File size: 76,168 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
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
{
    "paper_id": "P02-1017",
    "header": {
        "generated_with": "S2ORC 1.0.0",
        "date_generated": "2023-01-19T09:30:51.589343Z"
    },
    "title": "A Generative Constituent-Context Model for Improved Grammar Induction",
    "authors": [
        {
            "first": "Dan",
            "middle": [],
            "last": "Klein",
            "suffix": "",
            "affiliation": {
                "laboratory": "",
                "institution": "Stanford University Stanford",
                "location": {
                    "postCode": "94305-9040",
                    "region": "CA"
                }
            },
            "email": "klein@cs.stanford.edu"
        },
        {
            "first": "Christopher",
            "middle": [
                "D"
            ],
            "last": "Manning",
            "suffix": "",
            "affiliation": {
                "laboratory": "",
                "institution": "Stanford University Stanford",
                "location": {
                    "postCode": "94305-9040",
                    "region": "CA"
                }
            },
            "email": "manning@cs.stanford.edu"
        }
    ],
    "year": "",
    "venue": null,
    "identifiers": {},
    "abstract": "We present a generative distributional model for the unsupervised induction of natural language syntax which explicitly models constituent yields and contexts. Parameter search with EM produces higher quality analyses than previously exhibited by unsupervised systems, giving the best published unsupervised parsing results on the ATIS corpus. Experiments on Penn treebank sentences of comparable length show an even higher F 1 of 71% on nontrivial brackets. We compare distributionally induced and actual part-of-speech tags as input data, and examine extensions to the basic model. We discuss errors made by the system, compare the system to previous models, and discuss upper bounds, lower bounds, and stability for this task.",
    "pdf_parse": {
        "paper_id": "P02-1017",
        "_pdf_hash": "",
        "abstract": [
            {
                "text": "We present a generative distributional model for the unsupervised induction of natural language syntax which explicitly models constituent yields and contexts. Parameter search with EM produces higher quality analyses than previously exhibited by unsupervised systems, giving the best published unsupervised parsing results on the ATIS corpus. Experiments on Penn treebank sentences of comparable length show an even higher F 1 of 71% on nontrivial brackets. We compare distributionally induced and actual part-of-speech tags as input data, and examine extensions to the basic model. We discuss errors made by the system, compare the system to previous models, and discuss upper bounds, lower bounds, and stability for this task.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Abstract",
                "sec_num": null
            }
        ],
        "body_text": [
            {
                "text": "The task of inducing hierarchical syntactic structure from observed yields alone has received a great deal of attention (Carroll and Charniak, 1992; Pereira and Schabes, 1992; Brill, 1993; Stolcke and Omohundro, 1994) . Researchers have explored this problem for a variety of reasons: to argue empirically against the poverty of the stimulus (Clark, 2001) , to use induction systems as a first stage in constructing large treebanks (van Zaanen, 2000) , or to build better language models (Baker, 1979; Chen, 1995) .",
                "cite_spans": [
                    {
                        "start": 120,
                        "end": 148,
                        "text": "(Carroll and Charniak, 1992;",
                        "ref_id": "BIBREF4"
                    },
                    {
                        "start": 149,
                        "end": 175,
                        "text": "Pereira and Schabes, 1992;",
                        "ref_id": "BIBREF14"
                    },
                    {
                        "start": 176,
                        "end": 188,
                        "text": "Brill, 1993;",
                        "ref_id": "BIBREF3"
                    },
                    {
                        "start": 189,
                        "end": 217,
                        "text": "Stolcke and Omohundro, 1994)",
                        "ref_id": "BIBREF17"
                    },
                    {
                        "start": 342,
                        "end": 355,
                        "text": "(Clark, 2001)",
                        "ref_id": "BIBREF7"
                    },
                    {
                        "start": 432,
                        "end": 450,
                        "text": "(van Zaanen, 2000)",
                        "ref_id": "BIBREF18"
                    },
                    {
                        "start": 488,
                        "end": 501,
                        "text": "(Baker, 1979;",
                        "ref_id": "BIBREF2"
                    },
                    {
                        "start": 502,
                        "end": 513,
                        "text": "Chen, 1995)",
                        "ref_id": "BIBREF5"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Introduction",
                "sec_num": "1"
            },
            {
                "text": "In previous work, we presented a conditional model over trees which gave the best published results for unsupervised parsing of the ATIS corpus (Klein and Manning, 2001b) . However, it suffered from several drawbacks, primarily stemming from the conditional model used for induction. Here, we improve on that model in several ways. First, we construct a generative model which utilizes the same features. Then, we extend the model to allow multiple constituent types and multiple prior distribu-tions over trees. The new model gives a 13% reduction in parsing error on WSJ sentence experiments, including a positive qualitative shift in error types. Additionally, it produces much more stable results, does not require heavy smoothing, and exhibits a reliable correspondence between the maximized objective and parsing accuracy. It is also much faster, not requiring a fitting phase for each iteration. Klein and Manning (2001b) and Clark (2001) take treebank part-of-speech sequences as input. We followed this for most experiments, but in section 4.3, we use distributionally induced tags as input. Performance with induced tags is somewhat reduced, but still gives better performance than previous models.",
                "cite_spans": [
                    {
                        "start": 144,
                        "end": 170,
                        "text": "(Klein and Manning, 2001b)",
                        "ref_id": "BIBREF11"
                    },
                    {
                        "start": 903,
                        "end": 928,
                        "text": "Klein and Manning (2001b)",
                        "ref_id": "BIBREF11"
                    },
                    {
                        "start": 933,
                        "end": 945,
                        "text": "Clark (2001)",
                        "ref_id": "BIBREF7"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Introduction",
                "sec_num": "1"
            },
            {
                "text": "Early work on grammar induction emphasized heuristic structure search, where the primary induction is done by incrementally adding new productions to an initially empty grammar (Olivier, 1968; Wolff, 1988) . In the early 1990s, attempts were made to do grammar induction by parameter search, where the broad structure of the grammar is fixed in advance and only parameters are induced (Lari and Young, 1990; Carroll and Charniak, 1992) . 1 However, this appeared unpromising and most recent work has returned to using structure search. Note that both approaches are local. Structure search requires ways of deciding locally which merges will produce a coherent, globally good grammar. To the extent that such approaches work, they work because good local heuristics have been engineered (Klein and Manning, 2001a; Clark, 2001) . September 5   5  4  3  2  1  0   5   4   3   2   1   0   Start   End   5  4  3  2  1  0   5   4   3   2   1   0   Start   End   5  4  3  2  1  0   5   4   3   2   1   0   Start   End   Span Label  Constituent  Context  0,5   S  NN NNS VBD IN NN   -0,2   NP  NN NNS   -VBD  2,5   VP  VBD IN NN   NNS -3,5   PP  IN NN   VBD -0,1   NN  NN   -NNS  1,2   NNS  NNS   NN -VBD  2,3   VBD VBD Figure 1 : (a) Example parse tree with (b) its associated bracketing and (c) the yields and contexts for each constituent span in that bracketing. Distituent yields and contexts are not shown, but are modeled.",
                "cite_spans": [
                    {
                        "start": 177,
                        "end": 192,
                        "text": "(Olivier, 1968;",
                        "ref_id": "BIBREF13"
                    },
                    {
                        "start": 193,
                        "end": 205,
                        "text": "Wolff, 1988)",
                        "ref_id": "BIBREF19"
                    },
                    {
                        "start": 385,
                        "end": 407,
                        "text": "(Lari and Young, 1990;",
                        "ref_id": "BIBREF12"
                    },
                    {
                        "start": 408,
                        "end": 435,
                        "text": "Carroll and Charniak, 1992)",
                        "ref_id": "BIBREF4"
                    },
                    {
                        "start": 787,
                        "end": 813,
                        "text": "(Klein and Manning, 2001a;",
                        "ref_id": "BIBREF10"
                    },
                    {
                        "start": 814,
                        "end": 826,
                        "text": "Clark, 2001)",
                        "ref_id": "BIBREF7"
                    }
                ],
                "ref_spans": [
                    {
                        "start": 829,
                        "end": 1208,
                        "text": "September 5   5  4  3  2  1  0   5   4   3   2   1   0   Start   End   5  4  3  2  1  0   5   4   3   2   1   0   Start   End   5  4  3  2  1  0   5   4   3   2   1   0   Start   End   Span Label  Constituent  Context  0,5   S  NN NNS VBD IN NN   -0,2   NP  NN NNS   -VBD  2,5   VP  VBD IN NN   NNS -3,5   PP  IN NN   VBD -0,1   NN  NN   -NNS  1,2   NNS  NNS   NN -VBD  2,3   VBD",
                        "ref_id": null
                    },
                    {
                        "start": 1213,
                        "end": 1221,
                        "text": "Figure 1",
                        "ref_id": null
                    }
                ],
                "eq_spans": [],
                "section": "Previous Work",
                "sec_num": "2"
            },
            {
                "text": "S NP NN 0 Factory NNS 1 payrolls VP VBD 2 fell PP IN 3 in NN 4",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Previous Work",
                "sec_num": "2"
            },
            {
                "text": "NNS -IN 3,4 IN IN VBD -NN 4,5 NN NNS IN - (a) (b) (c)",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Previous Work",
                "sec_num": "2"
            },
            {
                "text": "Parameter search is also local; parameters which are locally optimal may be globally poor. A concrete example is the experiments from (Carroll and Charniak, 1992) . They restricted the space of grammars to those isomorphic to a dependency grammar over the POS symbols in the Penn treebank, and then searched for parameters with the inside-outside algorithm (Baker, 1979) starting with 300 random production weight vectors. Each seed converged to a different locally optimal grammar, none of them nearly as good as the treebank grammar, measured either by parsing performance or data-likelihood.",
                "cite_spans": [
                    {
                        "start": 134,
                        "end": 162,
                        "text": "(Carroll and Charniak, 1992)",
                        "ref_id": "BIBREF4"
                    },
                    {
                        "start": 357,
                        "end": 370,
                        "text": "(Baker, 1979)",
                        "ref_id": "BIBREF2"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Previous Work",
                "sec_num": "2"
            },
            {
                "text": "However, parameter search methods have a potential advantage. By aggregating over only valid, complete parses of each sentence, they naturally incorporate the constraint that constituents cannot cross -the bracketing decisions made by the grammar must be coherent. The Carroll and Charniak experiments had two primary causes for failure. First, random initialization is not always good, or necessary. The parameter space is riddled with local likelihood maxima, and starting with a very specific, but random, grammar should not be expected to work well. We duplicated their experiments, but used a uniform parameter initialization where all productions were equally likely. This allowed the interaction between the grammar and data to break the initial symmetry, and resulted in an induced grammar of higher quality than Carroll and Charniak reported. This grammar, which we refer to as DEP-PCFG will be evaluated in more detail in section 4. The second way in which their experiment was guaranteed to be somewhat unencouraging is that a delexicalized dependency grammar is a very poor model of language, even in a supervised setting. By the F 1 measure used in the experiments in section 4, an induced dependency PCFG scores 48.2, compared to a score of 82.1 for a supervised PCFG read from local trees of the treebank. However, a supervised dependency PCFG scores only 53.5, not much better than the unsupervised version, and worse than a right-branching baseline (of 60.0). As an example of the inherent shortcomings of the dependency grammar, it is structurally unable to distinguish whether the subject or object should be attached to the verb first. Since both parses involve the same set of productions, both will have equal likelihood.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Previous Work",
                "sec_num": "2"
            },
            {
                "text": "To exploit the benefits of parameter search, we used a novel model which is designed specifically to enable a more felicitous search space. The fundamental assumption is a much weakened version of classic linguistic constituency tests (Radford, 1988) : constituents appear in constituent contexts. A particular linguistic phenomenon that the system exploits is that long constituents often have short, common equivalents, or proforms, which appear in similar contexts and whose constituency is easily discovered (or guaranteed). Our model is designed to transfer the constituency of a sequence directly to its containing context, which is intended to then pressure new sequences that occur in that context into being parsed as constituents in the next round. The model is also designed to exploit the successes of distributional clustering, and can equally well be viewed as doing distributional clustering in the presence of no-overlap constraints.",
                "cite_spans": [
                    {
                        "start": 235,
                        "end": 250,
                        "text": "(Radford, 1988)",
                        "ref_id": "BIBREF15"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "A Generative Constituent-Context Model",
                "sec_num": "3"
            },
            {
                "text": "Unlike a PCFG, our model describes all contiguous subsequences of a sentence (spans), including empty spans, whether they are constituents or nonconstituents (distituents). A span encloses a sequence of terminals, or yield, \u03b1, such as DT JJ NN. A span occurs in a context x, such as -VBZ, where x is the ordered pair of preceding and following ter-minals ( denotes a sentence boundary). A bracketing of a sentence is a boolean matrix B, which indicates which spans are constituents and which are not. Figure 1 shows a parse of a short sentence, the bracketing corresponding to that parse, and the labels, yields, and contexts of its constituent spans. Figure 2 shows several bracketings of the sentence in figure 1. A bracketing B of a sentence is non-crossing if, whenever two spans cross, at most one is a constituent in B. A non-crossing bracketing is tree-equivalent if the size-one terminal spans and the full-sentence span are constituents, and all size-zero spans are distituents. Figure 2 (a) and (b) are tree-equivalent. Tree-equivalent bracketings B correspond to (unlabeled) trees in the obvious way. A bracketing is binary if it corresponds to a binary tree. Figure 2 (b) is binary. We will induce trees by inducing tree-equivalent bracketings.",
                "cite_spans": [],
                "ref_spans": [
                    {
                        "start": 501,
                        "end": 509,
                        "text": "Figure 1",
                        "ref_id": null
                    },
                    {
                        "start": 652,
                        "end": 660,
                        "text": "Figure 2",
                        "ref_id": "FIGREF0"
                    },
                    {
                        "start": 988,
                        "end": 996,
                        "text": "Figure 2",
                        "ref_id": "FIGREF0"
                    },
                    {
                        "start": 1171,
                        "end": 1179,
                        "text": "Figure 2",
                        "ref_id": "FIGREF0"
                    }
                ],
                "eq_spans": [],
                "section": "Constituents and Contexts",
                "sec_num": "3.1"
            },
            {
                "text": "Our generative model over sentences S has two phases. First, we choose a bracketing B according to some distribution P(B) and then generate the sentence given that bracketing:",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Constituents and Contexts",
                "sec_num": "3.1"
            },
            {
                "text": "P(S, B) = P(B)P(S|B)",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Constituents and Contexts",
                "sec_num": "3.1"
            },
            {
                "text": "Given B, we fill in each span independently. The context and yield of each span are independent of each other, and generated conditionally on the constituency B i j of that span.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Constituents and Contexts",
                "sec_num": "3.1"
            },
            {
                "text": "= i, j \u2208spans(S) P(\u03b1 i j , x i j |B i j ) = i, j P(\u03b1 i j |B i j )P(x i j |B i j )",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "P(S|B)",
                "sec_num": null
            },
            {
                "text": "The distribution P(\u03b1 i j |B i j ) is a pair of multinomial distributions over the set of all possible yields: one for constituents (B i j = c) and one for distituents (B i j = d). Similarly for P(x i j |B i j ) and contexts. The marginal probability assigned to the sentence S is given by summing over all possible bracketings of S: P(S) = B P(B)P(S|B). 2 To induce structure, we run EM over this model, treating the sentences S as observed and the bracketings B as unobserved. The parameters of the model are the constituency-conditional yield and context distributions P(\u03b1|b) and P(x|b). If P(B) is uniform over all (possibly crossing) bracketings, then this procedure will be equivalent to softclustering with two equal-prior classes.",
                "cite_spans": [
                    {
                        "start": 354,
                        "end": 355,
                        "text": "2",
                        "ref_id": null
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "P(S|B)",
                "sec_num": null
            },
            {
                "text": "There is reason to believe that such soft clusterings alone will not produce valuable distinctions, even with a significantly larger number of classes. The distituents must necessarily outnumber the constituents, and so such distributional clustering will result in mostly distituent classes. Clark (2001) finds exactly this effect, and must resort to a filtering heuristic to separate constituent and distituent clusters. To underscore the difference between the bracketing and labeling tasks, consider figure 3. In both plots, each point is a frequent tag sequence, assigned to the (normalized) vector of its context frequencies. Each plot has been projected onto the first two principal components of its respective data set. The left plot shows the most frequent sequences of three constituent types. Even in just two dimensions, the clusters seem coherent, and it is easy to believe that they would be found by a clustering algorithm in the full space. On the right, sequences have been labeled according to whether their occurrences are constituents more or less of the time than a cutoff (of 0.2). The distinction between constituent and distituent seems much less easily discernible.",
                "cite_spans": [
                    {
                        "start": 293,
                        "end": 305,
                        "text": "Clark (2001)",
                        "ref_id": "BIBREF7"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "P(S|B)",
                "sec_num": null
            },
            {
                "text": "We can turn what at first seems to be distributional clustering into tree induction by confining P(B) to put mass only on tree-equivalent bracketings. In particular, consider P bin (B) which is uniform over binary bracketings and zero elsewhere. If we take this bracketing distribution, then when we sum over data completions, we will only involve bracketings which correspond to valid binary trees. This restriction is the basis for our algorithm.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "P(S|B)",
                "sec_num": null
            },
            {
                "text": "Usually a Constituent Rarely a Constituent (a) Constituent Types (b) Constituents vs. Distituents Figure 3 : The most frequent yields of (a) three constituent types and (b) constituents and distituents, as context vectors, projected onto their first two principal components. Clustering is effective at labeling, but not detecting constituents.",
                "cite_spans": [],
                "ref_spans": [
                    {
                        "start": 98,
                        "end": 106,
                        "text": "Figure 3",
                        "ref_id": null
                    }
                ],
                "eq_spans": [],
                "section": "NP VP PP",
                "sec_num": null
            },
            {
                "text": "We now essentially have our induction algorithm. We take P(B) to be P bin (B), so that all binary trees are equally likely. We then apply the EM algorithm: E-Step: Find the conditional completion likelihoods P(B|S, ) according to the current .",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "The Induction Algorithm",
                "sec_num": "3.2"
            },
            {
                "text": "Step: Fix P(B|S, ) and find the which maximizes B P(B|S, ) log P(S, B| ).",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "M-",
                "sec_num": null
            },
            {
                "text": "The completions (bracketings) cannot be efficiently enumerated, and so a cubic dynamic program similar to the inside-outside algorithm is used to calculate the expected counts of each yield and context, both as constituents and distituents. Relative frequency estimates (which are the ML estimates for this model) are used to set .",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "M-",
                "sec_num": null
            },
            {
                "text": "To begin the process, we did not begin at the Estep with an initial guess at . Rather, we began at the M-step, using an initial distribution over completions. The initial distribution was not the uniform distribution over binary trees P bin (B). That was undesirable as an initial point because, combinatorily, almost all trees are relatively balanced. On the other hand, in language, we want to allow unbalanced structures to have a reasonable chance to be discovered. Therefore, consider the following uniformsplitting process of generating binary trees over k terminals: choose a split point at random, then recursively build trees by this process on each side of the split. This process gives a distribution P split which puts relatively more weight on unbalanced trees, but only in a very general, non language-specific way. This distribution was not used in the model itself, however. It seemed to bias too strongly against balanced structures, and led to entirely linear-branching structures.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "M-",
                "sec_num": null
            },
            {
                "text": "The smoothing used was straightforward. For each yield \u03b1 or context x, we added 10 counts of that item as a constituent and 50 as a distituent. This reflected the relative skew of random spans being more likely to be distituents. This contrasts with our previous work, which was sensitive to smoothing method, and required a massive amount of it.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "M-",
                "sec_num": null
            },
            {
                "text": "We performed most experiments on the 7422 sentences in the Penn treebank Wall Street Journal section which contained no more than 10 words after the removal of punctuation and null elements (WSJ-10). Evaluation was done by measuring unlabeled precision, recall, and their harmonic mean F 1 against the treebank parses. Constituents which could not be gotten wrong (single words and entire sentences) were discarded. 3 The basic experiments, as described above, do not label constituents. An advantage to having only a single constituent class is that it encourages constituents of one type to be found even when they occur in a context which canonically holds another type. For example, NPs and PPs both occur between a verb and the end of the sentence, and they can transfer constituency to each other through that context. Figure 4 shows the F 1 score for various methods of parsing. RANDOM chooses a tree uniformly ",
                "cite_spans": [
                    {
                        "start": 416,
                        "end": 417,
                        "text": "3",
                        "ref_id": null
                    }
                ],
                "ref_spans": [
                    {
                        "start": 825,
                        "end": 833,
                        "text": "Figure 4",
                        "ref_id": null
                    }
                ],
                "eq_spans": [],
                "section": "Experiments",
                "sec_num": "4"
            },
            {
                "text": "\" ! $ # % Percent & $ & ( ' 0 ) 1 2 4 3 55 & $ & ( ' 0 6 7 & $ & ( ' 0 8 91 2 4 @A 4 @B C D $ E F 8 H G I 8 H & P 6 I Q R 6 7",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Experiments",
                "sec_num": "4"
            },
            {
                "text": "Figure 5: Accuracy scores for CCM-induced structures by span size. The drop in precision for span length 2 is largely due to analysis inside NPs which is omitted by the treebank. Also shown is F 1 for the induced PCFG. The PCFG shows higher accuracy on small spans, while the CCM is more even.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Experiments",
                "sec_num": "4"
            },
            {
                "text": "at random from the set of binary trees. 4 This is the unsupervised baseline. DEP-PCFG is the result of duplicating the experiments of Carroll and Charniak (1992) , using EM to train a dependencystructured PCFG. LBRANCH and RBRANCH choose the left-and right-branching structures, respectively. RBRANCH is a frequently used baseline for supervised parsing, but it should be stressed that it encodes a significant fact about English structure, and an induction system need not beat it to claim a degree of success. CCM is our system, as described above. SUP-PCFG is a supervised PCFG parser trained on a 90-10 split of this data, using the treebank grammar, with the Viterbi parse rightbinarized. 5 UBOUND is the upper bound of how well a binary system can do against the treebank sentences, which are generally flatter than binary, limiting the maximum precision. CCM is doing quite well at 71.1%, substantially better than right-branching structure. One common issue with grammar induction systems is a tendency to chunk in a bottom-up fashion. Especially since the CCM does not model recursive structure explicitly, one might be concerned that the high overall accuracy is due to a high accuracy on short-span constituents. Figure 5 shows that this is not true. Recall drops slightly for mid-size constituents, but longer constituents are as reliably proposed as short ones. Another effect illustrated in this graph is that, for span 2, constituents have low precision for their recall. This contrast is primarily due to the single largest difference between the system's induced structures and those in the treebank: the treebank does not parse into NPs such as DT JJ NN, while our system does, and generally does so correctly, identifying N units like JJ NN. This overproposal drops span-2 precision. In contrast, figure 5 also shows the F 1 for DEP-PCFG, which does exhibit a drop in F 1 over larger spans.",
                "cite_spans": [
                    {
                        "start": 134,
                        "end": 161,
                        "text": "Carroll and Charniak (1992)",
                        "ref_id": "BIBREF4"
                    }
                ],
                "ref_spans": [
                    {
                        "start": 1224,
                        "end": 1232,
                        "text": "Figure 5",
                        "ref_id": null
                    }
                ],
                "eq_spans": [],
                "section": "Experiments",
                "sec_num": "4"
            },
            {
                "text": "The top row of figure 8 shows the recall of nontrivial brackets, split according the brackets' labels in the treebank. Unsurprisingly, NP recall is highest, but other categories are also high. Because we ignore trivial constituents, the comparatively low S represents only embedded sentences, which are somewhat harder even for supervised systems.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Experiments",
                "sec_num": "4"
            },
            {
                "text": "To facilitate comparison to other recent work, figure 6 shows the accuracy of our system when trained on the same WSJ data, but tested on the ATIS corpus, and evaluated according to the EVALB program. 6 The F 1 numbers are lower for this corpus and evaluation method. 7 Still, CCM beats not only RBRANCH (by 8.3%), but also the previous conditional COND-CCM and the next closest unsupervised system (which does not beat RBRANCH in F 1 ).",
                "cite_spans": [
                    {
                        "start": 201,
                        "end": 202,
                        "text": "6",
                        "ref_id": null
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Experiments",
                "sec_num": "4"
            },
            {
                "text": "6 EMILE and ABL are lexical systems described in (van Zaanen, 2000; Adriaans and Haas, 1999) . CDC-40, from (Clark, 2001) , reflects training on much more data (12M words).",
                "cite_spans": [
                    {
                        "start": 49,
                        "end": 67,
                        "text": "(van Zaanen, 2000;",
                        "ref_id": "BIBREF18"
                    },
                    {
                        "start": 68,
                        "end": 92,
                        "text": "Adriaans and Haas, 1999)",
                        "ref_id": "BIBREF1"
                    },
                    {
                        "start": 108,
                        "end": 121,
                        "text": "(Clark, 2001)",
                        "ref_id": "BIBREF7"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Experiments",
                "sec_num": "4"
            },
            {
                "text": "7 The primary cause of the lower F 1 is that the ATIS corpus is replete with span-one NPs; adding an extra bracket around all single words raises our EVALB recall to 71.9; removing all unaries from the ATIS gold standard gives an F 1 of 63.3%. Figure 7 : Constituents most frequently over-and underproposed by our system.",
                "cite_spans": [],
                "ref_spans": [
                    {
                        "start": 244,
                        "end": 252,
                        "text": "Figure 7",
                        "ref_id": null
                    }
                ],
                "eq_spans": [],
                "section": "Experiments",
                "sec_num": "4"
            },
            {
                "text": "Parsing figures can only be a component of evaluating an unsupervised induction system. Low scores may indicate systematic alternate analyses rather than true confusion, and the Penn treebank is a sometimes arbitrary or even inconsistent gold standard. To give a better sense of the kinds of errors the system is or is not making, we can look at which sequences are most often over-proposed, or most often under-proposed, compared to the treebank parses. Figure 7 shows the 10 most frequently over-and under-proposed sequences. The system's main error trends can be seen directly from these two lists. It forms MD VB verb groups systematically, and it attaches the possessive particle to the right, like a determiner, rather than to the left. 8 It provides binarybranching analyses within NPs, normally resulting in correct extra N constituents, like JJ NN, which are not bracketed in the treebank. More seriously, it tends to attach post-verbal prepositions to the verb and gets confused by long sequences of nouns. A significant improvement over earlier systems is the absence of subject-verb groups, which disappeared when we switched to P split (B) for initial completions; the more balanced subject-verb analysis had a substantial combinatorial advantage with P bin (B).",
                "cite_spans": [],
                "ref_spans": [
                    {
                        "start": 455,
                        "end": 463,
                        "text": "Figure 7",
                        "ref_id": null
                    }
                ],
                "eq_spans": [],
                "section": "Error Analysis",
                "sec_num": "4.1"
            },
            {
                "text": "We also ran the system with multiple constituent classes, using a slightly more complex generative model in which the bracketing generates a labeling which then generates the constituents and contexts. The set of labels for constituent spans and distituent spans are forced to be disjoint.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Multiple Constituent Classes",
                "sec_num": "4.2"
            },
            {
                "text": "Intuitively, it seems that more classes should help, by allowing the system to distinguish different types of constituents and constituent contexts. However, it seemed to slightly hurt parsing accuracy overall. Figure 8 compares the performance for 2 versus 12 classes; in both cases, only one of the classes was allocated for distituents. Overall F 1 dropped very slightly with 12 classes, but the category recall numbers indicate that the errors shifted around substantially. PP accuracy is lower, which is not surprising considering that PPs tend to appear rather optionally and in contexts in which other, easier categories also frequently appear. On the other hand, embedded sentence recall is substantially higher, possibly because of more effective use of the top-level sentences which occur in the signature context -. The classes found, as might be expected, range from clearly identifiable to nonsense. Note that simply directly clustering all sequences into 12 categories produced almost entirely the latter, with clusters representing various distituent types. Figure 9 shows several of the 12 classes. Class 0 is the model's distituent class. Its most frequent members are a mix of obvious distituents (IN DT, DT JJ, IN DT, NN VBZ) and seemingly good sequences like NNP NNP. However, there are many sequences of 3 or more NNP tags in a row, and not all adjacent pairs can possibly be constituents at the same time. Class 1 is mainly common NP sequences, class 2 is proper NPs, class 3 is NPs which involve numbers, and class 6 is N sequences, which tend to be linguistically right but unmarked in the treebank. Class 4 is a mix of seemingly good NPs, often from positions like VBZ-NN where they were not constituents, and other sequences that share such contexts with otherwise good NP sequences. This is a danger of not jointly modeling yield and context, and of not modeling any kind of recursive structure. Class 5 is mainly composed of verb phrases and verb groups. No class corresponded neatly to PPs: perhaps because they have no signature contexts. The 2-class model is effective at identifying them only because they share contexts with a range of other constituent types (such as NPs and VPs).",
                "cite_spans": [],
                "ref_spans": [
                    {
                        "start": 211,
                        "end": 219,
                        "text": "Figure 8",
                        "ref_id": null
                    },
                    {
                        "start": 1073,
                        "end": 1081,
                        "text": "Figure 9",
                        "ref_id": "FIGREF2"
                    }
                ],
                "eq_spans": [],
                "section": "Multiple Constituent Classes",
                "sec_num": "4.2"
            },
            {
                "text": "A reasonable criticism of the experiments presented so far, and some other earlier work, is that we assume treebank part-of-speech tags as input. This criticism could be two-fold. First, state-of-the-art supervised PCFGs do not perform nearly so well with their input delexicalized. We may be reducing data sparsity and making it easier to see a broad picture of the grammar, but we are also limiting how well we can possibly do. It is certainly worth exploring methods which supplement or replace tagged input with lexical input. However, we address here the more serious criticism: that our results stem from clues latent in the treebank tagging information which are conceptually posterior to knowledge of structure. For instance, some treebank tag distinctions, such as particle (RP) vs. preposition (IN) or predeterminer (PDT) vs. determiner (DT) or adjective (JJ), could be said to import into the tagset distinctions that can only be made syntactically.",
                "cite_spans": [
                    {
                        "start": 804,
                        "end": 808,
                        "text": "(IN)",
                        "ref_id": null
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Induced Parts-of-Speech",
                "sec_num": "4.3"
            },
            {
                "text": "To show results from a complete grammar induction system, we also did experiments starting with a clustering of the words in the treebank. We used basically the baseline method of word type clustering in (Sch\u00fctze, 1995) (which is close to the methods of (Finch, 1993) ). For (all-lowercased) word types in the Penn treebank, a 1000 element vector was made by counting how often each co-occurred with each of the 500 most common words immediately to the left or right in Treebank text and additional 1994-96 WSJ newswire. These vectors were length-normalized, and then rank-reduced by an SVD, keeping the 50 largest singular vectors. The resulting vectors were clustered into 200 word classes by a weighted k-means algorithm, and then grammar induction operated over these classes. We do not believe that the quality of our tags matches that of the better methods of Sch\u00fctze (1995) , much less the recent results of Clark (2000) . Nevertheless, using these tags as input still gave induced structure substantially above right-branching. the performance with induced tags compared to correct tags. Overall F 1 has dropped, but, interestingly, VP and S recall are higher. This seems to be due to a marked difference between the induced tags and the treebank tags: nouns are scattered among a disproportionally large number of induced tags, increasing the number of common NP sequences, but decreasing the frequency of each.",
                "cite_spans": [
                    {
                        "start": 204,
                        "end": 219,
                        "text": "(Sch\u00fctze, 1995)",
                        "ref_id": "BIBREF16"
                    },
                    {
                        "start": 254,
                        "end": 267,
                        "text": "(Finch, 1993)",
                        "ref_id": "BIBREF8"
                    },
                    {
                        "start": 866,
                        "end": 880,
                        "text": "Sch\u00fctze (1995)",
                        "ref_id": "BIBREF16"
                    },
                    {
                        "start": 915,
                        "end": 927,
                        "text": "Clark (2000)",
                        "ref_id": "BIBREF6"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Induced Parts-of-Speech",
                "sec_num": "4.3"
            },
            {
                "text": "Another issue with previous systems is their sensitivity to initial choices. The conditional model of Klein and Manning (2001b) had the drawback that the variance of final F 1 , and qualitative grammars found, was fairly high, depending on small differences in first-round random parses. The model presented here does not suffer from this: while it is clearly sensitive to the quality of the input tagging, it is robust with respect to smoothing parameters and data splits. Varying the smoothing counts a factor of ten in either direction did not change the overall F 1 by more than 1%. Training on random subsets of the training data brought lower performance, but constantly lower over equal-size splits. Moreover, there are no first-round random decisions to be sensitive to; the soft EM procedure is deterministic. Figure 10 shows the overall F 1 score and the data likelihood according to our model during convergence. 9 Surprisingly, both are non-decreasing as the system iterates, indicating that data likelihood in this model corresponds well with parse accuracy. 10 Figure 11 shows recall for various categories by iteration. NP recall exhibits the more typical pattern of a sharp rise followed by a slow fall, but the other categories, after some initial drops, all increase until convergence. These graphs stop at 40 iterations. The system actually converged in both likelihood and F 1 by iteration 38, to within a tolerance of 10 \u221210 . The time to convergence varied according to smoothing amount, number of classes, and tags used, but the system almost always converged within 80 iterations, usually within 40.",
                "cite_spans": [
                    {
                        "start": 102,
                        "end": 127,
                        "text": "Klein and Manning (2001b)",
                        "ref_id": "BIBREF11"
                    }
                ],
                "ref_spans": [
                    {
                        "start": 819,
                        "end": 828,
                        "text": "Figure 10",
                        "ref_id": "FIGREF4"
                    },
                    {
                        "start": 1075,
                        "end": 1084,
                        "text": "Figure 11",
                        "ref_id": "FIGREF5"
                    }
                ],
                "eq_spans": [],
                "section": "Convergence and Stability",
                "sec_num": "4.4"
            },
            {
                "text": "We have presented a simple generative model for the unsupervised distributional induction of hierarchical linguistic structure. The system achieves the best published unsupervised parsing scores on the WSJ-10 and ATIS data sets. The induction algorithm combines the benefits of EM-based parameter search and distributional clustering methods. We have shown that this method acquires a substantial amount of correct structure, to the point that the most frequent discrepancies between the induced trees and the treebank gold standard are systematic alternate analyses, many of which are linguistically plausible. We have shown that the system is not reliant on supervised POS tag input, and demonstrated increased accuracy, speed, simplicity, and stability compared to previous systems.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Conclusions",
                "sec_num": "5"
            },
            {
                "text": "On this approach, the question of which rules are included or excluded becomes the question of which parameters are zero.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "",
                "sec_num": null
            },
            {
                "text": "Viewed as a model generating sentences, this model is deficient, placing mass on yield and context choices which will not tile into a valid sentence, either because specifications for positions conflict or because yields of incorrect lengths are chosen. However, we can renormalize by dividing by the mass placed on proper sentences and zeroing the probability of improper bracketings. The rest of the paper, and results, would be unchanged except for notation to track the renormalization constant.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "",
                "sec_num": null
            },
            {
                "text": "Since reproducible evaluation is important, a few more notes: this is different from the original (unlabeled) bracketing measures proposed in the PARSEVAL standard, which did not count single words as constituents, but did give points for putting a bracket over the entire sentence. Secondly, bracket labels and multiplicity are just ignored. Below, we also present results using the EVALB program for comparability, but we note that while one can get results from it that ignore bracket labels, it never ignores bracket multiplicity. Both these alternatives seem less satisfactory to us as measures for evaluating unsupervised constituency decisions.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "",
                "sec_num": null
            },
            {
                "text": "This is different from making random parsing decisions, which gave a higher score of 35%.5 Without post-binarization, the F 1 score was 88.9.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "",
                "sec_num": null
            },
            {
                "text": "Linguists have at times argued for both analyses:Halliday (1994) andAbney (1987), respectively.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "",
                "sec_num": null
            }
        ],
        "back_matter": [
            {
                "text": " 9 The data likelihood is not shown exactly, but rather we show the linear transformation of it calculated by the system. 10 Pereira and Schabes (1992) find otherwise for PCFGs.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "acknowledgement",
                "sec_num": null
            }
        ],
        "bib_entries": {
            "BIBREF0": {
                "ref_id": "b0",
                "title": "The English Noun Phrase in its Sentential Aspect",
                "authors": [
                    {
                        "first": "P",
                        "middle": [],
                        "last": "Stephen",
                        "suffix": ""
                    },
                    {
                        "first": "",
                        "middle": [],
                        "last": "Abney",
                        "suffix": ""
                    }
                ],
                "year": 1987,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Stephen P. Abney. 1987. The English Noun Phrase in its Sen- tential Aspect. Ph.D. thesis, MIT.",
                "links": null
            },
            "BIBREF1": {
                "ref_id": "b1",
                "title": "Grammar induction as substructural inductive logic programming",
                "authors": [
                    {
                        "first": "Pieter",
                        "middle": [],
                        "last": "Adriaans",
                        "suffix": ""
                    },
                    {
                        "first": "Erik",
                        "middle": [],
                        "last": "Haas",
                        "suffix": ""
                    }
                ],
                "year": 1999,
                "venue": "Proceedings of the 1st Workshop on Learning Language in Logic",
                "volume": "",
                "issue": "",
                "pages": "117--127",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Pieter Adriaans and Erik Haas. 1999. Grammar induction as substructural inductive logic programming. In James Cussens, editor, Proceedings of the 1st Workshop on Learn- ing Language in Logic, pages 117-127, Bled, Slovenia.",
                "links": null
            },
            "BIBREF2": {
                "ref_id": "b2",
                "title": "Trainable grammars for speech recognition",
                "authors": [
                    {
                        "first": "James",
                        "middle": [
                            "K"
                        ],
                        "last": "Baker",
                        "suffix": ""
                    }
                ],
                "year": 1979,
                "venue": "Speech Communication Papers for the 97th Meeting of the",
                "volume": "",
                "issue": "",
                "pages": "547--550",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "James K. Baker. 1979. Trainable grammars for speech recogni- tion. In D. H. Klatt and J. J. Wolf, editors, Speech Communi- cation Papers for the 97th Meeting of the Acoustical Society of America, pages 547-550.",
                "links": null
            },
            "BIBREF3": {
                "ref_id": "b3",
                "title": "Automatic grammar induction and parsing free text: A transformation-based approach",
                "authors": [
                    {
                        "first": "Eric",
                        "middle": [],
                        "last": "Brill",
                        "suffix": ""
                    }
                ],
                "year": 1993,
                "venue": "ACL 31",
                "volume": "",
                "issue": "",
                "pages": "259--265",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Eric Brill. 1993. Automatic grammar induction and parsing free text: A transformation-based approach. In ACL 31, pages 259-265.",
                "links": null
            },
            "BIBREF4": {
                "ref_id": "b4",
                "title": "Two experiments on learning probabilistic dependency grammars from corpora",
                "authors": [
                    {
                        "first": "Glenn",
                        "middle": [],
                        "last": "Carroll",
                        "suffix": ""
                    },
                    {
                        "first": "Eugene",
                        "middle": [],
                        "last": "Charniak",
                        "suffix": ""
                    },
                    {
                        "first": ";",
                        "middle": [
                            "C"
                        ],
                        "last": "Weir",
                        "suffix": ""
                    },
                    {
                        "first": "S",
                        "middle": [],
                        "last": "Abney",
                        "suffix": ""
                    },
                    {
                        "first": "R",
                        "middle": [],
                        "last": "Grishman",
                        "suffix": ""
                    },
                    {
                        "first": "R",
                        "middle": [],
                        "last": "Weischedel",
                        "suffix": ""
                    }
                ],
                "year": 1992,
                "venue": "editors, Working Notes of the Workshop Statistically-Based NLP Techniques",
                "volume": "",
                "issue": "",
                "pages": "1--13",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Glenn Carroll and Eugene Charniak. 1992. Two experiments on learning probabilistic dependency grammars from corpora. In C. Weir, S. Abney, R. Grishman, and R. Weischedel, edi- tors, Working Notes of the Workshop Statistically-Based NLP Techniques, pages 1-13. AAAI Press.",
                "links": null
            },
            "BIBREF5": {
                "ref_id": "b5",
                "title": "Bayesian grammar induction for language modeling",
                "authors": [
                    {
                        "first": "F",
                        "middle": [],
                        "last": "Stanley",
                        "suffix": ""
                    },
                    {
                        "first": "",
                        "middle": [],
                        "last": "Chen",
                        "suffix": ""
                    }
                ],
                "year": 1995,
                "venue": "ACL 33",
                "volume": "",
                "issue": "",
                "pages": "228--235",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Stanley F. Chen. 1995. Bayesian grammar induction for lan- guage modeling. In ACL 33, pages 228-235.",
                "links": null
            },
            "BIBREF6": {
                "ref_id": "b6",
                "title": "Inducing syntactic categories by context distribution clustering",
                "authors": [
                    {
                        "first": "Alexander",
                        "middle": [],
                        "last": "Clark",
                        "suffix": ""
                    }
                ],
                "year": 2000,
                "venue": "The Fourth Conference on Natural Language Learning",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Alexander Clark. 2000. Inducing syntactic categories by con- text distribution clustering. In The Fourth Conference on Natural Language Learning.",
                "links": null
            },
            "BIBREF7": {
                "ref_id": "b7",
                "title": "Unsupervised induction of stochastic context-free grammars using distributional clustering",
                "authors": [
                    {
                        "first": "Alexander",
                        "middle": [],
                        "last": "Clark",
                        "suffix": ""
                    }
                ],
                "year": 2001,
                "venue": "The Fifth Conference on Natural Language Learning",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Alexander Clark. 2001. Unsupervised induction of stochastic context-free grammars using distributional clustering. In The Fifth Conference on Natural Language Learning.",
                "links": null
            },
            "BIBREF8": {
                "ref_id": "b8",
                "title": "Finding Structure in Language",
                "authors": [
                    {
                        "first": "Steven",
                        "middle": [
                            "Paul"
                        ],
                        "last": "Finch",
                        "suffix": ""
                    }
                ],
                "year": 1993,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Steven Paul Finch. 1993. Finding Structure in Language. Ph.D. thesis, University of Edinburgh.",
                "links": null
            },
            "BIBREF9": {
                "ref_id": "b9",
                "title": "An introduction to functional grammar",
                "authors": [
                    {
                        "first": "M",
                        "middle": [
                            "A K"
                        ],
                        "last": "Halliday",
                        "suffix": ""
                    }
                ],
                "year": 1994,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "M. A. K. Halliday. 1994. An introduction to functional gram- mar. Edward Arnold, London, 2nd edition.",
                "links": null
            },
            "BIBREF10": {
                "ref_id": "b10",
                "title": "Distributional phrase structure induction",
                "authors": [
                    {
                        "first": "Dan",
                        "middle": [],
                        "last": "Klein",
                        "suffix": ""
                    },
                    {
                        "first": "Christopher",
                        "middle": [
                            "D"
                        ],
                        "last": "Manning",
                        "suffix": ""
                    }
                ],
                "year": 2001,
                "venue": "Proceedings of the Fifth Conference on Natural Language Learning",
                "volume": "",
                "issue": "",
                "pages": "113--120",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Dan Klein and Christopher D. Manning. 2001a. Distribu- tional phrase structure induction. In Proceedings of the Fifth Conference on Natural Language Learning (CoNLL 2001), pages 113-120.",
                "links": null
            },
            "BIBREF11": {
                "ref_id": "b11",
                "title": "Natural language grammar induction using a constituent-context model",
                "authors": [
                    {
                        "first": "Dan",
                        "middle": [],
                        "last": "Klein",
                        "suffix": ""
                    },
                    {
                        "first": "Christopher",
                        "middle": [
                            "D"
                        ],
                        "last": "Manning",
                        "suffix": ""
                    }
                ],
                "year": 2001,
                "venue": "Advances in Neural Information Processing Systems",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Dan Klein and Christopher D. Manning. 2001b. Natural lan- guage grammar induction using a constituent-context model. In Advances in Neural Information Processing Systems, vol- ume 14. MIT Press.",
                "links": null
            },
            "BIBREF12": {
                "ref_id": "b12",
                "title": "The estimation of stochastic context-free grammars using the inside-outside algorithm",
                "authors": [
                    {
                        "first": "K",
                        "middle": [],
                        "last": "Lari",
                        "suffix": ""
                    },
                    {
                        "first": "S",
                        "middle": [
                            "J"
                        ],
                        "last": "Young",
                        "suffix": ""
                    }
                ],
                "year": 1990,
                "venue": "Computer Speech and Language",
                "volume": "4",
                "issue": "",
                "pages": "35--56",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "K. Lari and S. J. Young. 1990. The estimation of stochastic context-free grammars using the inside-outside algorithm. Computer Speech and Language, 4:35-56.",
                "links": null
            },
            "BIBREF13": {
                "ref_id": "b13",
                "title": "Stochastic Grammars and Language Acquisition Mechanisms",
                "authors": [
                    {
                        "first": "Donald Cort",
                        "middle": [],
                        "last": "Olivier",
                        "suffix": ""
                    }
                ],
                "year": 1968,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Donald Cort Olivier. 1968. Stochastic Grammars and Language Acquisition Mechanisms. Ph.D. thesis, Harvard University.",
                "links": null
            },
            "BIBREF14": {
                "ref_id": "b14",
                "title": "Inside-outside reestimation from partially bracketed corpora",
                "authors": [
                    {
                        "first": "Fernando",
                        "middle": [],
                        "last": "Pereira",
                        "suffix": ""
                    },
                    {
                        "first": "Yves",
                        "middle": [],
                        "last": "Schabes",
                        "suffix": ""
                    }
                ],
                "year": 1992,
                "venue": "ACL 30",
                "volume": "",
                "issue": "",
                "pages": "128--135",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Fernando Pereira and Yves Schabes. 1992. Inside-outside rees- timation from partially bracketed corpora. In ACL 30, pages 128-135.",
                "links": null
            },
            "BIBREF15": {
                "ref_id": "b15",
                "title": "Transformational Grammar. Cambridge University Press",
                "authors": [
                    {
                        "first": "Andrew",
                        "middle": [],
                        "last": "Radford",
                        "suffix": ""
                    }
                ],
                "year": 1988,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Andrew Radford. 1988. Transformational Grammar. Cam- bridge University Press, Cambridge.",
                "links": null
            },
            "BIBREF16": {
                "ref_id": "b16",
                "title": "Distributional part-of-speech tagging",
                "authors": [
                    {
                        "first": "Hinrich",
                        "middle": [],
                        "last": "Sch\u00fctze",
                        "suffix": ""
                    }
                ],
                "year": 1995,
                "venue": "EACL 7",
                "volume": "",
                "issue": "",
                "pages": "141--148",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Hinrich Sch\u00fctze. 1995. Distributional part-of-speech tagging. In EACL 7, pages 141-148.",
                "links": null
            },
            "BIBREF17": {
                "ref_id": "b17",
                "title": "Inducing probabilistic grammars by Bayesian model merging",
                "authors": [
                    {
                        "first": "Andreas",
                        "middle": [],
                        "last": "Stolcke",
                        "suffix": ""
                    },
                    {
                        "first": "Stephen",
                        "middle": [
                            "M"
                        ],
                        "last": "Omohundro",
                        "suffix": ""
                    }
                ],
                "year": 1994,
                "venue": "Grammatical Inference and Applications: Proceedings of the Second International Colloquium on Grammatical Inference",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Andreas Stolcke and Stephen M. Omohundro. 1994. Induc- ing probabilistic grammars by Bayesian model merging. In Grammatical Inference and Applications: Proceedings of the Second International Colloquium on Grammatical Infer- ence. Springer Verlag.",
                "links": null
            },
            "BIBREF18": {
                "ref_id": "b18",
                "title": "ABL: Alignment-based learning",
                "authors": [
                    {
                        "first": "M",
                        "middle": [],
                        "last": "Van Zaanen",
                        "suffix": ""
                    }
                ],
                "year": 2000,
                "venue": "COLING 18",
                "volume": "",
                "issue": "",
                "pages": "961--967",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "M. van Zaanen. 2000. ABL: Alignment-based learning. In COLING 18, pages 961-967.",
                "links": null
            },
            "BIBREF19": {
                "ref_id": "b19",
                "title": "Learning syntax and meanings through optimization and distributional analysis",
                "authors": [
                    {
                        "first": "J",
                        "middle": [
                            "G"
                        ],
                        "last": "Wolff",
                        "suffix": ""
                    }
                ],
                "year": 1988,
                "venue": "Categories and processes in language acquisition",
                "volume": "",
                "issue": "",
                "pages": "179--215",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "J. G. Wolff. 1988. Learning syntax and meanings through optimization and distributional analysis. In Y. Levy, I. M. Schlesinger, and M. D. S. Braine, editors, Categories and processes in language acquisition, pages 179-215. Lawrence Erlbaum, Hillsdale, NJ.",
                "links": null
            }
        },
        "ref_entries": {
            "FIGREF0": {
                "type_str": "figure",
                "text": "Three bracketings of the sentence in figure 1: constituent spans in black. (b) corresponds to the binary parse infigure 1; (a)does not contain the 2,5 VP bracket, while (c) contains a 0,3 bracket crossing that VP bracket.",
                "uris": null,
                "num": null
            },
            "FIGREF1": {
                "type_str": "figure",
                "text": "Figure 4: F 1 for various models on WSJ-10.",
                "uris": null,
                "num": null
            },
            "FIGREF2": {
                "type_str": "figure",
                "text": "Most frequent members of several classes found.",
                "uris": null,
                "num": null
            },
            "FIGREF4": {
                "type_str": "figure",
                "text": "F 1 is non-decreasing until convergence.",
                "uris": null,
                "num": null
            },
            "FIGREF5": {
                "type_str": "figure",
                "text": "Recall by category during convergence.",
                "uris": null,
                "num": null
            },
            "TABREF2": {
                "html": null,
                "num": null,
                "type_str": "table",
                "content": "<table><tr><td colspan=\"2\">Classes Tags</td><td colspan=\"2\">Precision Recall</td><td>F 1</td><td colspan=\"4\">NP Recall PP Recall VP Recall S Recall</td></tr><tr><td/><td>2 Treebank</td><td>63.8</td><td>80.2</td><td>71.1</td><td>83.4</td><td>78.5</td><td>78.6</td><td>40.7</td></tr><tr><td colspan=\"2\">12 Treebank</td><td>63.6</td><td>80.0</td><td>70.9</td><td>82.2</td><td>59.1</td><td>82.8</td><td>57.0</td></tr><tr><td/><td>2 Induced</td><td>56.8</td><td>71.1</td><td>63.2</td><td>52.8</td><td>56.2</td><td>90.0</td><td>60.5</td></tr><tr><td colspan=\"2\">Class 0</td><td>Class 1</td><td>Class 2</td><td/><td>Class 3</td><td>Class 4</td><td>Class 5</td><td>Class 6</td></tr><tr><td colspan=\"2\">NNP NNP NN VBD</td><td>DT NN</td><td colspan=\"2\">NNP NNP</td><td>CD CD</td><td>VBN IN</td><td>MD VB</td><td>JJ NN</td></tr><tr><td>NN IN</td><td>NN NN</td><td>JJ NNS</td><td colspan=\"2\">NNP NNP NNP</td><td>CD NN</td><td>JJ IN</td><td colspan=\"2\">MD RB VB JJ NNS</td></tr><tr><td>IN DT</td><td colspan=\"2\">NNS VBP DT NNS</td><td colspan=\"2\">CC NNP</td><td>IN CD CD</td><td>DT NN</td><td>VBN IN</td><td>JJ JJ NN</td></tr><tr><td>DT JJ</td><td colspan=\"2\">NNS VBD DT JJ NN</td><td colspan=\"2\">POS NN</td><td>CD NNS</td><td>JJ CC</td><td colspan=\"2\">WDT VBZ CD NNS</td></tr><tr><td>NN VBZ</td><td>TO VB</td><td colspan=\"5\">NN NNS NNP NNP NNP NNP CD CD IN CD CD DT JJ NN</td><td>JJ IN</td><td>NNP NN</td></tr></table>",
                "text": "Figure 8: Scores for the 2-and 12-class model with Treebank tags, and the 2-class model with induced tags."
            }
        }
    }
}