File size: 72,507 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 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 | {
"paper_id": "S13-1013",
"header": {
"generated_with": "S2ORC 1.0.0",
"date_generated": "2023-01-19T15:43:04.179050Z"
},
"title": "HENRY-CORE: Domain Adaptation and Stacking for Text Similarity *",
"authors": [
{
"first": "Michael",
"middle": [],
"last": "Heilman",
"suffix": "",
"affiliation": {},
"email": "mheilman@ets.org"
},
{
"first": "Nitin",
"middle": [],
"last": "Madnani",
"suffix": "",
"affiliation": {},
"email": "nmadnani@ets.org"
}
],
"year": "",
"venue": null,
"identifiers": {},
"abstract": "This paper describes a system for automatically measuring the semantic similarity between two texts, which was the aim of the 2013 Semantic Textual Similarity (STS) task (Agirre et al., 2013). For the 2012 STS task, Heilman and Madnani (2012) submitted the PERP system, which performed competitively in relation to other submissions. However, approaches including word and n-gram features also performed well (B\u00e4r et al., 2012; Sari\u0107 et al., 2012), and the 2013 STS task focused more on predicting similarity for text pairs from new domains. Therefore, for the three variations of our system that we were allowed to submit, we used stacking (Wolpert, 1992) to combine PERP with word and ngram features and applied the domain adaptation approach outlined by Daume III (2007) to facilitate generalization to new domains. Our submissions performed well at most subtasks, particularly at measuring the similarity of news headlines, where one of our submissions ranked 2nd among 89 from 34 teams, but there is still room for improvement. * System description papers for this task were required to have a team ID and task ID (e.g., \"HENRY-CORE\") as a prefix.",
"pdf_parse": {
"paper_id": "S13-1013",
"_pdf_hash": "",
"abstract": [
{
"text": "This paper describes a system for automatically measuring the semantic similarity between two texts, which was the aim of the 2013 Semantic Textual Similarity (STS) task (Agirre et al., 2013). For the 2012 STS task, Heilman and Madnani (2012) submitted the PERP system, which performed competitively in relation to other submissions. However, approaches including word and n-gram features also performed well (B\u00e4r et al., 2012; Sari\u0107 et al., 2012), and the 2013 STS task focused more on predicting similarity for text pairs from new domains. Therefore, for the three variations of our system that we were allowed to submit, we used stacking (Wolpert, 1992) to combine PERP with word and ngram features and applied the domain adaptation approach outlined by Daume III (2007) to facilitate generalization to new domains. Our submissions performed well at most subtasks, particularly at measuring the similarity of news headlines, where one of our submissions ranked 2nd among 89 from 34 teams, but there is still room for improvement. * System description papers for this task were required to have a team ID and task ID (e.g., \"HENRY-CORE\") as a prefix.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Abstract",
"sec_num": null
}
],
"body_text": [
{
"text": "We aim to develop an automatic measure of the semantic similarity between two short texts (e.g., sentences). Such a measure could be useful for various applications, including automated short answer scoring (Leacock and Chodorow, 2003; Nielsen et al., 2008) , question answering (Wang et al., 2007) , and machine translation evaluation (Przybocki et al., 2009) .",
"cite_spans": [
{
"start": 207,
"end": 235,
"text": "(Leacock and Chodorow, 2003;",
"ref_id": "BIBREF11"
},
{
"start": 236,
"end": 257,
"text": "Nielsen et al., 2008)",
"ref_id": "BIBREF12"
},
{
"start": 279,
"end": 298,
"text": "(Wang et al., 2007)",
"ref_id": "BIBREF18"
},
{
"start": 336,
"end": 360,
"text": "(Przybocki et al., 2009)",
"ref_id": "BIBREF15"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "In this paper, we describe our submissions to the 2013 Semantic Textual Similarity (STS) task (Agirre et al., 2013) , which evaluated implementations of text-to-text similarity measures. Submissions were evaluated according to Pearson correlations between gold standard similarity values acquired from human raters and machine-produced similarity values. Teams were allowed to submit up to three submissions. For each submission, correlations were calculated separately for four subtasks: measuring similarity between news headlines (\"headlines\"), between machine translation outputs and human reference translations (\"SMT\"), between word glosses from OntoNotes (Pradhan and Xue, 2009) and WordNet (Fellbaum, 1998) (\"OnWN\") , and between frame descriptions from FrameNet (Fillmore et al., 2003) and glosses from WordNet (\"FNWN\"). A weighted mean of the correlations was also computed as an overall evaluation metric (the OnWn and FNWN datasets were smaller than the headlines and SMT datasets).",
"cite_spans": [
{
"start": 94,
"end": 115,
"text": "(Agirre et al., 2013)",
"ref_id": "BIBREF1"
},
{
"start": 662,
"end": 685,
"text": "(Pradhan and Xue, 2009)",
"ref_id": "BIBREF14"
},
{
"start": 698,
"end": 714,
"text": "(Fellbaum, 1998)",
"ref_id": "BIBREF8"
},
{
"start": 715,
"end": 723,
"text": "(\"OnWN\")",
"ref_id": null
},
{
"start": 771,
"end": 794,
"text": "(Fillmore et al., 2003)",
"ref_id": "BIBREF9"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "The suggested training data for the 2013 STS task was the data from the 2012 STS task (Agirre et al., 2012) , including both the training and test sets for that year. The 2012 task was similar except that the data were from a different set of subtasks: measuring similarity between sentences from the Microsoft Research Paraphrase corpus (Dolan et al., 2004) (\"MSRpar\") , between sentences from the Microsoft Research Video Description corpus (Chen and Dolan, 2011) (\"MSRvid\") , and between human and machine translations of parliamentary proceedings (\"SMTeuroparl\"). The 2012 task provided training and test sets for those three subtasks and also included two additional tasks with just test sets: a similar OnWN task, and measuring similarity between human and machine translations of news broadcasts (\"SMTnews\").",
"cite_spans": [
{
"start": 86,
"end": 107,
"text": "(Agirre et al., 2012)",
"ref_id": "BIBREF0"
},
{
"start": 338,
"end": 369,
"text": "(Dolan et al., 2004) (\"MSRpar\")",
"ref_id": null
},
{
"start": 443,
"end": 465,
"text": "(Chen and Dolan, 2011)",
"ref_id": "BIBREF5"
},
{
"start": 466,
"end": 476,
"text": "(\"MSRvid\")",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "Heilman and Madnani (2012) described the PERP system and submitted it to the 2012 STS task. PERP measures the similarity of a sentence pair by finding a sequence of edit operations (e.g., insertions, deletions, substitutions, and shifts) that converts one sentence to the other. It then uses various features of the edits, with weights learned from labeled sentence pairs, to assign a similarity score. PERP performed well, ranking 7th out of 88 submissions from 35 teams according to the weighted mean correlation. However, PERP lacked some of the useful word and n-gram overlap features included in some of the other top-performing submissions. In addition, domain adaptation seemed more relevant for the STS 2013 task since in-domain data was available only for one (OnWN) of the four subtasks.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "Therefore, in this work, we combine the PERP system with various word and n-gram features. We also apply the domain adaptation technique of Daume III (2007) to support generalization beyond the domains in the training data.",
"cite_spans": [
{
"start": 140,
"end": 156,
"text": "Daume III (2007)",
"ref_id": "BIBREF6"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "In this section, we describe the system we developed, and the variations of it that comprise our submissions to the 2013 STS task.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "System Details",
"sec_num": "2"
},
{
"text": "Our system is a linear model estimated using ridge regression, as implemented in the scikit-learn toolkit (Pedregosa et al., 2011) . The system uses a 5-fold cross-validation grid search to tune the \u03b1 penalty for ridge regression (with \u03b1 \u2208 2 {\u22125,\u22124,...,4} ). During development, we evaluated its performance on the full STS 2012 data (training and test) using 10-fold cross-validation, with the 5-fold crossvalidation being used to tune within each training partition.",
"cite_spans": [
{
"start": 106,
"end": 130,
"text": "(Pedregosa et al., 2011)",
"ref_id": "BIBREF13"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "System Details",
"sec_num": "2"
},
{
"text": "Our full system uses the following features computed from an input sentence pair (s 1 , s 2 ).",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Features",
"sec_num": "2.1"
},
{
"text": "The system standardizes feature values to zero mean and unit variance by subtracting the feature's mean and dividing by its standard deviation. The means and standard deviations are estimated from the training set, or from each training partition during cross-validation.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Features",
"sec_num": "2.1"
},
{
"text": "The system computes Jaccard similarity (i.e., the ratio of the sizes of the set intersection to the set union) for the following overlap features:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "n-gram Overlap Features",
"sec_num": "2.1.1"
},
{
"text": "\u2022 character n-gram overlap (n = 1 . . . 12). Note that this is computed from the entire original texts for a pair, including punctuation, whitespace, etc.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "n-gram Overlap Features",
"sec_num": "2.1.1"
},
{
"text": "\u2022 word n-gram overlap (n = 2 . . . 8). We do not include n = 1 here because it would be identical to the n = 1 version for the unordered word n-gram feature described next.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "n-gram Overlap Features",
"sec_num": "2.1.1"
},
{
"text": "\u2022 unordered word n-gram overlap features (n = 1 . . . 3). By unordered, we mean combinations (in the mathematical sense of \"combinations\") of word tokens, regardless of order. Note that these features are similar to the word n-gram overlap features except that the words need not be contiguous to match. For example, the text \"John saw Mary\" would result in the following unordered word n-grams: {john}, {mary}, {saw}, {john, saw}, {mary, saw}, {john, mary}, and {john, mary, saw}.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "n-gram Overlap Features",
"sec_num": "2.1.1"
},
{
"text": "For the word and unordered n-gram overlap features, we computed two variants: one based on all tokens and one based on just content words, which we define as words that are not punctuation and do not appear in the NLTK (Bird et al., 2009) English stopword list. We lowercase everything for the word overlap measures but not for character overlap.",
"cite_spans": [
{
"start": 219,
"end": 238,
"text": "(Bird et al., 2009)",
"ref_id": "BIBREF4"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "n-gram Overlap Features",
"sec_num": "2.1.1"
},
{
"text": "The system includes various length-related features, where",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Length Features",
"sec_num": "2.1.2"
},
{
"text": "L max = max(length(s 1 ), length(s 2 )), L min = min(length(s 1 ), length(s 2 ))",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Length Features",
"sec_num": "2.1.2"
},
{
"text": ", and length(x) denotes the number of tokens in x. log denotes the natural logarithm.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Length Features",
"sec_num": "2.1.2"
},
{
"text": "\u2022 log( Lmax L min ) \u2022 Lmax\u2212L min Lmax \u2022 log(L min ) \u2022 log(L max ) \u2022 log(|L max \u2212 L min | + 1) 2.1.3 Sentiment Features",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Length Features",
"sec_num": "2.1.2"
},
{
"text": "The system includes various features based on the proprietary sentiment lexicon described by Beigman Klebanov et al. (2012) . Each word in this lexicon is associated with a 3-tuple specifying a distribution over three classes: positive, negative, and neutral. These distributions were estimated via crowdsourcing. If a word is not in the lexicon, we assume its positivity and negativity are zero.",
"cite_spans": [
{
"start": 101,
"end": 123,
"text": "Klebanov et al. (2012)",
"ref_id": "BIBREF3"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Length Features",
"sec_num": "2.1.2"
},
{
"text": "We define the set of sentiment words in a sentence s as \u03c3(s) = {w : positivity(w) > 0.5 \u2228 negativity(w) > 0.5}. We also define the positivity, negativity, and neutrality of a sentence as the sum over the corresponding values of individual words w. For example, positivity(s) = w\u2208s positivity(w). The system includes the following features:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Length Features",
"sec_num": "2.1.2"
},
{
"text": "\u2022 \u03c3(s 1 )\u2229\u03c3(s 2 )",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Length Features",
"sec_num": "2.1.2"
},
{
"text": "\u03c3(s 1 )\u222a\u03c3(s 2 ) (i.e., the Jaccard similarity of the sentiment words)",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Length Features",
"sec_num": "2.1.2"
},
{
"text": "\u2022 The cosine distance between (positivity(s 1 ), negativity(s 1 )) and (positivity(s 2 ), negativity(s 2 ))",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Length Features",
"sec_num": "2.1.2"
},
{
"text": "\u2022 |positivity(s 1 ) \u2212 positivity(s 2 )| \u2022 |negativity(s 1 ) \u2212 negativity(s 2 )| \u2022 |neutrality(s 1 ) \u2212 neutrality(s 2 )|",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Length Features",
"sec_num": "2.1.2"
},
{
"text": "The system also incorporates the PERP system (Heilman and Madnani, 2012) (as briefly described in \u00a71) as a feature in its model by using 10-fold stacking (Wolpert, 1992) . Stacking is a procedure similar to k-fold cross-validation that allows one to use the output of one model as the input to another model, without requiring multiple training sets. A PERP model is iteratively trained on nine folds and then the PERP feature is computed for the tenth, producing PERP features for the whole training set, which are then used in the final regression model.",
"cite_spans": [
{
"start": 154,
"end": 169,
"text": "(Wolpert, 1992)",
"ref_id": "BIBREF19"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "PERP with Stacking",
"sec_num": "2.1.4"
},
{
"text": "We trained PERP in a general manner using data from all the STS 2012 subtasks rather than training subtask-specific models. PERP was trained for 100 iterations.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "PERP with Stacking",
"sec_num": "2.1.4"
},
{
"text": "We refer readers to Heilman and Madnani (2012) for a full description of PERP. Next, we provide details about modifications made to PERP since STS 2012. Although these details are not necessary to understand how the system works in general, we include them here for completeness.",
"cite_spans": [
{
"start": 20,
"end": 46,
"text": "Heilman and Madnani (2012)",
"ref_id": "BIBREF10"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "PERP with Stacking",
"sec_num": "2.1.4"
},
{
"text": "\u2022 We extended PERP to model abbreviations as zero cost edits, using a list of common abbreviations extracted from Wikipedia. 1",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "PERP with Stacking",
"sec_num": "2.1.4"
},
{
"text": "\u2022 In a similar vein, we also extended PERP to model multiword sequences with differing punctuation (e.g., \"Built-In Test\" \u2192 \"Built In Test\") as zero cost edits.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "PERP with Stacking",
"sec_num": "2.1.4"
},
{
"text": "\u2022 We changed the stemming and synonymy edits in the original PERP (Heilman and Madnani, 2012) to be substitution edits that activate additional stemming and synonymy indicator features.",
"cite_spans": [
{
"start": 66,
"end": 93,
"text": "(Heilman and Madnani, 2012)",
"ref_id": "BIBREF10"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "PERP with Stacking",
"sec_num": "2.1.4"
},
{
"text": "\u2022 We added an incentive to TERp's (Snover et al., 2009) original inference algorithm to prefer matching words when searching for a good edit sequence. We added this to avoid rare cases where other edits would have a negative costs, and then the same word in a sentence pair would be, for example inserted and deleted rather than matched.",
"cite_spans": [
{
"start": 34,
"end": 55,
"text": "(Snover et al., 2009)",
"ref_id": "BIBREF16"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "PERP with Stacking",
"sec_num": "2.1.4"
},
{
"text": "\u2022 We fixed a minor bug in the inference algorithm, which appeared to only affect results on the MSRvid subtask in the STS 2012 task.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "PERP with Stacking",
"sec_num": "2.1.4"
},
{
"text": "\u2022 We tweaked the learning algorithm by increasing the learning rate and not performing weight averaging.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "PERP with Stacking",
"sec_num": "2.1.4"
},
{
"text": "The system also uses the domain adaptation technique described by Daume III (2007) to facilitate generalization to new domains. Instead of having a single weight for each of the features described above, the system maintains a generic and a subtaskspecific copy. For example, the content bigram overlap feature had six copies: a generic copy and one for each of the five subtasks in the training data from STS 2012 (i.e., OnWN, MSRpar, MSRvid, SMTeuroparl, SMTnews). And then for an instance from MSRpar, only the generic and MSRpar-specific versions of the feature will be active. For an instance from a new subtask (e.g., a test set instance), only the generic feature will be active. We also included a generic intercept feature and intercept features for each subtask (these always had a value of 1). These help the model capture, for example, whether high or low similarities are more frequent in general, without having to use the other feature weights to do so.",
"cite_spans": [
{
"start": 66,
"end": 82,
"text": "Daume III (2007)",
"ref_id": "BIBREF6"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Domain Adaptation",
"sec_num": "2.2"
},
{
"text": "We submitted three variations of the system.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Submissions",
"sec_num": "2.3"
},
{
"text": "\u2022 Run 1: This run used all the features described above. In addition, we mapped the test subtasks to the training subtasks as follows so that the specific features would be active for test data from previously unseen but related subtasks: headlines to MSRpar, SMT to SMTnews, and FNWN to OnWN.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Submissions",
"sec_num": "2.3"
},
{
"text": "\u2022 Run 2: As in Run 1, this run used all the features described above. However, we did not map the STS 2013 subtasks to STS 2012 subtasks. Thus, the specific copies of features were only active for OnWN test set examples.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Submissions",
"sec_num": "2.3"
},
{
"text": "\u2022 Run 3: This run used all the features except for the PERP and sentiment features. Like Run 2, this run did not map subtasks.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Submissions",
"sec_num": "2.3"
},
{
"text": "This section presents results on the STS 2012 data (our development set) and results for our submissions to STS 2013.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Results",
"sec_num": "3"
},
{
"text": "Although we used cross-validation on the entire STS 2012 dataset during preliminary experiments ( \u00a72), in this section, we train the system on the original STS 2012 training set and report performance on the original STS 2012 test set, in order to facilitate comparison to submissions to that task. It is important to note that our system's results here may be somewhat optimistic since we had access to the STS 2012 test data and were using it for development, whereas the participants in the 2012 task only had access to the training data. Table 1 presents the results. We include the results for our three submissions, the results for the topranked submission according to the weighted mean (\"UKP\"), the results for the best submission from Heilman and Madnani (2012) (\"PERPphrases\"), and the mean across all submissions. Note that while we compare to the PERP submission from Heilman and Madnani (2012) , the results are not directly comparable since the version of PERP is not the same and since PERP was trained differently.",
"cite_spans": [
{
"start": 880,
"end": 906,
"text": "Heilman and Madnani (2012)",
"ref_id": "BIBREF10"
}
],
"ref_spans": [
{
"start": 542,
"end": 549,
"text": "Table 1",
"ref_id": null
}
],
"eq_spans": [],
"section": "STS 2012 (development set)",
"sec_num": "3.1"
},
{
"text": "For Run 1 on the STS 2012 data, we mapped OnWN to MSRpar, and SMTnews to SMTeuroparl, similar to Heilman and Madnani (2012) . Table 2 presents results for our submissions to the 2013 STS task. We include results for our three submissions, results for the top-ranked submission according to the weighted mean, results for the baseline provided by the task organizers, and the mean across all submissions and the baseline from the organizers. 2 Note that while our Run 2 submission outperformed the top-ranked UMBC submission on the headlines subtask, as shown in 2, there was another UMBC submission that performed better than Run 2 for the headlines subtask.",
"cite_spans": [
{
"start": 97,
"end": 123,
"text": "Heilman and Madnani (2012)",
"ref_id": "BIBREF10"
},
{
"start": 441,
"end": 442,
"text": "2",
"ref_id": null
}
],
"ref_spans": [
{
"start": 126,
"end": 133,
"text": "Table 2",
"ref_id": null
}
],
"eq_spans": [],
"section": "STS 2012 (development set)",
"sec_num": "3.1"
},
{
"text": "The weighted mean correlation across tasks for our submissions was relatively poor compared to the top-ranked systems for STS 2013: our Run 1, Run 2, and Run 3 submissions beat the baseline and ranked 41st, 26th, and 48th, respectively, out of 89 submissions.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Discussion",
"sec_num": "4"
},
{
"text": "The primary reason for this result is that performance of our submissions was poor for the OnWN subtask, where, e.g., our Run 2 submission's correlation was r = .4631, compared to r = .8431 for the top-ranked submission for that subtask (\"deftbaseline\"). Upon investigation, we found that OnWN training and test data were very different in terms of their score distributions. Table 1 : Pearson correlations for STS 2012 data for each subtask and then the weighted mean across subtasks. \"UKP\" was submitted by B\u00e4r et al. (2012) , \"PERPphrases\" was submitted by Heilman and Madnani (2012) , and \"mean-2012\" is the mean of all submissions to STS 2012. Table 2 : Pearson correlations for STS 2013 data for each subtask and then the weighted mean across subtasks. \"UMBC\" = \"UMBC EBIQUITY-ParingWords\", and \"mean-2013\" is the mean of the submissions to STS 2013 and the baseline.",
"cite_spans": [
{
"start": 509,
"end": 526,
"text": "B\u00e4r et al. (2012)",
"ref_id": "BIBREF2"
},
{
"start": 560,
"end": 586,
"text": "Heilman and Madnani (2012)",
"ref_id": "BIBREF10"
}
],
"ref_spans": [
{
"start": 376,
"end": 383,
"text": "Table 1",
"ref_id": null
},
{
"start": 649,
"end": 656,
"text": "Table 2",
"ref_id": null
}
],
"eq_spans": [],
"section": "Discussion",
"sec_num": "4"
},
{
"text": "standard similarity value for the STS 2012 OnWN data was 3.87 (with a standard deviation of 1.02), while the mean for the 2013 OnWN data was 2.31 (with a standard deviation of 1.76). We speculate that our system performed relatively poorly because it was expecting the OnWN data to include many highly similar sentences (as in the 2012 data). We hypothesize that incorporating more detailed Word-Net information (only the PERP feature used Word-Net, and only in a limited fashion, to check synonymy) and task-specific features for comparing definitions might have helped performance for the OnWN subtask. If we ignore the definition comparison subtasks, and consider performance on just the headlines and SMT subtasks, the system performed quite well. Our Run 2 submission had a mean correlation of r = .5619 for those two subtasks, which would rank 5th among all submissions.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Submission",
"sec_num": null
},
{
"text": "We have not fully explored the effects on performance of the domain adaptation approach used in the system, but our approach of mapping tasks used for our Run 1 submission did not seem to help. It seems better to keep a general model, as in Runs 2 and 3.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Submission",
"sec_num": null
},
{
"text": "Additionally, we observe that the performance of Run 3, which did not use the PERP and sentiment features, was relatively good compared to Runs 1 and 2, which used all the features. This indicates that if speed and implementation simplicity are important concerns for an application, it may suffice to use relatively simple overlap and length features to measure semantic similarity.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Submission",
"sec_num": null
},
{
"text": "The contribution of domain adaptation is not clear. Mapping novel subtasks to tasks for which training data is available ( \u00a72.3), in combination with the domain adaptation technique we used, did not generally improve performance. However, we leave to future work a detailed analysis of whether the domain adaptation approach (without mapping) is better than simply training a separate system for each subtask and using out-of-domain data when indomain data is unavailable.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Submission",
"sec_num": null
},
{
"text": "In this paper, we described a system for predicting the semantic similarity of two short texts. The system uses stacking to combine a trained edit-based similarity model (Heilman and Madnani, 2012) with simple features such as word and n-gram overlap, and it uses the technique described by Daume III (2007) to support generalization to domains not represented in the training data. We also presented evaluation results, using data from the STS 2012 and STS 2013 shared tasks, that indicate that the system performs competitively relative to other approaches for many tasks. In particular, we observed very good performance on the news headline similarity and MT evaluation subtasks of the STS 2013 shared task.",
"cite_spans": [
{
"start": 170,
"end": 197,
"text": "(Heilman and Madnani, 2012)",
"ref_id": "BIBREF10"
},
{
"start": 291,
"end": 307,
"text": "Daume III (2007)",
"ref_id": "BIBREF6"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Conclusion",
"sec_num": "5"
},
{
"text": "http://en.wikipedia.org/wiki/List_of_ acronyms_and_initialisms, downloaded April 27, 2012",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
},
{
"text": "The STS 2013 results are from http://ixa2.si. ehu.es/sts/.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
}
],
"back_matter": [
{
"text": "We would like to thank the STS 2013 task organizers for facilitating this research and Dan Blanchard for helping with scikit-learn.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Acknowledgments",
"sec_num": null
}
],
"bib_entries": {
"BIBREF0": {
"ref_id": "b0",
"title": "Semeval-2012 task 6: A pilot on semantic textual similarity",
"authors": [
{
"first": "Eneko",
"middle": [],
"last": "Agirre",
"suffix": ""
},
{
"first": "Daniel",
"middle": [],
"last": "Cer",
"suffix": ""
},
{
"first": "Mona",
"middle": [],
"last": "Diab",
"suffix": ""
},
{
"first": "Aitor",
"middle": [],
"last": "Gonzalez-Agirre",
"suffix": ""
}
],
"year": 2012,
"venue": "*SEM 2012: The First Joint Conference on Lexical and Computational Semantics",
"volume": "1",
"issue": "",
"pages": "7--8",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Eneko Agirre, Daniel Cer, Mona Diab, and Aitor Gonzalez-Agirre. 2012. Semeval-2012 task 6: A pilot on semantic textual similarity. In *SEM 2012: The First Joint Conference on Lexical and Computa- tional Semantics -Volume 1: Proceedings of the main conference and the shared task, and Volume 2: Pro- ceedings of the Sixth International Workshop on Se- mantic Evaluation (SemEval 2012), pages 385-393, Montr\u00e9al, Canada, 7-8 June. Association for Compu- tational Linguistics.",
"links": null
},
"BIBREF1": {
"ref_id": "b1",
"title": "*sem 2013 shared task: Semantic textual similarity, including a pilot on typed-similarity",
"authors": [
{
"first": "Eneko",
"middle": [],
"last": "Agirre",
"suffix": ""
},
{
"first": "Daniel",
"middle": [],
"last": "Cer",
"suffix": ""
},
{
"first": "Mona",
"middle": [],
"last": "Diab",
"suffix": ""
},
{
"first": "Aitor",
"middle": [],
"last": "Gonzalez-Agirre",
"suffix": ""
},
{
"first": "Weiwei",
"middle": [],
"last": "Guo",
"suffix": ""
}
],
"year": 2013,
"venue": "*SEM 2013: The Second Joint Conference on Lexical and Computational Semantics",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Eneko Agirre, Daniel Cer, Mona Diab, Aitor Gonzalez- Agirre, and Weiwei Guo. 2013. *sem 2013 shared task: Semantic textual similarity, including a pilot on typed-similarity. In *SEM 2013: The Second Joint Conference on Lexical and Computational Semantics. Association for Computational Linguistics.",
"links": null
},
"BIBREF2": {
"ref_id": "b2",
"title": "UKP: Computing semantic textual similarity by combining multiple content similarity measures",
"authors": [
{
"first": "Daniel",
"middle": [],
"last": "B\u00e4r",
"suffix": ""
},
{
"first": "Chris",
"middle": [],
"last": "Biemann",
"suffix": ""
},
{
"first": "Iryna",
"middle": [],
"last": "Gurevych",
"suffix": ""
},
{
"first": "Torsten",
"middle": [],
"last": "Zesch",
"suffix": ""
}
],
"year": 2012,
"venue": "*SEM 2012: The First Joint Conference on Lexical and Computational Semantics",
"volume": "1",
"issue": "",
"pages": "7--8",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Daniel B\u00e4r, Chris Biemann, Iryna Gurevych, and Torsten Zesch. 2012. UKP: Computing semantic textual simi- larity by combining multiple content similarity mea- sures. In *SEM 2012: The First Joint Conference on Lexical and Computational Semantics -Volume 1: Proceedings of the main conference and the shared task, and Volume 2: Proceedings of the Sixth Inter- national Workshop on Semantic Evaluation (SemEval 2012), pages 435-440, Montr\u00e9al, Canada, 7-8 June. Association for Computational Linguistics.",
"links": null
},
"BIBREF3": {
"ref_id": "b3",
"title": "Building sentiment lexicon(s) from scratch for essay data",
"authors": [
{
"first": "Jill",
"middle": [],
"last": "Beata Beigman Klebanov",
"suffix": ""
},
{
"first": "Nitin",
"middle": [],
"last": "Burstein",
"suffix": ""
},
{
"first": "Adam",
"middle": [],
"last": "Madnani",
"suffix": ""
},
{
"first": "Joel",
"middle": [],
"last": "Faulkner",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Tetreault",
"suffix": ""
}
],
"year": 2012,
"venue": "Proceedings of the 13th International Conference on Intelligent Text Processing and Computational Linguistics (CICLing)",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Beata Beigman Klebanov, Jill Burstein, Nitin Madnani, Adam Faulkner, and Joel Tetreault. 2012. Build- ing sentiment lexicon(s) from scratch for essay data. In Proceedings of the 13th International Conference on Intelligent Text Processing and Computational Lin- guistics (CICLing), New Delhi, India, March.",
"links": null
},
"BIBREF4": {
"ref_id": "b4",
"title": "Natural Language Processing with Python",
"authors": [
{
"first": "S",
"middle": [],
"last": "Bird",
"suffix": ""
},
{
"first": "E",
"middle": [],
"last": "Klein",
"suffix": ""
},
{
"first": "E",
"middle": [],
"last": "Loper",
"suffix": ""
}
],
"year": 2009,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "S. Bird, E. Klein, and E. Loper. 2009. Natural Language Processing with Python. O'Reilly Media.",
"links": null
},
"BIBREF5": {
"ref_id": "b5",
"title": "Collecting highly parallel data for paraphrase evaluation",
"authors": [
{
"first": "David",
"middle": [],
"last": "Chen",
"suffix": ""
},
{
"first": "William",
"middle": [],
"last": "Dolan",
"suffix": ""
}
],
"year": 2011,
"venue": "Proceedings of the 49th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies",
"volume": "",
"issue": "",
"pages": "190--200",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "David Chen and William Dolan. 2011. Collecting highly parallel data for paraphrase evaluation. In Pro- ceedings of the 49th Annual Meeting of the Associa- tion for Computational Linguistics: Human Language Technologies, pages 190-200, Portland, Oregon, USA, June. Association for Computational Linguistics.",
"links": null
},
"BIBREF6": {
"ref_id": "b6",
"title": "Frustratingly easy domain adaptation",
"authors": [
{
"first": "Hal",
"middle": [],
"last": "Daume",
"suffix": ""
},
{
"first": "Iii",
"middle": [],
"last": "",
"suffix": ""
}
],
"year": 2007,
"venue": "Proceedings of the 45th Annual Meeting of the Association of Computational Linguistics",
"volume": "",
"issue": "",
"pages": "256--263",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Hal Daume III. 2007. Frustratingly easy domain adapta- tion. In Proceedings of the 45th Annual Meeting of the Association of Computational Linguistics, pages 256-263, Prague, Czech Republic, June. Association for Computational Linguistics.",
"links": null
},
"BIBREF7": {
"ref_id": "b7",
"title": "Unsupervised construction of large paraphrase corpora: Exploiting massively parallel news sources",
"authors": [
{
"first": "Bill",
"middle": [],
"last": "Dolan",
"suffix": ""
},
{
"first": "Chris",
"middle": [],
"last": "Quirk",
"suffix": ""
},
{
"first": "Chris",
"middle": [],
"last": "Brockett",
"suffix": ""
}
],
"year": 2004,
"venue": "Proceedings of Coling",
"volume": "",
"issue": "",
"pages": "350--356",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Bill Dolan, Chris Quirk, and Chris Brockett. 2004. Un- supervised construction of large paraphrase corpora: Exploiting massively parallel news sources. In Pro- ceedings of Coling 2004, pages 350-356, Geneva, Switzerland, Aug 23-Aug 27. COLING.",
"links": null
},
"BIBREF8": {
"ref_id": "b8",
"title": "WordNet: An Electronic Lexical Database",
"authors": [
{
"first": "C",
"middle": [],
"last": "Fellbaum",
"suffix": ""
}
],
"year": 1998,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "C. Fellbaum. 1998. WordNet: An Electronic Lexical Database. Bradford Books.",
"links": null
},
"BIBREF9": {
"ref_id": "b9",
"title": "Background to Framenet",
"authors": [
{
"first": "Charles",
"middle": [
"J"
],
"last": "Fillmore",
"suffix": ""
},
{
"first": "Christopher",
"middle": [
"R"
],
"last": "Johnson",
"suffix": ""
},
{
"first": "Miriam",
"middle": [
"R L"
],
"last": "Petruck",
"suffix": ""
}
],
"year": 2003,
"venue": "International Journal of Lexicography",
"volume": "16",
"issue": "3",
"pages": "235--250",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Charles J. Fillmore, Christopher R. Johnson, and Miriam R.L. Petruck. 2003. Background to Framenet. International Journal of Lexicography, 16(3):235- 250.",
"links": null
},
"BIBREF10": {
"ref_id": "b10",
"title": "ETS: Discriminative edit models for paraphrase scoring",
"authors": [
{
"first": "Michael",
"middle": [],
"last": "Heilman",
"suffix": ""
},
{
"first": "Nitin",
"middle": [],
"last": "Madnani",
"suffix": ""
}
],
"year": 2012,
"venue": "*SEM 2012: The First Joint Conference on Lexical and Computational Semantics",
"volume": "1",
"issue": "",
"pages": "7--8",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Michael Heilman and Nitin Madnani. 2012. ETS: Dis- criminative edit models for paraphrase scoring. In *SEM 2012: The First Joint Conference on Lexi- cal and Computational Semantics -Volume 1: Pro- ceedings of the main conference and the shared task, and Volume 2: Proceedings of the Sixth International Workshop on Semantic Evaluation (SemEval 2012), pages 529-535, Montr\u00e9al, Canada, 7-8 June. Associa- tion for Computational Linguistics.",
"links": null
},
"BIBREF11": {
"ref_id": "b11",
"title": "c-rater: Scoring of short-answer questions. Computers and the Humanities",
"authors": [
{
"first": "C",
"middle": [],
"last": "Leacock",
"suffix": ""
},
{
"first": "M",
"middle": [],
"last": "Chodorow",
"suffix": ""
}
],
"year": 2003,
"venue": "",
"volume": "37",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "C. Leacock and M. Chodorow. 2003. c-rater: Scoring of short-answer questions. Computers and the Humani- ties, 37.",
"links": null
},
"BIBREF12": {
"ref_id": "b12",
"title": "Classification errors in a domain-independent assessment system",
"authors": [
{
"first": "Rodney",
"middle": [
"D"
],
"last": "Nielsen",
"suffix": ""
},
{
"first": "Wayne",
"middle": [],
"last": "Ward",
"suffix": ""
},
{
"first": "James",
"middle": [
"H"
],
"last": "Martin",
"suffix": ""
}
],
"year": 2008,
"venue": "Proceedings of the Third Workshop on Innovative Use of NLP for Building Educational Applications",
"volume": "",
"issue": "",
"pages": "10--18",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Rodney D. Nielsen, Wayne Ward, and James H. Martin. 2008. Classification errors in a domain-independent assessment system. In Proceedings of the Third Work- shop on Innovative Use of NLP for Building Educa- tional Applications, pages 10-18, Columbus, Ohio, June. Association for Computational Linguistics.",
"links": null
},
"BIBREF13": {
"ref_id": "b13",
"title": "Scikit-learn: Machine learning in Python",
"authors": [
{
"first": "F",
"middle": [],
"last": "Pedregosa",
"suffix": ""
},
{
"first": "G",
"middle": [],
"last": "Varoquaux",
"suffix": ""
},
{
"first": "A",
"middle": [],
"last": "Gramfort",
"suffix": ""
},
{
"first": "V",
"middle": [],
"last": "Michel",
"suffix": ""
},
{
"first": "B",
"middle": [],
"last": "Thirion",
"suffix": ""
},
{
"first": "O",
"middle": [],
"last": "Grisel",
"suffix": ""
},
{
"first": "M",
"middle": [],
"last": "Blondel",
"suffix": ""
},
{
"first": "P",
"middle": [],
"last": "Prettenhofer",
"suffix": ""
},
{
"first": "R",
"middle": [],
"last": "Weiss",
"suffix": ""
},
{
"first": "V",
"middle": [],
"last": "Dubourg",
"suffix": ""
},
{
"first": "J",
"middle": [],
"last": "Vanderplas",
"suffix": ""
},
{
"first": "A",
"middle": [],
"last": "Passos",
"suffix": ""
},
{
"first": "D",
"middle": [],
"last": "Cournapeau",
"suffix": ""
},
{
"first": "M",
"middle": [],
"last": "Brucher",
"suffix": ""
},
{
"first": "M",
"middle": [],
"last": "Perrot",
"suffix": ""
},
{
"first": "E",
"middle": [],
"last": "Duchesnay",
"suffix": ""
}
],
"year": 2011,
"venue": "Journal of Machine Learning Research",
"volume": "12",
"issue": "",
"pages": "2825--2830",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "F. Pedregosa, G. Varoquaux, A. Gramfort, V. Michel, B. Thirion, O. Grisel, M. Blondel, P. Prettenhofer, R. Weiss, V. Dubourg, J. Vanderplas, A. Passos, D. Cournapeau, M. Brucher, M. Perrot, and E. Duches- nay. 2011. Scikit-learn: Machine learning in Python. Journal of Machine Learning Research, 12:2825- 2830.",
"links": null
},
"BIBREF14": {
"ref_id": "b14",
"title": "OntoNotes: The 90% solution",
"authors": [
{
"first": "S",
"middle": [
"S"
],
"last": "Pradhan",
"suffix": ""
},
{
"first": "N",
"middle": [],
"last": "Xue",
"suffix": ""
}
],
"year": 2009,
"venue": "The 2009 Annual Conference of the North American Chapter of the Association for Computational Linguistics, Companion Volume: Tutorial Abstracts",
"volume": "",
"issue": "",
"pages": "11--12",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "S. S. Pradhan and N. Xue. 2009. OntoNotes: The 90% solution. In Proceedings of Human Language Tech- nologies: The 2009 Annual Conference of the North American Chapter of the Association for Computa- tional Linguistics, Companion Volume: Tutorial Ab- stracts, pages 11-12.",
"links": null
},
"BIBREF15": {
"ref_id": "b15",
"title": "The NIST 2008 metrics for machine translation challenge -overview, methodology, metrics, and results. Machine Translation",
"authors": [
{
"first": "M",
"middle": [
"A"
],
"last": "Przybocki",
"suffix": ""
},
{
"first": "K",
"middle": [],
"last": "Peterson",
"suffix": ""
},
{
"first": "S",
"middle": [],
"last": "Bronsart",
"suffix": ""
},
{
"first": "G",
"middle": [
"A"
],
"last": "Sanders",
"suffix": ""
}
],
"year": 2009,
"venue": "",
"volume": "23",
"issue": "",
"pages": "71--103",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "M. A. Przybocki, K. Peterson, S. Bronsart, and G. A. Sanders. 2009. The NIST 2008 metrics for machine translation challenge -overview, methodology, met- rics, and results. Machine Translation, 23(2-3):71- 103.",
"links": null
},
"BIBREF16": {
"ref_id": "b16",
"title": "TER-Plus: Paraphrase, semantic, and alignment enhancements to translation edit rate",
"authors": [
{
"first": "G",
"middle": [],
"last": "Matthew",
"suffix": ""
},
{
"first": "Nitin",
"middle": [],
"last": "Snover",
"suffix": ""
},
{
"first": "Bonnie",
"middle": [],
"last": "Madnani",
"suffix": ""
},
{
"first": "Richard",
"middle": [],
"last": "Dorr",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Schwartz",
"suffix": ""
}
],
"year": 2009,
"venue": "Machine Translation",
"volume": "23",
"issue": "2-3",
"pages": "117--127",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Matthew G. Snover, Nitin Madnani, Bonnie Dorr, and Richard Schwartz. 2009. TER-Plus: Paraphrase, semantic, and alignment enhancements to translation edit rate. Machine Translation, 23(2-3):117-127, September.",
"links": null
},
"BIBREF17": {
"ref_id": "b17",
"title": "TakeLab: Systems for measuring semantic text similarity",
"authors": [
{
"first": "Goran",
"middle": [],
"last": "Frane\u0161ari\u0107",
"suffix": ""
},
{
"first": "Mladen",
"middle": [],
"last": "Glava\u0161",
"suffix": ""
},
{
"first": "Jan\u0161najder",
"middle": [],
"last": "Karan",
"suffix": ""
},
{
"first": "Bojana Dalbelo",
"middle": [],
"last": "Ba\u0161i\u0107",
"suffix": ""
}
],
"year": 2012,
"venue": "*SEM 2012: The First Joint Conference on Lexical and Computational Semantics",
"volume": "1",
"issue": "",
"pages": "7--8",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Frane\u0160ari\u0107, Goran Glava\u0161, Mladen Karan, Jan\u0160najder, and Bojana Dalbelo Ba\u0161i\u0107. 2012. TakeLab: Systems for measuring semantic text similarity. In *SEM 2012: The First Joint Conference on Lexical and Computa- tional Semantics -Volume 1: Proceedings of the main conference and the shared task, and Volume 2: Pro- ceedings of the Sixth International Workshop on Se- mantic Evaluation (SemEval 2012), pages 441-448, Montr\u00e9al, Canada, 7-8 June. Association for Compu- tational Linguistics.",
"links": null
},
"BIBREF18": {
"ref_id": "b18",
"title": "What is the Jeopardy model? a quasisynchronous grammar for QA",
"authors": [
{
"first": "Mengqiu",
"middle": [],
"last": "Wang",
"suffix": ""
},
{
"first": "Noah",
"middle": [
"A"
],
"last": "Smith",
"suffix": ""
},
{
"first": "Teruko",
"middle": [],
"last": "Mitamura",
"suffix": ""
}
],
"year": 2007,
"venue": "Proceedings of the 2007 Joint Conference on Empirical Methods in Natural Language Processing and Computational Natural Language Learning (EMNLP-CoNLL)",
"volume": "",
"issue": "",
"pages": "22--32",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Mengqiu Wang, Noah A. Smith, and Teruko Mita- mura. 2007. What is the Jeopardy model? a quasi- synchronous grammar for QA. In Proceedings of the 2007 Joint Conference on Empirical Methods in Natu- ral Language Processing and Computational Natural Language Learning (EMNLP-CoNLL), pages 22-32, Prague, Czech Republic, June. Association for Com- putational Linguistics.",
"links": null
},
"BIBREF19": {
"ref_id": "b19",
"title": "Stacked generalization",
"authors": [
{
"first": "H",
"middle": [],
"last": "David",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Wolpert",
"suffix": ""
}
],
"year": 1992,
"venue": "Neural Networks",
"volume": "5",
"issue": "",
"pages": "241--259",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "David H. Wolpert. 1992. Stacked generalization. Neural Networks, 5:241-259.",
"links": null
}
},
"ref_entries": {}
}
} |