File size: 86,463 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
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
{
    "paper_id": "O08-2006",
    "header": {
        "generated_with": "S2ORC 1.0.0",
        "date_generated": "2023-01-19T08:02:29.689303Z"
    },
    "title": "",
    "authors": [],
    "year": "",
    "venue": null,
    "identifiers": {},
    "abstract": "",
    "pdf_parse": {
        "paper_id": "O08-2006",
        "_pdf_hash": "",
        "abstract": [],
        "body_text": [
            {
                "text": "n \u2208 N \uff0c n \u6709 \u4e09 \u500b \u53c3 \u6578 \u5206 \u5225 \u662f n[STREE//] \u3001 n[/STC/] \u548c n[//ORDER] \uff1b \u6211 \u5011 \u4ee5 n[STREE/STC/ORDER]\u4f86\u8868\u793a\u3002\u70ba\u4e86\u65b9\u4fbf\u8aaa\u660e\uff0c\u82e5\u7bc0\u9ede n \u53ea\u6709 n[STREE//]\u548c n[/STC/]\uff0c \u5247\u4ee5 n[STREE/STC/]\u8868\u793a\u3002\u518d\u5047\u8a2d n C(n) \u70ba\u7bc0\u9ede n \u6709 1 \u5230 C(n)\u500b\u5b50\u7bc0\u9ede\u3002n[STREE//]\u70ba \u7bc0\u9ede n \u6240\u6db5\u84cb\u4f86\u6e90\u53e5\u7684\u7bc4\u570d\uff0c\u5c64\u6578\u6700\u5927\u7bc0\u9ede\u7684 n[STREE//]\u5fc5\u5b9a\u5c0d\u61c9\u5230\u4e00\u500b\u4f86\u6e90\u53e5\u55ae\u5b57\uff0c \u6b64\u53c3\u6578\u7684\u529f\u7528\u70ba\u7576\u4f5c\u6bcf\u500b\u7bc0\u9ede\u7684\u9375\u503c(primary key)\uff0c\u6545\u5728\u540c\u4e00\u68f5\u5256\u6790\u6a39\u4e2d n[STREE//]\u4e0d \u6703\u91cd\u8907\u3002n[/STC/]\u8868\u793a\u4ee5 n \u70ba\u6a39\u6839\u7684\u5b50\u6a39\uff0c\u6240\u6db5\u84cb\u4f86\u6e90\u53e5\u5b57\u4e32\u7684\u7bc4\u570d\u5c0d\u61c9\u5230\u76ee\u6a19\u53e5\u5b57\u4e32 \u7684\u7bc4\u570d\uff1bn[/STC/]\u4e5f\u53ef\u4ee5\u662f\u4e00\u500b\u6578\u5b57\uff0c\u8868\u793a\u6b64\u5b50\u6a39\u5305\u542b\u7684\u76ee\u6a19\u53e5\u5b57\u4e32\u70ba\u76ee\u6a19\u53e5\u5b57\u4e32\u4e2d\u7684 \u4e00\u500b\u5b57\uff1bn[/STC/]\u4e5f\u53ef\u80fd\u662f 0\uff0c\u4ee3\u8868\u4f86\u6e90\u53e5\u7121\u6cd5\u5c0d\u61c9\u5230\u76ee\u6a19\u53e5\u3002n[//ORDER]\u662f\u7531 n[/STC/] \u8a08\u7b97\u51fa\u4f86\uff0cn[//ORDER]\u662f\u7528\u4f86\u8868\u793a\u4f86\u6e90\u53e5\u8ddf\u76ee\u6a19\u53e5\u8a5e\u5e8f\u5c0d\u61c9\u7684\u95dc\u4fc2\uff0c\u82e5\u4f86\u6e90\u53e5\u8ddf\u76ee\u6a19 \u53e5\u6709\u8a5e\u5e8f\u4e0d\u540c\u7684\u60c5\u5f62\uff0c\u5c31\u53ef\u7531 n \u8207\u6240\u6709\u5144\u5f1f\u7bc0\u9ede\u7684 n[//ORDER]\u4f86\u5224\u65b7\u3002ORDER \u7684\u7bc4 \u570d\u7531 1 \u5230 C(n)\uff0c\u7576 ORDER \u8d8a\u5c0f\uff0c\u4ee3\u8868 n \u6240\u5c0d\u61c9\u76ee\u6a19\u53e5\u7bc4\u570d\uff0c\u6bd4\u5176\u4ed6\u5144\u5f1f\u7bc0\u9ede\u7684\u76ee\u6a19 \u53e5\u7bc4\u570d\u66f4\u9760\u8fd1\u53e5\u5b50\u7684\u524d\u6bb5\u3002",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "",
                "sec_num": null
            },
            {
                "text": "\u5716\u4e09\u662f\u4e00\u500b BSSTC \u7d50\u69cb\u7684\u4f8b\u5b50\uff0c\u4f86\u6e90\u53e5\u70ba\u82f1\u6587\uff1a\"Our experiments were simple in concept\"\uff1b\u76ee\u6a19\u53e5\u70ba\u4e2d\u6587\uff1a\"\u6211\u5011\u7684\u5be6\u9a57\u6982\u5ff5\u5f88\u7c21\u55ae\"\u3002\u9996\u5148\u82f1\u6587\u53e5\u5fc5\u9807\u5148\u5efa\u6210\u5256\u6790\u6a39\uff0c \u6bcf\u500b\u8449\u5b50\u7bc0\u9ede\u70ba\u4e00\u500b\u82f1\u6587\u55ae\u5b57\uff0c\u4e26\u4ee5\u82f1\u6587\u55ae\u5b57\u70ba\u55ae\u4f4d\u505a\u6a19\u865f\uff0c\u4f8b\u5982\uff1a\"Our(1)\", \"experiments(2)\", \"were(3)\", \"simple(4)\", \"in(5)\",\"concept(6)\"\u3002\u53e6\u5916\u4e2d\u6587\u53e5\u7d93\u904e\u65b7\u8a5e\u7684\u8655\uf9e4\u5f8c\uff0c \u4ee5\u65b7\u8a5e\u5f8c\u7684\u55ae\u4f4d\u505a\u6a19\u865f\uff0c\u4f8b\u5982\uff1a\"\u6211\u5011(1)\", \"\u7684(2)\", \"\u5be6\u9a57(3)\", \"\u6982\u5ff5(4)\", \"\u5f88(5)\", \"\u7c21\u55ae (1) ",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "",
                "sec_num": null
            },
            {
                "text": "\u4e2d\u6587\u7ffb\u8b6f\uff0c\u5c0d\u61c9\u7387\u70ba 0\uff0c\u6240\u4ee5\u5c0d\u61c9\u7387\u6703\u843d\u5728 0 \u5230 1 \u4e4b\u9593\uff0c\u503c\u8d8a\u5927\u4ee3\u8868\u5c0d\u61c9\u7387\u8d8a\u9ad8\u3002\u6211\u5011 \u9700\u8981\u5920\u5927\u7684\u5c0d\u61c9\u7387\uff0c\u624d\u80fd\u8a8d\u5b9a\u70ba\u7bc4\u4f8b\u6a39\u3002\u56e0\u6b64\uff0c\u9700\u8981\u5b9a\u4e00\u500b\u9580\u6abb\u503c\u4f86\u7be9\u9078\uff0c\u6839\u64da\u5be6\u9a57\u7d50 \u679c\u7576\u9580\u6abb\u503c\u8d8a\u9ad8\u7559\u4e0b\u4f86\u7684\u7bc4\u4f8b\u6a39\u8d8a\u5c11\uff0c\u800c\u9580\u6abb\u503c\u8d8a\u4f4e\u6703\u4f7f\u7ffb\u8b6f\u7684\u54c1\u8cea\u4e0b\u964d\u3002 \u5716\u56db\u3001\u50c5\u6a19\u8a18\u6700\u5e95\u5c64 \u5716\u4e94\u3001\u50c5\u6a19\u8a18 STREE \u53ca STC \u76ee\u524d\u7bc4\u4f8b\u6a39\u53ea\u5c07\u6700\u5e95\u5c64\u7684 n[STREE/STC/]\u6a19\u8a18\u5b8c\u6210\uff0c\u5982\u5716\u56db\uff0c\u73fe\u5728\u8981\u9010\u5c64\u5c07\u672a\u6a19\u8a18 n[STREE/STC/]\u7684\u7bc0\u9ede\u6a19\u8a18\u4e0a\u53bb\u3002n[STREE//]\u6a19\u8a18\u7684\u65b9\u5f0f\uff0c\u662f\u5c07 n 1 \u5230 n C(n) \u7684 STREE \u90fd \u52a0\u5165 ES \u4e2d\u3002ES \u70ba\u7528\u4f86\u5132\u5b58 n C(n) [STREE//]\u4e2d STREE \u7684\u96c6\u5408\u3002\u7576\u5c64\u6578\u8d8a\u5c0f\uff0c\u5247 n[STREE//] \u5c07\u6703\u6db5\u84cb 1 \u500b\u4ee5\u4e0a\u7684\u4f86\u6e90\u53e5\u7684\u8a5e\u5f59\u3002\u82e5 n C(n) [STREE//]\u70ba\u4e00\u500b\u7bc4\u570d\uff0c\u5247\u5c07\u6b64\u7bc4\u570d\u6700\u5927\u548c \u6700\u5c0f\u7684\u503c\u52a0\u5165 ES\uff0c\u6700\u5f8c ES \u5167\u53ef\u80fd\u70ba\u4e00\u500b\u6578\u5b57\u6216\u5169\u500b\u4ee5\u4e0a\u7684\u6578\u5b57\u9019\u5169\u7a2e\u60c5\u6cc1\uff0c\u5982\u53ea\u6709\u4e00 \u500b\u6578\u5b57\u5247 n[STREE//]\u53ea\u6a19\u8a18\u8a72\u6578\u5b57\uff0c\u5982\u6709\u5169\u500b\u4ee5\u4e0a\u7684\u6578\u5b57\u5247 ES \u4e2d\u6700\u5c0f\u548c\u6700\u5927\u7684\u6578\u503c\u6a19 \u8a18\u5728 n[STREE//]\u4e0a\uff0c\u683c\u5f0f\u70ba n[\u6700\u5c0f-\u6700\u5927//] \uff1bn[/STC/]\u6a19\u8a18\u7684\u65b9\u5f0f\uff0c\u662f\u5c07 n 1 \u5230 n C(n) \u7684 STC \u90fd\u52a0\u5165 CS \u4e2d\u3002CS \u70ba\u7528\u4f86\u5132\u5b58 n C(n) [/STC/]\u4e2d STC \u7684\u96c6\u5408\u3002\u7576\u5c64\u6578\u8d8a\u5c0f\uff0c\u5247 n[/STC/] \u5c07\u6703\u6db5\u84cb 1 \u500b\u4ee5\u4e0a\u76ee\u6a19\u53e5\u7684\u8a5e\u5f59\u3002\u5982 n C(n) [/STC/]\u70ba\u4e00\u500b\u7bc4\u570d\uff0c\u5247\u5c07\u6b64\u7bc4\u570d\u6700\u5927\u548c\u6700\u5c0f \u7684\u503c\u52a0\u5165 CS\u3002\u82e5 n C(n) [/STC/]\u51fa\u73fe 0 \u5247\u4e0d\u52a0\u5165 CS\u3002\u6700\u5f8c CS \u53ef\u80fd\u70ba\u7a7a\u3001\u4e00\u500b\u6578\u5b57\u6216\u5169\u500b \u4ee5\u4e0a\u7684\u6578\u5b57\u9019\u4e09\u7a2e\u60c5\u6cc1\uff0c\u5982\u70ba\u7a7a\u5247\u5c07 n[/STC/]\u6a19\u8a18\u70ba 0\uff0c\u82e5\u53ea\u6709\u4e00\u500b\u6578\u5b57\u5247 n[/STC/]\u53ea \u6a19\u8a18\u8a72\u6578\u5b57\uff0c\u5047\u5982\u6709\u5169\u500b\u4ee5\u4e0a\u7684\u6578\u5b57\u5c07 CS \u4e2d\u6700\u5c0f\u548c\u6700\u5927\u7684 STC \u6a19\u8a18\u5728 n[/STC/]\u4e0a\uff0c\u683c \u5f0f\u70ba n[/\u6700\u5c0f-\u6700\u5927/]\u3002 \u5047\u5982\u6211\u5011\u73fe\u5728\u8981\u6a19\u8a18\u5716\u4e94\u7b2c\u4e00\u5c64\u7684\u7bc0\u9ede VP\uff0c\u5247\u5fc5\u9700\u5c07\u7bc0\u9ede VP \u7684\u5b50\u7bc0\u9ede VBD \u548c ADJP \u7684 VBD[3//]\u53ca ADJP[4-6//]\u4e2d\u7684 STREE \u52a0\u5165 ES \u4e2d\uff0c\u56e0\u6b64 ES \u5305\u542b\u4e86 3\u30014 \u548c 6 \u4e09 \u500b\u6578\u5b57\uff0c\u6240\u4ee5 VP[STREE//]\u4e2d\u7684 STREE \u6a19\u8a18\u70ba 3-6\u3002\u63a5\u8457\u6a19\u8a18 STC\uff0c\u5c07\u7bc0\u9ede VP \u7684\u5b50\u7bc0 \u9ede VBD \u548c ADJP \u7684 VBD[3/0/]\u53ca ADJP[4-6/4-6/]\u4e2d\u7684 STC \u52a0\u5165 CS \u4e2d\uff0c\u56e0\u70ba 0 \u4e0d\u6703\u88ab\u52a0 \u5165 CS \u4e2d\uff0c\u56e0\u6b64 CS \u53ea\u6709 4 \u548c 6 \u5169\u500b\u6578\u5b57\uff0c\u6240\u4ee5 VP[3-6/STC/]\u4e2d\u7684 STC \u6a19\u8a18\u70ba 4-6\u3002 \u6700\u5f8c\uff0c\u6574\u68f5\u5256\u6790\u6a39\u7684 STREE \u8ddf STC \u90fd\u5df2\u7d93\u6a19\u8a18\u5b8c\u6210\uff0c\u5982\u5716\u4e94\uff0c\u53ea\u5269\u4e0b ORDER \u9084 \u6c92\u6a19\u8a18\u4e0a\u3002ORDER \u8655\uf9e4\u65b9\u5f0f\u5206\u70ba\u5169\u90e8\u5206\uff0c\u7b2c\u4e00\u90e8\u4efd\uff1aSTC \u70ba 0 \u4e4b\u5144\u5f1f\u7bc0\u9ede\u6309\u7167\u7531\u5de6\u81f3 \u53f3\u7684\u9806\u5e8f\u7de8\u865f\uff1b\u7b2c\u4e8c\u90e8\u5206\uff1a\u6bd4\u8f03 n \u8207 STC \u975e 0",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "",
                "sec_num": null
            },
            {
                "text": "\u7bc4\u4f8b\u6a39\u8cc7\u6599\u5eab D={D 1 , D 2 , ..., D m }\uff0cD i \u2208D D i \u5305\u542b T i \u8207 O i \uff0ci \u70ba 1 \u5230 m T i \u662f\u7b2c i \u68f5\u7bc4\u4f8b\u6a39\uff0cO i \u662f T i \u6240\u6a19\u8a18\u7684\u8a5e\u5e8f \u958b\u59cb \u8a2d\u4f47\uf99c Q \u7528\u4f86\u5132\u5b58\u5256\u6790\u6a39\uff0c\u521d\u59cb\u70ba NULL S \u52a0\u5165 Q \u7576 Q\u2260NULL \u5f9e Q \u4e2d pop \u4e00\u68f5\u7bc4\u4f8b\u6a39 \u5982\u679c\u5728 D \u4e2d\u641c\u5c0b\u5230\u76f8\u540c\u7684\u7bc4\u4f8b\u6a39 T i \u5247\u5c07 O i \u6a19\u8a18\u5728 S",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "",
                "sec_num": null
            },
            {
                "text": "\u516c\u5f0f(1)\u4e2d\u5b9a\u7fa9 C \u70ba\u4e2d\u6587\u7ffb\u8b6f\u8a5e\u5f59\uff0cE \u70ba\u82f1\u6587\u8a5e\u5f59\uff0c 1, \u70ba\u82f1\u6587\u53e5\u6709 1 \u5230 n \u500b\u82f1\u6587 \u8a5e\u5f59\uff0c\u4e2d\u6587\u7ffb\u8b6f\u8a5e\u5f59\u4e5f\u6703\u6709 1 \u5230 n \u500b\uff0c\u5373 1, \u3002\u5f9e\u516c\u5f0f\u4e2d\u53ef\u767c\u73fe\u4e2d\u6587\u8a5e\u5f59\u7ffb\u8b6f\u6210\u82f1\u6587\u8a5e \u5f59\u7684\u6a5f\u7387\uff0c\u7a31\u70ba\u4e2d\u82f1\u8a5e\u5f59\u5c0d\uf99c\uff0c\u5373Pr( | )\uff1b\u4ee5\u53ca\uf9dd\u7528\u524d\u4e00\u500b\u4e2d\u6587\u7ffb\u8b6f\u9078\u8a5e\u7684\u7d50\u679c \u22121 \uff0c \u627e\u51fa\u76ee\u524d\u4e2d\u6587\u7ffb\u8b6f\u8a5e\u5f59 \u5171\u540c\u51fa\u73fe\u7684\u6a5f\u7387\uff0c\u7a31\u70ba bi-gram \u8a9e\u8a00\u6a21\u578b\uff0c\u5373Pr( | \u22121 )\uff0c\u5c07 \u5169\u8005\u76f8\u4e58\u53d6\u8a08\u7b97\u5f8c\u6700\u5927\u7684\u6a5f\u7387\u503c\uff0c\u4ee5\u8fd1\u4f3cPr( 1, | 1, )\u7684\u6a5f\u7387\u503c\uff0c\u4f5c\u70ba\u6240\u9078\u64c7\u7684\u4e2d\u6587\u7ffb \u8b6f\u8a5e\u5f59\u3002\u5728\u9078\u8a5e\u7684\u904e\u7a0b\u4e2d\uff0cPr( | )\u8207Pr( | \u22121 )\u7684\u6a5f\u7387\u503c\u7686\u6709\u53ef\u80fd\u70ba 0\uff0c\u6211\u5011\u5c07\u4e58 0 \u63db\u6210\u4e58\u4e0a\u4e00\u500b\u6975\u5c0f\u6578(\u6211\u5011\u9810\u8a2d\u70ba 10 -6 ) \uff0c\u70ba\u4e86\u907f\u514d\u6a5f\u7387\u503c\u70ba 0 \u7684\u60c5\u5f62\uff0c\u6703\u5f71\u97ff\u9078\u8a5e\u7684\u7d50 \u679c\u3002\u4ee5\u4e0b\u5c07\u91dd\u5c0d\u4e2d\u82f1\u8a5e\u5f59\u5c0d\uf99c\u548c bi-",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "",
                "sec_num": null
            }
        ],
        "back_matter": [],
        "bib_entries": {
            "BIBREF0": {
                "ref_id": "b0",
                "title": "\u4e2d\u7814\u9662\u4e2d\u6587\u5256\u6790\u5668\u6aa2\u7d22\u7cfb\u7d71",
                "authors": [],
                "year": 2008,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "\u4e2d\u7814\u9662\u4e2d\u6587\u5256\u6790\u5668\u6aa2\u7d22\u7cfb\u7d71, http://parser.iis.sinica.edu.tw/ [Accessed: Jun. 30, 2008].",
                "links": null
            },
            "BIBREF1": {
                "ref_id": "b1",
                "title": "\u81ea\u7531\u6642\u5831\u4e2d\u82f1\u5c0d\u7167\u8b80\u65b0\u805e",
                "authors": [],
                "year": 2008,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "\u81ea\u7531\u6642\u5831\u4e2d\u82f1\u5c0d\u7167\u8b80\u65b0\u805e, http://www.libertytimes.com.tw/2008/new/jan/15/english.htm [Accessed: Jun. 30, 2008].",
                "links": null
            },
            "BIBREF4": {
                "ref_id": "b4",
                "title": "A synchronization structure of SSTC and its applications in machine translation",
                "authors": [
                    {
                        "first": "M",
                        "middle": [
                            "H"
                        ],
                        "last": "Al-Adhaileh",
                        "suffix": ""
                    },
                    {
                        "first": "T",
                        "middle": [
                            "E"
                        ],
                        "last": "Kong",
                        "suffix": ""
                    },
                    {
                        "first": "Y",
                        "middle": [],
                        "last": "Zaharin",
                        "suffix": ""
                    }
                ],
                "year": 2002,
                "venue": "Proceedings of the International Conference on Computational Linguistics -2002 Post-Conference Workshop on Machine Translation in Asia",
                "volume": "",
                "issue": "",
                "pages": "1--8",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "M. H. Al-Adhaileh, T. E. Kong and Y. Zaharin, \"A synchronization structure of SSTC and its applications in machine translation\", Proceedings of the International Confer- ence on Computational Linguistics -2002 Post-Conference Workshop on Machine Translation in Asia, 1-8, 2002.",
                "links": null
            },
            "BIBREF5": {
                "ref_id": "b5",
                "title": "A Statistical Approach to Machine Translation",
                "authors": [
                    {
                        "first": "P",
                        "middle": [
                            "F"
                        ],
                        "last": "Brown",
                        "suffix": ""
                    },
                    {
                        "first": "J",
                        "middle": [],
                        "last": "Cocke",
                        "suffix": ""
                    },
                    {
                        "first": "S",
                        "middle": [
                            "A D"
                        ],
                        "last": "Pietra",
                        "suffix": ""
                    },
                    {
                        "first": "V",
                        "middle": [
                            "J D"
                        ],
                        "last": "Pietra",
                        "suffix": ""
                    },
                    {
                        "first": "F",
                        "middle": [],
                        "last": "Jelinek",
                        "suffix": ""
                    },
                    {
                        "first": "J",
                        "middle": [
                            "D"
                        ],
                        "last": "Lafferty",
                        "suffix": ""
                    },
                    {
                        "first": "R",
                        "middle": [
                            "L"
                        ],
                        "last": "Mercer",
                        "suffix": ""
                    },
                    {
                        "first": "P",
                        "middle": [
                            "S"
                        ],
                        "last": "Roossin",
                        "suffix": ""
                    }
                ],
                "year": 1990,
                "venue": "Computational Linguistics",
                "volume": "",
                "issue": "",
                "pages": "79--85",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "P. F. Brown, J. Cocke, S. A. D. Pietra, V. J. D. Pietra, F. Jelinek, J. D. Lafferty, R. L. Mercer and P. S. Roossin, \"A Statistical Approach to Machine Translation\", Computa- tional Linguistics, 79-85, 1990.",
                "links": null
            },
            "BIBREF6": {
                "ref_id": "b6",
                "title": "Representation trees and string-tree correspondences",
                "authors": [
                    {
                        "first": "C",
                        "middle": [],
                        "last": "Boitet",
                        "suffix": ""
                    },
                    {
                        "first": "Y",
                        "middle": [],
                        "last": "Zaharin",
                        "suffix": ""
                    }
                ],
                "year": 1998,
                "venue": "Proceedings of the Twelfth International Conference on Computational Linguistics",
                "volume": "",
                "issue": "",
                "pages": "59--64",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "C. Boitet and Y. Zaharin, \"Representation trees and string-tree correspondences\", Pro- ceedings of the Twelfth International Conference on Computational Linguistics, 59-64, 1998.",
                "links": null
            },
            "BIBREF7": {
                "ref_id": "b7",
                "title": "A Survey of Current Paradigms in Machine Translation",
                "authors": [
                    {
                        "first": "B",
                        "middle": [
                            "J"
                        ],
                        "last": "Dorr",
                        "suffix": ""
                    },
                    {
                        "first": "P",
                        "middle": [
                            "W"
                        ],
                        "last": "Jordan",
                        "suffix": ""
                    },
                    {
                        "first": "J",
                        "middle": [
                            "W"
                        ],
                        "last": "Benoit",
                        "suffix": ""
                    }
                ],
                "year": 1999,
                "venue": "Advances in Computers",
                "volume": "",
                "issue": "",
                "pages": "1--68",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "B. J. Dorr, P. W. Jordan and J. W. Benoit, \"A Survey of Current Paradigms in Machine Translation\" Advances in Computers, London: Academic Press, 1-68, 1999.",
                "links": null
            },
            "BIBREF8": {
                "ref_id": "b8",
                "title": "Google Translate",
                "authors": [],
                "year": 2008,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Google Translate http://www.google.com/translate_t [Accessed: Jun. 30, 2008].",
                "links": null
            },
            "BIBREF9": {
                "ref_id": "b9",
                "title": "Building a large-scale knowledge base for machine translation",
                "authors": [
                    {
                        "first": "K",
                        "middle": [],
                        "last": "Knight",
                        "suffix": ""
                    },
                    {
                        "first": "S",
                        "middle": [
                            "K"
                        ],
                        "last": "Luk",
                        "suffix": ""
                    }
                ],
                "year": 1994,
                "venue": "Proceedings of the Twelfth National Conference on Artificial intelligence",
                "volume": "",
                "issue": "",
                "pages": "773--778",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "K. Knight and S. K. Luk, \"Building a large-scale knowledge base for machine transla- tion\", Proceedings of the Twelfth National Conference on Artificial intelligence, 773-778, 1994.",
                "links": null
            },
            "BIBREF10": {
                "ref_id": "b10",
                "title": "Statistical phrase-based translation",
                "authors": [
                    {
                        "first": "P",
                        "middle": [],
                        "last": "Koehn",
                        "suffix": ""
                    },
                    {
                        "first": "F",
                        "middle": [
                            "J"
                        ],
                        "last": "Och",
                        "suffix": ""
                    },
                    {
                        "first": "D",
                        "middle": [],
                        "last": "Marcu",
                        "suffix": ""
                    }
                ],
                "year": 2003,
                "venue": "Proceedings of the Human Language Technology Conference",
                "volume": "",
                "issue": "",
                "pages": "127--133",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "P. Koehn, F. J. Och and D. Marcu, \"Statistical phrase-based translation\", Proceedings of the Human Language Technology Conference, 127-133, 2003.",
                "links": null
            },
            "BIBREF11": {
                "ref_id": "b11",
                "title": "Example-based Machine Translation Based on TSC and Statistical Generation",
                "authors": [
                    {
                        "first": "Z",
                        "middle": [],
                        "last": "Liu",
                        "suffix": ""
                    },
                    {
                        "first": "H",
                        "middle": [],
                        "last": "Wang",
                        "suffix": ""
                    },
                    {
                        "first": "H",
                        "middle": [],
                        "last": "Wu",
                        "suffix": ""
                    }
                ],
                "year": 2005,
                "venue": "Proceedings of the Tenth Machine Translation Summit",
                "volume": "",
                "issue": "",
                "pages": "25--32",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Z. Liu, H. Wang and H. Wu, \"Example-based Machine Translation Based on TSC and Statistical Generation\", Proceedings of the Tenth Machine Translation Summit, 25-32, 2005.",
                "links": null
            },
            "BIBREF12": {
                "ref_id": "b12",
                "title": "MXPOST",
                "authors": [],
                "year": 2008,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "MXPOST, http://www.inf.ed.ac.uk/resources/nlp/local_doc/MXPOST.html [Accessed: Jun. 30, 2008].",
                "links": null
            },
            "BIBREF13": {
                "ref_id": "b13",
                "title": "An Efficient Method for Determining Bilingual Word Classes",
                "authors": [
                    {
                        "first": "F",
                        "middle": [
                            "J"
                        ],
                        "last": "Och",
                        "suffix": ""
                    }
                ],
                "year": 1999,
                "venue": "Proceedings of European Chapter of the Association for Computational Linguistics",
                "volume": "",
                "issue": "",
                "pages": "71--76",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "F. J. Och, \"An Efficient Method for Determining Bilingual Word Classes\", Proceedings of European Chapter of the Association for Computational Linguistics, 71-76, 1999.",
                "links": null
            },
            "BIBREF14": {
                "ref_id": "b14",
                "title": "Improved Statistical Alignment Models",
                "authors": [
                    {
                        "first": "F",
                        "middle": [
                            "J"
                        ],
                        "last": "Och",
                        "suffix": ""
                    },
                    {
                        "first": "H",
                        "middle": [],
                        "last": "Ney",
                        "suffix": ""
                    }
                ],
                "year": 2000,
                "venue": "Proceedings of the Thirty-eighth Annual Meeting of the Association for Computational Linguistics",
                "volume": "",
                "issue": "",
                "pages": "440--447",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "F. J. Och and H. Ney, \"Improved Statistical Alignment Models\", Proceedings of the Thirty-eighth Annual Meeting of the Association for Computational Linguistics, 440-447, 2000.",
                "links": null
            },
            "BIBREF15": {
                "ref_id": "b15",
                "title": "The Stanford Parser: A statistical parser",
                "authors": [],
                "year": 2008,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "The Stanford Parser: A statistical parser, http://nlp.stanford.edu/software/ [Accessed: Jun. 30, 2008].",
                "links": null
            },
            "BIBREF16": {
                "ref_id": "b16",
                "title": "SRILM -an extensible language modeling toolkit",
                "authors": [
                    {
                        "first": "A",
                        "middle": [],
                        "last": "Stolcke",
                        "suffix": ""
                    }
                ],
                "year": 2002,
                "venue": "Proceedings of the intelligence Conference on Spoken Language Processing",
                "volume": "",
                "issue": "",
                "pages": "901--904",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "A. Stolcke. SRILM -an extensible language modeling toolkit. Proceedings of the intel- ligence Conference on Spoken Language Processing, 901-904, 2002. http://www.speech.sri.com/projects/srilm/ [Accessed: Jun. 30, 2008].",
                "links": null
            },
            "BIBREF17": {
                "ref_id": "b17",
                "title": "Toward Memory-Based Translation",
                "authors": [
                    {
                        "first": "S",
                        "middle": [],
                        "last": "Sato",
                        "suffix": ""
                    },
                    {
                        "first": "M",
                        "middle": [],
                        "last": "Nagao",
                        "suffix": ""
                    }
                ],
                "year": 1990,
                "venue": "Proceedings of International Conference on Computational Linguistics",
                "volume": "",
                "issue": "",
                "pages": "247--252",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "S. Sato and M. Nagao, Toward Memory-Based Translation\", Proceedings of Interna- tional Conference on Computational Linguistics, 247-252, 1990.",
                "links": null
            },
            "BIBREF18": {
                "ref_id": "b18",
                "title": "The International Association for the Evaluation of Education Achievement",
                "authors": [],
                "year": 2008,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "The International Association for the Evaluation of Education Achievement, http://www.iea.nl/ [Accessed: Jun. 30, 2008].",
                "links": null
            },
            "BIBREF20": {
                "ref_id": "b20",
                "title": "The Porter Stemming Algorithm",
                "authors": [],
                "year": 2008,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "The Porter Stemming Algorithm, http://www.tartarus.org/martin/PorterStemmer/ [Ac- cessed: Jun. 30, 2008].",
                "links": null
            },
            "BIBREF22": {
                "ref_id": "b22",
                "title": "Machine Translation Based on Translation Corresponding Tree Structure",
                "authors": [
                    {
                        "first": "F",
                        "middle": [],
                        "last": "Wong",
                        "suffix": ""
                    },
                    {
                        "first": "M",
                        "middle": [],
                        "last": "Dong",
                        "suffix": ""
                    },
                    {
                        "first": "D",
                        "middle": [],
                        "last": "Hu",
                        "suffix": ""
                    }
                ],
                "year": 2006,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "25--31",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "F. Wong, M. Dong and D. Hu, Machine Translation Based on Translation Corresponding Tree Structure, Tsinghua Science & Technology, 25-31, 2006.",
                "links": null
            }
        },
        "ref_entries": {
            "FIGREF0": {
                "uris": null,
                "text": "\u4e4b\u5144\u5f1f\u7bc0\u9ede\u7684\u5927\u5c0f\uff0c\u4e26\u63a5\u5728\u7b2c\u4e00\u90e8\u4efd\u7684 \u7de8\u865f\u5f8c\uff0c\u7531\u5c0f\u5230\u5927\u7e7c\u7e8c\u6a19\u8a18\u7de8\u865f\u3002\u4f8b\u5982\u5716\u4e94\u82e5\u8981\u6a19\u8a18 JJ[4/6/ORDER]\u548c PP[5-6/4/ORDER] \u7684 ORDER\uff0c\u5247\u5c07 JJ[4/STC/]\u4e2d\u7684 STC=6 \u548c PP[5-6/STC/]\u4e2d\u7684 STC=4 \u7531\u5c0f\u6392\u5230\u5927\uff0c\u6240\u4ee5 PP[5-6/4/ORDER]\u4e2d\u7684 ORDER \u6a19\u8a18\u70ba 1\uff0cJJ[4/6/ ORDER] \u4e2d\u7684 ORDER \u6a19\u8a18\u70ba 2\u3002 \uf9dd\u7528\u4e0a\u8ff0\u7684\u65b9\u6cd5\u5f97\u5230\u7bc4\u4f8b\u6a39\uff0c\u5982\u5716\u4e09\u3002\u5982\u76f4\u63a5\u7528\u6574\u500b\u53e5\u5b50\u7684\u7bc4\u4f8b\u6a39\u5230\u8cc7\u6599\u5eab\u4e2d\u4f5c\u641c \u5c0b\uff0c\u5c07\u5f88\u96e3\u641c\u5c0b\u5230\u76f8\u540c\u7684\u7bc4\u4f8b\u6a39\uff0c\u56e0\u70ba\u53e5\u5b50\u8d8a\u9577\u53e5\u5b50\u7684\u7d50\u69cb\u6703\u8d8a\u8907\u96dc\uff0c\u6240\u4ee5\u76f8\u540c\u7d50\u69cb\u7684 \u53e5\u5b50\u91cd\u8907\u51fa\u73fe\u7684\u53ef\u80fd\u5f88\u4f4e\u3002\u56e0\u6b64\uff0c\u6211\u5011\u5c07\u7bc4\u4f8b\u6a39\u7684\u6240\u6709\u5b50\u6a39\u5206\u5225\u53d6\u51fa\u4f86\uff0c\u6bcf\u4e00\u500b\u5b50\u6a39\u6240 \u5305\u542b\u7684\u7bc4\u570d\u7684\u90fd\u662f\u82f1\u6587\u53e5\u7684\u5b50\u53e5\uff0c\u5728\u4e0d\u540c\u7684\u53e5\u5b50\u88e1\u53ef\u80fd\u6703\u6709\u76f8\u540c\u7d50\u69cb\u7684\u5b50\u53e5\uff0c\u4e0d\u4f46\u53ef\u4ee5 \u589e\u52a0\u6bd4\u5c0d\u5230\u7684\u6a5f\u7387\uff0c\u4e5f\u80fd\u589e\u52a0\u7bc4\u4f8b\u6a39\u7684\u6578\u91cf\u3002\u6700\u5f8c\u8a18\u9304\u5728\u7bc4\u4f8b\u6a39\u8cc7\u6599\u5eab\u7684\u5167\u5bb9\uff0c\u53ea\u6709\u7bc4 \u4f8b\u6a39\u548c ORDER \u53c3\u6578\u3002STREE \u548c STC \u4e0d\u9700\u8a18\u9304\u7684\u539f\u56e0\u662f\u6bcf\u4e00\u500b\u53e5\u5b50\u7684\u6bcf\u500b\u8a5e\u5f59\u90fd\u5728\u4e0d \u540c\u7684\u4f4d\u7f6e\u4e0a\uff0c\u5247\u5728\u8cc7\u6599\u5eab\u4e2d\u4e0d\u9700\u8981\u8a18\u9304 STREE \u548c STC\u3002 \u7bc4\u4f8b\u6a39\u7684\u7d50\u69cb\u6709\u53ef\u80fd\u76f8\u540c\uff0c\u800c\u8a5e\u5e8f\u4e0d\u540c\u3002\u4f8b\u5982\"NP(NP(NN fork))(PP(IN of)(NP(DT the)(NN road)))\"\uff0c\u4e2d\u6587\u7ffb\u8b6f\u70ba\"\u5c94\u8def\"\uff0c\u800c\"NP(NP(NN leader))(PP(IN of)(NP(DT a)(NN company)))\"\uff0c \u4e2d\u6587\u7ffb\u8b6f\u70ba\"\u4e00\u9593\u516c\u53f8\u7684\u9818\u5c0e\u8005\"NP(NP(DT)(NNS))(PP(IN)(NP(CD)(NNS))))\"\uff0c\u65b9\u5f62\u6846\u70ba\u7bc4\u4f8b\u6a39\u8cc7\u6599\u5eab\u4e2d\u5176\u4e2d\u4e00 \u68f5 \u7bc4 \u4f8b \u6a39 \u7d50 \u69cb \u70ba \"(NP(NP[//2](DT[//1]) (NNS[//2])) (PP[//1](IN[//1]) (NP[//2](CD[//1]) (NNS[//2]))))\"\uff0c\u6211\u5011\u53ef\u4ee5\u767c\u73fe\u7bc4\u4f8b\u6a39\u53bb\u9664 ORDER \u5f8c\u7684\u7d50\u69cb\uff0c\u6703\u8ddf\u5b50\u6a39\u7684\u7d50\u69cb\u5b8c\u5168\u76f8\u540c\uff0c \u6545\u5c07\u6b64\u7bc4\u4f8b\u6a39\u8a8d\u5b9a\u70ba\u5339\u914d\u5b50\u6a39\u3002 \u6839\u64da\u641c\u5c0b\u7bc4\u4f8b\u6a39\u6f14\u7b97\u6cd5\u7684\u6d41\u7a0b\uff0c\u5982\u5716\u4e03\u3002\u9996\u5148\u5c07\u4f86\u6e90\u53e5\u7684\u5256\u6790\u6a39\u52a0\u5230\u4f47\uf99c(queue)",
                "num": null,
                "type_str": "figure"
            },
            "FIGREF1": {
                "uris": null,
                "text": "graph shows the heights of four girls\"\uff0c \u5256\u6790\u6a39\u70ba\"(S(NP(DT The)(NN graph))(VP(VBZ shows)(NP(NP(DT the)(NNS \u5716\u516d\u3001\u5256\u6790\u6a39\u8207\u7bc4\u4f8b\u6a39\u7684\u5c0d\u61c9\u95dc\u4fc2 heights))(PP(IN of)(NP(CD four)(NNS girls)))))\"\u3002\u900f\u904e\u641c\u5c0b\u7bc4\u4f8b\u6a39\u6f14\u7b97\u6cd5\u627e\u51fa\u5339\u914d\u5b50\u6a39\uff0c \u9996\u5148\u4ee5\u7bc0\u9ede S \u70ba\u6a39\u6839\u7684\u5256\u6790\u6a39\u5230\u8cc7\u6599\u5eab\u4f5c\u641c\u5c0b\uff0c\u641c\u5c0b\u6642\u4e0d\u5305\u542b\u8449\u5b50\u7bc0\u9ede\uff0c\u6b64\u4f8b\u5b50\u6c92\u641c\u5c0b \u5230\u5339\u914d\u5b50\u6a39\uff0c\u5247\u5c07\u7bc0\u9ede S \u7684\u5b50\u6a39 NP \u548c VP \u52a0\u5165\u4f47\uf99c\u4e2d\u3002\u63a5\u4e0b\u4f86\u5c07\u5f9e\u4f47\uf99c\u4e2d\u53d6\u51fa\u7684\u5b50\u6a39 \u70ba NP\uff0c\u5230\u7bc4\u4f8b\u6a39\u8cc7\u6599\u5eab\u641c\u5c0b\u5339\u914d\u5b50\u6a39\uff0c\u4f46\u8cc7\u6599\u5eab\u4e2d\u6c92\u6709\u76f8\u540c\u7684\u7bc4\u4f8b\u6a39\uff0c\u6b64\u6642 NP \u7684\u5b50 \u6a39\u7686\u70ba\u8449\u5b50\u7bc0\u9ede\uff0c\u6240\u4ee5\u4e26\u7121\u5b50\u6a39\u5728\u52a0\u5165\u4f47\uf99c\u4e2d\u3002\u4f9d\u7167\u5148\u9032\u5148\u51fa\u7684\u539f\u5247\u4e0b\u4e00\u500b\u5f9e\u4f47\uf99c\u53d6\u51fa \u7684\u662f S \u7684\u53f3\u5b50\u6a39 VP\uff0c\u5728\u7bc4\u4f8b\u6a39\u8cc7\u6599\u5eab\u4e2d\u9084\u662f\u641c\u5c0b\u4e0d\u5230\uff0c\u56e0\u6b64\u8981\u5c07 VP \u7684\u5b50\u6a39 VBZ \u548c NP \u52a0\u5165\u4f47\uf99c\u4e2d\uff0c\u4f46 VBZ \u70ba\u8449\u5b50\u7bc0\u9ede\uff0c\u6545\u53ea\u6709 NP \u52a0\u5165\u4f47\uf99c\u4e2d\u3002\u63a5\u4e0b\u4f86\u662f\u5b50\u6a39 NP \u5f9e\u4f47 \uf99c\u4e2d\u88ab\u53d6\u51fa\u4f86\uff0c\u5b50\u6a39 NP \u5728\u8cc7\u6599\u5eab\u4e2d\u641c\u5c0b\u5230\u76f8\u540c\u7684\u7bc4\u4f8b\u6a39\uff0c\u5982\u5716\u516d\u7684\u7bc4\u4f8b\u6a39\u5c31\u662f\u6240\u641c\u5c0b \u5230\u7684\u5339\u914d\u5b50\u6a39\uff0c\u56e0\u6b64\u5c07\u7bc4\u4f8b\u6a39\u7684 ORDER \u6a19\u8a18\u4e0a\u53bb\uff0c\u6a19\u8a18\u5f8c\u7684\u5256\u6790\u6a39\u5c07\u5982\u5716\u516b\u6240\u793a\u3002\u6b64 \u6642\u4f47\uf99c\u4e2d\u5df2\u7d93\u70ba\u7a7a\uff0c\u641c\u5c0b\u7bc4\u4f8b\u6a39\u7684\u6d41\u7a0b\u5230\u6b64\u70ba\u6b62\u3002 \u6a19\u8a18\u5b8c ORDER \u4e4b\u5f8c\uff0c\u5c07\u6c92\u6709\u6a19\u8a18\u7684\u5b50\u6a39\u4f5c\u4fee\u526a\uff0c\u4e5f\u5c31\u662f\u5c07\u4e0d\u7528\u4f5c\u8a5e\u5e8f\u4ea4\u63db\u7684\u5b50\u6a39 \u4fee\u526a\u5230\u6700\u5c0f\u5c64\u6a39\u3002\u5982\u5716\u516b\u7bc0\u9ede S \u7684\u53f3\u5b50\u6a39\u3001NP[2]\u548c NP[1]\u7684\u5b50\u6a39\u7686\u4e0d\u9700\u8981\u4f5c\u8a5e\u5e8f\u4ea4\u63db\uff0c \u56e0\u6b64\u4fee\u526a\u7684\u7d50\u679c\u70ba\"(S(NP The graph)(VP(VBZ shows)(NP(NP[2] the heights)(PP[1](IN[2] of)(NP[1] four girls))))) \"\uff0c\u5982\u5716\u4e5d\u6240\u793a\u3002\u6700\u5f8c\u5f9e\u5c64\u6578\u6700\u5927\u7684\u6bcf\u500b\u5144\u5f1f\u7bc0\u9ede\u958b\u59cb\u9010\u5c64\u5f80\u4e0a \u4f9d\u7167\u512a\u5148\u6b0a\u9806\u5e8f\u8abf\u6574\u5256\u6790\u6a39\u7684\u7d50\u69cb\uff1b\u8abf\u6574\u5f8c\u7684\u7d50\u679c\u5c07\u6703\u8f38\u5165\u5230\u7ffb\u8b6f\u6a21\u7d44\u7522\u751f\u7ffb\u8b6f\u3002\u82e5\u6211 \u5011\u76f4\u63a5\u53d6\u4f86\u6e90\u53e5\u5256\u6790\u6a39\u7684\u8449\u5b50\u7bc0\u9ede\u4f5c\u7ffb\u8b6f\uff0c\u5c07\u6703\u6210\u70ba\u55ae\u5b57\u5f0f\u7684\u7ffb\u8b6f\uff0c\u6211\u5011\u5c07\u7121\u6cd5\u5c0d\u8a5e\u7d44 \u6216\u7247\u8a9e\u4f5c\u7ffb\u8b6f\u3002\u7ffb\u8b6f\u7684\u90e8\u5206\u6703\u5728\u4e0b\u4e00\u7bc0\u6703\u4f5c\u8a73\u7d30\u8aaa\u660e\u3002 \u5716\u4e5d\u7684\u5256\u6790\u6a39\u6709\u56db\u5c64\uff0c\u9996\u5148\u5c07\u7b2c\u56db\u5c64\u7684\u5144\u5f1f\u7bc0\u9ede\"(IN[2] of)(NP[1] four girls)\"\uff0c\u4f9d\u7167 ORDER \u7684\u9806\u5e8f\u8abf\u6574\u5f8c\u7684\u9806\u5e8f\u70ba\"(NP[1] four girls) (IN[2] of) \"\uff0c\u63a5\u4e0b\u4f86\u7b2c\u4e09\u5c64\u7684\u5144\u5f1f\u7bc0\u9ede \"(NP [2] the heights)(PP[1] (NP[1] four girls)(IN[2] of))\"\u4ea4\u63db\u5f8c\u7684\u9806\u5e8f\u70ba\"(PP[1] (NP[1] four girls)(IN[2] of)) (NP [2] the heights)\uff0c\u6b64\u4f8b\u5b50\u63a5\u4e0b\u4f86\u8a5e\u5e8f\u6c92\u6709\u518d\u8abf\u52d5\uff0c\u5982\u5716\u5341\u6240\u793a\uff1b \u6700\u5f8c\u8f38\u5165\u7ffb\u8b6f\u6a21\u7d44\u7684\u9806\u5e8f\u70ba\"The graph\"\u3001\"shows\"\u3001\"four girls\"\u3001\"of\"\u3001\"the heights\"\uff0c\u7531 \u6b64\u9806\u5e8f\u5206\u5225\u4f5c\u7ffb\u8b6f\u8655\uf9e4\u3002 \u5716\u516b\u3001\u5b8c\u6210 ORDER \u6a19\u8a18 \u5716\u4e5d\u3001\u5256\u6790\u6a39\u4fee\u526a\u5f8c\u7684\u7d50\u679c \u5716\u5341\u3001\u8abf\u6574\u8a5e\u5e8f\u5f8c\u7684\u7d50\u679c 3.4 \u7ffb\u8b6f\u8655\uf9e4 \u7d93\u904e\u4e0a\u4e00\u7bc0\u8655\uf9e4\u6700\u5f8c\u5f97\u5230\u4fee\u526a\u6a39\uff0c\u4fee\u526a\u6a39\u7684\u8449\u5b50\u7bc0\u9ede\u53ef\u80fd\u70ba\u82f1\u6587\u55ae\u5b57(word)\u3001\u8a5e\u7d44(term)\u3002\u8a5e\u7d44\u5373\u70ba\u6578\u500b\u55ae\u5b57\u7d50\u5408\u7684\u5b57\u4e32\uff0c\u4e0d\u4e00\u5b9a\u70ba\u5b8c\u6574\u7684\u53e5\u5b50\uff0c\u5982 \"would be left on the floor\"\u6216\u7247 \u8a9e(phrase\uff0c\u5982\u540d\u8a5e\u7247\u8a9e\u3001\u52d5\u8a5e\u7247\u8a9e\u3001\u5f62\u5bb9\u8a5e\u7247\u8a9e\u7b49) \uff0c\u5982\"in order to\"\u3002\u5728\u7ffb\u8b6f\u8655\uf9e4\u4e0a\u6703 shown in diagram\u2026\"\uff0c\u540c\u6642\u6eff \u8db3\u898f\u5247\u8a5e\u5178\u6a94\u5167\u7684\"as shown in diagram\"\u548c\"in diagram\"\u7247\u8a9e\u53e5\u578b\uff0c\u5247\u6211\u5011\u6703\u9078\u64c7\u9577\u5ea6\u8f03 \u9577\u7684\"as shown in diagram\"\u800c\u4e0d\u662f\u9078\u64c7\"in diagram\"\u52a0\u4e0a\"as show\"\u4f5c\u70ba\u65b7\u8a5e\u7684\u7d50\u679c\u3002",
                "num": null,
                "type_str": "figure"
            },
            "TABREF0": {
                "html": null,
                "num": null,
                "text": "The International Association for the Evaluation of Education Achievement, \u4ee5\u4e0b\u7c21\u7a31 IEA)[20]\u4e3b\u8981\u76ee\u7684\u5728\u65bc\u4e86\u89e3\u5404\u570b\u5b78\u751f\u6578\u5b78\u53ca\u79d1\u5b78(\u542b\u7269\uf9e4\u3001\u5316 \u5b78\u3001\u751f\u7269\u3001\u53ca\u5730\u7403\u79d1\u5b78)\u65b9\u9762\u5b78\u7fd2\u6210\u5c31\u3001\u6559\u80b2\u74b0\u5883\u7b49\uff0c\u5f71\u97ff\u5b78\u751f\u7684\u56e0\u7d20\uff0c\u627e\u51fa\u95dc\u806f\u6027\uff0c \u4e26\u5728\u570b\u969b\u9593\u76f8\u4e92\u4f5c\u6bd4\u8f03\u3002\u81ea 1970 \u5e74\u8d77\u958b\u59cb\u7b2c\u4e00\u6b21\u570b\u969b\u6578\u5b78\u8207\u79d1\u5b78\u6559\u80b2\u6210\u5c31\u8abf\u67e5\u5f8c\uff0c\u4e16 \u754c\u5404\u570b\u9010\u6f38\u5c0d\u570b\u969b\u6578\u5b78\u8207\u79d1\u5b78\u6559\u80b2\u6210\u5c31\u7814\u7a76\u611f\u5230\u8208\u8da3\uff0cIEA \u4fbf\u5728 1995 \u5e74\u958b\u59cb\u6bcf\u56db\u5e74\u8fa6 \uf9e4\u570b\u969b\u6578\u5b78\u8207\u79d1\u5b78\u6559\u80b2\u6210\u5c31\u7814\u7a76\u4e00\u6b21\uff0c\u7a31\u70ba\u570b\u969b\u6578\u5b78\u8207\u79d1\u5b78\u6559\u80b2\u6210\u5c31\u8da8\u52e2\u8abf\u67e5(Trends in International Mathematics and Science Study\uff0c\u4ee5\u4e0b\u7c21\u7a31 TIMSS )\uff0c\u81f3\u4eca\u5df2\u8fa6\uf9e4\u904e 1995\u3001 1999\u30012003 \u548c 2007 \u5171\u56db\u5c46\uff0c\u5171\u6709 38 \u500b\u570b\u5bb6\u53c3\u52a0\u3002 \u6211\u570b\u65bc 1999 \u5e74\u958b\u59cb\u52a0\u5165 TIMSS \u5f8c\uff0c\u7531\u570b\u79d1\u6703\u59d4\u8a17\u570b\u7acb\u53f0\u7063\u5e2b\u7bc4\u5927\u5b78\u79d1\u5b78\u6559\u80b2\u4e2d\u5fc3 (\u4ee5\u4e0b\u7c21\u7a31\u5e2b\u5927\u79d1\u6559\u4e2d\u5fc3)\u8ca0\u8cac\u8a74\u984c\u7ffb\u8b6f\u53ca\u6e2c\u9a57\u5de5\u4f5c\u30021999 \u5e74\u7684\u8abf\u67e5\u5c0d\u8c61\u53ea\u6709\u570b\u4e2d\u4e8c\u5e74\u7d1a \u5b78\u751f\uff0c 2003 \u5e74\u7684\u8abf\u67e5\u5c0d\u8c61\u5305\u62ec\u56db\u5e74\u7d1a\u53ca\u516b\u5e74\u7d1a\u5b78\u751f\u3002\u7ffb\u8b6f\u8a74\u984c\u4e3b\u8981\u7684\u6d41\u7a0b\u5305\u542b\uff1a\u5f9e IEA \u53d6\u5f97\u8a74\u984c\u5167\u5bb9\uff0c\u7531\u5e2b\u5927\u79d1\u6559\u4e2d\u5fc3\u6c7a\u8b70\u9032\u884c\u7ffb\u8b6f\u5de5\u4f5c\u5206\u914d\u3001\u4e2d\u6587\u8a74\u984c\u4ea4\u63db\u5be9\u7a3f\u6821\u6b63\u53ca\u7ffb\u8b6f \u554f\u984c\u8a0e\u8ad6\uff0c\u6700\u5f8c\u5c07\u4e2d\u6587\u7ffb\u8b6f\u8a74\u984c\u5b9a\u7a3f\u3002\u81f3\u76ee\u524d\u70ba\u6b62\uff0c\u5e2b\u5927\u79d1\u6559\u4e2d\u5fc3\u5df2\u5c07 1999 \u548c 2003 \u5e74\u8a74\u984c\u5167\u5bb9\u548c\u8a55\u91cf\u7d50\u679c\uff0c\u516c\u5e03\u65bc\u53f0\u7063 TIMSS \u5b98\u65b9\u7db2\u7ad9[21] \uff0c\u4ee5\u63d0\u4f9b\u7814\u7a76\u4e4b\u53c3\u8003\u3002\u5728 TIMSS",
                "content": "<table><tr><td>\u5f59\uff0c\u7ffb\u8b6f\u7684\u7d50\u679c\u53ef\u80fd\u6703\u4e0d\u7b26\u5408\u4e00\u822c\u4eba\u7684\u7528\u8a5e\u9806\u5e8f\u3002\u53e6\u5916\u4e2d\u6587\u7684\u81ea\u7531\u5ea6\u8f03\u9ad8\uff0c\u5f88\u5bb9\u6613\u9020\u6210 \u6a19\u53e5\u7684\u8868\u793a\u5f0f\uff1b\u5408\u6210\u968e\u6bb5\u5c07\u76ee\u6a19\u53e5\u7684\u8868\u793a\u5f0f\u5c55\u958b\u70ba\u76ee\u6a19\u53e5\u7684 word-dependency tree\uff0c\u4e26 \u672c\u7cfb\u7d71\u7684\u67b6\u69cb\u5982\u5716\u4e00\u6240\u793a\u3002\u6211\u5011\u91dd\u5c0d\u7bc4\u4f8b\u6a39\u7522\u751f\u7cfb\u7d71\u548c\u82f1\u6587\u53e5\u7ffb\u8b6f\u7cfb\u7d71\u9019\u5169\u90e8\u4efd\u5206\u5225\u7c21 3.1 \u96d9\u8a9e\u6a39\u5c0d\u61c9\u5b57\u4e32\u7684\u7d50\u69cb(BSSTC)</td></tr><tr><td>\u7ffb\u8b6f\u4e0a\u7528\u8a5e\u9806\u5e8f\u7684\u4e0d\u540c\u3002\u4f8b\u5982\uff1a\"\u4e0b\u5716\u986f\u793a\u67d0\u4e00\u500b\u570b\u5bb6\u6240\u7a2e\u7a40\u7269\u7684\u5206\u5e03\u5716\"\uff0c\u4e5f\u53ef\u7ffb\u8b6f\u70ba \u4e14\u8f38\u51fa\u7ffb\u8b6f\u7d50\u679c\u3002Al-Adhaileh \u7b49\u5b78\u8005[5]\u5c07 structured string tree correspondence(SSTC) [7] \u4ecb\u5982\u4e0b\u3002</td></tr><tr><td>\"\u67d0\u4e00\u500b\u570b\u5bb6\u6240\u7a2e\u7a40\u7269\u7684\u5206\u5e03\u5716\uff0c\u5982\u4e0b\u5716\u986f\u793a\"\u3002\u53ef\u80fd\u6703\u5f71\u97ff\u5230\u53d7\u6e2c\u8005\u7684\u601d\u7dd2\uff0c\u4f7f\u4f5c\u7b54\u6642 \u904b\u7528\u5728\u82f1\u6587\u7ffb\u8b6f\u6210\u99ac\u4f86\u897f\u4e9e\u6587\u4e0a\uff0cSSTC \u662f\u4e00\u7a2e\u80fd\u5c07\u82f1\u6587\u5c0d\u61c9\u99ac\u4f86\u897f\u4e9e\u6587\u7684\u7d50\u69cb\uff0c\u4f46\u6b64 \uf06c \u7bc4\u4f8b\u6a39\u7522\u751f\u7cfb\u7d71\uff1a \u9019\u500b\u7cfb\u7d71\uf9dd\u7528\u4e2d\u82f1\u5e73\u884c\u8a9e\u6599\uff0c\u9019\u88e1\u7684\u4e2d\u82f1\u5e73\u884c\u8a9e\u6599\u5fc5\u9700\u8981\u4e00\u53e5</td></tr><tr><td>\u7c97\u5fc3\u7684\u60c5\u5f62\u6703\u589e\u52a0\u3002\u56e0\u6b64\uff0c\u82e5\u80fd\uf9dd\u7528\u6a5f\u5668\u7ffb\u8b6f(machine translation)\u7684\u6280\u8853\u4f86\u8f14\u52a9\u7ffb\u8b6f\u4ee5 \u7d50\u69cb\u4e26\u6c92\u6709\u89e3\u6c7a\u8a5e\u5e8f\u4ea4\u63db\u7684\u554f\u984c\u3002\u76ee\u524d\u8f03\u5b8c\u6574\u7684 EBMT \u7cfb\u7d71\u6709 Liu \u7b49\u5b78\u8005\u6240\u63d0\u51fa \u82f1\u6587\u53e5\u5c0d\u61c9\u4e00\u53e5\u4e2d\u6587\u53e5\uff0c\u4e14\u6bcf\u4e00\u7d44\u4e2d\u82f1\u6587\u53e5\u90fd\u8981\u662f\u4e92\u70ba\u7ffb\u8b6f\u7684\u53e5\u5b50\u3002\u4e2d\u6587\u53e5\u7d93\u904e\u65b7</td></tr><tr><td>\u53ca\u8abf\u6574\u8a5e\u5e8f\uff0c\u4ee5\u671f\u63d0\u9ad8\u7ffb\u8b6f\u7684\u54c1\u8cea\u548c\u6548\u7387\u3002 tree-string correspondence (TSC)\u7d50\u69cb\u548c\u7d71\u8a08\u5f0f\u6a21\u578b\u6240\u7d44\u6210\u7684 EBMT \u7cfb\u7d71[13] \uff0c\u5728\u6bd4\u5c0d TSC \u8a5e\u8655\uf9e4\u5f8c\uff0c\u88ab\u65b7\u6210\u6578\u500b\u4e2d\u6587\u8a5e\u5f59\uff0c\u4ee5\u7a7a\u767d\u9694\u958b\uff1b\u82f1\u6587\u53e5\u7d93\u904e\u82f1\u6587\u5256\u6790\u5668\u5efa\u6210\u82f1\u6587\u5256</td></tr><tr><td>\u5728\u4eba\u5de5\u667a\u6167\u9818\u57df\uff0c\u6a5f\u5668\u7ffb\u8b6f\u662f\u4e00\u500b\u5f88\u56f0\u96e3\u7684\u554f\u984c\u3002\u6a5f\u5668\u7ffb\u8b6f\u662f\u6307\u5c07\u4e00\u7a2e\u81ea\u7136\u8a9e\u8a00\u7d93 \u7d50\u69cb\u7684\u6a5f\u5236\u662f\u8a08\u7b97\u4f86\u6e90\u53e5\u5256\u6790\u6a39\u548c TSC \u6bd4\u5c0d\u7684\u5206\u6578\uff0c\u7522\u751f\u7ffb\u8b6f\u7684\u662f\u7531\u4f86\u6e90\u8a5e\u5f59\u7ffb\u8b6f\u6210 \u6790\u6a39\u3002\u5c07\u65b7\u8a5e\u5f8c\u7684\u7d50\u679c\u548c\u82f1\u6587\u5256\u6790\u6a39\u7d93\u904e\u5256\u6790\u6a39\u5c0d\u61c9\u5b57\u4e32\u6a21\u7d44\u8655\uf9e4\uff0c\u5efa\u6210\u82f1\u6587\u5256\u6790</td></tr><tr><td>\u6458\u8981 \u904e\u96fb\u8166\u904b\u7b97\u7ffb\u8b6f\u6210\u53e6\u4e00\u7a2e\u8a9e\u8a00\uff0c\u56f0\u96e3\u7a0b\u5ea6\u4e5f\u8ddf\u4f86\u6e90\u53e5\u548c\u76ee\u6a19\u53e5\u6709\u95dc\uff0c\u50cf\u662f\u82f1\u6587\u548c\u8461\u8404\u7259 \u76ee\u6a19\u8a5e\u5f59\u7684\u6a5f\u7387\u548c\u76ee\u6a19\u53e5\u7684\u8a9e\u8a00\u6a21\u578b\u6240\u7d44\u6210\u3002 \u6a39\u5c0d\u61c9\u5b57\u4e32\u7684\u7d50\u69cb\u6a39\uff0c\u6b64\u7d50\u69cb\u6a39\u7a31\u70ba\u7bc4\u4f8b\u6a39\u3002\u518d\u5c07\u6bcf\u500b\u7bc4\u4f8b\u6a39\u53d6\u51fa\u5b50\u6a39\uff0c\u4e26\u4e14\u5224\u65b7</td></tr><tr><td>\u6587\u8a9e\u8a00\u7684\u7279\u6027\u8f03\u76f8\u8fd1\uff0c\u8f03\u5bb9\u6613\u7ffb\u8b6f\u3002\u800c\u4e2d\u6587\u8ddf\u82f1\u6587\u8a5e\u5e8f\u5dee\u7570\u5f88\u5927\uff0c\u4e14\u4e2d\u6587\u6bd4\u8f03\u6c92\u6709\u7279\u5b9a \u662f\u5426\u6709\u8a5e\u5e8f\u4ea4\u63db\uff0c\u5c07\u9700\u8981\u8a5e\u5e8f\u4ea4\u63db\u7684\u7bc4\u4f8b\u6a39\u5168\u90e8\u5b58\u5165\u7bc4\u4f8b\u6a39\u8cc7\u6599\u5eab\u4e2d\u65b9\u4fbf\u641c\u5c0b\u3002 \u9ec3\u8f1d\u7b49\u5b78\u8005\u6240\u63d0\u51fa\u7684 translation corresponding tree (TCT) [24]\uff0cTCT \u662f\u91dd\u5c0d\u82f1\u6587\u7ffb\u8b6f \u672c\u8ad6\u6587\u61c9\u7528\u4ee5\u7bc4\u4f8b\u70ba\u57fa\u790e\u7684\u6a5f\u5668\u7ffb\u8b6f\u6280\u8853\uff0c\u61c9\u7528\u82f1\u6f22\u96d9\u8a9e\u5c0d\u61c9\u7684\u7d50\u69cb\u8f14\u52a9\u82f1\u6f22\u55ae\u53e5\u8a9e\u6599 \u7684\u8a9e\u6cd5\uff0c\u5beb\u6cd5\u8f03\u81ea\u7531\uff0c\u5c0d\u7ffb\u8b6f\u4f86\u8aaa\u8f03\u70ba\u56f0\u96e3\u3002\u6a5f\u5668\u7ffb\u8b6f\u767c\u5c55\u81f3\u4eca\u5df2\u7d93\u8d85\u904e 50 \u5e74\u3002Dorr \u6210\u8461\u8404\u7259\u6587\u7684\u7cfb\u7d71\uff0c\u5728 TCT \u7d50\u69cb\u4e0a\u53ef\u4ee5\u8a18\u9304\u4f86\u6e90\u53e5\u8a5e\u5f59\u548c\u76ee\u6a19\u53e5\u8a5e\u5f59\u5c0d\u61c9\u7684\u95dc\u4fc2\u3001\u4f86 \uf06c \u82f1\u6587\u53e5\u7ffb\u8b6f\u7cfb\u7d71\uff1a\u7576\u8f38\u5165\u82f1\u6587\u53e5\u5f8c\uff0c\u5148\u5c07\u53e5\u5b50\u900f\u904e\u82f1\u6587\u5256\u6790\u5668\uff0c\u5efa\u6210\u82f1\u6587\u5256\u6790\u6a39\u3002 \u7684\u7ffb\u8b6f\u3002\u7ffb\u8b6f\u7bc4\u4f8b\u662f\u904b\u7528\u4e00\u7a2e\u7279\u6b8a\u7684\u7d50\u69cb\uff0c\u6b64\u7d50\u69cb\u5305\u542b\u4f86\u6e90\u53e5\u7684\u5256\u6790\u6a39\u3001\u76ee\u6a19\u53e5\u7684\u5b57\u4e32\u3001 \u7b49\u5b78\u8005[9] \u5c07\u73fe \u5728\u6a5f \u5668 \u7ffb\u8b6f\u4f9d \u64da\u7cfb \u7d71\u8655 \uf9e4\u7684 \u65b9\u5f0f\u4f86 \u5206 \u985e \uff0c\u5206 \u6210 \u4ee5 \u8a9e\u8a00\u5b78 \u70ba\u57fa \u7ffb\u8b6f \u6e90\u53e5\u8a5e\u5f59\u548c\u76ee\u6a19\u53e5\u8a5e\u5f59\u5c0d\u61c9\u7684\u7ffb\u8b6f\u7d50\u679c\u548c\u8a5e\u5e8f\uff0c\u4f46\u662f TCT \u662f\u4e8c\u5143\u7684\u5256\u6790\u6a39\uff0c\u4e5f\u5c31\u662f\u6bcf \u6709\u4e86\u82f1\u6587\u5256\u6790\u6a39\u5c31\u53ef\u4ee5\u900f\u904e\u641c\u5c0b\u7bc4\u4f8b\u6a39\u6a21\u7d44\uff0c\u6a19\u8a18\u82f1\u6587\u5256\u6790\u6a39\u4e0a\u9700\u8981\u8abf\u52d5\u8a5e\u5e8f\u7684\u7d50 \u4ee5\u53ca\u76ee\u6a19\u53e5\u548c\u4f86\u6e90\u53e5\u8a5e\u5f59\u5c0d\u61c9\u95dc\u4fc2\u3002\u5c07\u7ffb\u8b6f\u7bc4\u4f8b\u5efa\u7acb\u8cc7\u6599\u5eab\uff0c\u4ee5\u63d0\u4f9b\u4f86\u6e90\u53e5\u4f5c\u8a5e\u5e8f\u4ea4\u63db (linguistic-based paradigms)\uff0c\u4f8b\u5982\u57fa\u65bc\u77e5\u8b58(knowledge-based)\u548c\u57fa\u65bc\u898f\u5247(rule-based)\u7b49\uff1b \u500b\u7bc0\u9ede\u90fd\u53ea\u6709\u5169\u9846\u5b50\u6a39\uff0c\u5728 TCT \u4e0a\u8a5e\u5e8f\u53ea\u7528\u5e03\u6797\u503c(boolean value)\u4f86\u8a18\u9304\uff0c\u6240\u4ee5 TCT \u53ea \u69cb\uff0c\u4e26\u4f9d\u7167\u6240\u6a19\u8a18\u7684\u8a5e\u5e8f\u4f5c\u8abf\u6574\u3002\u8a5e\u5e8f\u8abf\u6574\u5b8c\u6210\u5f8c\u518d\u5c07\u82f1\u6587\u7d50\u69cb\u6a39\u4e2d\u7684\u82f1\u6587\u55ae\u5b57\u6216 \u7684\u4f9d\u64da\uff0c\u6700\u5f8c\u900f\u904e\u5b57\u5178\u7ffb\u8b6f\uff0c\u4ee5\u53ca\uf9dd\u7528\u7d71\u8a08\u5f0f\u4e2d\u82f1\u8a5e\u5f59\u5c0d\uf99c\u548c\u8a9e\u8a00\u6a21\u578b\u4f86\u9078\u8a5e\uff0c\u7522\u751f\u5efa \u4ee5\u53ca\u975e\u8a9e\u8a00\u5b78\u70ba\u57fa\u7ffb\u8b6f(non-linguistic-based paradigms) \uff0c\u4f8b\u5982\u57fa\u65bc\u7d71\u8a08(statistical-based) \u80fd\u904b\u7528\u5728\u4e8c\u5143\u5256\u6790\u6a39\u4e0a\u3002\u4f46\u662f\u6709\u4e9b\u5256\u6790\u5668\u6240\u7522\u751f\u5256\u6790\u6a39\u662f\u591a\u5143\u6a39\uff0c\u56e0\u6b64\u6211\u5011\u63d0\u51fa\u96d9\u8a9e\u6a39 \u7247\u8a9e\u900f\u904e\u7ffb\u8b6f\u6a21\u7d44\u505a\u7ffb\u8b6f\u3002\u5176\u4e2d\u7ffb\u8b6f\u6a21\u7d44\u5305\u542b\u4e86\u5927\u5c0f\u5beb\u8f49\u63db\u3001\u65b7\u8a5e\u8655\uf9e4\u3001stop word \u8b70\u7684\u7ffb\u8b6f\u3002\u6211\u5011\u662f\u4ee5 2003 \u5e74\u570b\u969b\u6578\u5b78\u8207\u79d1\u5b78\u6559\u80b2\u6210\u5c31\u8da8\u52e2\u8abf\u67e5\u6e2c\u9a57\u8a74\u984c\u70ba\u4e3b\u8981\u7ffb\u8b6f\u7684 \u548c\u57fa\u65bc\u7bc4\u4f8b(example-based)\u7b49\u3002 \u5c0d\u61c9\u5b57\u4e32\u7684\u7d50\u69cb(bilingual structured string tree correspondence\uff0c\u7c21\u7a31\u70ba BSSTC)\u53ef\u4ee5\u904b\u7528 filtering\u53castemming\uff0c\u4e4b\u5f8c\u5c07\u8655\uf9e4\u904e\u7684\u8a5e\u5f59\u900f\u904e\u5b57\u5178\u6a94\u505a\u7ffb\u8b6f[3]\u3002\u6bcf\u500b\u82f1\u6587\u55ae\u5b57\u6216\u7247 \u5c0d\u8c61\uff0c\u4ee5\u671f\u63d0\u5347\u7ffb\u8b6f\u7684\u4e00\u81f4\u6027\u548c\u6548\u7387\u3002\u4ee5 NIST \u548c BLEU \u7684\u8a55\u6bd4\u65b9\u5f0f\uff0c\u4f86\u8a55\u4f30\u548c\u6bd4\u8f03 \u4ee5\u77e5\u8b58\u70ba\u57fa\u790e\u7684\u6a5f\u5668\u7ffb\u8b6f(knowledge-based machine translation)\u7cfb\u7d71\u662f\u904b\u7528\u5b57\u5178\u3001\u6587 \u5728\u591a\u5143\u5256\u6790\u6a39\u4e0a\uff0c\u4e26\u4e14 BSSTC \u53ef\u5728\u7ffb\u8b6f\u904e\u7a0b\u4e2d\u7576\u4f5c\u8a5e\u5e8f\u4ea4\u63db\u7684\u53c3\u8003\uff0c\u6839\u64da\u6211\u5011\u5be6\u9a57\u7d50 \u8a9e\u90fd\u53ef\u80fd\u6709\u4e00\u500b\u4ee5\u4e0a\u7684\u4e2d\u6587\u7ffb\u8b6f\uff0c\u56e0\u6b64\u9700\u8981\u9078\u8a5e\u7684\u6a5f\u5236\u4f86\u7522\u751f\u521d\u6b65\u7ffb\u8b6f\u7d50\u679c\uff0c\u6b64\u7ffb \u7dda\u4e0a\u7ffb\u8b6f\u7cfb\u7d71\u548c\u672c\u7cfb\u7d71\u6240\u9054\u6210\u7684\u7ffb\u8b6f\u54c1\u8cea\u3002 \u6cd5\u898f\u5247\u6216\u662f\u8a9e\u8a00\u5b78\u5bb6\u7684\u77e5\u8b58\u4f86\u5e6b\u52a9\u7ffb\u8b6f\u3002Knight \u7b49\u5b78\u8005[11]\u7d50\u5408 Longman \u5b57\u5178\u3001WordNet \u679c\uff0c\u6211\u5011\u80fd\u6709\u6548\u7684\u8abf\u52d5\u8a5e\u5e8f\uff0c\u4ee5\u63d0\u5347\u7ffb\u8b6f\u7684\u54c1\u8cea\u3002\u5b8c\u6210\u8a5e\u5e8f\u4ea4\u63db\u5f8c\uff0c\u518d\u900f\u904e\u5b57\u5178\u7ffb\u8b6f\u6210 \u8b6f\u7d50\u679c\u5c1a\u9700\u8981\u4eba\u5de5\u4f5c\u5f8c\u7e8c\u7684\u7de8\u4fee\u3002</td></tr><tr><td>\u95dc\u9375\u8a5e\uff1a\u81ea\u7136\u8a9e\u8a00\u8655\uf9e4\uff0c\u8a74\u984c\u7ffb\u8b6f\uff0c\u6a5f\u5668\u7ffb\u8b6f\uff0cTIMSS \u548c Collins \u96d9\u8a9e\u5b57\u5178\u5efa\u7acb\u4e00\u500b\u77e5\u8b58\u5eab\uff0c\u904b\u7528\u5728\u897f\u73ed\u7259\u6587\u7ffb\u8b6f\u6210\u82f1\u6587\u3002\u9019\u7a2e\uf9dd\u7528\u5b57\u5178\u4f86\u5e6b \u52a9\u7ffb\u8b6f\u7684\u7cfb\u7d71\uff0c\u6703\u6709\u4e00\u5b57\u591a\u7fa9\u7684\u60c5\u5f62\u767c\u751f\uff0c\u4e00\u500b\u8a5e\u5f59\u5728\u5b57\u5178\u4e2d\u901a\u5e38\u6709\u4e00\u500b\u4ee5\u4e0a\u7684\u7ffb\u8b6f\u3002 \u4e2d\u6587\uff0c\u6700\u5f8c\u904b\u7528\u7d71\u8a08\u5f0f\u9078\u8a5e\u6a21\u578b\uff0c\u7522\u751f\u4e86\u521d\u6b65\u7ffb\u8b6f\u7d50\u679c\uff0c\u4f46\u672c\u7cfb\u7d71\u5c1a\u5c6c\u65bc\u534a\u81ea\u52d5\u7ffb\u8b6f\u7cfb 3. \u7cfb\u7d71\u76f8\u95dc\u6280\u8853 \u7d71\uff0c\u6545\u9700\u8981\u4eba\u5de5\u52a0\u4ee5\u4fee\u98fe\u7de8\u8f2f\u3002 \u6839\u64da\u4e0a\u4e00\u7bc0\u7cfb\u7d71\u67b6\u69cb\u7684\u63cf\u8ff0\u5206\u70ba\u7bc4\u4f8b\u6a39\u7522\u751f\u7cfb\u7d71\u548c\u82f1\u6587\u53e5\u7ffb\u8b6f\u7cfb\u7d71\u5169\u5927\u7cfb\u7d71\u3002\u7bc4\u4f8b\u7522\u751f 1. \u7dd2\u8ad6 \u4ee5\u82f1\u7ffb\u4e2d\u70ba\u4f8b\"current\"\u9019\u500b\u5b57\u5728\u5b57\u5178\u88e1\u5c31\u6709\u5341\u591a\u7a2e\u4e0d\u540c\u7684\u7ffb\u8b6f\uff0c\u5373\u4f7f\u5c08\u5bb6\u4e5f\u7121\u6cd5\u627e\u51fa\u4e00 \u500b\u7d71\u4e00\u7684\u898f\u5247\uff0c\u5728\u4f55\u7a2e\u60c5\u6cc1\u4e0b\u8981\u7528\u4f55\u7a2e\u7ffb\u8b6f\uff0c\u6240\u4ee5\u5728\u7ffb\u8b6f\u7684\u54c1\u8cea\u548c\u6b63\u78ba\u6027\u4e0a\u5f88\u96e3\u6eff\u8db3\u4f7f \u7528\u8005\u3002\u56e0\u6b64\uff0c\u7ffb\u8b6f\u7cfb\u7d71\u901a\u5e38\u90fd\u6703\u9650\u5b9a\u9818\u57df\u4f86\u6e1b\u5c11\u4e00\u5b57\u591a\u7fa9\uff0c\u4f8b\u5982 current \u5728\u96fb\u5b50\u96fb\u6a5f\u985e \u7684\u6587\u7ae0\u4e2d\u51fa\u73fe\uff0c\u6700\u5e38\u88ab\u7ffb\u8b6f\u70ba\u96fb\u6d41\uff0c\u5728\u6587\u5b78\u985e\u7684\u6587\u7ae0\u4e2d\uff0c\u6700\u5e38\u88ab\u7ffb\u8b6f\u70ba\u73fe\u4ee3\u3002 \u7d71\u8a08\u5f0f\u6a5f\u5668\u7ffb\u8b6f(statistical machine translation\uff0c\u4ee5\u4e0b\u7c21\u7a31 SMT)\u662f\u5c07\u8a9e\u6599\u5728\u7ffb\u8b6f\u4e4b\u524d \u80fd\u3002Brown \u7b49\u5b78\u8005[6]\u65bc 1990 \u5e74\u4ee5\u82f1\u6587\u53ca\u6cd5\u6587\u7684\u96d9\u8a9e\u8a9e\u6599\u70ba\u4f86\u6e90\uff0c\u63d0\u51fa\u7d71\u8a08\u5f0f\u96d9\u8a9e\u7ffb \u8b6f\u67b6\u69cb\u3002\u5047\u8a2d\u76ee\u6a19\u8a9e\u8a00\u70ba T \u53ca\u4f86\u6e90\u8a9e\u8a00\u70ba S\uff0cP(T)\u70ba\u76ee\u6a19\u8a9e\u8a00 T \u5728\u8a9e\u6599\u5eab\u4e2d\u51fa\u73fe\u7684\u6a5f\u7387\uff0c \u7a31\u70ba\u8a9e\u8a00\u6a21\u578b(language model)\uff0cP(S|T)\u70ba\u76ee\u6a19\u8a9e\u8a00 T \u7ffb\u8b6f\u6210\u4f86\u6e90\u8a9e\u8a00 S \u7684\u6a5f\u7387\uff0c\u7a31\u70ba\u7ffb \u8b6f\u6a21\u578b(translation model)\u3002SMT \u7cfb\u7d71\u9700\u8981\u5927\u91cf\u7684\u8a9e\u6599\u5eab\u8f14\u52a9\uff0c\u5927\u591a\u90fd\u9700\u8981\u5177\u5099\u96d9\u8a9e\u5c0d\u61c9 \u7684\u8a9e\u6599\u5eab(parallel corpora \u6216\u7a31 bilingual corpora)\uff0c\u518d\u900f\u904e\u6a5f\u7387\u516c\u5f0f\u8a08\u7b97\u51fa\u6a5f\u7387\u6a21\u578b\u3002\u5176 \u4e2d SMT \u56f0\u96e3\u7684\u5730\u65b9\u5728\u65bc\u9700\u8981\u6536\u96c6\u5927\u91cf\u53ef\u7528\u7684\u96d9\u8a9e\u8a9e\u6599\uff0c\u7576\u8a9e\u6599\u8d8a\u591a\u5efa\u7acb\u6a21\u578b\u6240\u82b1\u8cbb\u7684 \u6642\u9593\u8d8a\u591a\u3002Oct \u7b49\u5b78\u8005[16]\u63d0\u51fa\u55ae\u5b57\u5f0f(word-based)\u7ffb\u8b6f\u6a21\u578b\u904b\u7528\u5728\u8a5e\u5f59\u5c0d\u6e96 (word alignment)\uff0c\u4e26\u4e14\u767c\u5c55\u51fa GIZA++\u9019\u5957\u7cfb\u7d71\u3002Koehn \u7b49\u5b78\u8005[12]\u9032\u4e00\u6b65\u5c07\u55ae\u5b57\u5f0f\u8f49\u8b8a\u6210\u7247 \u8a9e\u5f0f(phrase-based)\u7ffb\u8b6f\u6a21\u578b\uff0c\u904b\u7528\u7247\u8a9e\u5f0f\u7ffb\u8b6f\u6a21\u578b\u7ffb\u8b6f\u7684\u7d50\u679c\u6703\u6bd4\u55ae\u5b57\u5f0f\u7ffb\u8b6f\u7684\u7d50\u679c\u8981 \u6b63\u78ba\u3002 \u9664\u4e86\u672c\u7bc0\u7c21\u55ae\u4ecb\u7d39\u672c\u7814\u7a76\u4ee5\u5916\uff0c\u6211\u5011\u5c07\u5728\u7b2c\u4e8c\u7bc0\u63cf\u8ff0\u6574\u500b\u7cfb\u7d71\u7684\u67b6\u69cb\uff0c\u7b2c\u4e09\u7bc0\u8aaa\u660e \u672c\u7bc7\u8ad6\u6587\u6240\u904b\u7528\u7684\u6280\u8853\uff0c\u7b2c\u56db\u7bc0\u5247\u5448\u73fe\u51fa\u6211\u5011\u7684\u5be6\u9a57\u7d50\u679c\uff0c\u7b2c\u4e94\u7bc0\u5247\u662f\u7d50\u8ad6\u3002 2. \u7cfb\u7d71\u67b6\u69cb \u7531\u65bc\u6211\u5011\u7684\u76ee\u7684\u5728\u65bc\uf9dd\u7528\u4e2d\u82f1\u4e92\u70ba\u7ffb\u8b6f\u7684\u53e5\u5b50\u627e\u51fa\u8a5e\u5e8f\u95dc\u4fc2\uff0c\u4e26\u4e14\u5c07\u82f1\u6587\u53e5\u548c\u4e2d\u6587\u53e5\u8a5e \u4e2d\uff0c\u6b64\u7d50\u69cb\u5c07\u6210\u70ba\u4e4b\u5f8c\u82f1\u6587\u53e5\u7684\u7d50\u69cb\u8abf\u6574\u70ba\u9069\u5408\u4e2d\u6587\u7684\u7d50\u69cb\u7684\u53c3\u8003\u3002\u6700\u5f8c\u518d\u5c07\u82f1\u6587\u8a5e\u5f59 \u7ffb\u8b6f\u6210\u4e2d\u6587\u8a5e\u5f59\uff0c\u4e26\uf9dd\u7528\u7d71\u8a08\u5f0f\u9078\u8a5e\u9078\u51fa\u6700\u6709\u53ef\u80fd\u7ffb\u8b6f\u6210\u7684\u4e2d\u6587\u8a5e\u5f59\uff0c\u8b93\u7ffb\u8b6f\u7684\u7d50\u679c\u66f4 \u7b26\u5408\u4e00\u822c\u4eba\u7684\u7528\u8a5e\u548c\u9806\u5e8f\u3002 \u7522\u751f\u7bc4\u4f8b\u6a39\u7cfb\u7d71 \u4e2d\u82f1\u5e73\u884c\u8a9e\u6599 \u4e2d\u6587\u53e5 \u82f1\u6587\u53e5 \u65b7\u8a5e \u82f1\u6587\u5256\u6790\u5668 \u8655\u7406 \u5256\u6790\u6a39\u5c0d\u61c9\u5b57\u4e32 \u6a21\u7d44 \u7bc4\u4f8b\u6a39\u8cc7\u6599 \u5eab \u67e5\u8a62 \u6a39\u7cfb\u7d71\u7684\u57f7\u884c\u6d41\u7a0b\u70ba\u5148\u5c07\u4e2d\u6587\u53e5\u65b7\u8a5e\u548c\u5256\u6790\u82f1\u6587\u53e5\uff0c\u518d\u5c07\u65b7\u8a5e\u548c\u5256\u6790\u5f8c\u7684\u7d50\u679c\u8f38\u5165\u81f3\u5256 \u6790\u6a39\u5c0d\u61c9\u5b57\u4e32\u6a21\u7d44\uff0c\u4e26\u5c07\u8655\uf9e4\u5f8c\u7684\u7bc4\u4f8b\u6a39\u5b58\u5165\u8cc7\u6599\u5eab\u4e2d\u3002\u82f1\u6587\u53e5\u7ffb\u8b6f\u7cfb\u7d71\u7684\u57f7\u884c\u6d41\u7a0b\u5340 \u5206\u70ba\u4e09\u5927\u90e8\u5206\uff0c\u7b2c\u4e00\u90e8\u5206\u662f\u7531\u641c\u5c0b\u7bc4\u4f8b\u6a39\u6a21\u7d44\uff0c\u5c07\u82f1\u6587\u5256\u6790\u6a39\u8ddf\u7bc4\u4f8b\u6a39\u8cc7\u6599\u5eab\u4f5c\u6bd4\u5c0d\uff0c \u4e26\u4e14\u5c07\u672a\u6bd4\u5c0d\u5230\u7684\u5b50\u6a39\u505a\u4fee\u526a\uff1b\u7b2c\u4e8c\u90e8\u5206\u5c07\u4fee\u526a\u5f8c\u7684\u5256\u6790\u6a39\u8f38\u5165\u5230\u7ffb\u8b6f\u6a21\u7d44\u7ffb\u6210\u4e2d\u6587\uff1b \u7b2c\u4e09\u90e8\u5206\u4ee5\u4e2d\u82f1\u8a5e\u5f59\u5c0d\uf99c\u5de5\u5177\u53ca bi-gram \u8a9e\u8a00\u6a21\u578b\uff0c\u8a08\u7b97\u51fa\u4e2d\u82f1\u8a5e\u5f59\u9593\u6700\u6709\u53ef\u80fd\u4e4b\u7ffb\u8b6f \u570b\u969b\u6559\u80b2\u5b78\u7fd2\u6210\u5c31\u8abf\u67e5\u59d4\u54e1\u6703(\u7684\u8a74\u984c\u5167\u5bb9\u4e0a\uff0c\u4e3b\u8981\u7684\u984c\u578b\u7a2e\u985e\u6709\u9078\u64c7\u984c\u548c\u554f\u7b54\u984c\uff0c\u8a74\u984c\u53e5\u578b\u5927\u591a\u70ba\u76f4\u8ff0\u53e5\u548c\u554f\u53e5\u7d50\u69cb \u5c31\u5df2\u7d93\u904e\u8a08\u7b97\u8f49\u63db\u6210\u7d71\u8a08\u6578\u64da\uff0c\u4e0d\u9700\u8981\u5728\u7ffb\u8b6f\u904e\u7a0b\u4e2d\u4f5c\u9f90\u5927\u7684\u6578\u5b78\u904b\u7b97\uff0c\u80fd\u6709\u8f03\u9ad8\u7684\u6548 \u5e8f\u7684\u8cc7\u8a0a\u5132\u5b58\u5728\u96fb\u8166\u4e2d\uff0c\u5132\u5b58\u7684\u683c\u5f0f\u662f\u5c07\u4e2d\u82f1\u6587\u53e5\u7684\u8a5e\u5e8f\u95dc\u4fc2\u8a18\u9304\u5728\u82f1\u6587\u5256\u6790\u6a39\u7684\u7d50\u69cb \u7d44\u5408\u3002</td></tr><tr><td>\u6240\u7d44\u6210\uff0c\u9078\u64c7\u984c\u5247\u591a\u4e86\u8a98\u7b54\u9078\u9805\u3002 \u4ee5\u7bc4\u4f8b\u70ba\u57fa\u790e\u7684\u6a5f\u5668\u7ffb\u8b6f(example-based machine translation\uff0c\u4ee5\u4e0b\u7c21\u7a31\u70ba EBMT) \u82f1\u6587\u53e5\u7ffb\u8b6f\u7cfb\u7d71</td></tr><tr><td>\u4ee5\u5f80\u4f7f\u7528\u4eba\u5de5\u7ffb\u8b6f\u96d6\u7136\u53ef\u4ee5\u9054\u5230\u5f88\u9ad8\u7684\u7ffb\u8b6f\u54c1\u8cea\uff0c\u4f46\u662f\u9700\u8981\u8017\u8cbb\u76f8\u7576\u591a\u7684\u4eba\u529b\u8cc7\u6e90 \u548c\u6642\u9593\uff0c\u800c\u4e14\u5728\u7ffb\u8b6f\u904e\u7a0b\u4e2d\u4e0d\u540c\u7684\u7ffb\u8b6f\u8005\u6703\u6709\u4e0d\u540c\u7684\u7ffb\u8b6f\u6a19\u6e96(\u4f8b\u5982\uff1a\u76f8\u540c\u7684\u53e5\u5b50\uff0c\u7ffb \u8b6f\u5f8c\u7684\u7d50\u679c\u4e0d\u540c)\uff1b\u76f8\u540c\u7684\u7ffb\u8b6f\u8005\u4e5f\u53ef\u80fd\u5728\u6587\u7ae0\u524d\u5f8c\u7ffb\u8b6f\u65b9\u5f0f\u4e0d\u4e00\u81f4\u800c\u7522\u751f\u8a9e\u610f\u4e0a\u7684\u6df7 \u7684\u76f8\u95dc\u7814\u7a76\u5df2\u6709\u76f8\u7576\u591a\u5e74\u6b77\u53f2\uff0c\u5728 1990 \u5e74\u65e5\u672c\u5b78\u8005 Sato \u548c Nagao[19]\u6240\u63d0\u51fa\u7684 EBMT \u82f1\u6587\u53e5\u8f38\u5165 \u641c\u5c0b\u7bc4\u4f8b\u6a39 \u6a21\u7d44 \u82f1\u6587\u5256\u6790\u5668 \u8655\u7406 \u7ffb\u8b6f\u6a21 \u7d44 \u7d71\u8a08\u5f0f \u4e2d\u6587\u53e5\u8f38\u51fa \u9078\u8a5e \u662f\u5c07\u7ffb\u8b6f\u904e\u7a0b\u5206\u70ba\u5206\u89e3(decomposition)\u3001\u8f49\u63db(transfer)\u548c\u5408\u6210(composition)\u4e09\u6b65\u9a5f\u3002\u5206\u89e3 \u968e\u6bb5\u662f\u5c07\u4f86\u6e90\u53e5\u5230\u7bc4\u4f8b\u5eab\u4e2d\u641c\u5c0b\uff0c\u4e26\u5c07\u6240\u641c\u5c0b\u5230 word-dependency tree \u7576\u4f5c\u4f86\u6e90\u53e5\u7684 \u5716\u4e8c\u3001\u82f1\u6587\u5256\u6790\u6a39 \u5716\u4e09\u3001BSSTC \u7d50\u69cb\u7684\u8868\u793a\u6cd5</td></tr><tr><td>\u6dc6\u3002\u56e0\u6b64\u9593\u63a5\u5f71\u97ff\u8a74\u984c\u96e3\u6613\u7a0b\u5ea6\u3002\u82e5\u76f4\u63a5\u5c07\u82f1\u6587\u8a5e\u5f59\u900f\u904e\u82f1\u6f22\u5b57\u5178\u7ffb\u8b6f\u6210\u76f8\u5c0d\u7684\u4e2d\u6587\u8a5e word-dependency tree\uff0c\u4e26\u4e14\u5f62\u6210\u4f86\u6e90\u53e5\u7684\u8868\u793a\u5f0f\uff1b\u8f49\u63db\u968e\u6bb5\u5c07\u4f86\u6e90\u53e5\u7684\u8868\u793a\u5f0f\u8f49\u63db\u6210\u76ee \u5716\u4e00\u3001\u7cfb\u7d71\u67b6\u69cb\u5716</td></tr></table>",
                "type_str": "table"
            },
            "TABREF1": {
                "html": null,
                "num": null,
                "text": "\uff0c\u5247 STREE \u4ee3\u8868\u7bc0\u9ede VP \u5c0d\u61c9\u4f86\u6e90\u53e5\u7b2c\u4e09\u5230 \u7b2c\u516d\u500b\u5b57 \"were simple in concept\"\uff1bSTC \u4ee3\u8868\"were simple in concept\"\u5c0d\u61c9\u76ee\u6a19\u53e5\u7684\u7b2c\u56db \u6240\u5c0d\u61c9\u8449\u5b50\u7bc0\u9ede\u7684\u7de8\u865f\u6a19\u8a18\u5728 n[STREE//]\u3002\u5982\u5716\u4e09\u7bc0\u9ede NNS \u6240\u5c0d\u61c9\u4f86\u6e90\u53e5\u7684\"experiments\"\u7684\u7de8\u865f\u70ba 2\uff0c\u6545 NNS[STREE//]\u4e2d\u7684 STREE \u6a19\u8a18\u70ba 2\u3002 \u63a5\u8457\u6a19\u8a18\u6700\u5e95\u5c64 n[/STC/]\u7684\u65b9\u6cd5\u662f\u5c0b\u627e\u4e2d\u82f1\u5c0d\u61c9\u53e5\u4e2d\u4e92\u70ba\u7ffb\u8b6f\u7684\u4e2d\u6587\u8a5e\u5f59\u548c\u82f1\u6587\u8a5e\u5f59\uff0c",
                "content": "<table><tr><td>\u4e5f\u5c31\u662f\u8a5e\u5f59\u5c0d\u6e96\u3002\u8a5e\u5f59\u5c0d\u6e96\u82e5\u63a1\u7528\u4eba\u5de5\u65b9\u5f0f\uff0c\u5247\u76f8\u7576\u8017\u6642\u8cbb\u529b\uff0c\u5176\u672c\u8eab\u4e5f\u662f\u4e00\u9805\u56f0\u96e3\u7684</td></tr><tr><td>\u7814\u7a76\u3002\u56e0\u6b64\uff0c\u6211\u5011\u5728\u6b64\u7528\u4e00\u500b\u7c21\u55ae\u7684\u65b9\u6cd5\uff0c\u9996\u5148\u5148\u5c07\u4e2d\u6587\u53e5\u7d93\u904e\u65b7\u8a5e\u8655\uf9e4\uff0c\u9019\u88e1\u6211\u5011\u4f7f</td></tr><tr><td>\u7528\u4e2d\u7814\u9662 CKIP \u65b7\u8a5e\u7cfb\u7d71[1]\uff1b\u5c07\u82f1\u6587\u53e5\u6bcf\u500b\u82f1\u6587\u5b57\u67e5\u5c0b\u5b57\u5178\u6a94\uff0c\u67e5\u5c0b\u5f8c\u53ef\u80fd\u6703\u6709\u8d85\u904e\u4e00</td></tr><tr><td>\u500b\u7684\u4e2d\u6587\u7ffb\u8b6f\uff0c\u5c07\u9019\u4e9b\u4e2d\u6587\u7ffb\u8b6f\u8ddf\u65b7\u8a5e\u5f8c\u7684\u4e2d\u6587\u8a5e\u5f59\u4e00\u500b\u4e00\u500b\u4f5c\u6bd4\u5c0d\uff0c\u5982\u6709\u6bd4\u5c0d\u5230\u5247\u8a8d</td></tr><tr><td>\u5b9a\u4e92\u70ba\u7ffb\u8b6f\uff0c\u4e26\u4e14\u6a19\u8a18 n[/STC/]\u5728\u5256\u6790\u6a39\u4e0a\u3002\u5982\u5716\u4e09\u4f86\u6e90\u53e5\u7684\"experiments\"\u5728\u5b57\u5178\u4e2d\u7684</td></tr><tr><td>\u7ffb\u8b6f\u6709\"\u5be6\u9a57\"\u3001\"\u7d93\u9a57\"\u548c\"\u8a74\u9a57\"\uff0c\u5c07\u9019\u4e09\u500b\u4e2d\u6587\u7ffb\u8b6f\u5230\u76ee\u6a19\u53e5\u53bb\u6bd4\u5c0d\uff0c\u6b64\u4f8b\u5b50\u5c07\u6703\u6bd4\u5c0d</td></tr><tr><td>\u5230\u76ee\u6a19\u53e5\u7b2c\u4e09\u500b\u8a5e\u5f59\"\u5be6\u9a57\"\uff0c\u63a5\u8457\u5c07\u76ee\u6a19\u53e5\"\u5be6\u9a57\"\u7684\u7de8\u865f\u6a19\u8a18\u5728 NNS[2/STC/]\u4e2d\u7684 STC</td></tr><tr><td>\u4e0a\u3002\u6700\u5f8c\u5c07\u6bd4\u5c0d\u5230\u7684\u500b\u6578\u9664\u4ee5\u82f1\u6587\u53e5\u55ae\u5b57\u7684\u500b\u6578\uff0c\u7a31\u70ba\u5c0d\u61c9\u7387\u3002\u6700\u4f73\u60c5\u6cc1\u4e0b\u662f\u6bcf\u500b\u82f1\u6587</td></tr><tr><td>\u55ae\u5b57\u90fd\u6709\u76f8\u5c0d\u61c9\u7684\u4e2d\u6587\u7ffb\u8b6f\uff0c\u5c0d\u61c9\u7387\u70ba 1\uff1b\u6700\u5dee\u7684\u60c5\u6cc1\u4e0b\u6bcf\u500b\u82f1\u6587\u55ae\u5b57\u90fd\u6c92\u6709\u76f8\u5c0d\u61c9\u7684</td></tr><tr><td>(6)\"\u3002\u4e2d\u82f1\u5c0d\u61c9\u53e5\u90fd\u6a19\u865f\u5f8c\uff0c\u4ee5\u6a19\u865f\u70ba\u55ae\u4f4d\u958b\u59cb\u505a\u8a5e\u5f59\u5c0d\u6e96(word alignment)\uff0c\u4e26\u6a19\u8a18\u5728</td></tr><tr><td>\u5256\u6790\u6a39\u7684\u7bc0\u9ede\u4e0a\u3002\u5256\u6790\u6a39\u662f\u7528\u6587\u6cd5\u7d50\u69cb\u4f86\u5206\u5c64\uff0c\u4e0d\u540c\u5c64\u7bc0\u9ede\u80fd\u5c0d\u61c9\u5230\u4e0d\u540c\u7684\u7bc4\u570d\u7684\u76ee\u6a19</td></tr><tr><td>\u53e5\u5b57\u4e32\u3002n[STREE/STC/]\u82e5\u70ba VP[3-6/4-6/]\u5230\u7b2c\u516d\u500b\u5b57\"\u6982\u5ff5\u5f88\u7c21\u55ae\"\u3002n C(n) [STREE/STC/ORDER]\u7684\u5144\u5f1f\u7bc0\u9ede(sibling node)\u82e5\u70ba</td></tr><tr><td>JJ[4/6/2]\u548c PP[5-6/4/1] \uff0c\u6211\u5011\u53ef\u4ee5\u89c0\u5bdf\u5230 JJ \u7684 ORDER \u5927\u65bc PP \u7684 ORDER\uff0c\u6545 PP[5-6/4/1]</td></tr><tr><td>\u7684\u4e2d\u6587\u5c0d\u61c9\u300c\u6982\u5ff5\u300d\u5728 JJ[4/6/2] \u7684\u4e2d\u6587\u5c0d\u61c9\u300c\u7c21\u55ae\u300d\u4e4b\u524d\u3002</td></tr><tr><td>3.2 \u5efa\u7acb BSSTC \u7d50\u69cb\u548c\u7522\u751f\u7bc4\u4f8b\u6a39</td></tr><tr><td>\u5efa\u7acb BSSTC \u7d50\u69cb\u5fc5\u9700\u8981\u6709\u82f1\u6587\u8ddf\u4e2d\u6587\u4e92\u70ba\u7ffb\u8b6f\u7684\u53e5\u5b50\uff0c\u5efa\u69cb\u7684\u9806\u5e8f\u662f\u5f9e\u6700\u5e95\u5c64\u4e5f\u5c31\u662f</td></tr><tr><td>\u5c64\u6578\u6700\u5927\u7684\u958b\u59cb\u6a19\u8a18\uff0c\u518d\u4e00\u5c64\u4e00\u5c64\u5f80\u4e0a\u5efa\u7f6e\u5230\u7b2c 0 \u5c64\u70ba\u6b62\uff0c\u6a19\u8a18\u53c3\u6578\u9806\u5e8f\u662f\u5148\u5c07\u6240\u6709\u7bc0</td></tr></table>",
                "type_str": "table"
            },
            "TABREF3": {
                "html": null,
                "num": null,
                "text": "(\u4ee5\u4e0b\u7c21\u7a31\u570b\u4e2d\u88dc\u5145\u8cc7\u6599\u984c\u5eab)\u53ca\u79d1\u5b78\u4eba\u96dc\u8a8c\u3002\u570b\u4e2d\u88dc\u5145\u8cc7\u6599\u984c\u5eab\u4ee5\u4eba\u5de5\u65b9\u5f0f\u5b8c\u6210\u4e2d \u82f1\u8a9e\u53e5\u5c0d\uf99c(sentence alignment) \uff0c\u518d\u7d93\u904e\u7bc4\u4f8b\u6a39\u7684\u7be9\u9078\u9580\u6abb\u503c\u70ba 0.6 \u7684\u60c5\u6cc1\u4e0b\u6709 565 \u53e5\u3002 \u7528\u4f86\u8a13\u7df4\u9078\u8a5e\u6a5f\u7387\u6a21\u578b\u7684\u4f86\u6e90\u6709\u81ea\u7531\u6642\u5831\u4e2d\u82f1\u5c0d\u7167\u8b80\u65b0\u805e\u53ca\u79d1\u5b78\u4eba\u96dc\u8a8c\u3002\u81ea\u7531\u6642\u5831 \u4e2d\u82f1\u5c0d\u7167\u8b80\u65b0\u805e\u5f9e 2005 \u5e74 2 \u6708 14 \u65e5\u81f3 2007 \u5e74 10 \u6708 31 \u65e5\uff0c\u800c\u81ea\u7531\u6642\u5831\u4e2d\u82f1\u5c0d\u7167\u8b80\u65b0 \u805e\u672c\u8eab\u5c31\u5df2\u7d93\u4f5c\u597d\u4e2d\u82f1\u8a9e\u53e5\u5c0d\uf99c\u3002\u79d1\u5b78\u4eba\u96dc\u8a8c\u662f\u5f9e 2002 \u5e74 3 \u6708\u5275\u520a\u865f\u81f3 2006 \u5e74 12 \u6708 NIST \u6a19\u6e96\u7684 mteval-10\uff0c\u4e26\u4e14\u6211\u5011\u5c07\u53c3\u8003 \u7684\u4e2d\u6587\u6a19\u6e96\u7ffb\u8b6f\u548c\u7cfb\u7d71\u5efa\u8b70\u7ffb\u8b6f\uff0c\u6bcf\u500b\u4e2d\u6587\u5b57\u8ddf\u4e2d\u6587\u5b57\u4e4b\u9593\u7528\u7a7a\u767d\u4f5c\u5206\u9694\uff0c\u8a08\u7b97\u51fa\u5404\u5225 n-gram \u53ca\u7d2f\u52a0\u5404\u500b n-gram \u7684 BLEU \u53ca NIST \u503c\u3002\u4e3b\u8981\u8a55\u4f30\u7684\u5c0d\u8c61\u6709 Google \u7dda\u4e0a\u7ffb\u8b6f\u3001 \u70ba\u4ee3\u865f\uff1b\u79d1\u5b78\u9818\u57df\u4ee5 S \u70ba\u4ee3\u865f\uff0c\u7576 \u2020 \u672c\u7bc7\u8ad6\u6587 TIMSS \u8a74\u984c\u5be6\u9a57\u7d44\uff0c\u50c5\u5305\u542b 2003 \u5e74\u8a74\u984c\uff0c\u8207\u5442\u660e\u6b23\u5b78\u8005\u7684\u5be6\u9a57\u7d44\u4e26\u4e0d\u76f8\u540c\u3002 \u4f5c\u5be6\u9a57\u7d44\u5225\u7684\u540d\u7a31\u3002\u53ef\u4ee5 TIMSS2003 \u5206\u70ba\u516b\u5e74\u7d1a 2003 M \u7d44\u3001\u516b\u5e74\u7d1a 2003 S \u7d44\u3001\u56db\u5e74 \u7d1a 2003 M \u7d44\u53ca\u4ee5\u56db\u5e74\u7d1a 2003 S \u7d44\u56db\u7d44\uff1b\u5728\u52a0\u4e0a TIMSS 2003 \u6578\u5b78\u53ca\u79d1\u5b78\u9818\u57df\u4e4b\u516b\u5e74\u7d1a \u8a74\u984c\uff0c\u548c TIMSS 2003 \u6578\u5b78\u53ca\u79d1\u5b78\u9818\u57df\u4e4b\u56db\u5e74\u7d1a\u8a74\u984c\uff0c\u5206\u5225\u70ba\u516b\u5e74\u7d1a 2003 MS \u7d44\u53ca\u56db\u5e74 \u7d1a 2003 MS \u7d44\uff0c\u7e3d\u5171\u516d\u7d44\uff0c\u5982\u8868\u4e8c\u6240\u793a\u3002 4.3 \u5be6\u9a57\u7d50\u679c \u4f9d\u7167\u4e0a\u4e00\u7bc0\u7684\u5be6\u9a57\u8a2d\u8a08\uff0c\u6211\u5011\u91dd\u5c0d TIMSS2003 \u8a74\u984c\u9a57\u8b49\u672c\u7cfb\u7d71\u3001Lu \u7cfb\u7d71\u53ca\u7dda\u4e0a\u7ffb\u8b6f\u7cfb \u7d71\u5728 BLEU \u548c NIST \u6bd4\u8f03\u6578\u64da\u3002\u5f9e\u8868\u4e09\u662f\u4ee5 cumulative n-gram scoring \u4e4b 4-gram \u70ba\u5e73\u5747 \u503c\uff0c\u6574\uf9e4\u4e4b\u5404\u7d44 NIST \u53ca BLEU \u503c\u4e4b\u6bd4\u8f03\u8868\u3002NIST \u8ddf BLEU \u6700\u5927\u7684\u4e0d\u540c\u5728\u65bc\uff0cNIST \u5c07 \u5f9e\u8868\u4e09\u53ef\u89c0\u5bdf\u5230\uff0c\u516b\u5e74\u7d1a 2003 M \u7d44 NIST \u5206\u6578\u4ee5 Yahoo!\u6700\u9ad8\u5206\uff0c\u4f46 BLEU \u5206\u6578\u8207 \u672c\u7cfb\u7d71\u5dee\u4e0d\u591a\uff0c\u53ef\u77e5 Yahoo!\u5c0d\u516b\u5e74\u7d1a 2003 M \u7d44\u6240\u7ffb\u8b6f\u7684\u8a5e\u5f59\u8ddf\u53c3\u8003\u7ffb\u8b6f\u8f03\u76f8\u540c\uff0c\u4f46 Yahoo!\u548c\u672c\u7cfb\u7d71\u7ffb\u8b6f\u5f8c\u8a5e\u5e8f\u7684\u6b63\u78ba\u6027\u662f\u5dee\u4e0d\u591a\u7684\u3002\u56db\u5e74\u7d1a 2003 M \u7d44\u8a74\u984c\u4e2d\u6709\u8f03\u591a\u7279\u6b8a \u7b26\u865f\uff0c\u4f8b\u5982\u25cb\u548c\u2022\u7b49\uff0cYahoo!\u53ca Google \u7dda\u4e0a\u7ffb\u8b6f\u7cfb\u7d71\u6703\u5c07\u9019\u4e9b\u7279\u6b8a\u7b26\u865f\u8655\uf9e4\u6210\u4e82\u78bc\uff0c \u4f46\u672c\u7cfb\u7d71\u53ef\u4ee5\u5c07\u7279\u6b8a\u7b26\u865f\u4fdd\u7559\u4e0b\u4f86\uff0c\u6545\u56db\u5e74\u7d1a\u548c\u516b\u5e74\u7d1a 2003 M \u7d44\u8207\u6700\u9ad8\u5206\u7cfb\u7d71\u7684\u5dee\u8ddd \u8f03\u5c0f\u3002\u5148\u524d\u6211\u5011\u5047\u8a2d\u7ffb\u8b6f\u54c1\u8cea\u662f\u5426\u6703\u6309\u7167\u8d8a\u4f4e\u5e74\u7d1a\u5176\u7ffb\u8b6f\u54c1\u8cea\u8d8a\u597d\u7684\u8da8\u52e2\uff0c\u89c0\u5bdf\u516b\u5e74\u7d1a 2003MS \u7d44\u53ca\u5c0f\u56db MS \u7d44\uff0c\u53ef\u767c\u73fe\u8207\u5047\u8a2d\u76f8\u53cd\uff0c\u5404\u7cfb\u7d71\u5728\u516b\u5e74\u7d1a 2003 MS \u7d44\u7684\u8868\u73fe\u90fd\u6bd4 \u56db\u5e74\u7d1a 2003 MS \u7d44\u8981\u597d\u3002\u53ef\u63a8\u6e2c\u51fa\u672c\u7cfb\u7d71\u5176\u4e2d\u4e00\u7a2e\u8a9e\u6599\u70ba\u570b\u4e2d\u88dc\u5145\u8cc7\u6599\u984c\u5eab\u8f03\u7b26\u5408 TIMSS \u516b\u5e74\u7d1a 2003 \u7684\u8a74\u984c\u3002 \u6211\u5011\u5c07\u516b\u5e74\u7d1a 2003M \u7d44\u548c\u516b\u5e74\u7d1a 2003S \u7d44\u4f5c\u6bd4\u8f03\uff0c\u56db\u5e74\u7d1a 2003 M \u7d44\u548c\u56db\u5e74\u7d1a 2003 S \u7d44\u4f5c\u6bd4\u8f03\uff0c\u53ef\u4ee5\u767c\u73fe\u5404\u7cfb\u7d71\u9664\u4e86 Google \u4e4b\u5916\uff0c\u5728 M \u7d44\u4e0a\u8868\u73fe\u90fd\u6bd4 S \u7d44\u597d\uff0c\u56e0\u70ba M \u7d44 \u7684\u8a74\u984c\u5167\u5bb9\u5305\u542b\u8f03\u591a\u7684\u6578\u5b57\uff0c\u5c0d\u65bc\u7ffb\u8b6f\u7cfb\u7d71\u8f03\u5bb9\u6613\u8655\uf9e4\uff0c\u800c S \u7d44\u5247\u5305\u542b\u8f03\u591a\u5c08\u6709\u540d\u8a5e\uff0c",
                "content": "<table><tr><td>\u6240\u6709\u5be6\u9a57\u8a9e\u6599\u53e5\u5c0d\u6578\u3001\u4e2d\u82f1\u8a5e\u5f59\u6578\u3001\u4e2d\u82f1\u7e3d\u8a5e\u5f59\u500b\u6578\u53ca\u5e73\u5747\u53e5\u9577\uff0c\u7686\u5982\u8868\u4e00\u6240\u793a\u3002 5. \u7d50\u8ad6</td></tr><tr><td>\u7528\u4f86\u5efa\u7acb\u7bc4\u4f8b\u6a39\u7684\u4f86\u6e90\u6709\u6559\u80b2\u90e8\u59d4\u8a17\u5b9c\u862d\u7e23\u5efa\u7f6e\u8a9e\u6587\u5b78\u7fd2\u9818\u57df\u570b\u4e2d\u6559\u79d1\u66f8\u88dc\u5145\u8cc7\u6599\u984c \u672c\u8ad6\u6587\u63d0\u51fa BSSTC \u7d50\u69cb\uff0c\u6b64\u7d50\u69cb\u80fd\u5920\u8a18\u9304\u4f86\u6e90\u53e5\u8a5e\u5f59\u7684\u4f4d\u7f6e\u3001\u76ee\u6a19\u53e5\u8a5e\u5f59\u7684\u4f4d\u7f6e\u53ca\u4f86</td></tr><tr><td>\u6e90\u53e5\u8207\u76ee\u6a19\u53e5\u8a5e\u5f59\u5c0d\u61c9\u7684\u95dc\u4fc2\uff1b\u4e26\u4e14\u5c07 BSSTC \u7d50\u69cb\u904b\u7528\u5728\u6211\u5011\u5be6\u4f5c\u7684\u7ffb\u8b6f\u7cfb\u7d71\u4e0a\u3002\u672c \u7cfb\u7d71\u662f\uf9dd\u7528 BSSTC \u7d50\u69cb\u5efa\u7acb\u7bc4\u4f8b\u6a39\uff0c\u5c07\u4f86\u6e90\u53e5\u7d93\u904e\u641c\u5c0b\u7bc4\u4f8b\u6a39\u6f14\u7b97\u6cd5\uff0c\u4f86\u9054\u5230\u4fee\u6b63\u8a5e \u5e8f\u7684\u76ee\u7684\u3002\u6700\u5f8c\uff0c\u5728\u4f9d\u64da\u4fee\u6b63\u5f8c\u7684\u8a5e\u5e8f\u9032\u884c\u7ffb\u8b6f\uff0c\u7ffb\u8b6f\u6642\u518d\uf9dd\u7528\u4e2d\u82f1\u8a5e\u5f59\u5c0d\uf99c\u5de5\u5177\u53ca bi-gram \u8a9e\u8a00\u6a21\u578b\uff0c\u9078\u51fa\u6700\u9069\u5408\u7684\u4e2d\u6587\u7ffb\u8b6f\uff0c\u7522\u751f\u5efa\u8b70\u7684\u7ffb\u8b6f\uff0c\u6b64 \u7ffb\u8b6f\u9084\u9700\u8981\u4eba\u5de5\u7de8\u4fee\u3002 TIMSS \u7684\u8a74\u984c\u70ba\u6578\u5b78\u53ca\u79d1\u5b78\u985e\uff0c\u61c9\u8a72\u8981\u7528\u5927\u91cf\u6578\u5b78\u53ca\u79d1\u5b78\u985e\u7684\u8a9e\u6599\uff0c\u4f46\u5be6\u969b\u4e0a\u6211 \u5eab[4] \u5171 110 \u7bc7\u70ba\u8a9e\u6599\u4f86\u6e90\u3002 \u5011\u4e26\u7121\u6cd5\u627e\u5230\u5920\u591a\u7684\u6578\u5b78\u53ca\u79d1\u5b78\u985e\u8a9e\u6599\uff0c\u5c24\u5176\u4ee5\u4e2d\u82f1\u5c0d\u61c9\u7684\u8a9e\u6599\u6700\u5c11\uff0c\u6240\u4ee5\u6211\u5011\u9078\u7528\u65b0</td></tr><tr><td>gram \u6a21\u578b\u8a73\u7d30\u4ecb\u7d39\u3002 \u4e2d\u82f1\u8a5e\u5f59\u5c0d\uf99c\uff1a\u5c07\u4e2d\u82f1\u8a9e\u6599\u96d9\u8a9e\u8a9e\u6599\uff0c\u7d93\u904e\u4eba\u5de5\u7684\u4e2d\u82f1\u8a9e\u53e5\u5c0d\uf99c(sentence alignment)\u6280 \u8853\uff0c\u63a5\u8457\u5c07\u4e2d\u6587\u8a9e\u6599\uf9dd\u7528\u4e2d\u7814\u9662 CKIP \u65b7\u8a5e\u7cfb\u7d71[1]\u52a0\u4ee5\u65b7\u8a5e\uff1b\u82f1\u6587\u8a9e\u6599\u5247\u662f\u7d93\u904e\u5927\u5c0f\u5beb \u8f49\u63db\u53ca\uf9dd\u7528\u5b57\u548c\u5b57\u4e4b\u9593\u7a7a\u767d\u65b7\u8a5e\uff0c\u6700\u5f8c\u8f38\u5165\u81f3 GIZA++[16]\u53ca mkcls[15]\u7b49\u5de5\u5177\uff0c\u7522\u751f\u4e2d \u82f1\u8a5e\u5f59\u5c0d\uf99c\u7d50\u679c\u4ee5\u53ca\u4e2d\u82f1\u8a5e\u5f59\u5c0d\u7167\u6a5f\u7387\u8868\u3002 bi-gram \u8a9e\u8a00\u6a21\u578b\uff1a\u5c07\u4e2d\u6587\u8a9e\u6599\u7d71\u8a08\u5404\u4e2d\u6587\u8a5e\u5f59\u548c\u4e0b\u4e00\u500b\u4e2d\u6587\u8a5e\u5f59\u51fa\u73fe\u7684\u6b21\u6578\uff0c\u8a08\u7b97\u5176 \u51fa\u73fe\u6a5f\u7387\u3002\u6211\u5011\u662f\uf9dd\u7528 SRI Speech Technology and Research Laboratory \u6240\u958b\u767c\u7684\u81ea\u7136\u8a9e \u8a00\u5de5\u5177 SRILM[18]\u4f86\u5efa\u7acb bi-gram \u8a9e\u8a00\u6a21\u578b\u3002 4. \u7cfb\u7d71\u7ffb\u8b6f\u6548\u679c\u8a55\u4f30 \u672c\u7bc0\u4e3b\u8981\u4ecb\u7d39\uf9dd\u7528\u672c\u7cfb\u7d71\u7ffb\u8b6f\u570b\u969b\u6578\u5b78\u8207\u79d1\u5b78\u6559\u80b2\u6210\u5c31\u8da8\u52e2\u8abf\u67e5 2003 \u5e74\u8003\u984c\uff0c\u7c21\u7a31 TIMSS2003\uff0c\u4e26\u5c07\u8a74\u984c\u4f9d\u7167\u5e74\u9f61\u5225\u548c\u79d1\u76ee\u5225\uff0c\u5206\u5225\u6bd4\u8f03\u7ffb\u8b6f\u7684\u54c1\u8cea\u3002\u6700\u5f8c\u5c07\u8207\u7dda\u4e0a\u7ffb\u8b6f \u4ee5\u53ca\u5442\u660e\u6b23\u7b49\u5b78\u8005\u7814\u767c\u7684\u7ffb\u8b6f\u7cfb\u7d71\u4f5c\u6bd4\u8f03\u3002\u8a55\u4f30\u65b9\u5f0f\u70ba\uf9dd\u7528 BLEU \u53ca NIST \u6307\u6a19\u3002 4.1 \u5be6\u9a57\u4f86\u6e90 \u6211\u5011\u4e3b\u8981\u7528\u4f86\u7ffb\u8b6f\u7684\u4f86\u6e90\u70ba TIMSS2003 \u8a74\u984c\uff0c\u5340\u5206\u6578\u5b78\u53ca\u79d1\u5b78\u985e\u5225\uff0c\u4e26\u4e14\u4ee5\u56db\u5e74\u7d1a\u53ca \u516b\u5e74\u7d1a\u70ba\u8003\u8a74\u5c0d\u8c61\uff0c\u5171\u6709\u56db\u7a2e\u8a74\u984c\u5206\u5225\u70ba\u56db\u5e74\u7d1a\u6578\u5b78\u9818\u57df 31 \u984c\uff1b\u56db\u5e74\u7d1a\u79d1\u5b78\u9818\u57df 70 \u8868\u4e00\u3001\u5be6\u9a57\u8a9e\u6599\u4f86\u6e90\u7d71\u8a08 \u8a9e\u6599 \u8a9e\u8a00 \u53e5\u5c0d\u6578 \u8fad\u5f59\u6578 \u7e3d\u8a5e\u5f59\u500b\u6578(tokens) \u5e73\u5747\u53e5\u9577 \u570b\u4e2d\u88dc\u5145\u8cc7\u6599\u984c\u5eab \u4e2d\u6587 2059 \u53e5 2333 12460 6.1 \u82f1\u6587 2887 13170 6.4 \u79d1\u5b78\u4eba \u4e2d\u6587 4247 \u53e5 9279 70411 16.6 \u82f1\u6587 10504 68434 16.1 \u81ea\u7531\u6642\u5831\u4e2d\u82f1\u5c0d \u7167\u8b80\u65b0\u805e \u4e2d\u6587 4248 \u53e5 19188 145336 34.2 \u82f1\u6587 25782 133123 31.3 4.2 \u5be6\u9a57\u8a2d\u8a08 \u9996\u5148\uff0c\u5c07 TIMSS2003 \u8a74\u984c\u554f\u53e5\u4ee5\u9017\u865f\u3001\u554f\u865f\u6216\u9a5a\u5606\u865f\u505a\u70ba\u65b7\u53e5\u7684\u55ae\u4f4d\uff0c\u6bcf\u500b\u8a98\u7b54\u9078\u9805 \u505a\u70ba\u65b7\u53e5\u7684\u55ae\u4f4d\uff0c\u82e5\u4e00\u9053\u984c\u76ee\u70ba\u4e00\u53e5\u8a74\u984c\u554f\u53e5\u53ca\u56db\u9805\u8a98\u7b54\u9078\u9805\u6240\u7d44\u6210\uff0c\u5247\u4e00\u9053\u984c\u76ee\u53ef\u65b7 \u51fa\u4e94\u53e5\u3002\u7d93\u904e\u4eba\u5de5\u65b7\u53e5\u8655\uf9e4 TIMSS2003 \u8a74\u984c\uff0c\u56db\u5e74\u7d1a\u6578\u5b78\u9818\u57df\u6709 165 \u53e5\uff1b\u56db\u5e74\u7d1a\u79d1\u5b78 \u9818\u57df\u6709 262 \u53e5\uff1b\u516b\u5e74\u7d1a\u6578\u5b78\u9818\u57df\u6709 439 \u53e5\uff1b\u516b\u5e74\u7d1a\u79d1\u5b78\u9818\u57df\u6709 236 \u53e5\uff0c\u4e26\u6574\uf9e4\u70ba\u6587\u5b57\u6a94\u3002 \u7ffb\u8b6f\u6642\u4e2d\u6587\u8a74\u984c\u6240\u904b\u7528\u7684\u4e2d\u6587\u65b7\u8a5e\u70ba\u4e2d\u7814\u9662 CKIP \u65b7\u8a5e\u7cfb\u7d71[1]\uff0c\u82f1\u6587\u8a74\u984c\u6240\u904b\u7528\u7684\u5256\u6790 \u5668\u70ba StanfordLexParser-1.6[17]\uff0c\u5efa\u7acb\u7bc4\u4f8b\u6a39\u8cc7\u6599\u5eab\u6240\u4f7f\u7528\u7684\u8a9e\u6599\u70ba\u570b\u4e2d\u88dc\u5145\u8cc7\u6599\u984c\u5eab\uff0c \u8a13\u7df4\u6a5f\u7387\u6a21\u578b\u6240\u4f7f\u7528\u7684\u8a9e\u6599\u81ea\u7531\u6642\u5831\u4e2d\u82f1\u5c0d\u7167\u8b80\u65b0\u805e\u52a0\u4e0a\u79d1\u5b78\u4eba\u96dc\u8a8c\uff0c\u5176\u4e2d\u8a13\u7df4\u8a9e\u8a00\u6a21 \u578b\u5f97\u5230\u7684 bi-gram \u5171\u6709 134435 \u500b\uff1bGIZA++\u7522\u751f\u4e2d\u82f1\u8a5e\u5f59\u5c0d\uf99c\u7d50\u679c\u6709 128551 \u7d44\u3002 \u8868\u4e8c\u3001TIMSS \u8a74\u984c\u5be6\u9a57\u7d44\u5225\u8868  \u2020 \u516b\u5e74\u7d1a 2003 M \u7d44 \u516b\u5e74\u7d1a 2003 S \u7d44 \u56db\u5e74\u7d1a 2003 M \u7d44 \u56db\u5e74\u7d1a 2003 S \u7d44 \u516b\u5e74\u7d1a 2003 MS \u7d44 \u56db\u5e74\u7d1a 2003 MS \u7d44 TIMSS2003 \u570b\u4e2d \u6578\u5b78\u9818\u57df\u8a74\u984c TIMSS2003 \u570b\u4e2d \u79d1\u5b78\u9818\u57df\u8a74\u984c TIMSS2003 \u570b\u5c0f \u6578\u5b78\u9818\u57df\u8a74\u984c TIMSS2003 \u570b\u5c0f \u79d1\u5b78\u9818\u57df\u8a74\u984c TIMSS2003 \u570b\u4e2d \u6578\u5b78\u53ca\u79d1\u5b78\u9818\u57df \u8a74\u984c TIMSS200 \u570b\u5c0f \u6578\u5b78\u53ca\u79d1\u5b78\u9818\u57df \u8a74\u984c \u6211\u5011\u8a55\u4f30\u6240\u4f7f\u7528\u7684\u5de5\u5177\u70ba\u4f9d\u7167 BLEU \u53ca Yahoo!\u7dda\u4e0a\u7ffb\u8b6f\u3001\u5442\u660e\u6b23\u5b78\u8005\u7684\u7cfb\u7d71(Lu)\u53ca\u672c\u7cfb\u7d71\u4e92\u76f8\u505a\u6bd4\u8f03\uff0c\u4e26\u4e14\u8a55\u4f30\u7ffb\u8b6f\u7cfb\u7d71\u5728\u4e0d \u540c\u5e74\u7d1a\u7684\u8a74\u984c\u5167\u5bb9\u4e0a\uff0c\u7ffb\u8b6f\u54c1\u8cea\u662f\u5426\u6703\u6309\u7167\u8d8a\u4f4e\u5e74\u7d1a\u5176\u7ffb\u8b6f\u54c1\u8cea\u8d8a\u597d\u7684\u8da8\u52e2\u3002\u56e0\u6b64\uff0c\u6211 \u7684\u5927\u5c0f\uff0c\u800c BLEU \u91dd\u5c0d\u5404 n-gram \u5339\u914d\u6b63\u78ba\u7387\u53ca\u76f8\u4f3c\u5ea6\u9032\u884c\u8a08\u5206\u3002\u7531\u6b64\u53ef\u77e5\u7576\u53c3\u8003\u7ffb\u8b6f \u53e5\u5b50\u548c\u7cfb\u7d71\u7ffb\u8b6f\u53e5\u5b50\u7528\u7684\u8a5e\u5f59\u76f8\u540c\u6642\uff0cNIST \u5206\u6578\u6703\u6bd4\u8f03\u9ad8\uff1b\u7576\u53c3\u8003\u7ffb\u8b6f\u53e5\u5b50\u548c\u7cfb\u7d71\u7ffb \u8b6f\u53e5\u5b50\u7528\u7684\u8a5e\u5f59\u9806\u5e8f\u8f03\u76f8\u8fd1\u6642\uff0cBLEU \u5206\u6578\u6703\u6bd4\u8f03\u9ad8\u3002 \u8868\u4e09\u3001\u672c\u7cfb\u7d71\u3001Lu \u7cfb\u7d71\u53ca\u7dda\u4e0a\u7ffb\u8b6f\u7cfb\u7d71\u4e4b NIST \u53ca BLEU \u503c\u6bd4\u8f03\u8868 \u7d44\u5225 \u516b\u5e74\u7d1a 2003 M \u7d44 \u516b\u5e74\u7d1a 2003 S \u7d44 \u56db\u5e74\u7d1a 2003 M \u7d44 \u6307\u6a19 NIST BLEU NIST BLEU NIST BLEU \u672c\u7cfb\u7d71 4.7002 0.1440 4.4089 0.1254 3.9819 0.1304 Lu 3.6185 0.1007 3.5831 0.0890 3.3319 0.0983 Google 4.5268 0.1467 4.8587 0.1848 3.7573 0.1016 Yahoo! 4.8793 0.1455 4.6136 0.1396 4.0457 0.1419 \u7d44\u5225 \u56db\u5e74\u7d1a 2003 S \u7d44 \u516b\u5e74\u7d1a 2003 MS \u7d44 \u56db\u5e74\u7d1a 2003 MS \u7d44 \u6307\u6a19 NIST BLEU NIST BLEU NIST BLEU \u672c\u7cfb\u7d71 4.2228 0.1018 4.8613 0.1309 4.4400 0.1138 Lu 3.2495 0.0682 3.8031 0.0966 3.4970 0.0803 Google 4.4445 0.1527 4.9343 0.1611 4.4720 0.1344 Yahoo! 4.4361 0.1442 5.0755 0.1435 4.6070 0.1436 \u5c0d\u65bc\u7ffb\u8b6f\u7cfb\u7d71\u8f03\u70ba\u56f0\u96e3\u3002\u63a5\u8457\u5c07\u672c\u7cfb\u7d71\u8207 Lu \u7cfb\u7d71\u4f5c\u6bd4\u8f03\uff0cLu \u7cfb\u7d71\u548c\u672c\u7cfb\u7d71\u7684\u5dee\u5225\u70ba\u6c92 \u6709\u4f5c\u8a5e\u5e8f\u7684\u4ea4\u63db\u3002\u7d93\u904e\u8a5e\u5e8f\u4ea4\u63db\u5f8c\uff0c\u5f97\u5230\u6b63\u78ba\u7684\u4e2d\u6587\u8a5e\u5e8f\uff0c\u56e0\u6b64\u9078\u8a5e\u7684\u6b63\u78ba\u6027\u76f8\u5c0d\u6703\u63d0 \u805e\u53ca\u570b\u4e2d\u88dc\u5145\u8cc7\u6599\u984c\u5eab\u4f86\u64ec\u88dc\u8a9e\u6599\u7684\u4e0d\u8db3\u3002\u4e0d\u904e\u8a13\u7df4\u91cf\u9084\u7b97\u662f\u4e0d\u8db3\u5920\uff0c\u5728\u9078\u8a5e\u4e0a\u6703\u6709\u8a31 \u591a\u6a5f\u7387\u70ba 0 \u7684\u60c5\u6cc1\uff0c\u9020\u6210\u9078\u8a5e\u932f\u8aa4\u3002\u672a\u4f86\u5c07\u76e1\u91cf\u627e\u5c0b\u76f8\u95dc\u9818\u57df\u7684\u8a9e\u6599\uff0c\u4f86\u5efa\u7acb\u7bc4\u4f8b\u6a39\u548c \u8a13\u7df4\u8a9e\u8a00\u6a21\u578b\uff0c\u5c31\u80fd\u91dd\u5c0d\u4e0d\u540c\u9818\u57df\u7684\u4f86\u5ba2\u88fd\u5316\u7ffb\u8b6f\uff0c\u4f7f\u7ffb\u8b6f\u7684\u7d50\u679c\u66f4\u70ba\u7cbe\u78ba\u3002 \u8a13\u7df4\u8a9e\u6599\u4e2d\u7684\u65b7\u8a5e\u662f\u4f7f\u7528\u4e2d\u7814\u9662 CKIP \u7cfb\u7d71\uff0c\u800c\u6211\u5011\u7ffb\u8b6f\u4f7f\u7528\u7684\u5b57\u5178\u70ba\u725b\u6d25\u5b57\u5178\uff0c \u5169\u8005\u6240\u4f7f\u7528\u7684\u5b57\u5178\u4e26\u4e0d\u76f8\u540c\uff0c\u6703\u4f7f\u65b7\u8a5e\u5f8c\u7684\u8a5e\u5f59\u53ef\u80fd\u7121\u6cd5\u5728\u725b\u6d25\u5b57\u5178\u4e2d\u627e\u5230\uff0c\u9020\u6210\u9078\u8a5e \u932f\u8aa4\u3002\u672a\u4f86\u53ef\u5c07\u7ffb\u8b6f\u5f8c\u7684\u8a5e\u5f59\uff0c\u627e\u51fa\u540c\u7fa9\u8a5e\u4f86\u64f4\u5145\u8a5e\u5f59\u6578\uff0c\u4fbf\u80fd\u589e\u52a0\u88ab\u627e\u5230\u7684\u53ef\u80fd\u6027\u3002 \u82f1\u6587\u7684\u8a9e\u8a00\u7279\u6027\u4e0a\u4e26\u6c92\u6709\u91cf\u8a5e\uff0c\u800c\u4e2d\u6587\u53e5\u4e2d\u904b\u7528\u4e86\u5f88\u591a\u7684\u91cf\u8a5e\uff0c\u5982\u7f3a\u5c11\u91cf\u8a5e\u4e5f\u6703\u4f7f \u4e2d\u6587\u7684\u6d41\u66a2\u5ea6\u4e0b\u5c07\u3002\u672c\u7cfb\u7d71\u7684\u7ffb\u8b6f\u7d50\u679c\u4e5f\u7f3a\u5c11\u4e2d\u6587\u7684\u91cf\u8a5e\u3002\u672a\u4f86\u82e5\u80fd\u5c07\u7ffb\u8b6f\u7d50\u679c\u586b\u88dc\u4e0a \u7f3a\u5c11\u7684\u91cf\u8a5e\uff0c\u4fbf\u53ef\u9054\u5230\u66f4\u597d\u7684\u54c1\u8cea\u3002 \u81f4\u8b1d \u672c\u7814\u7a76\u627f\u8499\u570b\u79d1\u6703\u7814\u7a76\u8a08\u756b NSC-95-2221-E-004-013-MY2 \u7684\u90e8\u5206\u88dc\u52a9\u8b39\u6b64\u81f4\u8b1d\u3002\u6211\u5011 \u611f\u8b1d\u533f\u540d\u8a55\u5be9\u5c0d\u65bc\u672c\u6587\u521d\u7a3f\u7684\u5404\u9805\u6307\u6b63\u8207\u6307\u5c0e\uff0c\u96d6\u7136\u6211\u5011\u5df2\u7d93\u5728\u5f9e\u4e8b\u76f8\u95dc\u7684\u90e8\u5206\u7814\u7a76\u8b70 \u984c\uff0c\u4e0d\u904e\u9650\u65bc\u7bc7\u5e45\u56e0\u6b64\u4e0d\u80fd\u5728\u672c\u6587\u4e2d\u5168\u9762\u4ea4\u4ee3\u76f8\u95dc\u7d30\u7bc0\u3002 \u5011\u5c07\u5be6\u9a57\u7d44\u5225\u5206\u70ba\u516b\u5e74\u7d1a\u548c\u56db\u5e74\u7d1a\uff1b\u6578\u5b78\u9818\u57df\u4ee5 M \u5404 n-gram \u8a5e\u5f59\u4e2d\u5171\u73fe(co-occurrence)\u7684\u6b21\u6578\u6bd4\u7684\u7d2f\u52a0\u503c\uff0c\u7576\u4f5c\u5404 n-gram \u5e73\u5747\u8cc7\u8a0a\u91cf \u53c3\u8003\u6587\u737b</td></tr><tr><td>\u984c\uff1b\u516b\u5e74\u7d1a\u6578\u5b78\u9818\u57df 41 \u984c\uff1b\u516b\u5e74\u7d1a\u79d1\u5b78\u9818\u57df 38 \u984c\u3002\u6240\u6709\u8a74\u984c\u90fd\u6709\u82f1\u6587\u539f\u6587\u8a74\u984c\u548c\u5e2b\u5927 \u5347\uff0c\u6240\u4ee5\u672c\u7cfb\u7d71\u5728\u5404\u7d44\u7684\u8868\u73fe\u90fd\u6bd4 Lu \u7cfb\u7d71\u8981\u597d\uff0c\u986f\u793a\u8a5e\u5e8f\u4ea4\u63db\u5f8c\u6703\u5f97\u5230\u54c1\u8cea\u8f03\u597d\u7684\u4e2d</td></tr><tr><td>\u79d1\u6559\u4e2d\u5fc3\u6240\u7ffb\u8b6f\u7684\u4e2d\u6587\u8a74\u984c\u3002 \u6587\u7ffb\u8b6f\u3002</td></tr></table>",
                "type_str": "table"
            }
        }
    }
}