File size: 62,864 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 | {
"paper_id": "P01-1005",
"header": {
"generated_with": "S2ORC 1.0.0",
"date_generated": "2023-01-19T09:30:13.408024Z"
},
"title": "Scaling to Very Very Large Corpora for Natural Language Disambiguation",
"authors": [
{
"first": "Michele",
"middle": [],
"last": "Banko",
"suffix": "",
"affiliation": {},
"email": "mbanko@microsoft.com"
},
{
"first": "Eric",
"middle": [],
"last": "Brill",
"suffix": "",
"affiliation": {},
"email": "brill@microsoft.com"
}
],
"year": "",
"venue": null,
"identifiers": {},
"abstract": "The amount of readily available on-line text has reached hundreds of billions of words and continues to grow. Yet for most core natural language tasks, algorithms continue to be optimized, tested and compared after training on corpora consisting of only one million words or less. In this paper, we evaluate the performance of different learning methods on a prototypical natural language disambiguation task, confusion set disambiguation, when trained on orders of magnitude more labeled data than has previously been used. We are fortunate that for this particular application, correctly labeled training data is free. Since this will often not be the case, we examine methods for effectively exploiting very large corpora when labeled data comes at a cost.",
"pdf_parse": {
"paper_id": "P01-1005",
"_pdf_hash": "",
"abstract": [
{
"text": "The amount of readily available on-line text has reached hundreds of billions of words and continues to grow. Yet for most core natural language tasks, algorithms continue to be optimized, tested and compared after training on corpora consisting of only one million words or less. In this paper, we evaluate the performance of different learning methods on a prototypical natural language disambiguation task, confusion set disambiguation, when trained on orders of magnitude more labeled data than has previously been used. We are fortunate that for this particular application, correctly labeled training data is free. Since this will often not be the case, we examine methods for effectively exploiting very large corpora when labeled data comes at a cost.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Abstract",
"sec_num": null
}
],
"body_text": [
{
"text": "Machine learning techniques, which automatic ally learn linguistic information from online text corpora, have been applied to a number of natural language problems throughout the last decade. A large percentage of papers published in this area involve comparisons of different learning approaches trained and tested with commonly used corpora. While the amount of available online text has been increasing at a dramatic rate, the size of training corpora typically used for learning has not. In part, this is due to the standardization of data sets used within the field, as well as the potentially large cost of annotating data for those learning methods that rely on labeled text.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "The empirical NLP community has put substantial effort into evaluating performance of a large number of machine learning methods over fixed, and relatively small, data sets. Yet since we now have access to significantly more data, one has to wonder what conclusions that have been drawn on small data sets may carry over when these learning methods are trained using much larger corpora.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "In this paper, we present a study of the effects of data size on machine learning for natural language disambiguation. In particular, we study the problem of selection among confusable words, using orders of magnitude more training data than has ever been applied to this problem. First we show learning curves for four different machine learning algorithms. Next, we consider the efficacy of voting, sample selection and partially unsupervised learning with large training corpora, in hopes of being able to obtain the benefits that come from significantly larger training corpora without incurring too large a cost.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "Confusion set disambiguation is the problem of choosing the correct use of a word, given a set of words with which it is commonly confused. Example confusion sets include: {principle , principal}, {then, t han}, {to,two,t oo}, and {weather,whether}. Numerous methods have been presented for confusable disambiguation. The more recent set of techniques includes mult iplicative weightupdate algorithms (Golding and Roth, 1998) , latent semantic analysis (Jones and Martin, 1997) , transformation-based learning (Mangu and Brill, 1997) , differential grammars (Powers, 1997) , decision lists (Yarowsky, 1994) , and a variety of Bayesian classifiers (Gale et al., 1993 , Golding, 1995 , Golding and Schabes, 1996 . In all of these approaches, the problem is formulated as follows:",
"cite_spans": [
{
"start": 401,
"end": 425,
"text": "(Golding and Roth, 1998)",
"ref_id": null
},
{
"start": 453,
"end": 477,
"text": "(Jones and Martin, 1997)",
"ref_id": "BIBREF10"
},
{
"start": 510,
"end": 533,
"text": "(Mangu and Brill, 1997)",
"ref_id": "BIBREF13"
},
{
"start": 558,
"end": 572,
"text": "(Powers, 1997)",
"ref_id": "BIBREF18"
},
{
"start": 590,
"end": 606,
"text": "(Yarowsky, 1994)",
"ref_id": "BIBREF21"
},
{
"start": 647,
"end": 665,
"text": "(Gale et al., 1993",
"ref_id": "BIBREF5"
},
{
"start": 666,
"end": 681,
"text": ", Golding, 1995",
"ref_id": "BIBREF6"
},
{
"start": 682,
"end": 709,
"text": ", Golding and Schabes, 1996",
"ref_id": "BIBREF8"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Confusion Set Disambiguation",
"sec_num": "2"
},
{
"text": "Given a specific confusion set (e.g. {to,two,too}), all occurrences of confusion set members in the test set are replaced by a marker; everywhere the system sees this marker, it must decide which member of the confusion set to choose.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Confusion Set Disambiguation",
"sec_num": "2"
},
{
"text": "Confusion set disambiguation is one of a class of natural language problems involving disambiguation from a relatively small set of alternatives based upon the string context in which the ambiguity site appears. Other such problems include word sense disambiguation, part of speech tagging and some formulations of phrasal chunking. One advantageous aspect of confusion set disambiguation, which allows us to study the effects of large data sets on performance, is that labeled training data is essentially free, since the correct answer is surface apparent in any collection of reasonably well-edited text.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Confusion Set Disambiguation",
"sec_num": "2"
},
{
"text": "This work was partially motivated by the desire to develop an improved grammar checker. Given a fixed amount of time, we considered what would be the most effective way to focus our efforts in order to attain the greatest performance improvement. Some possibilities included modifying standard learning algorithms, exploring new learning techniques, and using more sophisticated features. Before exploring these somewhat expensive paths, we decided to first see what happened if we simply trained an e xisting method with much more data. This led to the exploration of learning curves for various machine learning algorithms : winnow 1 , perceptron, na\u00efve Bayes, and a very simple memory-based learner. For the first three learners, we used the standard colle ction of features employed for this problem: the set of words within a window of the target word, and collocations containing words and/or parts of We collected a 1-billion-word training corpus from a variety of English texts, including news articles, scientific abstracts, government transcripts, literature and other varied forms of prose. This training corpus is three orders of magnitude greater than the largest training corpus previously used for this problem. We used 1 million words of Wall Street Journal text as our test set, and no data from the Wall Street Journal was used when constructing the training corpus. Each learner was trained at several cutoff points in the training corpus, i.e. the first one million words, the first five million words, and so on, until all one billion words were used for training. In order to avoid training biases that may result from merely concatenating the different data sources to form a larger training corpus, we constructed each consecutive training corpus by probabilistically sampling sentences from the different sources weighted by the size of each source.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Learning Curve Expe riments",
"sec_num": "3"
},
{
"text": "In Figure 1 , we show learning curves for each learner, up to one billion words of training data. Each point in the graph is the average performance over ten confusion sets for that size training corpus. Note that the curves appear to be log-linear even out to one billion words.",
"cite_spans": [],
"ref_spans": [
{
"start": 3,
"end": 11,
"text": "Figure 1",
"ref_id": "FIGREF0"
}
],
"eq_spans": [],
"section": "Learning Curve Expe riments",
"sec_num": "3"
},
{
"text": "Of course for many problems, additional training data has a non-zero cost. However, these results suggest that we may want to reconsider the trade-off between spending time and money on algorithm development versus spending it on corpus development. At least for the problem of confusable disambiguation, none of the learners tested is close to asymptoting in performance at the training corpus size commonly employed by the field.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Learning Curve Expe riments",
"sec_num": "3"
},
{
"text": "Such gains in accuracy, however, do not come for free. Figure 2 shows the size of learned representations as a function of training data size. For some applications, this is not necessarily a concern. But for others, where space comes at a premium, obtaining the gains that come with a billion words of training data may not be viable without an effort made to compress information. In such cases, one could look at numerous methods for compressing data (e.g. Dagan and Engleson, 1995, Weng, et al, 1998) .",
"cite_spans": [
{
"start": 460,
"end": 469,
"text": "Dagan and",
"ref_id": "BIBREF4"
},
{
"start": 470,
"end": 504,
"text": "Engleson, 1995, Weng, et al, 1998)",
"ref_id": null
}
],
"ref_spans": [
{
"start": 55,
"end": 63,
"text": "Figure 2",
"ref_id": "FIGREF1"
}
],
"eq_spans": [],
"section": "Learning Curve Expe riments",
"sec_num": "3"
},
{
"text": "Voting has proven to be an effective technique for improving classifier accuracy for many applications, including part-of-speech tagging (van Halteren, et al, 1998) , parsing (Henderson and Brill, 1999) , and word sense disambiguation (Pederson, 2000) . By training a set of classifiers on a single training corpus and then combining their outputs in classification, it is often possible to achieve a target accuracy with less labeled training data than would be needed if only one cla ssifier was being used. Voting can be effective in reducing both the bias of a particular training corpus and the bias of a specific learner. When a training corpus is very small, there is much more room for these biases to surface and therefore for voting to be effective. But does voting still offer performance gains when classifiers are trained on much larger corpora? The complementarity between two learners was defined by Brill and Wu (1998) in order to quantify the percentage of time when one system is wrong, that another system is correct, and therefore providing an upper bound on combination accuracy. As training size increases significantly, we would expect complementarity between classifiers to decrease. This is due in part to the fact that a l arger training corpus will reduce the data set variance and any bias arising from this. Also, some of the differences between classifiers might be due to how they handle a sparse training set. As a result of comparing a sample of two learners as a function of increasingly large training sets, we see in Table 1 that complementarity does indeed decrease as training size increases.",
"cite_spans": [
{
"start": 137,
"end": 164,
"text": "(van Halteren, et al, 1998)",
"ref_id": "BIBREF19"
},
{
"start": 175,
"end": 202,
"text": "(Henderson and Brill, 1999)",
"ref_id": "BIBREF9"
},
{
"start": 235,
"end": 251,
"text": "(Pederson, 2000)",
"ref_id": null
},
{
"start": 915,
"end": 934,
"text": "Brill and Wu (1998)",
"ref_id": "BIBREF2"
}
],
"ref_spans": [
{
"start": 1553,
"end": 1560,
"text": "Table 1",
"ref_id": null
}
],
"eq_spans": [],
"section": "The Efficacy of Voting",
"sec_num": "4"
},
{
"text": "Training Size (words) Complementarity(L1,L2) 10 6 0.2612 10 7 0.2410 10 8 0.1759 10 9 0.1612 Table 1 . Complementarity Next we tested whether this decrease in complementarity meant that voting loses its effectiveness as the training set increases. To examine the impact of voting when using a significantly larger training corpus, we ran 3 out of the 4 learners on our set of 10 confusable pairs, excluding the memory-based learner. Voting was done by combining the normalized score each learner assigned to a classification choice. In Figure 3 , we show the accuracy obtained from voting, along with the single best learner accuracy at each training set size. We see that for very small corpora, voting is beneficial, resulting in better performance than any single classifier. Beyond 1 million words, little is gained by voting, and indeed on the largest training sets voting actually hurts accuracy. ",
"cite_spans": [],
"ref_spans": [
{
"start": 93,
"end": 100,
"text": "Table 1",
"ref_id": null
},
{
"start": 536,
"end": 544,
"text": "Figure 3",
"ref_id": null
}
],
"eq_spans": [],
"section": "The Efficacy of Voting",
"sec_num": "4"
},
{
"text": "While the observation that learning curves are not asymptoting even with orders of magnitude more training data than is currently used is very exciting, this result may have somewhat limited ramifications. V ery few problems exist for which annotated data of this size is available for free. Surely we cannot reasonably expect that the manual annotation of one billion words along with corresponding parse trees will occur any time soon (but see (Banko and Brill 2001) for a discussion that this might not be completely infeasible). Despite this pitfall, there are techniques one can use to try to obtain the benefits of considerably larger training corpora without incurring significant additional costs. In the sections that follow, we study two such solutions: active learning and unsupervised learning.",
"cite_spans": [
{
"start": 446,
"end": 468,
"text": "(Banko and Brill 2001)",
"ref_id": "BIBREF0"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "When Annotated Data Is Not Free",
"sec_num": "5"
},
{
"text": "Active learning involves intelligently selecting a portion of samples for annotation from a pool of as-yet unannotated training samples. Not all samples in a training set are equally useful. By concentrating human annotation efforts on the samples of greatest utility to the machine learning algorithm, it may be possible to attain better performance for a fixed annotation cost than if samples were chosen randomly for human annotation. Most active learning approaches work by first training a seed learner (or family of learners) and then running the learner(s) over a set of unlabeled samples.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Active Learning",
"sec_num": "5.1"
},
{
"text": "A sample is presumed to be more useful for training the more uncertain its classification label is. Uncertainty can be judged by the relative weights assigned to different labels by a single classifier (Lewis and Catlett, 1994) . Another approach, committee-based sampling, first creates a committee of classifie rs and then judges classification uncertainty according to how much the learners differ among label assignments. For example, Dagan and Engelson (1995) describe a committee-based sampling technique where a part of speech tagger is trained using an annotated seed corpus. A family of taggers is then generated by randomly permuting the tagger probabilities, and the disparity among tags output by the committee members is used as a measure of classification uncertainty. Sentences for human annotation are drawn, biased to prefer those containing high uncertainty instances.",
"cite_spans": [
{
"start": 202,
"end": 227,
"text": "(Lewis and Catlett, 1994)",
"ref_id": "BIBREF12"
},
{
"start": 439,
"end": 464,
"text": "Dagan and Engelson (1995)",
"ref_id": "BIBREF4"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Active Learning",
"sec_num": "5.1"
},
{
"text": "While active learning has been shown to work for a number of tasks, the majority of active learning experiments in natural language processing have been conducted using very small seed corpora and sets of unlabeled examples. Therefore, we wish to explore situations where we have, or can afford, a nonnegligible sized training corpus (such as for part-of-speech tagging) and have access to very large amounts of unlabeled data.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Active Learning",
"sec_num": "5.1"
},
{
"text": "We can use bagging (Breiman, 1996) , a technique for generating a committee of classifiers, to assess the label uncertainty of a potential training instance. With bagging, a variant of the original training set is constructed by randomly sampling sentences with replacement from the source training set in order to produce N new training sets of size equal to the original. After the N models have been trained and run on the same test set, their classifications for each test sentence can be compared for classification agreement. The higher the disagreement between classifiers, the more useful it would be to have an instance We used the na\u00efve Bayes classifier, creating 10 classifiers each trained on bags generated from an initial one million words of labeled training data. We present the active learning algorithm we used below.",
"cite_spans": [
{
"start": 19,
"end": 34,
"text": "(Breiman, 1996)",
"ref_id": "BIBREF1"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Active Learning",
"sec_num": "5.1"
},
{
"text": "Initialize: Training data consists of X words correctly labeled Iterate : 1) Generate a committee of classifiers using bagging on the training set 2) Run the committee on unlabeled portion of the training set 3) Choose M instances from the unlabeled set for labeling -pick the M/2 with the greatest vote entropy and then pick another M/2 randomly -and add to training set",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Active Learning",
"sec_num": "5.1"
},
{
"text": "We initially tried selecting the M most uncertain examples, but this resulted in a sample too biased toward the difficult instances. Instead we pick half of our samples for annotation randomly and the other half from those whose labels we are most uncertain of, as judged by the entropy of the votes assigned to the instance by the committee. This is, in effect, biasing our sample toward instances the classifiers are most uncertain of.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Active Learning",
"sec_num": "5.1"
},
{
"text": "We show the results from sample selection for confusion set disambiguation in Figure 4 . The line labeled \"sequential\" shows test set accuracy achieved for different percentages of the one billion word training set, where training instances are taken at random. We ran three a ctive learning experiments, increasing the size of the total unlabeled training corpus from which we can pick samples to be annotated. In all three cases, sample selection outperforms sequential sampling.",
"cite_spans": [],
"ref_spans": [
{
"start": 78,
"end": 86,
"text": "Figure 4",
"ref_id": "FIGREF3"
}
],
"eq_spans": [],
"section": "Active Learning",
"sec_num": "5.1"
},
{
"text": "At the endpoint of each training run in the graph, the same number of samples has been annotated for training. However, we see that the larger the pool of candidate instances for annotation is, the better the resulting accuracy. By increasing the pool of unlabel ed training instances for active learning, we can improve accuracy with only a fixed additional annotation cost. Thus it is possible to benefit from the availability of extremely large corpora without incurring the full costs of annotation, training time, and representation size.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Active Learning",
"sec_num": "5.1"
},
{
"text": "While the previous section shows that we can benefit from substantially larger training corpora without needing significant additional manual annotation, it would be ideal if we could improve classification accuracy using only our seed annotated corpus and the large unlabeled corpus, without requiring any additional hand labeling. In this section we turn to unsupervised learning in an attempt to achieve this goal. Numerous approaches have been explored for exploiting situations where some a mount of annotated data is available and a much larger amount of data exists unannotated, e.g. Marialdo's HMM part-of-speech tagger training (1994), Charniak's parser retraining experiment (1996), Yarowsky's seeds for word sense disambiguation (1995) and Nigam et al's (1998) topic classifier learned in part from unlabelled documents. A nice discussion of this general problem can be found in Mitchell (1999) .",
"cite_spans": [
{
"start": 740,
"end": 746,
"text": "(1995)",
"ref_id": null
},
{
"start": 751,
"end": 771,
"text": "Nigam et al's (1998)",
"ref_id": "BIBREF16"
},
{
"start": 890,
"end": 905,
"text": "Mitchell (1999)",
"ref_id": "BIBREF15"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Weakly Supe rvised Learning",
"sec_num": "5.2"
},
{
"text": "The question we want to answer is whether there is something to be gained by combining unsupervised and supervised learning when we scale up both the seed corpus and the unlabeled corpus significantly. We can again use a committee of bagged classifiers, this time for unsupervised learning. Whereas with active learning we want to choose the most uncertain instances for human annotation, with unsupervised learning we want to choose the instances that have the highest probability of being correct for automatic labeling and inclusion in our labeled training data.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Weakly Supe rvised Learning",
"sec_num": "5.2"
},
{
"text": "In Table 2 , we show the test set accuracy (averaged over the four most frequently occurring confusion pairs) as a function of the number of classifiers that agree upon the label of an instance. For this experiment, we trained a collection of 10 na\u00efve Bayes classifiers, using bagging on a 1-million-word seed corpus. As can be seen, the greater the classifier agreement, the more likely it is that a test sample has been correctly labeled. Since the instances in which all bags agree have the highest probability of being correct, we attempted to automatically grow our labeled training set using the 1 -million-word labeled seed corpus along with the collection of na\u00efve Bayes classifiers described above. All instances from the remainder of the corpus on which all 10 classifiers agreed were selected, trusting the agreed-upon label. The classif iers were then retrained using the labeled seed corpus plus the new training material collected automatically during the previous step.",
"cite_spans": [],
"ref_spans": [
{
"start": 3,
"end": 10,
"text": "Table 2",
"ref_id": null
}
],
"eq_spans": [],
"section": "Weakly Supe rvised Learning",
"sec_num": "5.2"
},
{
"text": "In Table 3 we show the results from these unsupervised learning experiments for two confusion sets. In both cases we gain from unsupervised training compared to using only the seed corpus, but only up to a point. At this point, test set accuracy begins to decline as additional training instances are automatically harvested. We are able to attain improvements in accuracy for free using unsupervised learning, but unlike our learning curve experiments using correctly labeled data, accuracy does not continue to improve with additional data. Table 3 . Committee-Based Unsupervised Learning Charniak (1996) ran an experiment in which he trained a parser on one million words of parsed data, ran the parser over an additional 30 million words, and used the resulting parses to reestimate model probabilities. Doing so gave a small improvement over just using the manually parsed data.",
"cite_spans": [
{
"start": 591,
"end": 606,
"text": "Charniak (1996)",
"ref_id": "BIBREF3"
}
],
"ref_spans": [
{
"start": 3,
"end": 10,
"text": "Table 3",
"ref_id": null
},
{
"start": 543,
"end": 550,
"text": "Table 3",
"ref_id": null
}
],
"eq_spans": [],
"section": "Weakly Supe rvised Learning",
"sec_num": "5.2"
},
{
"text": "We repeated this experiment with our data, and show the outcome in Table 4 . Choosing only the labeled instances most likely to be correct as judged by a committee of classifiers results in higher accuracy than using all instances classified by a model trained with the labeled seed corpus. In applying unsupervised learning to improve upon a seed-trained method, we consistently saw an improvement in performance followed by a decline. This is likely due to eventually having reached a point where the gains from additional training data are offset by the sample bias in mining these instances. It may be possible to combine active learning with unsupervised learning as a way to reduce this sample bias and gain the benefits of both approaches.",
"cite_spans": [],
"ref_spans": [
{
"start": 67,
"end": 74,
"text": "Table 4",
"ref_id": null
}
],
"eq_spans": [],
"section": "Weakly Supe rvised Learning",
"sec_num": "5.2"
},
{
"text": "In this paper, we have looked into what happens when we begin to take advantage of the large amounts of text that are now readily available. We have shown that for a prototypical natural language classification task, the performance of learners can benefit significantly from much larger training sets. We have also shown that both active learning and unsupervised learning can be used to attain at least some of the advantage that comes with additional training data, while minimizing the cost of additional human annotation. We propose that a logical next step for the research community would be to direct efforts towards increasing the size of annotated training collections, while deemphasizing the focus on comparing different learning techniques trained only on small training corpora. While it is encouraging that there is a vast amount of on-line text, much work remains to be done if we are to learn how best to exploit this resource to improve natural language processing.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Conclusions",
"sec_num": "6"
}
],
"back_matter": [],
"bib_entries": {
"BIBREF0": {
"ref_id": "b0",
"title": "Mitigating the Paucity of Data Problem",
"authors": [
{
"first": "M",
"middle": [],
"last": "Banko",
"suffix": ""
},
{
"first": "E",
"middle": [],
"last": "Brill",
"suffix": ""
}
],
"year": 2001,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Banko, M. and Brill, E. (2001). Mitigating the Paucity of Data Problem. Human Language Technology.",
"links": null
},
"BIBREF1": {
"ref_id": "b1",
"title": "Bagging Predictors",
"authors": [
{
"first": "L",
"middle": [],
"last": "Breiman",
"suffix": ""
}
],
"year": 1996,
"venue": "Machine Learning",
"volume": "24",
"issue": "",
"pages": "123--140",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Breiman L., (1996). Bagging Predictors, Machine Learning 24 123-140.",
"links": null
},
"BIBREF2": {
"ref_id": "b2",
"title": "Classifier combination for improved lexical disambiguation",
"authors": [
{
"first": "E",
"middle": [],
"last": "Brill",
"suffix": ""
},
{
"first": "J",
"middle": [],
"last": "Wu",
"suffix": ""
}
],
"year": 1998,
"venue": "Proceedings of the 17th International Conference on Computational Linguistics",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Brill, E. and Wu, J. (1998). Classifier combination for improved lexical disambiguation. In Proceedings of the 17th International Conference on Computational Linguistics.",
"links": null
},
"BIBREF3": {
"ref_id": "b3",
"title": "Treebank Grammars, Proceedings AAAI-96",
"authors": [
{
"first": "E",
"middle": [],
"last": "Charniak",
"suffix": ""
}
],
"year": 1996,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Charniak, E. (1996). Treebank Grammars, Proceedings AAAI-96 , Menlo Park, Ca.",
"links": null
},
"BIBREF4": {
"ref_id": "b4",
"title": "Committee-based sampling for training probabilistic classifiers",
"authors": [
{
"first": "I",
"middle": [],
"last": "Dagan",
"suffix": ""
},
{
"first": "S",
"middle": [],
"last": "Engelson",
"suffix": ""
}
],
"year": 1995,
"venue": "Proc. ML-95, the 12th Int. Conf. on Machine Learning",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Dagan, I. and Engelson, S. (1995). Committee-based sampling for training probabilistic classifiers. In Proc. ML-95, the 12th Int. Conf. on Machine Learning.",
"links": null
},
"BIBREF5": {
"ref_id": "b5",
"title": "A method for disambiguating word senses in a large corpus",
"authors": [
{
"first": "W",
"middle": [
"A"
],
"last": "Gale",
"suffix": ""
},
{
"first": "K",
"middle": [
"W"
],
"last": "Church",
"suffix": ""
},
{
"first": "Yarowsky",
"middle": [],
"last": "",
"suffix": ""
},
{
"first": "D",
"middle": [],
"last": "",
"suffix": ""
}
],
"year": 1993,
"venue": "Computers and the Humanities",
"volume": "26",
"issue": "",
"pages": "415--439",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Gale, W. A., Church, K. W., and Yarowsky, D. (1993). A method for disambiguating word senses in a large corpus. Computers and the Humanities, 26:415--439.",
"links": null
},
"BIBREF6": {
"ref_id": "b6",
"title": "A Bayesian hybrid method for context-sensitive spelling correction",
"authors": [
{
"first": "A",
"middle": [
"R"
],
"last": "Golding",
"suffix": ""
}
],
"year": 1995,
"venue": "Proc. 3rd Workshop on Very Large Corpora",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Golding, A. R. (1995). A Bayesian hybrid method for context-sensitive spelling correction. In Proc. 3rd Workshop on Very Large Corpora, Boston, MA.",
"links": null
},
"BIBREF7": {
"ref_id": "b7",
"title": "A Winnow-Based Approach to Context-Sensitive Spelling Correction",
"authors": [
{
"first": "A",
"middle": [
"R"
],
"last": "Golding",
"suffix": ""
},
{
"first": "D",
"middle": [],
"last": "Roth",
"suffix": ""
}
],
"year": 1999,
"venue": "Machine Learning",
"volume": "34",
"issue": "",
"pages": "107--130",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Golding, A. R. and Roth, D.(1999), A Winnow- Based Approach to Context-Sensitive Spelling Correction. Machine Learning, 34:107--130.",
"links": null
},
"BIBREF8": {
"ref_id": "b8",
"title": "Combining trigram-based and feature-based methods for context-sensitive spelling correction",
"authors": [
{
"first": "A",
"middle": [
"R"
],
"last": "Golding",
"suffix": ""
},
{
"first": "Y",
"middle": [],
"last": "Schabes",
"suffix": ""
}
],
"year": 1996,
"venue": "Proc. 34th Annual Meeting of the Association for Computational Linguistics",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Golding, A. R. and Schabes, Y. (1996). Combining trigram-based and feature-based methods for context-sensitive spelling correction. In Proc. 34th Annual Meeting of the Association for Computational Linguistics, Santa Cruz, CA.",
"links": null
},
"BIBREF9": {
"ref_id": "b9",
"title": "Exploiting diversity in natural language processing: combining parsers",
"authors": [
{
"first": "J",
"middle": [
"C"
],
"last": "Henderson",
"suffix": ""
},
{
"first": "E",
"middle": [],
"last": "Brill",
"suffix": ""
}
],
"year": 1999,
"venue": "1999 Joint Sigdat Conference on Empirical Methods in Natural Language Processing and Very Large Corpora. ACL",
"volume": "",
"issue": "",
"pages": "187--194",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Henderson, J. C. and Brill, E . (1999). Exploiting diversity in natural language processing: combining parsers. In 1999 Joint Sigdat Conference on Empirical Methods in Natural Language Processing and Very Large Corpora. ACL, New Brunswick NJ. 187-194.",
"links": null
},
"BIBREF10": {
"ref_id": "b10",
"title": "Contextual spelling correction using latent semantic analysis",
"authors": [
{
"first": "M",
"middle": [
"P"
],
"last": "Jones",
"suffix": ""
},
{
"first": "J",
"middle": [
"H"
],
"last": "Martin",
"suffix": ""
}
],
"year": 1997,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Jones, M. P. and Martin, J. H. (1997). Contextual spelling correction using latent semantic analysis.",
"links": null
},
"BIBREF11": {
"ref_id": "b11",
"title": "Proc. 5th Conference on Applied Natural Language Processing",
"authors": [],
"year": null,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "In Proc. 5th Conference on Applied Natural Language Processing, Washington, DC.",
"links": null
},
"BIBREF12": {
"ref_id": "b12",
"title": "Heterogeneous uncertainty sampling",
"authors": [
{
"first": "D",
"middle": [
"D"
],
"last": "Lewis",
"suffix": ""
},
{
"first": "J",
"middle": [],
"last": "Catlett",
"suffix": ""
}
],
"year": 1994,
"venue": "Proceedings of the Eleventh International Conference on Machine Learning",
"volume": "",
"issue": "",
"pages": "148--156",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Lewis , D. D., & Catlett, J. (1994). Heterogeneous uncertainty sampling. Proceedings of the Eleventh International Conference on Machine Learning (pp. 148--156). New Brunswick, NJ: Morgan Kaufmann.",
"links": null
},
"BIBREF13": {
"ref_id": "b13",
"title": "Automatic rule acquisition for spelling correction",
"authors": [
{
"first": "L",
"middle": [],
"last": "Mangu",
"suffix": ""
},
{
"first": "E",
"middle": [],
"last": "Brill",
"suffix": ""
}
],
"year": 1997,
"venue": "Proc. 14th International Conference on Machine Learning",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Mangu, L. and Brill, E. (1997). Automatic rule acquisition for spelling correction. In Proc. 14th International Conference on Machine Learning. Morgan Kaufmann.",
"links": null
},
"BIBREF14": {
"ref_id": "b14",
"title": "Tagging English text with a probabilistic model",
"authors": [
{
"first": "B",
"middle": [],
"last": "Merialdo",
"suffix": ""
}
],
"year": 1994,
"venue": "Computational Linguistics",
"volume": "20",
"issue": "2",
"pages": "155--172",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Merialdo, B. (1994). Tagging English text with a probabilistic model. Computational Linguistics, 20(2):155--172.",
"links": null
},
"BIBREF15": {
"ref_id": "b15",
"title": "The role of unlabeled data in supervised learning",
"authors": [
{
"first": "T",
"middle": [
"M"
],
"last": "Mitchell",
"suffix": ""
}
],
"year": 1999,
"venue": "Proceedings of the Sixth International Colloquium on Cognitive Science",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Mitchell, T. M. (1999), The role of unlabeled data in supervised learning, in Proceedings of the Sixth International Colloquium on Cognitive Science, San Sebastian, Spain.",
"links": null
},
"BIBREF16": {
"ref_id": "b16",
"title": "Learning to classify text from labeled and unlabeled documents",
"authors": [
{
"first": "N",
"middle": [],
"last": "Nigam",
"suffix": ""
},
{
"first": "A",
"middle": [],
"last": "Mccallum",
"suffix": ""
},
{
"first": "S",
"middle": [],
"last": "Thrun",
"suffix": ""
},
{
"first": "T",
"middle": [],
"last": "Mitchell",
"suffix": ""
}
],
"year": 1998,
"venue": "Proceedings of the Fifteenth National Conference on Artificial Intelligence",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Nigam, N., McCallum, A., Thrun, S., and Mitchell, T. (1998). Learning to classify text from labeled and unlabeled documents. In Proceedings of the Fifteenth National Conference on Artificial Intelligence. AAAI Press..",
"links": null
},
"BIBREF17": {
"ref_id": "b17",
"title": "A simple approach to building ensembles of naive bayesian classifiers for word sense disambiguation",
"authors": [
{
"first": "T",
"middle": [],
"last": "Pedersen",
"suffix": ""
}
],
"year": 2000,
"venue": "Proceedings of the First Meeting of the North American Chapter of the Association for Computational Linguistics",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Pedersen, T. (2000). A simple approach to building ensembles of naive bayesian classifiers for word sense disambiguation. In Proceedings of the First Meeting of the North American Chapter of the Association for Computational Linguistics May 1- 3, 2000, Seattle, WA",
"links": null
},
"BIBREF18": {
"ref_id": "b18",
"title": "Learning and application of differential grammars",
"authors": [
{
"first": "D",
"middle": [],
"last": "Powers",
"suffix": ""
}
],
"year": 1997,
"venue": "Proc. Meeting of the ACL Special Interest Group in Natural Language Learning",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Powers, D. (1997). Learning and application of differential grammars. In Proc. Meeting of the ACL Special Interest Group in Natural Language Learning, Madrid.",
"links": null
},
"BIBREF19": {
"ref_id": "b19",
"title": "Improving data driven wordclass tagging by system combination",
"authors": [
{
"first": "H",
"middle": [],
"last": "Van Halteren",
"suffix": ""
},
{
"first": "J",
"middle": [],
"last": "Zavrel",
"suffix": ""
},
{
"first": "W",
"middle": [],
"last": "Daelemans",
"suffix": ""
}
],
"year": 1998,
"venue": "COLING-ACL'98",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "van Halteren, H. Zavrel, J. and Daelemans, W. (1998). Improving data driven wordclass tagging by system combination. In COLING-ACL'98, pages 491497, Montreal, Canada.",
"links": null
},
"BIBREF20": {
"ref_id": "b20",
"title": "Efficient lattice representation and generation",
"authors": [
{
"first": "F",
"middle": [],
"last": "Weng",
"suffix": ""
},
{
"first": "A",
"middle": [],
"last": "Stolcke",
"suffix": ""
},
{
"first": "A",
"middle": [],
"last": "Sankar",
"suffix": ""
}
],
"year": 1998,
"venue": "Proc. Intl. Conf. on Spoken Language Processing",
"volume": "6",
"issue": "",
"pages": "2531--2534",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Weng, F., Stolcke, A, & Sankar, A (1998). Efficient lattice representation and generation. Proc. Intl. Conf. on Spoken Language Processing, vol. 6, pp. 2531-2534. Sydney, Australia.",
"links": null
},
"BIBREF21": {
"ref_id": "b21",
"title": "Decision lists for lexical ambiguity resolution: Application to accent restoration in Spanish and French",
"authors": [
{
"first": "D",
"middle": [],
"last": "Yarowsky",
"suffix": ""
}
],
"year": 1994,
"venue": "Proc. 32nd Annual Meeting of the Association for Computational Linguistics",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Yarowsky, D. (1994). Decision lists for lexical ambiguity resolution: Application to accent restoration in Spanish and French. In Proc. 32nd Annual Meeting of the Association for Computational Linguistics, Las Cruces, NM.",
"links": null
},
"BIBREF22": {
"ref_id": "b22",
"title": "Unsupervised word sense disambiguation rivaling supervised methods",
"authors": [
{
"first": "D",
"middle": [],
"last": "Yarowsky",
"suffix": ""
}
],
"year": 1995,
"venue": "Proceedings of the 33rd Annual Meeting of the Association for Computational Linguistics",
"volume": "",
"issue": "",
"pages": "189--196",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Yarowsky, D. (1995) Unsupervised word sense disambiguation rivaling supervised methods. In Proceedings of the 33rd Annual Meeting of the Association for Computational Linguistics. Cambridge, MA, pp. 189-196, 1995.",
"links": null
}
},
"ref_entries": {
"FIGREF0": {
"text": "Learning Curves for Confusion Set Disambiguation",
"num": null,
"type_str": "figure",
"uris": null
},
"FIGREF1": {
"text": "Representation Size vs. Training Corpus Size",
"num": null,
"type_str": "figure",
"uris": null
},
"FIGREF2": {
"text": "Figure 3. Voting Among Classifiers",
"num": null,
"type_str": "figure",
"uris": null
},
"FIGREF3": {
"text": "Active Learning with Large Corpora manually labeled.",
"num": null,
"type_str": "figure",
"uris": null
},
"TABREF0": {
"content": "<table><tr><td/><td>1.00</td><td/><td/><td/></tr><tr><td/><td>0.95</td><td/><td/><td/></tr><tr><td/><td>0.90</td><td/><td/><td/></tr><tr><td>Test Accuracy</td><td>0.85</td><td/><td/><td/></tr><tr><td/><td>0.80</td><td/><td/><td/></tr><tr><td/><td/><td/><td/><td colspan=\"2\">Memory-Based</td></tr><tr><td/><td>0.75</td><td/><td/><td>Winnow</td></tr><tr><td/><td/><td/><td/><td colspan=\"2\">Perceptron</td></tr><tr><td/><td/><td/><td/><td colspan=\"2\">Na\u00efve Bayes</td></tr><tr><td/><td>0.70</td><td/><td/><td/></tr><tr><td/><td>0.1</td><td>1</td><td>10</td><td>100</td><td>1000</td></tr><tr><td/><td/><td/><td>Millions of Words</td><td/></tr></table>",
"type_str": "table",
"text": "Thanks to Dan Roth for making both Winnow and Perceptron available.speech. The memory-based learner used only the word before and word after as features.",
"num": null,
"html": null
}
}
}
} |