\u8cfd\uff0c\u77e5\u540d\u7684\u6bd4\u8cfd\u5982\uff1a SIGHAN \u6240\u8209\u8fa6\u7684\u570b\u969b\u4e2d\u6587\u5206\u8a5e\u7af6\u8cfd\u3002\u7b2c\u4e00\u5c46\u7af6\u8cfd\u8d77\u59cb\u65bc 2003 \u5e74\u5728\u65e5\u672c\u672d\u5e4c\u8209\u884c\u3002\u800c\u4e4b\u5f8c\u6bcf\u5e74 \u90fd\u6709\u76f8\u7576\u591a\u9ad8\u624b\u5171\u8944\u76db\u8209\u3002 \u5728\u9019\u9ebc\u591a\u7433\u746f\u6eff\u76ee\u7684\u65b7\u8a5e\u6280\u8853\u4e2d\uff0c\u5e38\u898b\u7684\u4e2d\u6587\u65b7\u8a5e\u6280\u8853\u53ef\u5206\u70ba\u4e09\u5927\u985e\uff1a(\u4e00)\u7d71\u8a08\u5f0f\u65b7\u8a5e \u6cd5\u3001 (\u4e8c)\u6cd5\u5247\u5f0f\u65b7\u8a5e\u6cd5\u548c\u3001(\u4e09)\u6df7\u5408\u5f0f\u65b7\u8a5e\u6cd5\u3002 (A) \u7d71\u8a08\u5f0f\u65b7\u8a5e\u6cd5 \u7d71\u8a08\u5f0f\u65b7\u8a5e\u6cd5\u85c9\u7531\u6536\u96c6\u8a5e\u5f59\u8cc7\uf9be\uff0c\u5982\u8a5e\u5f59\u9577\ufa01\u548c\u8a5e\u5f59\u51fa\u73fe\u7684\u983b\uf961\u6216\u6b21\u6578\u7b49\u7d71\u8a08\u4e0a\u7684\u8cc7\u8a0a\u3002 \u7136\u5f8c\u7cfb\u7d71\u904b\u7528\u6b64\u8a13\u7df4\u8cc7\u6599\u7d93\u7531\u6f14\u7b97\u6cd5\u5206\u6790\u4f86\u53d6\u5f97\u65b7\u8a5e\u5e8f\u5217\u3002\u5e38\u898b\u7684\u6f14\u7b97\u6cd5\u5982 Xue \u4f7f\u7528\u7684 Maximum Entropy [15]\uff0c\u6700\u5f8c\u5be6\u9a57\u5f97\u5230\u6700\u597d\u7684 F \u5206\u6578\u662f 94.98%\uff0c\u6216\u662f Lo \u4f7f\u7528\u7684 Conditional Random Field [4]\uff0c\u6700\u5f8c\u5be6\u9a57\u5f97\u5230\u6700\u597d\u7684 F \u5206\u6578\u662f 96.40%\u3002\u9019\u4e9b\u6f14\u7b97\u6cd5\u90fd\u662f\u5229\u7528\u5b57\u5143\u4e4b\u9593\u7684\u8cc7\u8a0a \u7576\u4f5c\u7279\u5fb5\u3002\u7136\u5f8c\u628a\u65b7\u8a5e\u554f\u984c\u8f49\u63db\u6210\u70ba\u5b57\u5143\u4e4b\u9593\u7684\u5206\u985e\u554f\u984c\u3002 \u800c\u904e\u53bb\u5e38\u88ab\u4f7f\u7528\u7684\u6f14\u7b97\u6cd5\u662f Hidden Markov Model\u3002\u5982 Fu \u548c Luck [5]\u5f9e\u8a13\u7df4\u8a9e\u6599\u4e2d\u7d71\u8a08\u8a5e \u983b\u3001\u5b57\u5143\u5728\u8a5e\u4e2d\u51fa\u73fe\u4f4d\u7f6e\u7684\u6b21\u6578\u7b49\u8cc7\u8a0a\uff0c\u7d44\u5408\u904e\u5f8c\u505a\u5be6\u9a57\u3002\u5f97\u5230 F \u5206\u6578\u6700\u597d\u53ef\u9054 93.7%\u3002\u800c Lin \u548c Chang [4]\u4f7f\u7528\u5169\u968e\u6bb5\u7279\u88fd\u5316\u7684\u65b9\u5f0f\uff0c\u85c9\u8457\u64f4\u5145\u89c0\u6e2c\u7b26\u865f\u53ca\u72c0\u614b\u7b26\u865f\u4f86\u6539\u5584\u96b1\u85cf\u5f0f\u99ac\u53ef\u592b\u6a21\u578b \u7684\u65b7\u8a5e\u6548\u679c\u3002\u5f97\u5230 F \u5206\u6578\u6700\u597d\u53ef\u9054 96.3%\u3002 (B) \u6cd5\u5247\u5f0f\u65b7\u8a5e\u6cd5 \u6cd5\u5247\u5f0f\u65b7\u8a5e\u6cd5\u4e3b\u8981\u662f\u6839\u64da\u4e00\u4e9b\u7d93\u9a57\u6cd5\u5247\u505a\u70ba\u65b7\u8a5e\u7684\u6a19\u6e96\uff0c\u85c9\u4ee5\u9054\u5230\u8f03\u597d\u7684\u65b7\u8a5e\u5e8f\u5217\u3002\u5e38\u898b \u7684\u6cd5\u5247\u5982\u300c\u9577\u8a5e\u512a\u65bc\u77ed\u8a5e\u300d \u3001 \u300c\u8207\u5de6\u908a\u8a5e\u7684\u7d50\u5408\u512a\u65bc\u8207\u53f3\u908a\u8a5e\u7684\u7d50\u5408\u300d \u3002\u800c\u9019\u985e\u578b\u7684\u65b7\u8a5e\u6cd5\u5e38\u88ab\u53c3 \u8003\u7684\u662f Chen \u548c Liu [17]\uff0c\u4ed6\u5011\u5728\u8a72\u7bc7\u8ad6\u6587\u4e2d\u63d0\u51fa\u4e86\u516d\u689d\u6cd5\u5247(heuristic rules)\uff0c\u4e26\u4e14\u6839\u64da\u9019\u4e9b\u6cd5\u5247 \u89e3\u6c7a\u4e86\u6b67\u7fa9\u6027\u7684\u554f\u984c\u53ca\u5254\u9664\u4e00\u4e9b\u8f03\u4e0d\u53ef\u80fd\u7684\u8a5e\u5f59\u7d44\u5408\uff0c\u4ee5\u5b8c\u6210\u4e2d\u6587\u65b7\u8a5e\u7684\u5de5\u4f5c\u3002\u5728\u5be6\u9a57\u4e0a\u6548\u679c \u76f8\u7576\u4e0d\u932f\u3002 \u4e0d\u904e\uff0c\u6b64\u7a2e\u65b7\u8a5e\u6cd5\u5bb9\u6613\u53d7\u5230\u8a5e\u5178\u7684\u597d\u58de\u800c\u5f71\u97ff\u6548\u80fd\u3002\u82e5\u662f\u53e5\u5b50\u4e2d\u51fa\u73fe\u672a\u77e5\u7684\u65b0\u8a5e\u5f59\u6642\uff0c\u5247 \u6b63\u78ba\u7387\u5c31\u53ef\u80fd\u4e0b\u6ed1\u3002 (C) \u6df7\u5408\u5f0f\u65b7\u8a5e\u6cd5 \u6bcf\u7a2e\u65b9\u5f0f\u7684\u65b7\u8a5e\u6cd5\u90fd\u6709\u597d\u58de\u3001\u512a\u7f3a\u9ede\uff0c\u56e0\u6b64\u5f8c\u4f86\u5b78\u8005\u5011\u624d\u6703\u5617\u8a66\u53bb\u6df7\u5408\u5169\u7a2e\u65b7\u8a5e\u65b9\u5f0f\u3002\u65e9 \u671f\u7684 Nie \u7b49\u4eba[18]\u63d0\u51fa\u7d50\u5408\u8a5e\u5178\u3001\u7d93\u9a57\u6cd5\u5247\u53ca\u7d71\u8a08\u8cc7\u8a0a\u4f86\u5c0d\u4e2d\u6587\u65b7\u8a5e\u3002\u800c\u4ee4\u4eba\u5370\u8c61\u6df1\u523b\u7684\u662f Wu 6 \u6642\u9593\u55ae\u4f4d\u70ba\u79d2\u3002 \u5716\u4e00\u3001\u5ba2\u8a9e\u65b7\u8a5e\u7b54\u6848\u6a19\u8a18\u5de5\u5177\u67b6\u69cb\u5716 \u8a5e\u4e26\u8f49\u63db\u70ba\u300c\u9c17\u9c0d\u4ed4/\u6ed1/\u6e9c\u6e9c/\uff0c\u300d \u3002\u4f46\u5176\u5be6\u6211\u5011\u7684\u570b\u5ba2\u8a9e\u5c0d\u7167\u8a5e\u5178\u4e2d\uff0c\u6709\u6536\u9304\u300c\u6ed1\u6e9c\u6e9c/\u6ed1\u6e9c\u6e9c 5 \u8655\u7406\u53e5\u6578\u7684\u7d71\u8a08\uff0c\u5305\u542b\u91cd\u65b0\u8655\u7406\u66fe\u7d93\u653e\u68c4\u7684\u53e5\u5b50\uff0c\u56e0\u6b64\u5be6\u969b\u8655\u7406\u53ef\u80fd\u6703\u6bd4\u5206\u914d\u5230\u7684\u7b46\u6578\u591a\u3002 \u548c Jiang [19]\u63d0\u51fa\u7d50\u5408\u5256\u6790\u5668\u548c\u65b7\u8a5e\u5668\u7684\u65b9\u6cd5\uff0c\u5728\u65b7\u8a5e\u6642\u5148\u4f7f\u7528\u67e5\u8a5e\u5178\u4f86\u7522\u751f\u6240\u6709\u53ef\u80fd\u7684\u65b7\u8a5e\u7d44 \u4e8c\u985e\u662f\u91dd\u5c0d\u4e2d\u6587\u6587\u53e5\u7ffb\u8b6f\u6210\u5ba2\u8a9e\u65b7\u8a5e\u7d50\u679c\u3002 \u800c\u76ee\u524d\u4ecd\u6c92\u6709\u4efb\u4f55\u4e00\u7bc7\u662f\u91dd\u5c0d\u5ba2\u8a9e\u65b7\u8a5e\u505a\u6df1\u5165\u7814\u7a76\u7684\u8ad6\u6587\uff0c\u5176\u4e2d\u5c0d\u5ba2\u8a9e\u65b7\u8a5e\u6709\u505a\u6548\u80fd\u8a55\u4f30 \u7684\u8ad6\u6587\uff0c\u4e5f\u50c5\u6709 Tsai \u7684\u78a9\u58eb\u8ad6\u6587 [1]\u2500\u57fa\u65bc\u96b1\u85cf\u5f0f\u99ac\u53ef\u592b\u6a21\u578b\u4e4b\u5ba2\u8a9e\u6587\u53e5\u8f49\u8a9e\u97f3\u7cfb\u7d71\u3002\u986f\u898b\u76ee \u524d\u5ba2\u8a9e\u65b7\u8a5e\u7684\u7814\u7a76\uff0c\u4e0d\u7ba1\u662f\u8a9e\u6599\u7684\u5efa\u7f6e\uff0c\u9084\u662f\u65b7\u8a5e\u7684\u65b9\u6cd5\uff0c\u4ecd\u975e\u5e38\u591a\u5f85\u63a2\u8a0e\u8207\u89e3\u6c7a\u7684\u554f\u984c\u3002 (A) \u8f38\u5165\u70ba\u5ba2\u8a9e \u9019\u4e00\u985e\u7684\u7cfb\u7d71\uff0c\u9069\u5408\u5177\u5099\u5ba2\u8a9e\u8f38\u5165\u80fd\u529b\u53ca\u719f\u6089\u5ba2\u8a9e\u7684\u4f7f\u7528\u8005\uff0c\u5c0d\u65bc\u4e00\u822c\u4e0d\u719f\u6089\u5ba2\u8a9e\u7684\u4f7f\u7528 \u8005\u800c\u8a00\uff0c\u8f03\u4e0d\u65b9\u4fbf\u3002\u9019\u985e\u7cfb\u7d71\u5e38\u898b\u7684\u505a\u6cd5\uff0c\u662f\u76f4\u63a5\u4f7f\u7528\u4e2d\u6587\u65b7\u8a5e\u7cfb\u7d71\uff0c\u5c0d\u5ba2\u6587\u505a\u65b7\u8a5e\u3002\u7576\u7136\uff0c \u9019\u6a23\u6703\u6709\u4e00\u4e9b\u5ba2\u8a9e\u9020\u5b57\u6216\u5ba2\u8a9e\u7528\u8a5e\u7121\u6cd5\u8fa8\u5225\u7684\u554f\u984c\uff0c\u91dd\u5c0d\u9019\u90e8\u4efd\uff0c\u662f\u4f7f\u7528\u570b\u5ba2\u8a9e\u5c0d\u7167\u8fad\u5178\uff0c\u4f86 \u89e3\u6c7a\u5ba2\u8a9e\u672a\u77e5\u8a5e(Out of Vocabulary, OOV)\u554f\u984c\u3002 \u5982 Tsai \u7684\u8ad6\u6587 [1]\uff0c\u4ed6\u5011\u900f\u904e Conditional Random Field \u65b9\u6cd5\u5be6\u505a\u4e2d\u6587\u65b7\u8a5e\u7cfb\u7d71\uff0c\u4e26\u52a0\u5165\u570b\u5ba2 \u8a9e\u5c0d\u7167\u5916\u90e8\u8fad\u5178\uff0c\u914d\u5408\u5ba2\u8a9e\u69cb\u8a5e\u898f\u5247\uff0c\u5be6\u505a\u51fa\u5ba2\u8a9e\u65b7\u8a5e\u6a21\u7d44\u3002\u6700\u5f8c\u7684\u5be6\u9a57\u6548\u80fd\uff0c\u5ba2\u8a9e\u65b7\u8a5e\u7684 F \u5206\u6578\u70ba 82.87%\uff0c\u5ba2\u8a9e\u8a5e\u6027\u6a19\u8a18\u7684 F \u5206\u6578\u70ba 77.14%\u3002 (B) \u8f38\u5165\u70ba\u4e2d\u6587 \u9019\u4e00\u985e\u7684\u7cfb\u7d71\uff0c\u4f7f\u7528\u8005\u4e0d\u9700\u4f7f\u7528\u5ba2\u8a9e\u8f38\u5165\u6cd5\uff0c\u4e5f\u4e0d\u9700\u719f\u6089\u5ba2\u8a9e\uff0c\u5f88\u9069\u5408\u5ba2\u8a9e\u521d\u5b78\u8005\u4f7f\u7528\u3002 \u9019\u985e\u7cfb\u7d71\u5e38\u898b\u7684\u505a\u6cd5\uff0c\u662f\u4f7f\u7528\u4e2d\u6587\u65b7\u8a5e\u7cfb\u7d71\uff0c\u5148\u5c07\u8f38\u5165\u7684\u4e2d\u6587\u6587\u53e5\u65b7\u8a5e\uff0c\u627e\u51fa\u8a5e\u8207\u8a5e\u6027\u5f8c\uff0c\u518d \u5c07\u8a5e\u900f\u904e\u570b\u5ba2\u8a9e\u7684\u5e73\u884c\u5c0d\u7167\u8fad\u5178\uff0c\u7ffb\u8b6f\u6210\u5ba2\u8a9e\u8a5e\u3002\u5982\u672c\u5be6\u9a57\u5ba4\u7684\u7dda\u4e0a\u5ba2\u8a9e\u8a9e\u97f3\u5408\u6210\u7cfb\u7d71\uff0cWu [5]\u3001Lo [6]\u7684\u65b7\u8a5e\u65b9\u6cd5\u7686\u76f8\u540c\uff0c\u90fd\u662f\u4f7f\u7528 Jiang [7]\u6240\u63d0\u51fa\u7684\u4e2d\u6587\u65b7\u8a5e\u7cfb\u7d71\uff0c\u5c07\u4e2d\u6587\u6587\u53e5\u65b7\u8a5e\u5f8c\uff0c \u518d\u900f\u904e\u570b\u5ba2\u8a9e\u5c0d\u7167\u8fad\u5178\uff0c\u5c07\u4e2d\u6587\u8a5e\u7ffb\u8b6f\u6210\u5ba2\u8a9e\u8a5e\u3002\u7d93\u6e2c\u8a66\u5f8c\uff0c\u5176\u4e0d\u542b\u8a5e\u6027\u6a19\u8a18\u7684\u5ba2\u8a9e\u65b7\u8a5e\u6548\u80fd \u7684 F \u5206\u6578\u5206\u5225\u70ba 69.82%\u53ca 66.72%\u3002 \u53e6\u4e00\u7a2e\u662f\u50c5\u900f\u904e\u570b\u5ba2\u8a9e\u5c0d\u7167\u8fad\u5178\uff0c\u5c07\u4e2d\u6587\u6587\u53e5\u76f4\u7ffb\u6210\u5ba2\u8a9e\u3002\u5982 Lee [8]\uff0c\u4ed6\u5011\u5efa\u7f6e\u51fa\u4e00\u5957\u570b \u5ba2\u8a9e\u5c0d\u7167\u8fad\u5178\uff0c\u5c07\u8f38\u5165\u7684\u4e2d\u6587\u6587\u53e5\u5b57\u4e32\u5207\u5272\u6210 1 \u5230 4 \u5b57\u8a5e\uff0c\u4e26\u67e5\u627e\u5c0d\u7167\u8fad\u5178\u3001\u7ffb\u8b6f\u6210\u5ba2\u8a9e\u8a5e\u3002 \u800c\u4ed6\u5011\u6c92\u6709\u91dd\u5c0d\u4e2d\u6587\u7ffb\u5ba2\u8a9e\u8a5e\u505a\u6548\u80fd\u8a55\u4f30\uff0c\u56e0\u6b64\u7121\u6cd5\u5f97\u77e5\u6548\u679c\u5982\u4f55\u3002 3. \u6e96\u5099\u5de5\u5177\u53ca\u8a9e\u6599 3.1 \u4e2d\u6587\u65b7\u8a5e\u5de5\u5177 \u672c\u8ad6\u6587\u7684\u4e2d\u6587\u65b7\u8a5e\u7cfb\u7d71\uff0c\u662f\u4f7f\u7528 Lai \u65bc 2011 \u63d0\u51fa\u7684\u300c\u61c9\u7528\u591a\u8a5e\u53ca\u591a\u8a5e\u6027\u8a9e\u8a00\u6a21\u578b\u7684\u4e2d\u6587\u65b7 \u8a5e\u53ca\u8a5e\u6027\u6a19\u8a18\u65b9\u6cd5 [9]\u300d \u3002\u6b64\u65b7\u8a5e\u65b9\u6cd5\u63a1\u7528\u5169\u968e\u65b7\u5f0f\uff0c\u7b2c\u4e00\u968e\u65b7\u662f\u65b7\u8a5e\uff0c\u7b2c\u4e8c\u968e\u65b7\u662f\u8a5e\u6027\u6a19\u8a18\u3002 \u5176\u65b7\u8a5e\u7684 F \u5206\u6578\u6709 96.69%\uff0c\u8a5e\u6027\u6a19\u8a18\u7684 F \u5206\u6578 92.04%\u3002 3.2 \u570b\u5ba2\u8a9e\u5c0d\u7167\u8fad\u5178 \u6211\u5011\u6240\u5efa\u7f6e\u7684\u570b\u5ba2\u8a9e\u5c0d\u7167\u8fad\u5178\uff0c\u4e3b\u8981\u4f86\u6e90\u6709\uff1a(\u4e00)\u5ba2\u59d4\u6703\u521d\u7d1a\u3001\u4e2d\u7d1a\u66a8\u4e2d\u9ad8\u7d1a\u8a8d\u8b49\u8a9e\u6599 [10, 11]\u3001(\u4e8c)\u53f0\u5317\u5e02\u5ba2\u59d4\u6703-\u73fe\u4ee3\u5ba2\u8a9e\u8a5e\u5f59\u5f59\u7de8\u3002\u5c0d\u65bc\u65b7\u8a5e\u7cfb\u7d71\u800c\u8a00\uff0c\u8fad\u5178\u662f\u6c7a\u5b9a\u6b63\u78ba\u7387\u7684\u91cd\u8981\u56e0 \u7d20\uff0c \u7406\u8ad6\u4e0a\u8fad\u5178\u8d8a\u5927\uff0c\u65b7\u8a5e\u6548\u80fd\u4e5f\u5c31\u8d8a\u597d\u3002\u56e0\u6b64\u9664\u4e86\u73fe\u6709\u7684\u8fad\u5178\u4f86\u6e90\u5916\uff0c\u6211\u5011\u4e5f\u5229\u7528\u6a19\u8a18\u5ba2\u8a9e Pinyin heu1 sang2 ngin3 \u5ba2\u8a9e\u62fc\u97f3 Pos Na \u5ba2\u8a9e\u8a5e\u6027 Pos_pattern Na \u4e2d\u6587\u8a5e\u6027\u7d44 Hakka_pos_feq 25 \u5ba2\u8a9e\u542b\u8a5e\u6027\u8a5e\u983b Hakka_nonpos_feq 25 \u5ba2\u8a9e\u4e0d\u542b\u8a5e\u6027\u8a5e\u983b Chinese_pos_feq 21138 \u5ba2\u8a9e\u62fc\u97f3 gong\u02cb sien\u02c7 \u4e2d\u6587\u4f8b\u53e5 \u9019\u500b\u623f\u9593\u7684\u5149\u7dda\u4e0d\u597d\uff0c\u4e0d\u9069\u5408\u505a\u66f8\u623f\u3002 \u5ba2\u8a9e\u4f8b\u53e5 \u9019\u96bb\u623f\u9593\u4e2a\u5149\u7dda\u6bcb\u597d\uff0c\u6bcb\u9069\u5408\u505a\u66f8\u623f\u3002 3.3.2 \u5ba2\u8a9e\u65b7\u8a5e\u6a19\u8a18\u5de5\u5177\u53ca\u4eba\u5de5\u6a19\u8a18\u539f\u5247 \u5ba2\u8a9e\u8207\u4e2d\u6587\u7684\u6587\u6cd5\u7d50\u69cb\u76f8\u8fd1\uff0c\u56e0\u6b64\u4e2d\u6587\u65b7\u8a5e\u548c\u8a5e\u6027\u7684\u6a19\u8a18\uff0c\u5927\u90e8\u5206\u90fd\u80fd\u8207\u5ba2\u8a9e\u5b8c\u5168\u5c0d\u61c9\uff0c \u50c5\u6709\u5c11\u90e8\u5206\u7684\u5ba2\u8a9e\u4fda\u8a9e\u6216\u7279\u6b8a\u7528\u8a5e\u4f8b\u5916\u3002\u800c\u4e2d\u6587\u8a9e\u6599\u7684\u8655\u7406\uff0c\u56e0\u76ee\u524d\u4e2d\u6587\u65b7\u8a5e\u7cfb\u7d71\u7684\u767c\u5c55\u5df2\u76f8 \u6a19\u8a18\u8005 1 2 3 4 5 \u7e3d \u5e73 \u5747 \u8a9e\u6599\u7de8\u865f A B1 B2 B3 B4 \u8655\u7406\u53e5\u6578 5 4500 615 622 611 646 \u5132\u5b58\u53e5\u6578 4018 346 451 504 419 \u7e3d\u6642\u9593 6 91798 28698 25297 18864 17219 \u5e73\u5747\u6bcf\u53e5 20.39 46.66 40.67 30.87 26.65 33.04 \u6b64\u4f8b\u5b50\u4e2d\uff0c \u300c\u5929\u90fd\u300d\u4e00\u8a5e\uff0c\u88ab\u8aa4\u65b7\u6210\u4e00\u500b\u666e\u901a\u540d\u8a5e\uff0c\u9019\u8ddf\u300c\u5929\u3001\u90fd\u300d\u610f\u601d\u4e0d\u540c\uff0c \u300c\u5929\u90fd\u300d\u6307 \u7684\u662f\u5730\u65b9\u540d\uff0c\u800c\u5176\u6b63\u78ba\u65b7\u8a5e\u61c9\u8a72\u65b7\u6210\u300c\u5929(Nf) \u90fd(Da)\u300d \u3002 \u539f\u5247 3\uff1a\u53ef\u5fae\u8abf\u4e2d\u6587\u53e5\u5b50\u7528\u8a5e\u53ca\u8a5e\u7684\u9806\u5e8f\u4ee5\u6c42\u5c0d\u61c9\u5230\u5ba2\u8a9e\uff0c\u4f46\u4fee\u6539\u5f8c\u7684\u6587\u610f\u4e0d\u80fd\u6539\u8b8a\u3002 \u8868\u5341\u4e00\u3001\u6a19\u8a18\u539f\u5247 3 \u7bc4\u4f8b 1 \u4e2d\u6587 \u4eca\u5929\u7684\u5929\u6c23\u5f88\u597d\uff0c\u592a\u967d\u4e0b\u5c71\u4ee5\u5f8c\u5c31\u53ef\u4ee5\u770b\u5f97\u5230\u6eff\u5929\u7684\u661f\u661f\u3002 \u5ba2\u8a9e \u4e00\u8eab\u814a\u98df\u76ae\u7121\u624d\u8abf\u8cb7\u3002 \u6b64\u4f8b\u5b50\u4e2d\uff0c\u6211\u5011\u767c\u73fe\u9019\u53e5\u9664\u4e86\u300c\u4e00/\u4e00\u3001\u8eab/\u8eab\u3001\u7121/\u6c92\u6709\u3001\u624d\u8abf/\u80fd\u529b\u3001\u8cb7/\u8cb7\u300d\u9019\u4e9b\u8a5e\u80fd\u5c0d \u61c9\u5916\uff0c\u5176\u4ed6\u8a5e\u7686\u7121\u6cd5\u975e\u5c0d\u61c9\u3002\u56e0\u6b64\u6b64\u6b21\u7684\u4e0d\u7d0d\u5165\u6b63\u78ba\u7b54\u6848\u7bc4\u672c\u5167\uff0c\u4f46\u6709\u90e8\u5206\u8a5e\u80fd\u5c0d\u61c9\uff0c\u4e5f\u5c07\u9019 \u7ae0 \u56e0\u6b64\uff0c\u4e00\u500b\u826f\u597d\u7684\u5ba2\u8a9e\u65b7\u8a5e\u7cfb\u7d71\u662f\u5ba2\u8a9e\u8a9e\u8a00\u8655\u7406\u6240\u4e0d\u53ef\u6216\u7f3a\u7684\uff0c\u61c9\u662f\u63d0\u5347\u5ba2\u8a9e\u8a9e\u97f3\u5408\u6210\u6548 \u679c\u7684\u91cd\u8981\u56e0\u7d20\u3002\u672c\u7cfb\u7d71\u4e2d\u5ba2\u8a9e\u65b7\u8a5e\u7684\u67b6\u69cb\uff0c\u53c3\u898b\u5716\u4e09\u3002 \u7b75\u5e2d \u5716\u56db\u70ba\u6b64\u65b7\u8a5e\u6cd5\u7684\u6d41\u7a0b\uff1a 178 \u8f38\u5165\u4e2d\u6587\u6587\u53e5 \u4e0a\u9762\u7684\u4f8b\u5b50\u300c\u8d70\u4e00\u8d70\u300d\u88ab\u65b7\u6210\u300c\u8d70/\u4e00/\u8d70\u300d \uff0c\u4f46\u5be6\u969b\u4e0a\u6211\u5011\u8a5e\u5178\u6709\u6536\u9304\u300c\u8d70\u4e00\u8d70/\u9076\u9076\u554a\u300d \u2026 \u2026 \u8f38\u5165\u4e2d\u6587\u6587\u53e5 \u9019\u500b\u5c0d\u7167\u8a5e\u7d44\uff0c\u4f46\u7919\u65bc\u4e2d\u6587\u65b7\u8a5e\u908a\u754c\uff0c\u6211\u5011\u6c92\u8fa6\u6cd5\u9078\u5230\u8a72\u8a5e\u7d44\u3002 3. \u5c07\u6b65\u9a5f 2 \u7522\u751f\u7684\u65b0\u589e\u5019\u9078\u5217\u8868\u7684\u7d50\u679c\uff0c\u52a0\u5165\u4e2d\u6587\u65b7\u8a5e\u8fad\u5178\uff0c\u82e5\u6709\u76f8\u540c\u7684\u4e2d\u6587\u8a5e\uff0c\u5247\u5176\u8a5e\u983b \u7d44 \u8f38\u5165\u4e2d\u6587\u53e5\u5b50\u6216\u6587 \u4e2d\u6587\u65b7\u8a5e\u6a21 \u76f8\u52a0\u3002 \u4e2d\u6587\u65b7\u8a5e\u6a21\u7d44 \u4e9b\u8a5e\u7684\u570b\u5ba2\u8a9e\u5c0d\u61c9\u7d50\u679c\u5132\u5b58\u8d77\u4f86\uff0c\u5f85\u5f8c\u7e8c\u7684\u5de5\u4f5c\u4e2d\uff0c\u4ecd\u53ef\u61c9\u7528\u3002 \u7d44 \u57fa\u65bc\u4ee5\u4e0a\u7684\u539f\u56e0\uff0c\u6211\u5011\u6c7a\u5b9a\u5c07\u570b\u5ba2\u8a9e\u5c0d\u7167\u8fad\u5178\u4e2d\u7684\u4e2d\u6587\u8a5e\u6b04\u4f4d\uff0c\u52a0\u5165\u5230\u4e2d\u6587\u65b7\u8a5e\u8fad\u5178\u3002\u800c \u4e2d\u6587\u65b7\u8a5e\u6a21\u7d44 j=1 \u5ba2\u8a9e\u69cb\u8a5e\u6a21 \u9019\u4e9b\u65b0\u52a0\u5165\u7684\u4e2d\u6587\u8a5e\uff0c\u8981\u6c7a\u5b9a\u51fa\u5b83\u5011\u7684\u8a5e\u983b\u5927\u5c0f\uff0c\u56e0\u70ba\u9019\u5169\u500b\u8a9e\u6599\u898f\u6a21\u76f8\u5dee\u975e\u5e38\u9f90\u5927\uff0c\u6211\u5011\u7684 \u6700\u5f8c\u6211\u5011\u8a55\u4f30\u4fee\u6539\u524d\u8207\u4fee\u6539\u5f8c\u7684\u4e2d\u6587\u65b7\u8a5e\u7cfb\u7d71\uff0c\u5176\u6b63\u78ba\u7387\u7684\u5dee\u7570\u3002\u6e2c\u8a66\u70ba\u5916\u90e8\u6e2c\u8a66\uff0c\u4f7f\u7528 \u4e2d\u6587\u542b\u8a5e\u6027\u8a5e\u983b Chinese_nonpos_feq 21138 \u4e2d\u6587\u4e0d\u542b\u8a5e\u6027\u8a5e\u983b \u8868\u4e8c\u30012014 \u7248\uff0c\u8208\u5927\u570b\u5ba2\u8a9e\u5c0d\u7167\u8fad\u5178\u5206\u4f48\u7d71\u8a08\u53ca\u6bd4\u8f03 \u5b57\u8a5e 2014 \u8208\u5927\u5ba2\u8a9e\u8fad\u5178 Wu[5] Lo[6] \u4ea4\u5927\u5ba2\u8a9e\u8fad\u5178[1] \u4e00\u5b57\u8a5e 3457 780 838 6747 \u4e8c\u5b57\u8a5e 20800 16362 16540 18078 \u4e09\u5b57\u8a5e 7309 5654 5826 5095 \u56db\u5b57\u8a5e 4093 3769 3861 4217 \u4e94\u5b57\u8a5e 312 273 283 250 \u516d\u5b57\u8a5e 84 80 84 \u4e03\u5b57\u8a5e 65 63 68 \u516b\u5b57\u8a5e 9 12 14 \u7e3d\u8a08 36129 26993 27514 34541 3.3 \u5ba2\u8a9e\u56db\u7e23\u8154\u8a9e\u8a00\u6a21\u578b\u7684\u5efa\u7f6e \u8a9e\u8a00\u6a21\u578b\u662f\u65b7\u8a5e\u7cfb\u7d71\u4e2d\uff0c\u7528\u4f86\u9078\u64c7\u65b7\u8a5e\u7d50\u679c\u7684\u91cd\u8981\u5143\u4ef6\u3002\u800c\u8a9e\u6599\u7d93\u904e\u7d71\u8a08\u8a5e\u983b\u5f8c\uff0c\u5c31\u80fd\u5f97 \u5230\u8a72\u4efd\u8a9e\u6599\u7684\u6a5f\u7387\u5206\u4f48\u6a21\u578b\uff0c\u5373\u662f\u8a9e\u8a00\u6a21\u578b\u3002\u56e0\u6b64\uff0c\u8a9e\u6599\u8d8a\u5927\uff0c\u80fd\u5f97\u5230\u7684\u7d71\u8a08\u8cc7\u8a0a\u8d8a\u591a\uff0c\u8a9e\u8a00 \u8cc7\u8a0a 4 \u7684\u5ba2\u8a9e\u8a9e\u6599\uff0c\u6a19\u8a18\u51fa\u5ba2\u8a9e\u65b7\u8a5e\u8cc7\u8a0a\uff0c\u518d\u900f\u904e\u8a72\u8a9e\u6599\u7d71\u8a08\u51fa\u5ba2\u8a9e Uni-gram \u53ca Bi-gram \u8a9e\u8a00 \u65b7\u8a5e\u7b54\u6848\u6a19\u8a18\u7d50\u679c \u6a19\u8a18\u6642\u9593 P) \u5ba2\u8a9e\u6539 \u800c\u7b2c\u4e00\u7a2e\u65b9\u6cd5\u9069\u5408\u7528\u5728\u8a13\u7df4\u8a9e\u6599\u7a00\u758f\u7684\u60c5\u6cc1\u3002 \u900f\u904e\u4e2d\u6587\u65b7\u8a5e\u5f97\u5230\u300c\u6ce5\u9c0d/\u6ed1/\u6e9c\u6e9c/\uff0c\u300d\u7684\u65b7\u8a5e\u908a\u754c\uff0c\u518d\u900f\u904e\u4ee5\u4e0a\u65b9\u6cd5\u627e\u51fa\u6700\u4f73\u8a5e\u983b\u7684\u5ba2\u8a9e \u4e0a\u65e9\u4e2a\u4eba\u4fc2\u770b\u8457\u661f\u4ed4\u7009\u5c4e\uff0c\u6703\u60f3\u8457\u9ebc\u4e2a\u58de\u5146\u982d\u3002 \u570b\u5ba2\u8a9e\u8a5e\u5c0d\u61c9\u7d50\u679c \u6cd5\u5728\u5167\u90e8\u6e2c\u8a66\u6709\u8f03\u4f73\u7684\u6b63\u78ba\u7387\u3002\u5176\u610f\u7fa9\u662f\uff1a\u5982\u679c\u5728\u8a13\u7df4\u8a9e\u6599\u5145\u8db3\u7684\u60c5\u6cc1\u4e0b\uff0c\u7b2c\u4e8c\u7a2e\u65b9\u6cd5\u6703\u66f4 \u597d\u3002 \u6a21 \u578b\u3002 \u8a5e\u6578 45304 17646 \u5b57\u6578 65572 25478 \u4e2d\u6587\u6539 \u53e4\u65e9\u7684\u4eba\u82e5\u662f\u770b\u5230\u6d41\u661f\u96e8\uff0c\u6703\u60f3\u5230\u4ec0\u9ebc\u58de\u5146\u982d\u3002 \u6700\u4f73\u7684\u65b7\u8a5e\u8def\u5f91\u3002\u5169\u7a2e\u65b9\u6cd5\u5176\u5be6\u5404\u6709\u597d\u8655\uff0c\u7b2c\u4e00\u7a2e\u65b9\u6cd5\u5728\u5916\u90e8\u6e2c\u8a66\u6709\u8f03\u4f73\u7684\u6b63\u78ba\u7387\uff0c\u7b2c\u4e8c\u7a2e\u65b9 \u6b63\u78ba\u7b54\u6848 \u9c17\u9c0d\u4ed4/\u6ed1\u6e9c\u6e9c\u4ed4/\uff0c \u5206\u6578\u7684\u8a08\u7b97\u65b9\u6cd5\uff1a \u8868\u4e5d\u3001\u7bc4\u4f8b 1 \u6a19\u8a18\u5f8c\u6a23\u8c8c \u5ba2\u8a9e \u4e0a\u65e9\u4e2a\u4eba\u4fc2\u770b\u8457\u661f\u4ed4\u7009\u5c4e\uff0c\u6703\u60f3\u8457\u9ebc\u4e2a\u58de\u5146\u982d\u3002 \u9650\uff0c\u65b9\u6cd5(\u4e8c)\u7684\u8a5e\u908a\u754c\u662f\u81ea\u7531\u7684\uff0c\u53ef\u5f9e\u570b\u5ba2\u8a9e\u5c0d\u7167\u8fad\u5178\u53ca\u539f\u4e2d\u6587\u65b7\u8a5e\u65b7\u7684\u4e2d\u6587\u8a5e\u4e2d\uff0c\u627e\u51fa\u4e00\u689d \u4e2d\u6587\u7ffb\u5ba2\u8a9e \u5408 Uni-gram \u53ca Bi-gram \u6df7\u5408\u5f0f\u7684\u5206\u6578\u7b97\u6cd5\uff0c\u627e\u51fa\u4e00\u689d\u5206\u6578\u6700\u4f73\u7684\u65b7\u8a5e\u5e8f\u5217\u3002 \u9c17\u9c0d\u4ed4/\u6ed1/\u6e9c\u6e9c/\uff0c \u6211\u5011\u958b\u767c\u4e86\u4e00\u500b\u534a\u81ea\u52d5\u5f0f\u7684\u5ba2\u8a9e\u8a9e\u6599\u5efa\u7f6e\u5de5\u5177\u4f86\u5efa\u7f6e\u5ba2\u8a9e\u8a9e\u6599\uff0c\u900f\u904e\u9019\u500b\u5de5\u5177\uff0c\u53ef\u4ee5\u5c07\u5177\u5e73\u884c \u7b54\u6848 \u679c \u68c4 \u5132\u5b58 \u53ea\u5b58\u8a5e\u5c0d\u61c9\u7d50 \u5b8c\u5168\u653e \u53e5\u6578 4018 1282 \u6b64\u4f8b\u5b50\u4e2d\uff0c\u7a9f\u9019\u500b\u5b57\u53ea\u662f\u6307\u4e00\u7a9f\u4e95\u6216\u4e00\u7a9f\u6c34\u6c60\u7684\u610f\u601d\uff0c\u7701\u7565\u5f8c\u4e5f\u4e0d\u81f3\u65bc\u5f71\u97ff\u6574\u53e5\u6587\u610f\u3002 \u4e2d\u6587 \u53e4\u65e9\u7684\u4eba\u82e5\u662f\u770b\u5230\u6d41\u661f\u96e8\uff0c\u6703\u60f3\u5230\u4e00\u4e9b\u58de\u5146\u982d\u3002 \u5169\u7a2e\u65b9\u6cd5\u7684\u4e0d\u540c\u4e4b\u8655\uff1a\u65b9\u6cd5(\u4e00)\u7684\u8a5e\u908a\u754c\u5df2\u88ab\u56fa\u5b9a\uff0c\u76f8\u5c0d\u80fd\u627e\u51fa\u7684\u5019\u9078\u5ba2\u8a9e\u8a5e\u4e5f\u8f03\u5c11\u3001\u8f03\u4fb7 \u4e2d\u6587\u65b7\u8a5e \u6587\u8a5e\u7576\u5019\u9078\u8a5e\u3002\u6700\u5f8c\u5c07\u9019\u4e9b\u5019\u9078\u8a5e\u5efa\u7acb\u6210\u4e00\u500b\u6240\u6709\u65b7\u8a5e\u8def\u5f91\u7684\u6709\u5411\u5716\uff0c\u518d\u4ee5\u6700\u77ed\u8def\u5f91\u6f14\u7b97\u6cd5\u914d \u6ce5\u9c0d/\u6ed1/\u6e9c\u6e9c/\uff0c \u4f46\u662f\uff0c\u73fe\u4eca\u96fb\u8166\u4e0a\u7684\u5ba2\u8a9e\u8a9e\u6599\u4ecd\u975e\u5e38\u5331\u4e4f\uff0c\u80fd\u7528\u4f86\u5efa\u7f6e\u8a9e\u8a00\u6a21\u578b\u7684\u8a9e\u6599\u975e\u5e38\u6709\u9650\u3002\u56e0\u6b64\uff0c \u8a13\u7df4 \u6e2c\u8a66 \u8868\u5341\u4e09\u3001\u6a19\u8a18\u539f\u5247 4 \u7bc4\u4f8b \u8a9e\u5c0d\u7167\u8fad\u5178\u4e2d\uff0c\u627e\u51fa\u6240\u6709\u53ef\u80fd\u88ab\u8f49\u63db\u7684\u5ba2\u8a9e\u8a5e\uff0c\u6700\u5f8c\u518d\u8207\u4e2d\u6587\u65b7\u8a5e\u908a\u754c\u7684\u8a5e\u4e00\u8d77\u505a\u70ba\u5019\u9078\u8a5e\u3002 \u4e2d\u6587\u53e5\u5b50 \u65b0\u8a5e\u983b C(W i ) * \u5f8c\uff0c\u518d\u5c07\u9019\u4e9b\u4e2d\u6587\u8a5e\u67e5\u8a62\u5ba2\u8a9e\u8fad\u5178\uff0c\u627e\u51fa\u6240\u6709\u7684\u53ef\u80fd\u5019\u9078\u8a5e\u3002\u82e5\u627e\u4e0d\u5230\u5ba2\u8a9e\u8a5e\uff0c\u5247\u4ee5\u50c5\u6709\u7684\u4e2d Ceil(log 2 (464) * 15 * \ufe5d3 + 1\ufe5e) = 532 \u6ce5\u9c0d\u6ed1\u6e9c\u6e9c\uff0c \u6a21\u578b\u80fd\u5305\u542b\u7684\u60c5\u6cc1\u8d8a\u591a\uff0c\u6548\u80fd\u4e5f\u6703\u8d8a\u597d\u3002 \u5ba2\u59d4\u6703\u4e2d\u9ad8\u7d1a \u8a8d\u8b49\u8a9e\u6599\u8a8d\u8b49\u8a9e\u97f3\u6a94 \u5b8c\u5168 \u5c0d\u61c9 N \u90e8\u5206 \u5c0d\u61c9 \u5c07\u9019\u4e9b\u6a19\u8a18\u5b8c\u6210\u7684\u8cc7\u6599\u518d\u7d93\u904e\u4e00\u6b21\u7d93\u4eba\u5de5\u7be9\u9078\u5f8c\uff0c\u6700\u5f8c\u78ba\u8a8d\u6709\u6548\u53e5\u6578\u70ba\uff1aA \u8a9e\u6599 4018 \u53e5\uff0cB1-B4 \u8a9e\u6599\u5171 1282 \u53e5\uff0c\u6211\u5011\u4f7f\u7528\u8a9e\u6599\u7684\u5206\u4f48\u5982\u8868\u6240\u793a\uff1a \u8868\u4e03\u3001\u5ba2\u8a9e\u8a9e\u6599\u7684\u4f7f\u7528\u5206\u4f48 \u5ba2\u8a9e \u9019\u7a9f\u6cc9\u6c34\u4e2a\u6c34\u8cea\u7576\u751c\u7576\u6e05\u3002 \u4e2d\u6587\u6539 \u9019\u6cc9\u6c34\u7684\u6c34\u8cea\u5f88\u751c\u5f88\u6e05\u6f88\u3002 \u5ba2\u8a9e\u6539 \u9019\u6cc9\u6c34\u4e2a\u6c34\u8cea\u7576\u751c\u7576\u6e05\u3002 \u539f\u5247 4\uff1a\u4ee5\u5ba2\u8a9e\u53e5\u5b50\u70ba\u4e3b\u9ad4\uff0c\u767c\u73fe\u4e2d\u6587\u8a5e\u548c\u5ba2\u8a9e\u8a5e\u7684\u5c0d\u61c9\u6709\u722d\u8b70\u6642\u6216\u592a\u904e\u6a21\u7cca\u6642\uff0c\u8981\u627e\u904e\u4e00\u500b \u6700\u4f73\u9078\u64c7\u7684\u8a5e\u66ff\u63db\u3002 \u5ba2\u8a9e\u8a5e\u7684\u65b9\u6cd5\u6709\u95dc\u3002\u4f46\u6587\u7ae0\u7bc7\u5e45\u6709\u9650\uff0c\u672c\u7bc7\u8ad6\u6587\u5beb\u7684\u5be6\u9a57\u6578\u64da\u90fd\u4ee5\u7b2c\u4e00\u7a2e\u65b9\u6cd5\u70ba\u57fa\u5e95\u3002 \u6240\u8b02(\u4e00)\u4e2d\u6587\u65b7\u8a5e\u908a\u754c\u512a\u5148\uff0c\u662f\u5148\u5c07\u8f38\u5165\u4e2d\u6587\u6587\u53e5\u505a\u65b7\u8a5e\uff0c\u5f97\u5230\u78ba\u5b9a\u7684\u8a5e\u908a\u754c\u5f8c\uff0c\u518d\u4ee5\u9019\u4e9b \u4e2d\u6587\u8a5e\u53bb\u67e5\u627e\u570b\u5ba2\u8a9e\u5c0d\u7167\u8fad\u5178\uff0c\u627e\u51fa\u53ef\u80fd\u7684\u5ba2\u8a9e\u8a5e\u8f49\u63db\u5019\u9078\u3002\u800c(\u4e8c)\u5ba2\u8a9e\u8a5e\u512a\u5148\uff0c\u662f\u5148\u5f9e\u570b\u5ba2 \u5165\u5230\u4e2d\u6587\u65b7\u8a5e\u8fad\u5178\u88e1\u3002\u5982\u6b64\u505a\u6cd5\u80fd\u63d0\u9ad8\u539f\u672c\u5b8c\u5168\u4e0d\u6703\u88ab\u627e\u5230\u7684\u5ba2\u8a9e\u8a5e\uff0c\u6709\u6a5f\u6703\u6210\u70ba\u88ab\u8f49\u63db\u7684\u5019 \u672a\u542b\u8a5e\u6027\u7279\u5fb5\u7684\u8a5e\u983b \u672c\u65b7\u8a5e\u6cd5\uff0c\u662f\u5148\u5c07\u8f38\u5165\u7684\u4e2d\u6587\u53e5\u5b50\uff0c\u4ee5\u4e2d\u6587\u65b7\u8a5e\u6a21\u7d44\u505a\u65b7\u8a5e\uff0c\u5f97\u5230\u4e00\u500b\u78ba\u5b9a\u7684\u4e2d\u6587\u8a5e\u908a\u754c 3 \u8868\u5341\u516d\u3001\u4e2d\u6587\u65b7\u8a5e\u908a\u754c\u9650\u5236\uff0c\u7bc4\u4f8b 1 \u5ba2\u8a9e \u300c\u6ed1\u6e9c\u6e9c\u4ed4\u300d \u3002 4.3.3 \u4e2d\u6587\u65b7\u8a5e\u642d\u914d\u5ba2\u8a9e Uni-gram \u53ca Bi-gram \u8a9e\u8a00\u6a21\u578b\u7684\u6df7\u5408\u5f0f\u5206\u6578\u7b97\u6cd5 \u690d\u7269 \u9078\u8a5e\u3002\u4ee5\u4e0b\u662f\u4e00\u500b\u4f8b\u5b50\uff1a \u4e2d\u6587 \u539f\u672c\u7684\u65b7\u8a5e\u7cfb\u7d71\u7121\u6cd5\u5c07\u300c\u6ed1\u6e9c\u6e9c\u300d\u5224\u65b7\u51fa\u4f86\uff0c\u4fee\u6b63\u8fad\u5178\u5f8c\u5df2\u80fd\u6b63\u78ba\u65b7\u51fa\uff0c\u4e26\u8f49\u70ba\u5ba2\u8a9e\u8a5e \u5716\u56db\u3001\u4e2d\u6587\u65b7\u8a5e\u642d\u914d\u570b\u5ba2\u8a9e\u5c0d\u7167\u8fad\u5178\u7684\u76f4\u7ffb\u6cd5\u6d41\u7a0b\u5716 \u690d\u7269 N (PERIODCATEGORY) \u53e5\u5b50\u7de8\u865f 01-001 \u539f\u5247 1\uff1a\u6a19\u8a18\u6642\uff0c\u90fd\u4ee5\u4e0d\u4fee\u6539\u5ba2\u8a9e\u53e5\u5b50\u70ba\u539f\u5247\uff0c\u4f46\u53ef\u8df3\u904e\u4e0d\u5f71\u97ff\u6587\u610f\u7684\u5b57\u4e32\u3002 \u8868\u516b\u3001\u6a19\u8a18\u539f\u5247 1 \u7bc4\u4f8b \u4e2d\u6587 \u9019\u6cc9\u6c34\u7684\u6c34\u8cea\u5f88\u751c\u5f88\u6e05\u6f88\u3002 \u6b64\u4f8b\u5b50\u4e2d\uff0c\u5ba2\u8a9e\u300c\u5929\u9802\u661f\u5bbf\u4e2a\u8b8a\u5316\u300d\u8207\u4e2d\u6587\u300c\u5929\u4e0a\u7684\u661f\u5bbf\u8b8a\u5316\u300d\u96d6\u7136\u8a5e\u7684\u8a5e\u9806\u5e8f\u4e0d\u540c\uff0c\u4f46 \u4e2d\u6587\u53e5\u5b50\u4e2d\u7684\u300c\u7684\u300d\u8abf\u63db\u5f8c\uff0c\u4e5f\u4e0d\u5f71\u97ff\u6587\u610f\u3002\u800c\u4fee\u6539\u53e5\u5b50\u7684\u52d5\u4f5c\uff0c\u6211\u5011\u53ea\u5efa\u8b70\u4fee\u6539\u4e2d\u6587\uff0c\u5ba2\u6587 \u82e5\u975e\u5fc5\u8981\u5118\u91cf\u4fdd\u6301\u539f\u53e5\u3002\u539f\u56e0\u662f\u8f03\u80fd\u4fdd\u6301\u5ba2\u8a9e\u53e5\u5b50\u539f\u4f86\u7684\u7279\u6027\u3001\u7d50\u69cb\u3001\u7528\u8a9e\u2026\u7b49\u8cc7\u8a0a\u3002 4. \u7814\u7a76\u65b9\u6cd5 \u672c\u7cfb\u7d71\u5be6\u9a57\u6709\u5305\u62ec\u5169\u5927\u7a2e\u57fa\u5e95\uff0c(\u4e00)\u4e2d\u6587\u65b7\u8a5e\u908a\u754c\u512a\u5148\u3001(\u4e8c)\u5ba2\u8a9e\u8a5e\u908a\u754c\u512a\u5148\uff0c\u7d93\u5be6\u9a57\u767c\u73fe \u7b2c\u4e00\u7a2e\u65b9\u6cd5\u7684\u5916\u90e8\u6e2c\u8a66\u6b63\u78ba\u7387\u8f03\u9ad8\u3002\u4f46\u9019\u53ef\u80fd\u8ddf\u6a19\u8a18\u8a9e\u6599\u6642\u63a1\u7528\u7684\uff1a\u5148\u4e2d\u6587\u65b7\u8a5e\u3001\u518d\u4eba\u5de5\u5c0d\u61c9 \u8868\u5341\u4e5d\u3001\u570b\u5ba2\u8a9e\u5c0d\u7167\u8fad\u5178\u7684\u4e2d\u6587\u8a5e\u8a5e\u983b\u63db\u7b97 \u5716\u4e94\u3001\u642d\u914d\u5ba2\u8a9e\u8a5e\u983b\u7684\u76f4\u63a5\u7ffb\u8b6f\u65b7\u8a5e\u6cd5 \u7528\u8a5e\u6709\u6a5f\u6703\u6210\u70ba\u88ab\u8f49\u63db\u7684\u5019\u9078\u8a5e\uff0c\u6211\u5011\u5c07\u570b\u5ba2\u8a9e\u5c0d\u7167\u8fad\u5178\u4e2d\uff0c\u6240\u6709\u4e2d\u6587\u8a5e\u6b04\u4f4d\u7684\u8a5e\u8cc7\u6599\uff0c\u90fd\u52a0 \u4e00\u500b\u570b\u5ba2\u5c0d\u7167\u8a5e\u300c\u690d\u7269/\u690d\u7269\u300d \uff0c\u5176\u8a5e\u983b\u8f49\u63db\u7684\u4f8b\u5b50\uff1a \u6b63\u78ba\u7b54\u6848 \u9c17\u9c0d\u4ed4/\u6ed1\u6e9c\u6e9c\u4ed4/\uff0c \u5ba2\u8a9e\u8f49\u63db\u7d50\u679c \u524d\u6587\u6709\u63d0\u5230\uff0c\u672c\u8ad6\u6587\u63a1\u7528\u7b2c\u4e00\u7a2e\u65b9\u6cd5\u300c\u4e2d\u6587\u65b7\u8a5e\u908a\u754c\u512a\u5148\u300d\u70ba\u57fa\u5e95\uff0c\u56e0\u6b64\u70ba\u4e86\u8b93\u4e00\u4e9b\u5ba2\u8a9e \u6539\u5584\u5f8c\u4e2d\u6587\u7ffb\u5ba2\u8a9e \u9c17\u9c0d\u4ed4/\u6ed1\u6e9c\u6e9c\u4ed4/\uff0c \u5ba2\u8a9e\u8f49\u63db\u7d50\u679c 4.2 \u4fee\u6539\u4e2d\u6587\u65b7\u8a5e\u8fad\u5178 14 \u5c0d\u61c9 \u807d\u5ba2\u8a9e\u4f8b\u53e5\u8a9e\u97f3 \u4eba\u5de5 C(W i ) * = log 2 (464) * 15 * \ufe5dC(W i ) + 1\ufe5e (1) \u539f\u4e2d\u6587\u7ffb\u5ba2\u8a9e \u9c17\u9c0d\u4ed4/\u6ed1/\u6e9c\u6e9c/\uff0c (COMMACATEGORY) \u6bcb(D) \u9069\u5408(VH) \u505a(VC) \u66f8\u623f(Nc) \u3002 60 \u5ba2\u8a9e\u65b7\u8a5e\u7d50\u679c \u9019(Nep) \u96bb(Nf) \u623f\u9593(Nc) \u4e2a(DE) \u5149\u7dda(Na) \u6bcb(D) \u597d(VH) \uff0c \u61c9\u5230\u7684\u8a5e\uff0c\u800c\u4e0d\u5c07\u9019\u4e9b\u672a\u5b8c\u6210\u5c0d\u61c9\u7684\u53e5\u5b50\u8996\u70ba\u65b7\u8a5e\u7b54\u6848\u3002\u4ee5\u4e0b\u70ba\u6a19\u8a18\u6642\u7684 6 \u5927\u6a19\u8a18\u539f\u5247\uff1a \u5ba2\u8a9e\u6539 \u4e0a\u65e9\u4e2a\u4eba\u6703\u89c0\u5bdf\u5929\u9802\u661f\u5bbf\u4e2a\u8b8a\u5316\uff0c\u4f86\u5224\u65b7\u4eba\u9593\u4e2a\u5409\u51f6\u3002 \u4ef0\u822c\u5f62\u3001\u56de\u5bb6/\u8f49\u5c4b\u4e0b\u300d \u3002\u56e0\u6b64\u9019\u53e5\u53ef\u5132\u5b58\u5176\u300c\u8a5e\u914d\u5c0d\u300d\u6210\u529f\u7684\u90e8\u4efd\uff0c\u4f46\u653e\u68c4\u5132\u5b58\u70ba\u65b7\u8a5e\u7b54\u6848\u3002 W i \u2208 (Wo\u03b3d Length > 1)\uff1a \u6539\u5584\u5f8c\u4e2d\u6587\u65b7\u8a5e \u6ce5\u9c0d/\u6ed1\u6e9c\u6e9c/\uff0c \u5ba2\u8a9e\u8a5e \u5716\u4e09\u3001\u5ba2\u8a9e\u65b7\u8a5e\u6a21\u7d44\u67b6\u69cb\u5716 80 \u53e5 \u53e5 \u8868 \u7576\u6210\u719f\uff0c\u56e0\u70ba\u4e2d\u6587\u8a9e\u6599\u7684\u9f90\u5927\uff0c\u4ee5\u898f\u5247\u6cd5\u914d\u5408\u6a5f\u7387\u6a21\u578b\u7684\u6df7\u5408\u5f0f\u65b7\u8a5e\u6cd5\u6240\u767c\u5c55\u51fa\u4f86\u7684\u4e2d\u6587\u65b7\u8a5e \u7cfb\u7d71\uff0c\u65b7\u8a5e\u6548\u80fd\u53ca\u8a5e\u6027\u6a19\u8a18\u7684 F \u5206\u6578\u5df2\u9054\u5230 96.69%\u53ca 92.04%\u3002\u56e0\u6b64\u90fd\u80fd\u76f4\u63a5\u4ee5\u4e2d\u6587\u65b7\u8a5e\u7cfb\u7d71\u4f86 \u5f97\u5230\u53ef\u9760\u7684\u65b7\u8a5e\u53ca\u8a5e\u6027\u7279\u5fb5\u6a19\u8a18\u7684\u7d50\u679c\u3002\u4f46\u5ba2\u8a9e\u6587\u7ae0\u7684\u8655\u7406\uff0c\u56e0\u76ee\u524d\u5e02\u9762\u4e0a\u53ca\u5b78\u754c\u7684\u5ba2\u8a9e\u65b7\u8a5e \u7cfb\u7d71\u4ecd\u8655\u65bc\u767c\u5c55\u4e2d\u7684\u968e\u6bb5\uff0c\u9084\u6c92\u8fa6\u6cd5\u4f9d\u8cf4\u4efb\u4f55\u5ba2\u8a9e\u65b7\u8a5e\u7cfb\u7d71\u4f86\u81ea\u52d5\u8655\u7406\u3002\u56e0\u6b64\u6211\u5011\u958b\u767c\u51fa\u4e00\u5957 \u5de5\u5177\uff0c\u4ee5\u534a\u81ea\u52d5\u7684\u65b9\u6cd5\uff0c\u5feb\u901f\u7684\u91dd\u5c0d\u5ba2\u8a9e\u53e5\u5b50\uff0c\u505a\u5ba2\u8a9e\u65b7\u8a5e\u7684\u6a19\u8a18\u3002 3.3.2.1 \u5ba2\u8a9e\u65b7\u8a5e\u6a19\u8a18\u5de5\u5177\u4ecb\u7d39\u8207\u64cd\u4f5c \u53d6\u51fa\u4e2d\u6587\u4f8b \u53d6\u51fa\u5ba2\u8a9e\u4f8b \u5ba2\u59d4\u6703\u521d\u7d1a\u53ca\u4e2d \u4e2d\u6587\u8a5e\u5e8f\u5217 \u9ad8\u7d1a\u8a8d\u8b49\u8a9e\u6599 (COMMACATEGORY) \u4e0d(D) \u9069\u5408(VH) \u505a(VC) \u66f8\u623f(Nc) \u3002 (PERIODCATEGORY) \u5ba2\u8a9e\u53e5\u5b50 \u9019\u96bb\u623f\u9593\u4e2a\u5149\u7dda\u6bcb\u597d\uff0c\u6bcb\u9069\u5408\u505a\u66f8\u623f\u3002 \u56e0\u6b64\uff0c\u6211\u5011\u5728\u9032\u884c\u4eba\u5de5\u6a19\u8a18\u6642\u4e5f\u767c\u73fe\uff0c\u5176\u5be6\u5927\u90e8\u5206\u51fa\u73fe\u7121\u6cd5\u5c0d\u61c9\u7684\u60c5\u6cc1\uff0c\u90fd\u53ef\u4ee5\u4ee5\u4eba\u5de5\u4fee \u6539\u4e2d\u6587\u53e5\u5b50\u6216\u7528\u8a5e\u7684\u65b9\u5f0f\uff0c\u505a\u9069\u7576\u7684\u4fee\u98fe\u8207\u8abf\u6574\uff0c\u4f86\u9054\u5230\u4e0d\u5f71\u97ff\u6587\u610f\u3001\u53c8\u80fd\u8207\u5ba2\u8a9e\u53e5\u5b50\u5c0d\u61c9\u5b8c \u5168\u7684\u76ee\u7684\u3002\u4f46\u67d0\u4e9b\u53e5\u5b50\u4ecd\u7121\u6cd5\u78ba\u8a8d\u5982\u4f55\u6a19\u8a18\u6642\uff0c\u6a19\u8a18\u8005\u4e5f\u53ef\u9078\u64c7\u653e\u68c4\u8a72\u53e5\u7684\u6a19\u8a18\uff0c\u6216\u53ea\u5b58\u80fd\u5c0d \u4e2d\u6587 \u53e4\u6642\u5019\u7684\u4eba\u6703\u89c0\u5bdf\u5929\u4e0a\u7684\u661f\u5bbf\u8b8a\u5316\uff0c\u4f86\u5224\u65b7\u4eba\u9593\u7684\u5409\u51f6\u3002 \u5ba2\u8a9e \u4e0a\u65e9\u4e2a\u4eba\u6703\u89c0\u5bdf\u5929\u9802\u661f\u5bbf\u4e2a\u8b8a\u5316\uff0c\u4f86\u5224\u65b7\u4eba\u9593\u4e2a\u5409\u51f6\u3002 \u4e2d\u6587\u6539 \u53e4\u6642\u5019\u7684\u4eba\u6703\u89c0\u5bdf\u5929\u4e0a\u661f\u5bbf\u7684\u8b8a\u5316\uff0c\u4f86\u5224\u65b7\u4eba\u9593\u7684\u5409\u51f6\u3002 \u6b64\u4f8b\u5b50\u4e2d\uff0c\u767c\u73fe\u5230\u300c\u65e2\u98b3\u98a8\u53c8\u4e0b\u96e8\u300d\u548c\u300c\u98a8\u5408\u96e8\u300d\u770b\u4f3c\u80fd\u5c0d\u61c9\uff0c\u4f46\u5be6\u969b\u4e0a\u53ef\u80fd\u53ea\u6709\u5728\u9019\u53e5 \u66fe\u51fa\u73fe\u9019\u7a2e\u60c5\u6cc1\u3002\u56e0\u6b64\uff0c\u9019\u7a2e\u975e\u6210\u8a9e\u7684\u7247\u8a9e\uff0c\u4e5f\u8a31\u53ea\u662f\u5076\u7136\u51fa\u73fe\uff0c\u6211\u5011\u4e0d\u5132\u5b58\u6b64\u985e\u7684\u7b54\u6848\u548c\u8a5e \u914d\u5c0d\u3002\u4f46\u4f9d\u7167\u539f\u5247 5\uff0c\u6211\u5011\u767c\u73fe\u300c\u8981\u600e\u9ebc\u6a23\u56de\u5bb6/\u611b\u4ef0\u822c\u5f62\u8f49\u5c4b\u4e0b\u300d\u53ef\u914d\u5c0d\uff0c\u6210\u300c\u8981/\u611b\u3001\u600e\u9ebc\u6a23/ \u8a9e \u5178 \u7406 \u5f8c\u7e8c\u6587\u53e5\u5206\u6790\u7684\u8655 \u5e73\u5747\u8a5e\u983b 464 2. \u6211\u5011\u4ee5\u4e0b\u5217\u516c\u5f0f\uff0c\u8a08\u7b97\u51fa\u6bcf\u500b\u8981\u52a0\u5165\u4e2d\u6587\u65b7\u8a5e\u8fad\u5178\u4e2d\u7684\u4e2d\u6587\u8a5e\uff0c\u5176\u65b0\u8a5e\u983b C(W i ) * \uff0c\u5176\u4e2d \u4e2d\u6587\u53e5\u5b50 \u539f\u4e2d\u6587\u65b7\u8a5e \u6ce5\u9c0d/\u6ed1/\u6e9c\u6e9c/\uff0c \u5ba2\u8a9e\u8a5e \u6709\u5ba2\u8a9e\u8a5e \u6ce5\u9c0d\u6ed1\u6e9c\u6e9c\uff0c \u5c0d\u7167\u8a5e \u7e3d\u8a5e\u983b 462729801 \u6709\u5ba2\u8a9e\u8a5e \u8868\u4e8c\u5341\u4e8c\u3001\u52a0\u5165\u5ba2\u8a9e\u8a5e\u5f8c\u7684\u4e2d\u6587\u65b7\u8a5e\u8fad\u5178\u7684\u6539\u5584 \u570b\u5ba2 \u4e2d\u6587\u65b7\u8a5e\u53ca\u6a19\u8a5e\u6027 \u8868\u4e2d\u53ef\u770b\u51fa\uff0c\u6a19\u8a18\u901f\u5ea6\u5e73\u5747 33.04 \u79d2\u80fd\u5b8c\u6210\u4e00\u53e5\u3002\u800c\u6a19\u8a18\u5b8c\u6210\u7684\u8cc7\u6599\uff0c\u6703\u5132\u5b58\u5176 1.\u4e2d\u6587\u53e5 \u5b50\u30012.\u4e2d\u6587\u65b7\u8a5e\u53ca\u8a5e\u6027\u6a19\u8a18\u7d50\u679c\u30013.\u5ba2\u8a9e\u53e5\u5b50\u30014.\u5ba2\u8a9e\u65b7\u8a5e\u53ca\u8a5e\u6027\u6a19\u8a18\u7d50\u679c\u30015.\u5ba2\u59d4\u6703\u8a9e\u6599\u4e2d\u7684 \u53e5\u5b50\u7de8\u865f\uff0c\u7b49\u4e94\u500b\u6b04\u4f4d\u8cc7\u6599\uff0c\u5132\u5b58\u7d50\u679c\u5982\u4e0b\u8868\uff1a \u8868\u516d\u3001\u5ba2\u8a9e\u65b7\u8a5e\u6a19\u8a18\u7d50\u679c\u7684\u8cc7\u6599\u6a23\u8c8c \u4e2d\u6587\u53e5\u5b50 \u9019\u500b\u623f\u9593\u7684\u5149\u7dda\u4e0d\u597d\uff0c\u4e0d\u9069\u5408\u505a\u66f8\u623f\u3002 \u4e2d\u6587\u65b7\u8a5e\u7d50\u679c \u9019(Nep) \u500b(Nf) \u623f\u9593(Nc) \u7684(DE) \u5149\u7dda(Na) \u4e0d(D) \u597d(VH) \uff0c \u5ba2\u8a9e \u4eca\u6661\u65e5\u4e2a\u5929\u6642\u7576\u597d\uff0c\u65e5\u982d\u843d\u5c71\u4ee5\u5f8c\u5c31\u770b\u5f97\u8457\u6eff\u5929\u4e2a\u661f\u4ed4\u3002 \u539f\u5247 6\uff1a\u914d\u5c0d\u6642\u4ee5\u300c\u8a5e\u300d\u70ba\u55ae\u4f4d\uff0c\u4e0d\u8981\u4ee5\u7247\u8a9e\u70ba\u55ae\u4f4d\u3002 \u901a\u5e38\uff0c\u975e\u6210\u8a9e\u7684\u7247\u8a9e\uff0c\u7528\u6cd5\u53ef\u80fd\u53ea\u662f\u5076 \u7d44 \u4e2d\u6587\u8f49\u5ba2\u6587\u6a21 \u4e2d\u6587\u65b7\u8a5e\u8fad\u5178\u7e3d\u8a5e\u983b\u6578\u9ad8\u9054 462729801 \u500b\u8a5e\uff0c\u800c\u5ba2\u8a9e\u8a13\u7df4\u8a9e\u6599\u50c5\u6709 47079 \u500b\u8a5e\u3002\u56e0\u6b64\u6211\u5011\u5c07\u4e2d \u2211 \u300c\u4e2d\u7814\u9662\u4e2d\u6587\u5e73\u8861\u8a9e\u6599\u5eab 3.0\u300d \u3002\u5982\u8868\u4e8c\u5341\u4e00\uff1a \u4e2d\u6587\u65b7\u8a5e\u7d50\u679c \u5716\u4e8c\u3001\u6a19\u8a18\u5de5\u5177\u64cd\u4f5c\u756b\u9762 3.3.2.2 \u6a19\u8a18\u539f\u5247 \u56e0\u70ba\u9019\u4efd\u5ba2\u8a9e\u8a8d\u8b49\u6559\u6750\u8a9e\u6599\uff0c\u5c08\u5bb6\u5011\u7de8\u5be9\u7684\u4e3b\u8981\u76ee\u7684\u662f\u70ba\u4e86\u5ba2\u8a9e\u6559\u5b78\u7528\uff0c\u4e26\u4e0d\u662f\u70ba\u4e86\u8981\u5efa \u7acb\u300c\u570b\u8a9e/\u5ba2\u8a9e\u65b7\u8a5e\u5c0d\u61c9\u300d\u7684\u8a9e\u6599\uff0c\u6240\u4ee5\u5728\u64b0\u5beb\u4f8b\u53e5\u6642\u4e26\u4e0d\u6703\u7279\u5225\u6c42\u5f37\u6216\u6ce8\u610f\u5230\u570b\u5ba2\u8a9e\u7684\u5b8c\u5168\u5c0d \u61c9\u3002 \u4e2d\u6587\u6539 \u4eca\u5929\u7684\u5929\u6c23\u5f88\u597d\uff0c\u592a\u967d\u4e0b\u5c71\u4ee5\u5f8c\u5c31\u770b\u5f97\u5230\u6eff\u5929\u7684\u661f\u661f\u3002 \u5ba2\u8a9e\u6539 \u4eca\u6661\u65e5\u4e2a\u5929\u6642\u7576\u597d\uff0c\u65e5\u982d\u843d\u5c71\u4ee5\u5f8c\u5c31\u770b\u5f97\u8457\u6eff\u5929\u4e2a\u661f\u4ed4\u3002 \u6b64\u4f8b\u5b50\u4e2d\uff0c\u82e5\u76f4\u63a5\u7701\u7565\u300c\u53ef\u4ee5\u300d\u9019\u500b\u8a5e\uff0c\u4e5f\u4e0d\u6703\u5f71\u97ff\u5230\u539f\u53e5\u7684\u6587\u610f\u3002 \u8868\u5341\u4e8c\u3001\u6a19\u8a18\u539f\u5247 3 \u7bc4\u4f8b 2 \u7136\u51fa\u73fe\uff0c\u56e0\u6b64\u5132\u5b58\u9019\u985e\u8cc7\u6599\u6c92\u610f\u7fa9\u3002 \u8868\u5341\u4e94\u3001\u6a19\u8a18\u539f\u5247 6 \u7bc4\u4f8b \u4e2d\u6587 \u73fe\u5728\u5916\u9762\u65e2\u98b3\u98a8\u53c8\u4e0b\u96e8\uff0c\u8981\u600e\u9ebc\u6a23\u56de\u5bb6\uff1f \u5ba2\u8a9e \u9019\u4e0b\u5916\u80cc\u98a8\u5408\u96e8\uff0c\u611b\u4ef0\u822c\u5f62\u8f49\u5c4b\u4e0b\uff1f \u578b \u6700\u4f73\u7684\u8f49\u63db\u7d50\u679c\u53ca\u8a5e\u6027\u6a19\u8a18\u7d50\u679c \u6587\u65b7\u8a5e\u8fad\u5178\u7684\u5e73\u5747\u8a5e\u983b\u53d6 Log \u4ee5 2 \u70ba\u5e95\u4e58\u4e0a 15(\u7d93\u5be6\u9a57\u5f97\u5230\uff0c15 \u6709\u6700\u4f73\u7684\u6b63\u78ba\u7387)\uff0c\u518d\u8207\u570b\u5ba2\u8a9e \u8868\u4e8c\u5341\u4e00\u3001\u4e2d\u6587\u65b7\u8a5e\u8fad\u5178\u4fee\u6539\u524d\u5f8c\u6bd4\u8f03 \u4e2d\u6587\u65b7\u8a5e\u7d50\u679c Uni-gram 1. \u7d71\u8a08\u51fa\u4e2d\u6587\u65b7\u8a5e\u8fad\u5178\u539f\u59cb\u7684\u5206\u4f48\uff0c\u6211\u5011\u5f97\u5230\u5176\u5e73\u5747\u8a5e\u983b\u70ba 464 \u6b21\u3002 \u8868\u5341\u516b\u3001\u4e2d\u6587\u65b7\u8a5e\u8fad\u5178\u8cc7\u6599\u5206\u4f48 \u7e3d\u8a5e\u6578 995642 \u4fee\u6539\u5f8c 97.15% 96.16% 96.65% \u53ef\u770b\u5230 F-Measure \u7565\u4f4e 0.04%\uff0c\u4f46\u6539\u5584\u4e86\u4ee5\u4e0b\u554f\u984c\uff1a \u67e5\u7121\u5ba2\u8a9e\u8a5e \u570b\u5ba2\u8a9e\u5c0d\u7167\u8fad\u5178 \u67e5\u8a62 \u8fad\u5178 \u4e2d\u6587\u8a5e \u570b\u5ba2\u8a9e\u5c0d\u7167\u8fad\u5178 \u8fad\u5178 \u4e2d\u6587\u8a5e \u67e5\u8a62 \u67e5\u7121\u5ba2\u8a9e\u8a5e \u5ba2\u8a9e\u8a9e\u8a00\u6a21 \u4fee\u6539\u524d 97.16% 96.21% 96.69% \u4e14\u5b57\u9577\u6700\u77ed\u8005 Bi-gram \u5c0d\u7167\u8fad\u5178\u7684\u8a5e\u983b\u76f8\u4e58\uff0c\u5f97\u5230\u8a72\u8a5e\u65b0\u7684\u8a5e\u983b\u3002\u6b65\u9a5f\u5982\u4e0b\uff1a Precision Recall F-Measure \u8a5e\u983b\u6700\u9ad8\u3001\u8a5e\u6027\u76f8\u540c i-1 i</td></tr></table>", | |
| "num": null, | |
| "html": null, | |
| "type_str": "table", | |
| "text": "\u7684\u5e73\u884c\u8a9e\u6599\u6a19\u5de5\u4f5c\u8a18\u4e0a\uff0c\u5982 Tsai \u7684\u78a9\u58eb\u8ad6\u6587 [1]\u4e5f\u662f\u7528\u6b64\u65b9\u6cd5\u3002\u56e0\u70ba\u4e2d\u6587\u6587\u53e5\u8655\u7406\u7cfb \u5ba2\u8a9e\u8207\u4e2d\u6587\u90fd\u5c6c\u65bc\u6f22\u8a9e\uff0c\u6587\u6cd5\u7d50\u69cb\u5e7e\u4e4e\u76f8\u540c\uff0c\u50c5\u6709\u5c11\u6578\u4fda\u8a9e\u3001\u7279\u6b8a\u7684\u5ba2\u8a9e\u69cb\u8a5e\u4e0d\u540c\u3002 \u672c\u7814\u7a76\u65e8\u5728\u63d0\u51fa\u4e00\u500b\u4e2d\u6587\u8f49\u5ba2\u6587\u65b7\u8a5e\u8cc7\u8a0a\u7684\u65b9\u6cd5\uff0c\u91dd\u5c0d\u56b4\u91cd\u8cc7\u6599\u7a00\u758f\u7684\u60c5\u6cc1\u4e0b\uff0c\u63d0\u51fa\u642d\u914d \u4e2d\u6587\u65b7\u8a5e\u6a21\u7d44\u8207\u5ba2\u8a9e\u8a9e\u8a00\u6a21\u578b\u7684\u6df7\u5408\u5f0f N-Gram \u5e8f\u5217\u5206\u6578\u7684\u7b97\u6cd5\u3002\u900f\u904e\u5169\u968e\u6bb5\u65b9\u5f0f\uff0c\u5c07\u4e2d\u6587\u6587 \u53e5\u4ee5\u4e2d\u6587\u65b7\u8a5e\u6a21\u7d44\u5f97\u5230\u7b2c\u4e00\u968e\u6bb5\u7684\u65b7\u8a5e\u53ca\u8a5e\u6027\u6a19\u8a18\u7d50\u679c\u5f8c\uff0c\u518d\u4ee5\u570b\u5ba2\u8a9e\u5c0d\u7167\u8fad\u5178\u627e\u51fa\u6240\u6709\u53ef\u80fd \u88ab\u8f49\u63db\u7684\u5ba2\u8a9e\u8a5e\u5e8f\u5217\uff0c\u4ee5\u5c11\u91cf\u5ba2\u8a9e Bi-gram\u3001Uni-gram \u8a9e\u8a00\u6a21\u578b\u70ba\u57fa\u5e95\uff0c\u642d\u914d\u6df7\u5408\u5f0f N-gram \u5e8f \u53ca\u8a9e\u6599 B\uff0c\u8a9e\u6599 A \u6709 4196 \u53e5\uff0c\u8a9e\u6599 B \u6709 2444 \u53e5\uff0c\u4e26\u518d\u5c07 B \u8a9e\u6599\u4f9d\u7de8\u865f\u9806\u4f4d\u5206\u70ba 4 \u4efd\uff0cB1-B4\uff0c\u6bcf\u4efd 611 \u53e5\u3002 \u5c07 A\u3001B \u8a9e\u6599\uff0c\u5206\u5225\u627e(\u6a19\u8a18\u8005 1)\u4e00\u4f4d\u78a9\u58eb\u751f\u3001(\u6a19\u8a18\u8005 2~5)\u56db\u4f4d\u5927\u5b78\u5728\u5b78\u751f\uff0c\u4f7f\u7528\u672c\u8ad6\u6587\u958b \ufe5dP(W 1 | < S >) * P(W 1 )\ufe5e + \u2211 n Log e \ufe5dP(W i |W i-1 ) * P(W i )\ufe5e\ufe5c" | |
| }, | |
| "TABREF1": { | |
| "content": "<table><tr><td colspan=\"5\">\u53e5\uff0c\u6e2c\u8a66(B1-B4)\u8a9e\u6599\u5171 1282 \u53e5\uff0c\u6211\u5011\u4f7f\u7528\u8a9e\u6599\u7684\u5206\u4f48\u5982\u8868\u6240\u793a\uff1a \u6e2c\u8a66 80.78% 79.53% 80.15%</td><td>84.04%</td></tr><tr><td>\u5be6\u9a57 B</td><td>\u8a13\u7df4</td><td colspan=\"3\">\u8868\u4e8c\u5341\u4e09\u3001\u5ba2\u8a9e\u8a9e\u6599\u7684\u4f7f\u7528\u5206\u4f48 93.96% 93.24% 93.60%</td><td>-</td></tr><tr><td/><td>\u6e2c\u8a66</td><td>80.37%</td><td>\u8a13\u7df4 79.11%</td><td>\u6e2c\u8a66 79.74%</td><td>-</td></tr><tr><td/><td/><td>\u53e5\u6578</td><td>4018</td><td>1282</td></tr><tr><td/><td/><td>\u8a5e\u6578</td><td>45304</td><td>17646</td></tr><tr><td colspan=\"6\">\u5b57\u6578 \u73fe\u5728 Bi-gram \u8a9e\u8a00\u6a21\u578b\u4e2d\uff0c\u4e5f\u4e0d\u7b26\u5408\u53e5\u5b50\u5be6\u969b\u7684\u72c0\u6cc1\uff0c\u9019\u662f\u8cc7\u6599\u7a00\u758f\u7684\u554f\u984c\u3002 65572 25478</td></tr><tr><td colspan=\"6\">5.3 \u8a55\u4f30\u7d50\u679c\u8207\u8a0e\u8ad6 \u800c\u5728\u4e0b\u4e00\u968e\u6bb5\u7684\u5de5\u4f5c\u4e2d\uff0c\u6211\u5011\u5c07\u8981\u589e\u52a0\u5ba2\u8a9e\u69cb\u8a5e\u898f\u5247\u4ee5\u53ca\u6301\u7e8c\u589e\u52a0\u5ba2\u8a9e\u8a9e\u6599\uff0c\u5ba2\u8a9e\u69cb\u8a5e\u5305</td></tr><tr><td colspan=\"5\">5.3.1 \u4e2d\u6587\u65b7\u8a5e\u642d\u914d\u570b\u5ba2\u8a9e\u5c0d\u7167\u8fad\u5178\u7684\u76f4\u7ffb\u6cd5 \u62ec\u4e86\uff1a\u91cd\u8907\u3001\u9644\u52a0\u3001\u9644\u5408\u3001\u5408\u4f75\u2026\u7b49\u69cb\u8a5e\u898f\u5247\u3002</td></tr><tr><td colspan=\"3\">\u5be6\u9a57 A\uff1a\u4f7f\u7528 Lo[6]\u7684\u570b\u5ba2\u8a9e\u5c0d\u7167\u8fad\u5178\u3002 \u5be6 6. \u7d50\u8ad6\u53ca\u672a\u4f86\u5de5\u4f5c</td><td/><td/></tr><tr><td colspan=\"6\">\u9a57 B\uff1a\u4f7f\u7528 Wu[5]\u7684\u570b\u5ba2\u8a9e\u5c0d\u7167\u8fad\u5178\u3002 \u672c\u8ad6\u6587\u91dd\u5c0d\u4e2d\u6587\u8f49\u5ba2\u6587\u6587\u8f49\u97f3\u7cfb\u7d71(Hakka Text-to-Speech System, HTTS)\u4e2d\u7684\u5ba2\u8a9e\u65b7\u8a5e\u8655\u7406\uff0c</td></tr><tr><td colspan=\"6\">\u5be6\u9a57 C\uff1a\u4f7f\u7528\u672c\u8ad6\u6587\u6240\u5efa\u7f6e\u7684\u570b\u5ba2\u8a9e\u5c0d\u7167\u8a5e\u5178\u3002 \u5df2\u63d0\u51fa\u4e00\u500b\u57fa\u790e\u7684\u7814\u7a76\u67b6\u69cb\u3002\u4f46\u56e0\u70ba\u76ee\u524d\u5ba2\u8a9e\u96fb\u5b50\u8a9e\u6599\u6709\u56b4\u91cd\u4e0d\u8db3\u7684\u554f\u984c\uff0c\u5c0d\u65bc\u672c\u8ad6\u6587\u6240\u63a2\u8a0e</td></tr><tr><td colspan=\"6\">\u8868\u4e8c\u5341\u56db\u3001\u76f4\u63a5\u7ffb\u8b6f\u65b7\u8a5e\u6cd5\u6548\u80fd\u8a55\u4f30-\u5167\u90e8\u6e2c\u8a66 \u7684\u4e3b\u984c\u800c\u8a00\uff0c\u662f\u4e00\u9805\u975e\u5e38\u8271\u56f0\u7684\u6311\u6230\u3002\u7814\u7a76\u4e4b\u521d\uff0c\u6211\u5011\u4e26\u6c92\u6709\u5ba2\u8a9e\u65b7\u8a5e\u8a9e\u6599\u53ef\u4f7f\u7528\uff0c\u50c5\u6709\u5c11\u91cf</td></tr><tr><td colspan=\"6\">Precision \u4e00\u53e5\u53e5\u672a\u8655\u7406\u7684\u570b\u5ba2\u8a9e\u5c0d\u7167\u6587\u53e5\u3002\u56e0\u6b64\uff0c\u6211\u5011\u6295\u5165\u4e86\u5927\u91cf\u6642\u9593\u5728\u5ba2\u8a9e\u8a9e\u6599\u7684\u6a19\u8a18\u3001\u5efa\u7f6e\u53ca\u8fad\u5178 Recall F-Measure \u5b57\u4e32\u76f8\u4f3c\u5ea6</td></tr><tr><td colspan=\"6\">\u5be6\u9a57 A \u7684\u6821\u6b63\u3001\u6a19\u97f3\u3002\u6211\u5011\u4e5f\u6301\u7e8c\u7684\u5f9e\u570b\u5c0f\u5ba2\u8a9e\u6559\u6750\u3001\u5ba2\u8a9e\u6717\u8b80\u6bd4\u8cfd\u6587\u7ae0\u2026\u7b49\u96fb\u5b50\u6587\u672c\u4e2d\uff0c\u4eba\u5de5\u5efa\u7f6e 68.41% 69.12% 68.76% 73.29%</td></tr><tr><td colspan=\"6\">\u5be6\u9a57 B \u51fa\u66f4\u591a\u7684\u5ba2\u8a9e\u65b7\u8a5e\u8a9e\u6599\u53ca\u5ba2\u8a9e\u65b0\u8a5e\u3002\u800c\u76ee\u524d\u7528\u4f86\u6e2c\u8a66\u7684\u8a9e\u6599\uff0c\u7686\u662f\u6211\u5011\u81ea\u884c\u5efa\u7f6e\u7684\uff0c\u6240\u4ee5\u6211\u5011 72.66% 73.42% 73.04% 75.68%</td></tr><tr><td colspan=\"6\">\u5be6\u9a57 C \u4e5f\u975e\u5e38\u9700\u8981\u66f4\u591a\u4e0d\u540c\u4f86\u6e90\u7684\u5ba2\u8a9e\u65b7\u8a5e\u8a9e\u6599\uff0c\u505a\u66f4\u516c\u8b49\u5ba2\u89c0\u7684\u6548\u80fd\u8a55\u4f30\u3001\u6bd4\u8f03\u3002 75.02% 75.80% 75.41% 78.36%</td></tr><tr><td colspan=\"6\">\u7531\u6b64\u5be6\u9a57\u53ef\u5f97\u77e5\uff0c\u8a5e\u5178\u7684\u6821\u6b63\u8207\u8a5e\u76ee\u7684\u589e\u52a0\uff0c\u80fd\u986f\u8457\u7684\u6539\u5584\u5ba2\u8a9e\u65b7\u8a5e\u7cfb\u7d71\u7684\u6548\u80fd\u3002 \u672c\u8ad6\u6587\u5728\u5ba2\u8a9e\u65b7\u8a5e\u65b9\u6cd5\u65b9\u9762\uff0c\u4e0d\u540c\u65bc\u904e\u53bb\u7684\u67b6\u69cb\uff0c\u6211\u5011\u63d0\u51fa\u4e86\u4f7f\u7528\u5ba2\u8a9e Uni-gram \u53ca Bi-gram</td></tr><tr><td colspan=\"6\">5.3.2 \u4e2d\u6587\u65b7\u8a5e\u642d\u914d\u5ba2\u8a9e\u8a5e\u983b\u7684\u76f4\u63a5\u7ffb\u8b6f\u6cd5 \u8a9e\u8a00\u6a21\u578b\u7684\u6df7\u5408\u5f0f\u65b7\u8a5e\u5e8f\u5217\u5206\u6578\u7b97\u6cd5\uff0c\u53ef\u770b\u898b\u5728\u5167\u90e8\u6e2c\u8a66\u7684\u8a55\u4f30\u4e0a\uff0c\u986f\u793a\u5728\u8a9e\u6599\u5145\u8db3\u7684\u60c5\u6cc1 \u4e0b\uff0c</td></tr><tr><td colspan=\"6\">\u5be6\u9a57 A\uff1a\u8f38\u5165\u4e2d\u6587\u6587\u53e5\uff0c\u8a55\u4f30\u5176\u4e2d\u6587\u8f49\u5ba2\u8a9e\u7684\u65b7\u8a5e\u6548\u80fd\u3002 \u5be6 \u5c07\u6703\u6709\u4e0d\u932f\u7684\u65b7\u8a5e\u8868\u73fe\u3002\u4f46\u8a5e\u6027\u6a19\u8a18\u7684\u6b63\u78ba\u7387\u50c5\u80fd\u505a\u70ba\u53c3\u8003\uff0c\u56e0\u70ba\u6211\u5011\u81ea\u5df1\u6a19\u8a18\u7522\u751f\u7684\u6a19 \u6e96\u7b54</td></tr><tr><td colspan=\"6\">\u9a57 B\uff1a\u8f38\u5165\u4e2d\u6587\u6587\u53e5\uff0c\u8a55\u4f30\u5176\u4e2d\u6587\u8f49\u5ba2\u8a9e\u53ca\u8a5e\u6027\u6a19\u8a18\u6548\u80fd\u3002 \u6848\uff0c\u5176\u8a5e\u6027\u5927\u90e8\u4efd\u90fd\u662f\u4f9d\u7167\u4e2d\u6587\u65b7\u8a5e\u7cfb\u7d71\u7d66\u51fa\u7684\u8a5e\u6027\u70ba\u4e3b\uff0c\u9664\u96e2\u8b5c\u7684\u932f\u8aa4\u5916\uff0c\u5f88\u5c11\u7576\u4e0b\u9032 \u884c\u4fee</td></tr><tr><td colspan=\"6\">\u8868\u4e8c\u5341\u4e94\u3001\u4e2d\u6587\u65b7\u8a5e\u642d\u914d\u5ba2\u8a9e\u8a5e\u983b\u7684\u76f4\u63a5\u7ffb\u8b6f\u6cd5-\u65b7\u8a5e\u53ca\u8a5e\u6027\u6a19\u8a18\u5167\u5916\u90e8\u6e2c\u8a66\u7d50\u679c \u6b63\u3002\u56e0\u6b64\uff0c\u8a9e\u6599\u7684\u65b7\u8a5e\u3001\u8a5e\u6027\u6a19\u8a18\u8cc7\u8a0a\uff0c\u4ecd\u9700\u8981\u7d93\u5c08\u5bb6\u518d\u4e00\u6b21\u505a\u66f4\u56b4\u8b39\u7684\u6821\u6b63\u3002</td></tr><tr><td colspan=\"6\">Precision \u672c\u8ad6\u6587\u63d0\u51fa\u6df7\u5408\u578b\u7684 N-Gram \u5e8f\u5217\u5206\u6578\u7b97\u6cd5\uff0c\u642d\u914d\u4e2d\u6587\u65b7\u8a5e\u6a21\u7d44\u53ca\u52d5\u614b\u898f\u5283\u6f14\u7b97\u6cd5\u7684\u5ba2\u8a9e\u65b7 Recall F-Measure \u5b57\u4e32\u76f8\u4f3c\u5ea6</td></tr><tr><td colspan=\"6\">\u5be6\u9a57 A \u8a5e\u65b9\u6cd5\u3002\u5728\u56b4\u91cd\u8cc7\u6599\u7a00\u758f\u7684\u5ba2\u8a9e\u8a9e\u6599\u4e0b\uff0c\u5c0d\u4e2d\u6587\u8f49\u5ba2\u6587\u7684\u5916\u90e8\u6e2c\u8a66\u7cbe\u78ba\u7387\u6709 80.78%\uff0c\u5167\u90e8\u6e2c\u8a66 \u8a13\u7df4 88.16% 87.48% 87.82% 91.08%</td></tr><tr><td colspan=\"6\">\u6e2c\u8a66 \u6709 94.46%\u3002\u76f8\u8f03\u65bc\u50b3\u7d71\u4e2d\u6587\u8a5e\u76f4\u7ffb\u5ba2\u8a9e\u8a5e\u7684\u65b9\u6cd5\uff0c\u5df2\u7372\u5f97\u63d0\u5347\u3002\u76f8\u4fe1\u672a\u4f86\u6301\u7e8c\u589e\u52a0\u5ba2\u8a9e\u8a9e\u6599\u7684 80.32% 79.08% 79.69% 83.68%</td></tr><tr><td colspan=\"5\">\u5be6\u9a57 B \u898f\u6a21\u5f8c\uff0c\u4f7f\u7528\u672c\u8ad6\u6587\u6240\u63d0\u51fa\u7684\u65b9\u6cd5\uff0c\u6548\u80fd\u6703\u6709\u66f4\u986f\u8457\u7684\u63d0\u5347\u3002 \u8a13\u7df4 87.46% 86.78% 87.12%</td><td>-</td></tr><tr><td colspan=\"6\">\u6e2c\u8a66 \u5ba2\u8a9e\u65b7\u8a5e\u7684\u61c9\u7528\u5c64\u9762\u6975\u5ee3\uff0c\u4e0d\u50c5\u6b62\u4f7f\u7528\u65bc\u6211\u5011\u4e2d\u6587\u8f49\u5ba2\u6587\u6587\u8f49\u97f3\u7cfb\u7d71\u4e2d\u7684\u6587\u53e5\u5206\u6790\u6a21\u7d44\uff0c 79.90% 78.66% 79.27% -</td></tr><tr><td colspan=\"6\">\u4ee5\u5be6\u9a57\u7d50\u679c\u4f86\u770b\uff0c\u986f\u793a\u5ba2\u8a9e\u8a5e\u983b\u7684\u61c9\u7528\u80fd\u986f\u8457\u7684\u63d0\u5347\u9078\u8a5e\u7684\u6b63\u78ba\u7387\u3002\u4f46\u5916\u90e8\u6e2c\u8a66\u7684\u6b63\u78ba\u7387 \u9084\u53ef\u7368\u7acb\u7528\u65bc\u5ba2\u8a9e\u7684\u6578\u4f4d\u5b78\u7fd2\u3001\u5ba2\u8a9e\u6587\u53e5\u8655\u7406\u3001\u5ba2\u8a9e\u8a9e\u97f3\u8fa8\u8b58\u2026\u7b49\u9818\u57df\u3002\u672c\u8ad6\u6587\u63d0\u51fa\u7684\u7814\u7a76\u65b9</td></tr><tr><td colspan=\"6\">\u4ecd\u504f\u4f4e\uff0c\u5df2\u63a5\u8fd1\u672a\u4f7f\u7528\u8a5e\u983b\u7279\u5fb5\u7684\u7d50\u679c\u3002\u6b64\u60c5\u6cc1\u7684\u539f\u56e0\uff0c\u662f\u56e0\u70ba\u7528\u4f86\u7d71\u8a08\u5ba2\u8a9e\u8a5e\u983b\u7684\u8a9e\u6599\u4ecd\u4e0d \u6cd5\uff0c\u61c9\u80fd\u63d0\u4f9b\u672a\u4f86\u5ba2\u8a9e\u65b7\u8a5e\u76f8\u95dc\u7814\u7a76\u505a\u70ba\u57fa\u790e\u8207\u53c3\u8003\u3002</td></tr><tr><td colspan=\"6\">\u8db3\uff0c\u56e0\u6b64\u9020\u6210\u7d71\u8a08\u8cc7\u6599\u7a00\u758f\u7684\u554f\u984c\u3002\u4f46\u5f9e\u5be6\u9a57\u7d50\u679c\u7684\u7279\u6027\u89c0\u5bdf\u5230\uff0c\u82e5\u80fd\u6301\u7e8c\u7684\u589e\u52a0\u5ba2\u8a9e\u8a9e\u6599\u3001 \u6211\u5011\u63a5\u4e0b\u4f86\u8981\u9032\u884c\u7684\u5de5\u4f5c\u6709\uff1a</td></tr><tr><td colspan=\"5\">\u5efa\u7f6e\u51fa\u66f4\u591a\u7684\u5ba2\u8a9e\u8a5e\u983b\uff0c\u80fd\u63d0\u5347\u4e2d\u6587\u8f49\u5ba2\u6587\u7cfb\u7d71\u7684\u6548\u80fd\u3002 1. \u6301\u7e8c\u64f4\u5145\u570b\u5ba2\u8a9e\u5c0d\u7167\u8fad\u5178\u3002</td></tr><tr><td colspan=\"6\">2. \u52a0\u5165\u5ba2\u8a9e\u69cb\u8a5e\u898f\u5247\u3002 \u800c\u5c31\u76ee\u524d\u7684\u7a98\u6cc1\u800c\u8a00\uff0c\u63d0\u5347\u6b63\u78ba\u7387\u7684\u65b9\u6cd5\uff0c\u50c5\u80fd\u9760\u898f\u5247\u6cd5(Rules-base)\uff0c\u5982\u627e\u51fa\u5ba2\u8a9e\u7684\u69cb\u8a5e ) \u898f\u5247\u548c\u6587\u6cd5\u898f\u5247\uff0c\u4f86\u63d0\u5347\u5ba2\u8a9e\u65b7\u8a5e\u7684\u6b63\u78ba\u7387\u3002\u5ba2\u8a9e\u69cb\u8a5e\u90e8\u4efd\uff0c\u662f\u4e0b\u4e00\u968e\u6bb5\u5373\u5c07\u9032\u884c\u7684\u5de5\u4f5c\u3002 3. \u6700\u4f73\u5316\u8a9e\u8a00\u6a21\u578b\u5e73\u6ed1\u5316\u554f\u984c\uff0c\u5982\uff1aGood-Turing Katz\u3001Kneser-Ney\u3002</td></tr><tr><td colspan=\"6\">5.3.3 \u4e2d\u6587\u65b7\u8a5e\u642d\u914d\u5ba2\u8a9e Uni-gram \u53ca Bi-gram \u8a9e\u8a00\u6a21\u578b\u7684\u6df7\u5408\u5f0f\u5206\u6578\u7b97\u6cd5 4. \u6301\u7e8c\u6a19\u8a18\u3001\u5efa\u7f6e\u5ba2\u8a9e\u8a9e\u6599\u3002</td></tr><tr><td colspan=\"5\">5.2 \u6e2c\u8a66\u8a9e\u6599 \u5be6\u9a57 A\uff1a\u8f38\u5165\u4e2d\u6587\u6587\u53e5\uff0c\u8a55\u4f30\u5176\u4e2d\u6587\u8f49\u5ba2\u8a9e\u8a5e\u7684\u65b7\u8a5e\u6548\u80fd\u3002 \u81f4\u8b1d</td></tr><tr><td colspan=\"6\">\u5be6\u9a57 B\uff1a\u8f38\u5165\u4e2d\u6587\u6587\u53e5\uff0c\u8a55\u4f30\u5176\u65b7\u8a5e\u53ca\u8a5e\u6027\u6a19\u8a18\u6548\u80fd\u3002 \u5ba2\u5bb6\u59d4\u54e1\u6703\u63d0\u4f9b\u7d66\u672c\u8ad6\u6587\u5be6\u9a57\u7528\u4e4b\u5ba2\u8a9e\u8a8d\u8b49\u8a9e\u6599\u4ee5\u53ca\u734e\u52a9\u90e8\u4efd\u7d93\u8cbb\uff0c\u7279\u6b64\u81f4\u8b1d\u3002</td></tr><tr><td colspan=\"6\">\u8868\u4e8c\u5341\u516d\u3001\u4e2d\u6587\u65b7\u8a5e\u642d\u914d\u5ba2\u8a9e Uni-gram \u53ca Bi-gram \u8a9e\u8a00\u6a21\u578b\u7684\u6df7\u5408\u5f0f\u5206\u6578\u7b97\u6cd5</td></tr><tr><td/><td/><td>Precision</td><td>Recall</td><td>F-Measure</td><td>\u5b57\u4e32\u76f8\u4f3c\u5ea6</td></tr><tr><td colspan=\"6\">\u800c\u9019\u4e9b\u6a19\u8a18\u5b8c\u6210\u7684\u8cc7\u6599\uff0c\u518d\u7d93\u904e\u4e00\u6b21\u7d93\u4eba\u5de5\u7be9\u9078\u5f8c\uff0c\u78ba\u8a8d\u6709\u6548\u53e5\u6578\u70ba\uff1a\u8a13\u7df4\u8a9e\u6599 4018 \u5be6\u9a57 A \u8a13\u7df4 94.46% 93.73% 94.10% 96.17%</td></tr></table>", | |
| "num": null, | |
| "html": null, | |
| "type_str": "table", | |
| "text": "\u52a0\u5165 Bi-gram \u5f8c\uff0c\u5167\u90e8\u6e2c\u8a66\u7684\u6548\u80fd\u6709\u986f\u8457\u7684\u63d0\u5347\uff0c\u4f46\u5916\u90e8\u6e2c\u8a66\u50c5\u7565\u5347 0.46%\u3002\u539f\u56e0\u662f\u56e0\u70ba\u5ba2 \u8a9e\u8a9e\u8a00\u6a21\u578b\u8cc7\u6599\u7a00\u758f\u7684\u95dc\u4fc2\uff0c\u8a31\u591a Bi-gram pattern \u672a\u51fa\u73fe\u5728\u5ba2\u8a9e Bi-gram \u8a9e\u8a00\u6a21\u578b\u4e2d\uff0c\u6216\u5c31\u7b97\u51fa" | |
| } | |
| } | |
| } | |
| } |