File size: 49,266 Bytes
6fa4bc9 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 |
{
"paper_id": "O07-1010",
"header": {
"generated_with": "S2ORC 1.0.0",
"date_generated": "2023-01-19T08:07:56.577175Z"
},
"title": "\u8a5e\u7fa9\u8fa8\u8b58:\u6a5f\u5668\u5b78\u7fd2\u6f14\u7b97\u6cd5\u7279\u5fb5\u7684\u9078\u53d6\u8207\u7d44\u5408",
"authors": [],
"year": "",
"venue": null,
"identifiers": {},
"abstract": "",
"pdf_parse": {
"paper_id": "O07-1010",
"_pdf_hash": "",
"abstract": [],
"body_text": [
{
"text": "\u95dc\u9375\u8a5e\uff1a\u8a5e\u7fa9\u8fa8\u8b58(word sense disambiguation)\uff0c\u642d\u914d\u8a9e(collocation)\uff0c\u8a9e\u6cd5\u4f9d\u5b58\u95dc\u4fc2 (dependency relations)\uff0csketch engine\uff0cStanford parser\uff0cHownet\uff0cNa\u00efve Bayes\uff0cForward Sequential Selection Algorithm \u4e00\u3001\u524d\u8a00 \u4e00\u500b\u82f1\u6587\u8a5e\u53ef\u80fd\u6709\u597d\u5e7e\u500b\u4e0d\u540c\u7684\u610f\u601d\uff0c\u4f8b\u5982 bank \u6709\u9280\u884c\uff0c\u6cb3\u5824\uff0c\u5eab\u7b49\u591a\u500b\u610f\u7fa9\u3002\u8a5e\u7fa9 \u8fa8\u8b58\u7684\u76ee\u7684\u5c31\u662f\u8981\u8b93\u96fb\u8166\u81ea\u52d5\u8fa8\u8b58\u4e00\u500b\u5c90\u7fa9\u8a5e\u5728\u67d0\u4e00\u500b\u8a9e\u5883\u88e1\u6b63\u78ba\u7684\u610f\u7fa9\u3002\u7531\u65bc\u73fe\u6709\u8a5e \u6027\u6a19\u8a18\u7684\u6f14\u7b97\u6cd5\u6b63\u78ba\u7387\u90fd\u76f8\u7576\u7684\u9ad8\uff0c\u5982\u679c\u5c90\u7fa9\u8a5e\u7684\u610f\u7fa9\u5177\u6709\u4e0d\u540c\u7684\u8a5e\u6027\u5f88\u5bb9\u6613\u900f\u904e\u8a5e\u6027 \u6a19\u8a18\u7a0b\u5f0f\u8fa8\u8b58\u51fa\u4e0d\u540c\u7684\u610f\u7fa9\u3002\u800c\u50cf\u524d\u9762\u7684\u4f8b\u5b50 bank \u4e0d\u540c\u7684\u610f\u7fa9\u5982\u9280\u884c\uff0c\u6cb3\u5824\uff0c\u5eab\u90fd\u662f \u540d\u8a5e\uff0c\u8fa8\u8b58\u7684\u56f0\u96e3\u5ea6\u589e\u9ad8\u8a31\u591a\u3002\u6211\u5011\u6240\u4f7f\u7528\u7684\u8a13\u7df4\u8a9e\u6599 Senseval-2 English lexical sample\uff0c\u662f\u5728 2001 \u5e74\u6240\u767c\u5e03\uff0c\u8a9e\u6599\u4e2d\u5305\u542b\u4e86 73 \u500b\u4e0d\u540c\u7684\u76ee\u6a19\u8a5e\uff0c\u8a5e\u6027\u6709\u540d\u8a5e\u3001\u52d5\u8a5e\u3001 \u5f62\u5bb9\u8a5e\uff0c\u4f46\u540c\u4e00\u500b\u76ee\u6a19\u8a5e\u7684\u4e0d\u540c\u610f\u7fa9\u8a5e\u6027\u90fd\u662f\u76f8\u540c\u7684\uff0c\u5c0d\u65bc\u8a5e\u7fa9\u8fa8\u8b58\u7684\u6f14\u7b97\u6cd5\u5f62\u6210\u5f88\u5927 \u7684\u6311\u6230\u3002 Senseval-2 \u7684\u8a13\u7df4\u4ee5\u53ca\u6e2c\u8a66\u8a9e\u6599\u662f\u4ee5 XML \u7684\u578b\u5f0f\u5132\u5b58\uff0c\u4ee5\u4e0b\u662f\u4e00\u7b46\u8a13\u7df4\u8a9e\u6599\u7684\u7bc4\u4f8b\uff1a <instance id=\"art.40001\" docsrc=\"bnc_ACN_245\"> <answer instance=\"art.40001\" senseid=\"art%1:06:00::\"/> <context> Their multiscreen projections of slides and film loops have featured in orbital parties, at the Astoria and Heaven, in Rifat Ozbek's 1988/89 fashion shows, and at Energy's recent Docklands all-dayer. From their residency at the Fridge during the first summer of love, Halo used slide and film projectors to throw up a collage of op-art patterns, film loops of dancers like E-Boy and Wumni, and unique fractals derived from video feedback. &bquo;We're not aware of creating a visual identify for the house scene, because we're right in there. We see a dancer at a rave, film him later that week, and project him at the next rave.&equo; Ben Lewis Halo can be contacted on 071 738 3248. <head>Art</head>you can dance to from the creative group called Halo </context> </instance> \u8a9e\u6599\u4e2d\u76ee\u6a19\u5b57\u6703\u7528<head>\u4ee5\u53ca</head>\u6a19\u51fa\uff0c\u6e2c\u8a66\u8a9e\u6599\u683c\u5f0f\u8207\u8a13\u7df4\u8207\u6599\u76f8\u4f3c\uff0c\u5176\u5dee\u5225\u5728 \u65bc\u6c92\u6709 senseid \u7684\u6a19\u8a18\u3002 \u4e8c\u3001\u6587\u737b\u56de\u9867 \u65e9\u671f\u8a5e\u7fa9\u8fa8\u8b58\u7684\u6f14\u7b97\u6cd5\u5927\u90fd\u5229\u7528\u5229\u7528\u8fad\u5178\u7684\u5b9a\u7fa9\u3001\u6216\u540c\u7fa9\u8a5e\u8fad\u5178(thesaurus)\u7684\u8a9e\u7fa9 \u5206\u985e\u8a0a\u606f\u3002\u4f8b\u5982 Lesk (1986) \u5224\u65b7\u76ee\u6a19\u8a5e\u7684\u8a9e\u5883\u8207\u8fad\u5178\u7684\u54ea\u4e00\u500b\u610f\u7fa9\u7684\u5b9a\u7fa9\u6700\u63a5\u8fd1\uff0c\u6240 \u63a1\u7528\u7684\u76f8\u4f3c\u5ea6\u8a08\u7b97\u65b9\u5f0f\u4ee5\u5169\u8005\u76f8\u540c\u7684\u975e\u529f\u80fd\u8a5e\u7684\u6578\u76ee\u70ba\u4e3b\u3002Walker (1987)\u5247\u5229\u7528\u540c\u7fa9\u8a5e \u8fad\u5178(thesaurus)\u7576\u4e2d\u7684\u8a9e\u7fa9\u985e\u5225\u3002\u9019\u4e9b\u6f14\u7b97\u6cd5\u8ddf\u76ee\u524d\u5e38\u7528\u7684\u6a5f\u5668\u5b78\u7fd2\u6f14\u7b97\u6cd5\u76f8\u6bd4\u6b63\u78ba\u7387 \u4f4e\u8a31\u591a(\u8acb\u53c3\u8003\u8868 16 Senseval-2 \u8a5e\u7fa9\u8fa8\u8b58\u7af6\u8cfd\u5404\u500b\u65b9\u6cd5\u7684\u6b63\u78ba\u7387) \u6a5f \u5668 \u5b78 \u7fd2 \u65b9 \u6cd5 \u4e3b \u8981 \u53ef\u5206 \u70ba \u76e3 \u7763 \u5f0f (supervised learning) \u53ca \u975e \u76e3 \u7763 \u5f0f (unsupervised learning)\u3002\u5169\u8005\u7684\u5dee\u5225\u5728\u65bc\u524d\u8005\u7684\u8a13\u7df4\u8a9e\u6599\u6709\u6a19\u8a18\u7b54\u6848\u7684\u800c\u5f8c\u8005\u6c92\u6709\uff0c\u6211\u5011\u6240\u63a1\u7528\u7684\u65b9 \u6cd5\u662f\u76e3\u7763\u5f0f\u7684\u65b9\u6cd5\u3002\u7121\u8ad6\u662f\u54ea\u4e00\u7a2e\u6a5f\u5668\u5b78\u7fd2\u7684\u8a5e\u7fa9\u8fa8\u8b58\u6f14\u7b97\u6cd5\u90fd\u9700\u8981\u5229\u7528\u8a9e\u5883\u7684\u8a0a\u606f\u3002 \u4f8b\u5982 Purandare and Pedersen (2004) ",
"cite_spans": [
{
"start": 1537,
"end": 1548,
"text": "Lesk (1986)",
"ref_id": "BIBREF8"
},
{
"start": 1864,
"end": 1893,
"text": "Purandare and Pedersen (2004)",
"ref_id": "BIBREF12"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
}
],
"back_matter": [],
"bib_entries": {
"BIBREF1": {
"ref_id": "b1",
"title": "\u69cb\u5c0d\u61c9\u6f14\u7b97\u6cd5\u7684\u7814\u7a76(I)(II)",
"authors": [],
"year": null,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "\u69cb\u5c0d\u61c9\u6f14\u7b97\u6cd5\u7684\u7814\u7a76(I)(II)\u300dNSC94-2411-H-002-043 NSC95-2411-H-002-045-MY2 \u53c3\u8003\u6587\u737b",
"links": null
},
"BIBREF2": {
"ref_id": "b2",
"title": "Word sense disambiguation using statistical methods",
"authors": [
{
"first": "Peter",
"middle": [],
"last": "Brown",
"suffix": ""
}
],
"year": 1991,
"venue": "ACL 29",
"volume": "",
"issue": "",
"pages": "264--270",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Brown, Peter et al. (1991) Word sense disambiguation using statistical methods. In ACL 29, pp. 264-270.",
"links": null
},
"BIBREF3": {
"ref_id": "b3",
"title": "Hownet and the Computation of Meaning",
"authors": [
{
"first": "Zhendong",
"middle": [],
"last": "Dong",
"suffix": ""
},
{
"first": "Qiang",
"middle": [],
"last": "Dong",
"suffix": ""
}
],
"year": 2006,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Dong, Zhendong and Dong, Qiang. (2006) Hownet and the Computation of Meaning. World Scientific.",
"links": null
},
"BIBREF4": {
"ref_id": "b4",
"title": "A method of disambiguating word senses in a large corpus",
"authors": [
{
"first": "William",
"middle": [],
"last": "Gale",
"suffix": ""
},
{
"first": "Kenneth",
"middle": [],
"last": "Church",
"suffix": ""
},
{
"first": "David",
"middle": [],
"last": "Yarowsky",
"suffix": ""
}
],
"year": 1992,
"venue": "Computers and the Humanties",
"volume": "26",
"issue": "",
"pages": "415--439",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Gale, William, Church, Kenneth, and Yarowsky, David. (1992) A method of disambiguating word senses in a large corpus. Computers and the Humanties 26:415-439.",
"links": null
},
"BIBREF5": {
"ref_id": "b5",
"title": "Speech and Language Processing: An Introduction to Natural Language Processing",
"authors": [
{
"first": "Daniel",
"middle": [],
"last": "Jurafsky",
"suffix": ""
},
{
"first": "James",
"middle": [
"H"
],
"last": "Martin",
"suffix": ""
}
],
"year": 2000,
"venue": "Speech Recognition, and Computational Linguistics",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Jurafsky, Daniel, and James H. Martin. (2000) Speech and Language Processing: An Introduction to Natural Language Processing, Speech Recognition, and Computational Linguistics. Prentice-Hall.",
"links": null
},
"BIBREF6": {
"ref_id": "b6",
"title": "Accurate Unlexicalized Parsing",
"authors": [
{
"first": "Dan",
"middle": [],
"last": "Klein",
"suffix": ""
},
{
"first": "Christopher",
"middle": [],
"last": "Manning",
"suffix": ""
}
],
"year": 2003,
"venue": "Proceedings of the 41st Meeting of the Association for Computational Linguistics",
"volume": "",
"issue": "",
"pages": "423--430",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Klein, Dan. and Manning, Christopher. (2003) Accurate Unlexicalized Parsing. Proceedings of the 41st Meeting of the Association for Computational Linguistics, pp. 423-430.",
"links": null
},
"BIBREF7": {
"ref_id": "b7",
"title": "High WSD Accuracy Using Na\u00efve Bayesian Classifier with Rich Features",
"authors": [
{
"first": "Cuong",
"middle": [],
"last": "Le",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Anh",
"suffix": ""
},
{
"first": "Akira",
"middle": [],
"last": "Shimazu",
"suffix": ""
}
],
"year": 2004,
"venue": "PACLIC",
"volume": "18",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Le, Cuong Anh and Shimazu, Akira. (2004) High WSD Accuracy Using Na\u00efve Bayesian Classifier with Rich Features. PACLIC 18, Tokyo. http://dspace.wul.waseda.ac.jp/dspace/bitstream/2065/564/1/oral-8.pdf",
"links": null
},
"BIBREF8": {
"ref_id": "b8",
"title": "Automatic Sense Disambiguation: How to tell a pine cone from an ice cream cone",
"authors": [
{
"first": "Michael",
"middle": [],
"last": "Lesk",
"suffix": ""
}
],
"year": 1986,
"venue": "Proceedings of the 1986 SIGDOC Conference",
"volume": "",
"issue": "",
"pages": "24--26",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Lesk, Michael. (1986) Automatic Sense Disambiguation: How to tell a pine cone from an ice cream cone. In Proceedings of the 1986 SIGDOC Conference, pp. 24-26, New York. Association for Computing Machinery.",
"links": null
},
"BIBREF9": {
"ref_id": "b9",
"title": "Using Syntactic Dependency as Local Context to Resolve Word Sense Ambiguity",
"authors": [
{
"first": "Dekang",
"middle": [],
"last": "Lin",
"suffix": ""
}
],
"year": 1997,
"venue": "Proceedings of ACL-97",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Lin, Dekang . (1997). Using Syntactic Dependency as Local Context to Resolve Word Sense Ambiguity In Proceedings of ACL-97, Madrid, Spain. July, 1997.",
"links": null
},
"BIBREF10": {
"ref_id": "b10",
"title": "Foundations of Statistical Natural Language Processing",
"authors": [
{
"first": "Christopher",
"middle": [],
"last": "Manning",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Schutze",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Hinrich",
"suffix": ""
}
],
"year": 1999,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Manning, Christopher, and Schutze, Hinrich. (1999) Foundations of Statistical Natural Language Processing. MIT Press.",
"links": null
},
"BIBREF11": {
"ref_id": "b11",
"title": "SenseRelate::TargetWord -A Generalized Framework for Word Sense Disambiguation",
"authors": [
{
"first": "Banerjee",
"middle": [],
"last": "Patwardhan",
"suffix": ""
},
{
"first": "Pedersen",
"middle": [],
"last": "",
"suffix": ""
}
],
"year": 2005,
"venue": "Appears in the Proceedings of the Twentieth National Conference on Artificial Intelligence",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Patwardhan, Banerjee, and Pedersen (2005) SenseRelate::TargetWord -A Generalized Framework for Word Sense Disambiguation. Appears in the Proceedings of the Twentieth National Conference on Artificial Intelligence, July 12, 2005, Pittsburgh, PA. (Intelligent Systems Demonstration)",
"links": null
},
"BIBREF12": {
"ref_id": "b12",
"title": "Improving Word Sense Discrimination with Gloss Augmented Feature Vectors",
"authors": [
{
"first": "Pedersen",
"middle": [],
"last": "Purandare",
"suffix": ""
}
],
"year": 2004,
"venue": "Appears in the Proceedings of the Workshop on Lexical Resources for the Web and Word Sense Disambiguation",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Purandare and Pedersen (2004) Improving Word Sense Discrimination with Gloss Augmented Feature Vectors. Appears in the Proceedings of the Workshop on Lexical Resources for the Web and Word Sense Disambiguation, November 22, 2004, Puebla Mexico.",
"links": null
},
"BIBREF13": {
"ref_id": "b13",
"title": "Decision Lists for Lexical Ambiguity Resolution: Application to Accent Restoration in Spanish and French",
"authors": [
{
"first": "D",
"middle": [],
"last": "Yarowsky",
"suffix": ""
}
],
"year": 1994,
"venue": "Proceedings of the 32nd Annual Meeting of the Association for Computational Linguistics",
"volume": "",
"issue": "",
"pages": "88--95",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Yarowsky, D. (1994) Decision Lists for Lexical Ambiguity Resolution: Application to Accent Restoration in Spanish and French.'' In Proceedings of the 32nd Annual Meeting of the Association for Computational Linguistics. Las Cruces, NM, pp. 88-95.",
"links": null
}
},
"ref_entries": {
"TABREF0": {
"content": "<table><tr><td>\u8a5e\u6703\u53cd\u6620\u51fa\u76ee\u6a19\u8a5e\u7684\u610f\u7fa9\uff0c\u56e0\u6b64\u5c07\u5468\u570d\u7684\u8a5e\u4ee5\u53ca\u76ee\u6a19\u8a5e\u505a\u7d71\u8a08\u518d\u5229\u7528\u6a5f\u7387\u9078\u64c7\u8a5e\u7fa9\uff0c\u5728 \u6cbb:agent={~}}} \u8868\u4e09\u3001F3 Window Size \u5be6\u9a57\u7d50\u679c \u8868\u516d\u3001F6 Window Size \u5be6\u9a57\u7d50\u679c object_of 50.1 pp* 50.9 object_of 51.9 pp* 51.9 5 th 60.3 59.4 60.8 60.4 61.1</td></tr><tr><td>\u63a1\u7528\u975e\u76e3\u7763\u5f0f\u7684\u65b9\u6cd5\uff0c\u5f9e\u6c92\u6709\u6a19\u793a\u8a5e\u7fa9\u7d14\u6587\u5b57\u8a9e\u6599 \u62bd\u51fa\u8a9e\u5883\u4e26\u5c07\u6a5f\u8b80\u8fad\u5178 Wordnet \u88e1\u9762\u4e0d\u540c\u8a5e\u7fa9\u7684\u5b9a\u7fa9\u53bb\u9664\u529f\u80fd\u8a5e\u5f8c\u5efa\u7acb\u5171\u73fe\u77e9\u9663 (co-occurrence matrix)\uff0c\u5229\u7528 Singular Value Decomposition (SVD)\u5c07\u7dad\u6578\u964d\u5230 100\uff0c\u6700\u5f8c \u7528 Latent Semantic Indexing (LSI)\u627e\u51fa\u67d0\u4e00\u53e5\u4e2d\u7684\u76ee\u6a19\u8a5e\u6700\u6709\u53ef\u80fd\u7684\u8a5e\u7fa9\u3002Jurafsky and Martin (2000)\u5c07\u5e38\u7528\u7684\u8a9e\u5883\u7279\u5fb5\u5206\u6210\u5169\u985e\u3002\u4e00\u985e\u662f\u642d\u914d\u8a9e\u7279\u5fb5(collocational features)\uff0c\u53e6 \u4e00\u985e\u662f bag of words information\u3002\u5169\u8005\u7684\u6700\u5927\u5dee\u5225\u5728\u65bc\u5f8c\u8005\u53ea\u8003\u616e\u67d0\u4e9b\u8a5e\u5728\u76ee\u6a19\u8a5e\u5de6\u53f3 \u4e00\u5b9a\u7bc4\u570d\u7684\u8a5e\u6709\u6c92\u6709\u51fa\u73fe\uff0c\u4e0d\u8003\u616e\u9019\u4e9b\u8a5e\u5f7c\u6b64\u6216\u8ddf\u76ee\u6a19\u8a5e\u524d\u5f8c\u7684\u95dc\u4fc2\uff0c\u800c\u524d\u8005\u5247\u7d0d\u5165\u8207 \u76ee\u6a19\u8a5e\u524d\u5f8c\u76f8\u5c0d\u4f4d\u7f6e\u7684\u8a0a\u606f\uff0c\u751a\u81f3\u7528\u8a9e\u6cd5\u5256\u6790\u5668\u5f97\u5230\u8a9e\u6cd5\u4f9d\u5b58\u95dc\u4fc2\u3002 \u8a5e\u7fa9\u8fa8\u8b58\u65b9\u6cd5\u9664\u4e86\u53ef\u4ee5\u5229\u7528 Semantic Concordancer \u6216 Senseval \u9019\u4e9b\u6709\u6a19\u793a\u8a5e\u7fa9\u7684 \u8a9e\u6599\u4e4b\u5916\uff0c\u9084\u53ef\u4ee5\u5229\u7528 pseudoword \u6216\u96d9\u8a9e\u8a9e\u6599\u3002pseudoword \u662f Gale et al. (1992)\u548c Schutze(1992)\u70ba\u4e86\u7701\u53bb\u6a19\u793a\u8a5e\u7fa9\u6240\u9700\u7684\u5927\u91cf\u4eba\u529b\u8207\u6642\u9593\u6240\u5275\u9020\u51fa\u4f86\u7684\u65b9\u6cd5\u3002\u900f\u904e\u4eba\u9020\u7684 \u5c90\u7fa9\u8a5e\u5982 banana-door\uff0c\u5c07\u8a9e\u6599\u4e2d\u6240\u6709\u51fa\u73fe banana \u6216 door \u90fd\u4ee3\u63db\u6210 banana-door\uff0c\u9019\u6a23 \u5c31\u53ef\u4ee5\u5f97\u5230\u985e\u4f3c\u4eba\u5de5\u6a19\u8a18\u8a5e\u7fa9\u7684\u8a13\u7df4\u8a9e\u6599\u3002\u6b64\u5916\uff0c\u67d0\u4e00\u500b\u6709\u5c90\u7fa9\u7684\u8a5e\u5728\u53e6\u4e00\u500b\u8a9e\u8a00\u901a\u5e38 \u6c92\u6709\u5c90\u7fa9\uff0c\u4f8b\u5982\u82f1\u6587\u7684 duty \u6709\u5169\u500b\u610f\u7fa9\uff0c\u4f46\u5728\u4e2d\u6587\u88e1\u5247\u7531\u6d77\u95dc\u548c\u8cac\u4efb\u5169\u500b\u8a5e\u4f86\u8868\u9054\u3002 Brown et al. (1991) \u53ca Gale et al. (1992)\u5229\u7528\u9019\u500b\u7279\u6027\uff0c\u4ee5\u82f1\u6cd5\u96d9\u8a9e\u8a9e\u6599\u5eab\u4f5c\u70ba\u8a13\u7df4\u8a9e \u6599\uff0c\u63a1\u53d6\u76ee\u6a19\u8a5e\u5de6\u53f3\u82e5\u5e72\u8a5e(\u4f8b\u5982 50 \u500b\u8a5e)\u69cb\u6210\u4e00\u500b\u8a9e\u5883\u5411\u91cf(context vector),\u518d\u5229\u7528 Bayesian classification \u4f86\u9078\u64c7\u5728\u67d0\u4e00\u500b\u8a9e\u5883\u7576\u4e2d\u54ea\u4e00\u500b\u8a5e\u7fa9\u7684\u6a5f\u7387\u6700\u5927\u3002\u6211\u5011\u4e5f\u63a1\u7528 Bayesian classification \u4f46\u642d\u914d\u4e0d\u540c\u7684\u7279\u5fb5\u3002Bayesian classification \u7684\u6982\u5ff5\u662f\u76ee\u6a19\u8a5e\u5468\u570d\u7684 \u7b2c\u4e09\u7bc0\u4e2d\u6703\u6709\u8a73\u7d30\u7684\u4ecb\u7d39\u3002 Yarowsky (1995)\u6ce8\u610f\u5230\u5728\u67d0\u4e00\u7bc7\u6587\u7ae0\u4e2d\u4e00\u500b\u76ee\u6a19\u8a5e\u7684\u8a5e\u7fa9\u901a\u5e38\u662f\u56fa\u5b9a\u67d0\u4e00\u500b\u8a5e\u7fa9 (One sense per discourse)\u3002\u4e14\u76ee\u6a19\u8a5e\u7684\u642d\u914d\u8a9e\u63d0\u793a\u4e86\u9019\u500b\u76ee\u6a19\u8a5e\u7684\u8a5e\u7fa9(One sense per collocation) \u3002\u672c\u6587\u6240\u63a1\u7528\u642d\u914d\u8a9e\u4f5c\u70ba\u6a5f\u5668\u5b78\u7fd2\u6f14\u7b97\u7684\u6cd5\u7279\u5fb5\u53d7\u5230 Yarowsky (1995)\u7684\u555f \u767c\u3002Lin (1997)\u6709\u9451\u65bc\u4ee5\u6a5f\u5668\u5b78\u7fd2\u5206\u985e\u5668(classifier)\u4f86\u8fa8\u8b58\u8a5e\u7fa9\u9700\u70ba\u4e0d\u540c\u7684\u8a5e\u5206\u5225\u8a13\u7df4\u51fa \u4e0d\u540c\u7684\u5206\u985e\u5668\uff0c\u9817\u4e0d\u65b9\u4fbf\uff0c\u56e0\u6b64\u63d0\u51fa\u4e00\u7a2e\u4f7f\u7528\u540c\u4e00\u7a2e\u77e5\u8b58\u4f86\u6e90(knowledge source)\u7684\u65b9 \u6cd5\u3002\u4ed6\u5229\u7528\u81ea\u5df1\u6240\u767c\u5c55\u7684 MINIPAR \u82f1\u6587\u5256\u6790\u5668\u5f97\u5230\u7684\u8a9e\u6cd5\u4f9d\u5b58\u95dc\u4fc2(dependency relations)\uff0c\u5982\u52d5\u8a5e\u8207\u53d7\u8a5e\u7684\u95dc\u4fc2\u4f5c\u70ba\u6a5f\u5668\u5b78\u7fd2\u6f14\u7b97\u6cd5\u7684\u7279\u5fb5\u3002\u6bd4\u8f03\u7279\u5225\u7684\u5730\u65b9\u5728\u65bc\u4ed6\u7684 \u65b9\u6cd5\u4e0d\u9700\u8981\u6a19\u793a\u8a5e\u7fa9\u7684\u8a9e\u6599\uff0c\u800c\u662f\u5229\u7528\u76f8\u540c\u8a9e\u610f\u7684\u8a5e\u6703\u51fa\u73fe\u5728\u5177\u6709\u76f8\u540c\u7684\u4f9d\u5b58\u95dc\u4fc2\u6240\u7d44 \u6210\u7684\u5c40\u90e8\u8a9e\u5883(local context)\u3002Lin (1997) \u7684\u6b63\u78ba\u7387\u9054\u5230\u8207\u5176\u5b83\u6a5f\u5668\u5b78\u7fd2\u6f14\u7b97\u6cd5\u76f8\u540c\u6c34 \u6e96\u3002\u672c\u6587\u63a1\u7528\u8a9e\u6cd5\u4f9d\u5b58\u95dc\u4fc2\u4f5c\u70ba\u8a5e\u7fa9\u8fa8\u8b58\u7684\u7279\u5fb5\u6e90\u81ea\u65bc Lin (1997)\u7684\u60f3\u6cd5\u3002\u6709\u95dc\u65bc\u7279\u5fb5 \u7684\u9078\u53d6\uff0cLe and Shimazu (2004)\u91dd\u5c0d\u82f1\u6587\u8a5e\u7fa9\u8fa8\u8b58\u63d0\u51fa\u6578\u500b\u7279\u5fb5\u4e26\u4ee5 Forward Sequential Selection Algorithm \u4f86\u5f97\u5230\u6700\u4f73\u7684\u7279\u5fb5\u7d44\u5408\uff0c\u672c\u6587\u63a1\u7528 Le and Shimazu (2004)\u6240\u63d0\u51fa\u7684 5 \u500b\u7279\u5fb5\u53e6\u5916\u52a0\u4e0a 4 \u500b\u7279\u5fb5\uff0c\u4e26\u4eff\u7167 Le and Shimazu (2004)\u6240\u4f7f\u7528\u7684 Forward Sequential Selection Algorithm \u5f97\u5230\u6700\u4f73\u7279\u5fb5\u7684\u7d44\u5408\u3002 \u9664\u4e86\u4e0a\u9762\u4ecb\u7d39\u7684\u65b9\u6cd5\uff0c\u9084\u6709\u8a31\u591a\u8a5e\u7fa9\u8fa8\u8b58\u7684\u65b9\u6cd5\uff0c\u4f8b\u5982\u5229\u7528 mutual information \u7684 Flip-Flop algorithm (Brown et al. (1991)),\u4f7f\u7528 decision list (Yarowsky (1994))\u7b49\uff0c\u9650\u65bc\u7bc7\u5e45 \u7121\u6cd5\u4e00\u4e00\u4ecb\u7d39\u3002\u8fd1\u5e7e\u5e74\u8a5e\u7fa9\u8fa8\u8b58\u7684\u6f14\u7b97\u6cd5\u9664\u4e86 Na\u00efve Bayes \u4e4b\u5916\uff0c\u8d8a\u4f86\u8d8a\u591a\u4eba\u4f7f\u7528 Maximum Entropy\uff0cSupport Vector Machine\uff0c\u53ca Conditional Random Field \u7b49\u8f03\u65b0\u7684\u6a5f\u5668 \u5b78\u7fd2\u6f14\u7b97\u6cd5\u3002\u672c\u6587\u6240\u9078\u53d6\u7279\u5fb5\u548c\u7d44\u5408\u7684\u65b9\u6cd5\u4e5f\u53ef\u4ee5\u8207\u9019\u4e9b\u65b9\u6cd5\u4e00\u8d77\u4f7f\u7528\u3002 \u4e09\u3001\u6211\u5011\u63a1\u53d6\u7684\u65b9\u6cd5 (\u4e00) \u3001Bayesian Classification \u5728\u6211\u5011\u7684\u5be6\u9a57\u4e2d\uff0c\u6211\u5011\u63a1\u7528 Bayesian Classification \u642d\u914d\u591a\u7a2e\u7279\u5fb5\u7684\u65b9\u6cd5\uff0c\u4e0b\u9762\u7c21\u8ff0 Bayesian Classification\u3002 \u5047\u8a2d\u6211\u5011\u73fe\u5728\u8981\u5c0d\u4e00\u500b\u76ee\u6a19\u8a5e\u505a\u8a5e\u7fa9\u8fa8\u8a8d\uff0c\u8a72\u76ee\u6a19\u8a5e\u7684\u8a5e\u7fa9\u6709 k \u500b\uff0c\u4f9d\u5e8f\u662f \uff0c \u5247\u76ee\u6a19\u5c31\u662f\u8981\u627e\u51fa\u4e00\u500b \uff0c\u4f7f\u5f97 \u70ba\u6700\u5927\uff0cc \u662f\u76ee\u6a19\u8a5e\u6240\u542b\u6709\u7684\u67d0\u7a2e\u7279\u5fb5\u3002\u6839\u64da\u8c9d \u5f0f\u5b9a\u7406\uff0c\u53ef\u4ee5\u5f97\u5230\u5982\u4e0b\u7684\u7b49\u5f0f\uff1a \u56e0\u6b64 \u6211\u5011\u6240\u6709\u7684\u5be6\u9a57\u90fd\u662f\u4f7f\u7528\u9019\u500b\u65b9\u6cd5\u4f86\u4f5c\u8a5e\u7fa9\u8fa8\u8a8d\uff0c\u5dee\u5225\u662f\u5728\u65bc\u9078\u53d6\u7684\u7279\u5fb5\u7684\u4e0d\u540c\u3002 (\u4e8c) \u3001Forward Sequential Selection Algorithm \u5728\u7279\u5fb5\u7684\u9078\u53d6\u65b9\u9762\uff0c\u7531\u65bc\u6211\u5011\u5617\u8a66\u4e86\u5f88\u591a\u7a2e\u7279\u5fb5\uff0c\u5047\u5982\u7279\u5fb5\u6709 7 \u7a2e\u90a3\u9ebc\u7279\u5fb5\u7d44\u5408\u7684\u7a2e\u985e \u5c31\u6709 127 \u7a2e\uff0c\u6578\u91cf\u975e\u5e38\u7684\u53ef\u89c0\uff0c\u4e00\u500b\u4e00\u500b\u5c07\u6240\u6709\u7684\u7d44\u5408\u505a\u5be6\u9a57\u975e\u5e38\u6c92\u6709\u6548\u7387\uff0c\u56e0\u6b64\u4f7f\u7528 Le and Shimazu (2004)\u6240\u63d0\u51fa\u7684 Forward Sequential Selection \u6f14\u7b97\u6cd5\u4f86\u6311\u9078\u7279\u5fb5\u3002\u9019\u500b\u65b9 \u6cd5\u5927\u81f4\u4e0a\u662f\u5148\u4ee4\u4e00\u500b\u7279\u5fb5\u7684\u96c6\u5408 S \u70ba\u7a7a\u96c6\u5408\uff0c\u9996\u5148\u6311\u4e00\u500b\u6700\u597d\u7684\u7279\u5fb5\u653e\u9032 S \u4e2d\uff0c\u63a5\u8457 \u5c07\u6bcf\u4e00\u500b\u7279\u5fb5\u90fd\u653e\u9032 S \u4e2d\u770b\u54ea\u500b\u5f97\u5230\u7684\u6b63\u78ba\u7387\u6700\u9ad8\u4f86\u6c7a\u5b9a\u7b2c\u4e8c\u500b\u8981\u653e\u5165 S \u4e2d\u7684\u7279\u5fb5\uff0c \u5982\u6b64\u53cd\u8986\u76f4\u5230\u6700\u5f8c\u6b63\u78ba\u7387\u4e0d\u518d\u589e\u52a0\u70ba\u6b62\uff0c\u6700\u5f8c\u96c6\u5408 S \u4e2d\u7684\u7279\u5fb5\u5c31\u6703\u662f\u4e00\u500b\u5f88\u4e0d\u932f\u7684\u7279\u5fb5 \u7d44\u5408\uff0c\u96d6\u7136\u672a\u5fc5\u771f\u7684\u662f\u6700\u4f73\u89e3\u4f46\u61c9\u7528\u5728\u82f1\u6587\u8a5e\u7fa9\u8fa8\u8a8d\u7684\u7279\u5fb5\u9078\u53d6\u4e0a\u8207\u771f\u6b63\u6700\u4f73\u89e3\u7684\u5dee\u7570 \u975e\u5e38\u5c0f\u3002 (\u4e09) \u3001\u7279\u5fb5 \u6211\u5011\u4e00\u5171\u5617\u8a66\u4e86 9 \u7a2e\u7279\u5fb5\uff0c\u5206\u5225\u4ee5 F1 \u5230 F9 \u547d\u540d\u4e4b\uff0c\u524d\u4e94\u500b\u4e3b\u8981\u662f\u91dd\u5c0d\u76ee\u6a19\u8a5e\u5468 \u570d\u7684\u8a5e\u4ee5\u53ca\u5176\u8a5e\u6027\uff0c\u9019 5 \u500b\u7279\u5fb5\u90fd\u662f Le and Shimazu (2004)\u6240\u4f7f\u7528\u7684\u7279\u5fb5\uff0c\u7b2c\u516d\u500b\u4ee5\u53ca \u7b2c\u4e03\u500b\u5247\u8457\u91cd\u5728\u8a5e\u7684\u4f9d\u5b58\u95dc\u4fc2\uff0c\u4f8b\u5982\u4e3b\u8a5e\u8207\u52d5\u8a5e\uff0c\u52d5\u8a5e\u8207\u53d7\u8a5e\u7684\u95dc\u4fc2\u7b49\u7b49\uff0c\u800c\u6700\u5f8c\u5169\u500b \u5247\u662f\u5229\u7528 HowNet \u7684\u53d6\u76ee\u6a19\u8a5e\u7684\u524d\u5f8c\u5169\u500b\u8a5e\u4ee5\u53ca\u8207\u76ee\u6a19\u8a5e\u6709\u4f9d\u5b58\u95dc\u4fc2\u7684 HowNet \u7fa9\u5143 (\u8a9e\u7fa9\u7279\u5fb5)\u7576\u4f5c\u7279\u5fb5\uff0c\u5728\u6b64\u4e00\u4e00\u4ecb\u7d39\u3002 F1 \u662f\u76f4\u63a5\u628a\u76ee\u6a19\u8a5e\u5468\u570d\u7684\u8a5e\u505a\u70ba\u7279\u5fb5\uff0c\u4f46\u662f\u6703\u6392\u9664\u3127\u4e9b\u5982 is, a \u4e4b\u985e\u7684\u529f\u80fd\u8a5e(stop words)\u3002 F2 \u4e5f\u662f\u76ee\u6a19\u8a5e\u5468\u570d\u7684\u8a5e\uff0c\u4f46\u6703\u52a0\u4e0a\u4f4d\u7f6e\u7684\u8cc7\u8a0a\uff0c\u4f8b\u5982\u76ee\u6a19\u8a5e\u662f art \u6642\uff0c \u301dThe art of design\u301e \u4e2d\u6703\u88ab\u53d6\u51fa\u7684\u7279\u5fb5\u6703\u662f{(The, -1), (of, 1), (design, 2)}\u3002 F3 \u8ddf (ROOT (S (NP (DT The) (NN government)) (ADVP (RB first)) (VP (VBD established) (NP (NP (JJ modern) (JJ criminal) (NN investigation) (NN system)) (PP (IN in) (NP (CD 1946))))) (. .))) det(government-2, The-1) nsubj(established-4, government-2) advmod(established-4, first-3) amod(system-8, modern-5) amod(system-8, criminal-6) nn(system-8, investigation-7) dobj(established-4, system-8) prep(system-8, in-9) pobj(in-9, 1946-10) \u77e5\u7db2 Hownet(http://www.keenage.com)\u662f\u7531\u8463\u632f\u6771\u6240\u767c\u5c55\u51fa\u4f86(\u53c3\u8003 Dong and Dong (2006)) \u3002Hownet \u67b6\u69cb\u4e0d\u540c\u65bc Wordnet\uff0cWordnet \u57fa\u672c\u4e0a\u662f\u4e00\u500b\u8a5e\u5f59\u7db2\u8def\uff0c\u540c\u6a23\u8a9e\u610f\u7684\u8a5e \u5c6c\u65bc\u540c\u4e00\u7d44\u7684 synset\uff0c\u88e1\u9762\u7684\u5b9a\u7fa9\uff0c\u4f8b\u53e5\u90fd\u76f8\u540c\u3002Wordnet \u88e1\u9762\u5305\u542b\u7684\u8a5e\u5f59\u8a9e\u610f\u95dc\u4fc2\u5305 \u62ec\u4e0a\u4f4d\u8a5e\uff0c\u4e0b\u4f4d\u8a5e\u7b49\u3002Hownet \u5247\u5229\u7528\u62bd\u8c61\u7684\u7fa9\u5143\u4f5c\u70ba\u8868\u9054\u6240\u6709\u6982\u5ff5\u7684\u5de5\u5177\u548c\u55ae\u4f4d\u3002 Hownet \u5305\u542b\u7684\u8a0a\u606f\u76f8\u7576\u7684\u591a\uff0c\u662f\u4e00\u500b\u4e2d\u82f1\u96d9\u8a9e\u7684\u77e5\u8b58\u5eab\uff0c\u5305\u62ec\u7fa9\u5143\uff0c\u8a9e\u610f\u89d2\u8272\uff0c\u4e0a\u4e0b \u4f4d\u95dc\u4fc2\uff0c\u90e8\u4ef6\u8207\u6574\u9ad4\u95dc\u4fc2\u7b49\u7b49\u8a9e\u610f\u8a0a\u606f\u3002\u7fa9\u5143\u985e\u4f3c\u4e00\u500b\u8a9e\u610f\u7279\u5fb5\u3002Hownet \u5c0d\u65bc\u91ab\u751f (doctor)\u7684\u7fa9\u5143\u8868\u793a\u6cd5\u70ba {human| \u4eba :HostOf={Occupation| \u8077 \u4f4d },domain={medical| \u91ab },{doctor| \u91ab Hownet \u88e1\u9762\u7684\u8a0a\u606f\u8868\u793a\u91ab\u751f\u662f\u4e00\u500b\u4eba\uff0c\u5177\u6709\u8077\u4f4d\uff0c\u5c6c\u65bc\u91ab\u5b78\u9818\u57df\uff0c\u91ab\u751f\u5728\u91ab\u6cbb\u9019 \u500b\u4e8b\u4ef6\u88e1\u626e\u6f14\u4e3b\u4e8b\u8005\u7684\u8a9e\u7fa9\u89d2\u8272\u3002\u5728\u6211\u5011\u7684\u5be6\u9a57\u4e2d\uff0c\u6211\u5011\u53ea\u4f7f\u7528 Hownet \u8868\u793a\u6cd5\u7576\u4e2d\u7b2c \u4e00\u500b\u7fa9\u5143\uff0c\u4f8b\u5982\uff1adoctor \u7684\u7b2c\u4e00\u500b\u7fa9\u5143\u662f human\u3002\u5c0d\u65bc\u540d\u8a5e\u800c\u8a00\uff0c\u7b2c\u4e00\u500b\u7fa9\u5143\u76f8\u7576\u65bc\u9019 \u500b\u8a5e\u7684\u8a9e\u610f\u985e\u5225\u6216\u672c\u9ad4 ontology\u3002 \u56db\u3001\u5be6\u9a57\u7d50\u679c \u5728 F1~F6 \u4e2d\uff0c\u90fd\u5fc5\u9808\u53d6\u4e00\u500b Window Size\uff0c\u5426\u5247\u6703\u5c0e\u81f4\u7279\u5fb5\u548c\u76ee\u6a19\u8a5e\u4ee5\u53ca\u8a5e\u7fa9\u7684\u76f8\u95dc\u806f \u6027\u8868\u73fe\u4e0d\u51fa\u4f86\uff0c\u56e0\u6b64\u9019\u516d\u7a2e\u7279\u5fb5\u90fd\u6703\u6709 Window Size \u7684\u5be6\u9a57\u3002\u800c\u6211\u5011\u7684\u5be6\u9a57\u662f\u5c0d\u5404\u7a2e\u7279 \u5fb5\u5148\u7368\u7acb\u7684\u4f86\u505a\u8a5e\u7fa9\u8fa8\u8a8d\u4ee5\u5f97\u5230\u5404\u7a2e\u7279\u5fb5\u7684\u6700\u4f73\u53c3\u6578\uff0c\u6bcf\u7a2e\u7279\u5fb5\u90fd\u6700\u4f73\u5316\u4ee5\u5f8c\uff0c\u6700\u5f8c\u518d \u4f7f\u7528 Forward Sequential Selection Algorithm \u4f86\u6c7a\u5b9a\u8981\u63a1\u7528\u54ea\u4e9b\u7279\u5fb5\u3002\u8655\u7406\u8a9e\u6599\u4ee5\u53ca\u8fa8\u8a8d \u7a0b\u5f0f\u662f\u4ee5 Perl \u53ca C++\u5beb\u6210\u3002 (\u4e00) \u3001F1 F1 \u662f\u6700\u7c21\u55ae\u7684\u76f4\u63a5\u628a\u76ee\u6a19\u8a5e\u5468\u570d\u7684\u8a5e\u505a\u70ba\u7279\u5fb5\uff0c\u4f46\u662f\u6703\u6392\u9664\u3127\u4e9b\u5982 is, a \u4e4b\u985e\u7684 stop words\uff0c\u539f\u56e0\u662f stop words \u901a\u5e38\u5c0d\u65bc\u8fa8\u8a8d\u4e00\u500b\u8a5e\u7684\u8a5e\u7fa9\u6c92\u6709\u4ec0\u9ebc\u5e6b\u52a9\u3002\u8868\u4e00\u986f\u793a F1 \u7684\u6700 \u4f73 Window Size \u70ba 3\u3002 \u8868\u4e00\u3001F1 Window Size \u5be6\u9a57\u7d50\u679c Window Size \u6b63\u78ba\u7387(%) 1 52.7 2 54.2 3 54.6 4 54.6 5 54.1 (\u4e8c) \u3001F2 F2 \u4e5f\u662f\u76ee\u6a19\u8a5e\u5468\u570d\u7684\u8a5e\uff0c\u4f46\u6703\u52a0\u4e0a\u4f4d\u7f6e\u7684\u8cc7\u8a0a\uff0c\u6703\u8a18\u9304\u67d0\u500b\u8a5e\u662f\u51fa\u73fe\u5728\u76ee\u6a19\u8a5e\u7684\u4ec0\u9ebc \u4f4d\u7f6e\uff0c\u8868\u4e8c\u986f\u793a F2 \u7684\u6700\u4f73 Window Size \u70ba 1 \u8868\u4e8c\u3001F2 Window Size \u5be6\u9a57\u7d50\u679c Window Size \u6b63\u78ba\u7387(%) 1 54.9 2 53.6 3 51.1 4 47.9 (\u4e09) \u3001F3 F2 \u662f\u76ee\u6a19\u8a5e\u5468\u570d\u7684\u8a5e\uff0c\u4f46\u6703\u52a0\u4e0a\u4f4d\u7f6e\u7684\u8cc7\u8a0a\uff0c F3 \u8ddf F2 \u985e\u4f3c\uff0c\u4f46\u4e0d\u540c\u7684\u662f F3 \u662f\u53d6\u51fa \u8a5e\u6027\u3002 Window Size \u6b63\u78ba\u7387(%) 1 Window Size \u6b63\u78ba\u7387(%) Window Size \u6b63\u78ba\u7387(%) Subject 50.2 possessor 50.1 Subject 51.9 possessor 51.9 44.6 2 1 50.5 11 51.6 subject_of 50.1 possessed 50.0 subject_of 51.9 possessed 51.9 35.5 3 2 51.1 12 51.4 a_modifier 50.3 Modifier 50.2 comp_of 51.9 Modifier 52.0 30.7 4 27.7 \u56e0\u6b64 F3 \u7684\u6700\u4f73 Window Size \u70ba 1 (\u56db) \u3001F4 F4 \u5247\u662f\u76ee\u6a19\u8a5e\u8207\u5468\u570d\u8a5e\u7684\u7d44\u5408\uff0c\u540c\u6a23\u4ee5\u301dThe art of design\u301e\u70ba\u4f8b\uff0c\u6703\u88ab\u53d6\u51fa\u7684\u7279\u5fb5\u6709 {The-art, art-of, The-art-of, art-of-design, The-art-of-design}\u3002 \u8868\u56db\u3001F4 Window Size \u5be6\u9a57\u7d50\u679c Window Size \u6b63\u78ba\u7387(%) 1 48.2 2 56.9 3 57.8 4 57.8 5 57.8 \u56e0\u6b64 F4 \u7684\u6700\u4f73 Window Size \u70ba 3\u3002 (\u4e94) \u3001F5 F5 \u5247\u662f\u76ee\u6a19\u8a5e\u8207\u5468\u570d\u8a5e\u6027\u7684\u7d44\u5408\u3002 \u8868\u4e94\u3001F5 Window Size \u5be6\u9a57\u7d50\u679c Window Size \u6b63\u78ba\u7387(%) 1 48.2 2 52.1 3 53.8 4 54.2 5 54.1 \u56e0\u6b64 F5 \u7684\u6700\u4f73 Window Size \u70ba 4\u3002 (\u516d) \u3001F6 F6 \u5247\u662f\u900f\u904e Sketch Engine \u5c07\u53ef\u80fd\u8207\u76ee\u6a19\u8a5e\u6709\u4f9d\u5b58\u95dc\u4fc2\u7684\u8a5e\u5217\u70ba\u7279\u5fb5\u3002Sketch Engine \u5728 \u4f7f\u7528\u6642\u6709\u6578\u500b\u53c3\u6578\u53ef\u4ee5\u505a\u8abf\u6574\uff0c\u5206\u5225\u662f\u6240\u8981\u5305\u542b\u7684\u4f9d\u5b58\u95dc\u4fc2\u7a2e\u985e\u3001minimum salience\uff0c 3 51.6 13 51.4 4 51.8 14 n_modifier 50.7 part* 50.2 Comp 51.9 Part 51.9 \u7684\u7279\u5fb5\u7d44\u5408\u70ba\u76ee\u6a19\u8a5e\u9031\u570d\u7684\u4e09\u500b\u8a5e\u3001\u76ee\u6a19\u8a5e\u9031\u570d\u4e00\u500b\u8a5e\u53ca\u5176\u4f4d\u7f6e\u95dc\u4fc2\u3001\u76ee\u6a19\u8a5e\u8207\u5468\u570d\u4e09 51.1 5 52.0 15 50.8 6 52.0 7 51.8 8 51.5 9 51.4 10 51.5 \u56e0\u6b64 F6 \u7684\u6700\u4f73 Window Size \u70ba 5\u3002 *comp 50.1 *comp_of 50.1 \u5728\u9019\u6b65\u4e2d\u9078\u64c7\u4e86 object \u500b\u8a5e\u7684\u9023\u7e8c\u7d44\u5408\u3001\u4ee5\u53ca\u5229\u7528 Stanford Parser \u6240\u5f97\u5230\u8207\u76ee\u6a19\u8a5e\u6709\u4f9d\u5b58\u95dc\u4fc2\u7684\u8a5e\uff0c\u6b63\u78ba\u7387\u662f \u5728\u9019\u6b65\u4e2d\u9078\u64c7\u4e86 modifier 61.2%\u3002 \u8868\u5341\u56db\u3001F6 \u4f9d\u5b58\u95dc\u4fc2\u7d44\u5408\u9078\u64c7(\u7b2c\u4e03\u6b65) \u8868\u5341\u3001F6 \u4f9d\u5b58\u95dc\u4fc2\u7d44\u5408\u9078\u64c7(\u7b2c\u4e09\u6b65) Type \u6b63\u78ba\u7387(%) Type Type \u6b63\u78ba\u7387(%) Type \u6b63\u78ba\u7387(%) \u7531\u5be6\u9a57\u7d50\u679c\u53ef\u770b\u51fa\uff0c\u7531\u65bc senseval-2 \u7684\u5c90\u7fa9\u76ee\u6a19\u8a5e\u8a5e\u6027\u90fd\u4e00\u6a23\uff0c\u63a1\u7528\u8a5e\u6027\u76f8\u95dc\u7684\u7279\u5fb5\u5c0d \u6b63\u78ba\u7387(%) object_of 51.2 and/or object_of 52.0 pp* 51.9 \u65bc\u8a5e\u7fa9\u8fa8\u8a8d\u6c92\u6709\u4ec0\u9ebc\u5e6b\u52a9\uff0c\u8f03\u6709\u5e6b\u52a9\u7684\u662f\u76ee\u6a19\u8a5e\u5468\u570d\u7684\u8a5e\u4ee5\u53ca\u8207\u5176\u6709\u4f9d\u5b58\u95dc\u4fc2\u7684\u8a5e\u3002 51.5 Subject 51.1 pp* Subject 52.0 possessor 51.9 51.4 subject_of 51.1 possessor 51.1 a_modifier 51.3 possessed 51.0 n_modifier 51.5 Modifier 51.2 Comp 51.1 Part 51.1 subject_of 51.9 possessed 51.9 Part \u4e94\u3001\u7d50\u8ad6 52.0 comp_of \u4e0b\u8868\u662f Senseval-2 \u7576\u6642\u7684\u7d50\u679c\uff0c\u9019\u4efd\u7d50\u679c\u6240\u4f7f\u7528\u7684\u8a13\u7df4\u53ca\u6e2c\u8a66\u8a9e\u6599\u548c\u6211\u5011\u4f7f\u7528\u7684\u662f\u76f8\u540c 52.0 Comp \u7684\uff1a 52.0 \u8868\u5341\u516d\u3001Senseval-2 English Lexical Sample Result \u8868\u4e03\u3001F6 minimum salience \u5be6\u9a57\u7d50\u679c Minimum salience \u6b63\u78ba\u7387(%) 0.0 52.0 1.0 51.8 2.0 51.8 3.0 51.3 \u56e0\u6b64 F6 \u7684\u6700\u4f73 Minimum Salience \u7d04\u70ba 0.0\u3002 \u5c0d\u65bc\u4f9d\u5b58\u95dc\u4fc2\u7684\u9078\u64c7\uff0c\u5247\u662f\u4f7f\u7528 Forward Sequential Selection Algorithm \u4f86\u9078\u51fa\u6700\u597d\u7684\u7d44 \u5408\u3002 comp_of 51.2 \u5728\u9019\u6b65\u4e2d\u9078\u64c7\u4e86 n_modifier \u6e96\u78ba\u5ea6 \u7cfb\u7d71 \u6e96\u78ba\u5ea6 \u7cfb\u7d71 \u5728\u9019\u6b65\u4e2d\u53ef\u4ee5\u770b\u5230\u7121\u8ad6\u52a0\u9032\u54ea\u7a2e\u4f9d\u5b58\u95dc\u4fc2\u6b63\u78ba\u7387\u90fd\u4e0d\u518d\u4e0a\u5347\u4e86\uff0c\u56e0\u6b64\u6700\u5f8c\u6240\u627e\u5230\u7684\u6700\u4f73 64.2 JHU (R) 51.2 Baseline Lesk Corpus \u4f9d\u5b58\u95dc\u4fc2\u7d44\u5408\u70ba{ modifies, object, n_modifier, a_modifier, and/or, modifier}\u3002\u9019\u500b\u7d50\u679c\u986f \u8868\u5341\u4e00\u3001F6 \u4f9d\u5b58\u95dc\u4fc2\u7d44\u5408\u9078\u64c7(\u7b2c\u56db\u6b65) Type \u6b63\u78ba\u7387(%) Type \u793a\u4e3b\u8a5e\u7684\u7279\u5fb5\u5c0d\u65bc\u8a5e\u7fa9\u8fa8\u8b58\u800c\u8a00\u4e0d\u662f\u5f88\u91cd\u8981\u3002\u6700\u91cd\u8981\u7684\u662f\u4fee\u98fe\u8a9e\u548c\u53d7\u8a5e\u3002 63.8 SMUls 50.8 Duluth B \u6b63\u78ba\u7387(%) object_of 51.6 and/or 51.7 Subject 51.5 pp* 62.9 KUNLP 49.8 UNED -LS-T (\u4e03) \u3001F7 61.7 Stanford -CS224N 47.6 Baseline Commonest 51.6 subject_of 51.5 possessor 51.5 a_modifier 51.7 possessed F7 \u662f\u5229\u7528 Stanford Parser \u6240\u5256\u6790\u51fa\u8207\u76ee\u6a19\u8a5e\u6709\u4f9d\u5b58\u95dc\u4fc2\u7684\u8a5e\u4ee5\u53ca\u4f9d\u5b58\u95dc\u4fc2\u7684\u985e\u5225(\u4f8b 61.3 Sinequa-LIA -SCT 43.7 Baseline Grouping Lesk Corpus \u5982\uff1aobject_of, modifies)\u5217\u70ba\u7279\u5fb5\u3002\u6e96\u78ba\u7387\u662f 54.6% 59.4 TALP 42.7 Baseline Grouping Commonest 51.5 comp_of 51.4 Modifier 51.4 Comp 51.4 Part 51.3 57.1 Duluth 3 41.1 Alicante (\u516b) \u3001F8 56.8 JHU 26.8 Baseline Grouping Lesk F8 \u662f\u53d6\u76ee\u6a19\u8a5e\u524d\u5f8c\u5169\u500b\u8a5e\u7684 HowNet \u7fa9\u5143\u505a\u70ba\u7279\u5fb5\u3002\u6e96\u78ba\u7387\u662f 47.2% 56.8 UMD -SST 24.9 IRST \u8868\u516b\u3001F6 \u4f9d\u5b58\u95dc\u4fc2\u7d44\u5408\u9078\u64c7(\u7b2c\u4e00\u6b65) Type \u6b63\u78ba\u7387(%) Type \u6b63\u78ba\u7387(%) Object 49.2 and/or (\u4e5d) \u3001F9 56.4 BCU -ehu-dlist-all 23.3 BCU -ehu-dlist-best \u5728\u9019\u6b65\u4e2d\u9078\u64c7\u4e86 a_modifier \u8868\u5341\u4e8c\u3001F6 \u4f9d\u5b58\u95dc\u4fc2\u7d44\u5408\u9078\u64c7(\u7b2c\u4e94\u6b65) F9 \u662f\u5148\u5229\u7528 Stanford Parser \u627e\u51fa\u8207\u76ee\u6a19\u8a5e\u6709\u4f9d\u5b58\u95dc\u4fc2\u7684\u8a5e\uff0c\u518d\u53d6\u51fa\u5176 HowNet \u7fa9\u5143\u505a\u70ba 55.4 Duluth 5 23.0 Baseline Grouping Lesk Def 49.4 object_of 47.5 pp* Type \u6b63\u78ba\u7387(%) Type \u6b63\u78ba\u7387(%) \u7279\u5fb5\u3002\u6e96\u78ba\u7387\u662f 54.1% 55.0 Duluth C 22.6 Baseline Lesk 49.4 Subject 48.4 possessor 46.6 subject_of 48.0 possessed object_of 51.6 and/or 54.2 Duluth 4 18.3 Baseline Grouping Random 51.9 Subject 51.7 pp* 51.8 (\u5341) \u3001\u7279\u5fb5\u9078\u53d6 53.9 Duluth 2 16.3 Baseline Lesk Def 47.6 a_modifier 48.1 Modifier 48.4 n_modifier 49.0 part* 48.5 Modifies 50.1 *comp_of 48.3 *comp 48.2 subject_of 51.7 possessor 51.6 comp_of 51.7 possessed 51.7 Comp 51.8 Modifier 53.4 Duluth 1 14.1 Baseline Random \u63a1\u7528 Forward Sequential Selection Algorithm \u4f86\u505a\u7279\u5fb5\u9078\u53d6\u3002 52.3 Duluth A \u8868\u5341\u4e94\u3001\u7279\u5fb5\u9078\u53d6\u7d50\u679c 51.8 Part 51.7 Step F1 F2 F3 F4 F5 F6 F7 F8 F9 1 st 54.6 54.9 44.6 57.8 54.2 52.0 54.6 47.2 \u7531\u8868\u4e2d\u6578\u64da\u53ef\u770b\u51fa\uff0c\u5982\u679c\u5c0d\u6bcf\u500b\u76ee\u6a19\u8a5e\u96a8\u6a5f\u9078\u4e00\u500b\u610f\u7fa9\u7684\u8a71\u6e96\u78ba\u5ea6\u662f 14.1%\u3001\u9078\u6700\u5e38\u898b 54.1 \u7684\u610f\u7fa9\u7684\u8a71\u662f 47.6%\u3002\u800c\u6211\u5011\u7684\u7d50\u679c 61.2%\u9060\u5927\u65bc baseline \u7684\u6578\u64da\u800c\u4e14\u4e5f\u6bd4\u5927\u90e8\u5206\u7684\u7cfb \u5728\u9019\u6b65\u4e2d\u9078\u64c7\u4e86 modifies \u8868\u4e5d\u3001F6 \u4f9d\u5b58\u95dc\u4fc2\u7d44\u5408\u9078\u64c7(\u7b2c\u4e8c\u6b65) Type \u6b63\u78ba\u7387(%) Type \u6b63\u78ba\u7387(%) Object 51.1 and/or 50.8 2 nd 58.9 58.5 56.2 56.8 58.2 59.6 56.8 59.2 \u7d71\u597d\uff0c\u8868\u793a\u9019\u4e9b\u7279\u5fb5\u61c9\u7528\u5728\u8a5e\u7fa9\u8fa8\u8a8d\u4e2d\u662f\u6709\u6548\u7684\u3002\u6211\u5011\u5be6\u9a57\u7684\u7d50\u679c\u986f\u793a\u4e3b\u8a5e\u7684\u7279\u5fb5\u5c0d\u65bc \u5728\u9019\u6b65\u4e2d\u9078\u64c7\u4e86 and/or Type \u6b63\u78ba\u7387(%) Type \u6b63\u78ba\u7387(%) 4 th 61.2 60.1 58.8 60.6 60.4 60.7 \u8868\u5341\u4e09\u3001F6 \u4f9d\u5b58\u95dc\u4fc2\u7d44\u5408\u9078\u64c7(\u7b2c\u516d\u6b65) 3 rd 60.7 60.1 58.2 58.1 60.2 59.1 59.7 \u8a5e\u7fa9\u8fa8\u8b58\u800c\u8a00\u4e0d\u662f\u5f88\u91cd\u8981\u3002\u6700\u91cd\u8981\u7684\u662f\u4fee\u98fe\u8a9e\u548c\u53d7\u8a5e\u3002\u96d6\u7136\u8207 Senseval 2 \u53c3\u8cfd\u88e1\u9762\u6700\u597d</td></tr></table>",
"text": "F2 \u985e\u4f3c\uff0c\u4f46\u4e0d\u540c\u7684\u662f F3 \u662f\u53d6\u51fa\u8a5e\u6027\u3002 F4 \u5247\u662f\u76ee\u6a19\u8a5e\u8207\u5468\u570d\u8a5e\u7684\u7d44\u5408\uff0c\u540c\u6a23\u4ee5\u301dThe art of design\u301e\u70ba\u4f8b\uff0c\u6703\u88ab\u53d6\u51fa\u7684\u7279\u5fb5\u6709 {The-art, art-of, The-art-of, art-of-design, The-art-of-design}\u3002 F5 \u8207 F4 \u985e\u4f3c\u4f46\u53d6\u8a5e\u6027\u7d44\u5408\u3002 F6 \u5247\u662f\u5229\u7528 Sketch Engine\uff0c\u5c07\u53ef\u80fd\u8207\u76ee\u6a19\u8a5e\u6709\u8a9e\u6cd5\u642d\u914d\u95dc\u4fc2\u7684\u8a5e\u5217\u70ba\u7279\u5fb5\u3002 F7 \u662f\u5229\u7528 Stanford Parser\uff0c\u5c07 Stanford Parser \u6240\u5256\u6790\u51fa\u7684\u8207\u76ee\u6a19\u8a5e\u6709\u4f9d\u5b58\u95dc\u4fc2\u7684\u8a5e\u4ee5\u53ca \u4f9d\u5b58\u95dc\u4fc2\u7684\u985e\u5225\u5217\u70ba\u7279\u5fb5\u3002 F8 \u662f\u53d6\u76ee\u6a19\u8a5e\u524d\u5f8c\u5169\u500b\u8a5e\u7684 HowNet \u7fa9\u5143\u505a\u70ba\u7279\u5fb5\u3002 F9 \u662f\u5148\u5229\u7528 Stanford Parser \u627e\u51fa\u8207\u76ee\u6a19\u8a5e\u6709\u4f9d\u5b58\u95dc\u4fc2\u7684\u8a5e\uff0c\u518d\u53d6\u51fa\u5176 HowNet \u7fa9\u5143\u505a\u70ba \u7279\u5fb5\u3002 \u6211\u5011\u4f7f\u7528 Sketch Engine (Kilgarriff et al. (2004))\u627e\u51fa\u8ddf\u76ee\u6a19\u8a5e\u5177\u6709\u8a9e\u6cd5\u642d\u914d\u95dc\u4fc2\u7684 \u6240\u6709\u8a5e\u3002\u5716\u4e00\u662f\u5229\u7528 Sketch Engine (http://www.sketchengine.co.uk/)\u7684 word sketch \u67e5 \u8a62 duty \u500b\u76ee\u6a19\u8a5e\u7684\u8f38\u51fa\uff0cobject_of \u9019\u4e00\u6b04\u8868\u793a\u76ee\u6a19\u8a5e\u53ef\u4ee5\u4f5c\u70ba\u9019\u4e9b\u8a5e\u7684\u53d7\u8a5e\u7684\u642d\u914d \u8a9e\uff0csubject_of \u8868\u793a\u76ee\u6a19\u8a5e\u53ef\u4ee5\u4f5c\u70ba\u9019\u4e9b\u8a5e\u7684\u4e3b\u8a5e\u7684\u642d\u914d\u8a9e, a_modifier \u662f\u53ef\u4ee5\u4fee\u98fe\u9019\u500b \u76ee\u6a19\u8a5e\u7684\u5f62\u5bb9\u8a5e\uff0cn_modifier \u662f\u53ef\u4ee5\u4fee\u98fe\u9019\u500b\u76ee\u6a19\u8a5e\u7684\u540d\u8a5e, modifies \u5247\u662f\u53ef\u4ee5\u88ab\u9019\u500b\u76ee \u6a19\u8a5e\u4fee\u98fe\u7684\u8a5e\u3002\u6211\u5011\u9078\u64c7\u7684\u82f1\u6587\u8a9e\u6599\u8d85\u904e 20 \u5104\uff0c\u5982\u6b64\u9f90\u5927\u7684\u8a9e\u6599\u53ef\u4ee5\u78ba\u4fdd\u5f97\u5230\u5927\u90e8\u5206 \u7684\u642d\u914d\u8a9e\u3002 \u5716\u4e00 Sketch Engine Word Sketch \u7684\u8f38\u51fa\u7d50\u679c Stanford Parser \u662f\u53f2\u4e39\u798f\u5927\u5b78 Klein and Manning (2003)\u767c\u5c55\u51fa\u4f86\u7684\u591a\u570b\u8a9e\u8a00\u5256\u6790 \u5668\uff0c\u53ea\u8981\u8f38\u5165\u7b26\u5408 Pen Treebank \u683c\u5f0f\u7684\u8a9e\u6cd5\u6a39\u5eab\uff0c\u5373\u53ef\u81ea\u52d5\u5f9e\u8a9e\u6cd5\u6a39\u5eab\u4e2d\u8a13\u7df4\u5f97\u5230\u8a72\u8a9e \u8a00\u7684\u8a9e\u6cd5\u5256\u6790\u5668\u3002\u4e0b\u9762\u7684\u4f8b\u5b50\u662f Stanford Parser \u7684\u8f38\u51fa\u7d50\u679c\u3002\u9664\u4e86\u6a19\u793a\u8a5e\u6027\uff0c\u8a9e\u6cd5\u7d50\u69cb\uff0c \u6700\u7279\u5225\u7684\u662f\u9084\u5c07\u8a9e\u6cd5\u7684\u4f9d\u5b58\u95dc\u4fc2\u5217\u51fa\u4f86,\u4f8b\u5982,nsubj \u8868\u793a\u52d5\u8a5e\u548c\u4e3b\u8a5e\u7684\u95dc\u4fc2\uff0cdobj \u8868\u793a\u52d5 \u8a5e\u548c\u53d7\u8a5e\u7684\u95dc\u4fc2\uff0cadvmod \u8868\u793a\u52d5\u8a5e\u548c\u526f\u8a5e\u4fee\u98fe\u8a9e\u7684\u95dc\u4fc2\uff0camod \u8868\u793a\u540d\u8a5e\u548c\u540d\u8a5e\u4fee\u98fe\u8a9e \u7684\u95dc\u4fc2\u3002\u5fc5\u9808\u5f37\u8abf\u7684\u662f Stanford Parser \u7684\u8a9e\u6cd5\u4f9d\u5b58\u95dc\u4fc2\u662f\u5f9e\u8a9e\u6cd5\u6a39\u5eab\u6b78\u7d0d\u51fa\u4f86\u5f8c\u5229\u7528 regular expression \u62bd\u53d6\u51fa\u4f86\u7684\uff0c\u56e0\u6b64\u5373\u4f7f\u8a9e\u6cd5\u5256\u6790\u7684\u7d50\u679c\u6b63\u78ba\uff0c\u8a9e\u6cd5\u4f9d\u5b58\u95dc\u4fc2\u4e0d\u4e00\u5b9a\u6b63 \u78ba\u3002\u4e0b\u9762\u662f Stanford Parser \u7684\u8f38\u51fa\u7d50\u679c\u3002 The/DT government/NN first/RB established/VBD modern/JJ criminal/JJ investigation/NN system/NN in/IN 1946/CD ./. \u6bcf\u500b step \u6703\u6709\u4e00\u6b04\u662f\u7c97\u9ad4, \u4ee3\u8868\u8a72 step \u9078\u53d6\u7684\u7279\u5fb5\u3002\u4f8b\u5982\uff0c\u7b2c\u4e00\u6b65\u9078\u51fa F4\uff0c\u63a5\u4e0b\u9078\u51fa F7,\u63db\u8a00\u4e4b\uff0c\u7b2c\u4e8c\u6b65\u7684 F1 \u5176\u5be6\u662f\u4ee3\u8868 F4+F1, F2 \u4ee3\u8868 F4+F2..\u4f9d\u6b64\u985e\u63a8\u3002\u800c\u7b2c\u4e8c\u6b65\u9a5f\u7d50\u675f \u5f8c\u9078\u53d6\u7684\u7279\u5fb5\u5c31\u662f F4+F7\u3002\u5728\u7b2c\u4e09\u6b65\u9a5f\u6642\u7684 F1 \u662f\u4ee3\u8868 F4+F7+F1\uff0c\u4ee5\u6b64\u985e\u63a8\u3002\u56e0\u6b64\u6700\u597d",
"num": null,
"type_str": "table",
"html": null
}
}
}
} |