File size: 86,931 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 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 |
{
"paper_id": "2021",
"header": {
"generated_with": "S2ORC 1.0.0",
"date_generated": "2023-01-19T02:10:07.747124Z"
},
"title": "Exploring Methodologies for Collecting High-Quality Implicit Reasoning in Arguments",
"authors": [
{
"first": "Keshav",
"middle": [],
"last": "Singh",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "Tohoku University",
"location": {}
},
"email": ""
},
{
"first": "Farjana",
"middle": [],
"last": "Sultana",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "Tohoku University",
"location": {}
},
"email": ""
},
{
"first": "Naoya",
"middle": [],
"last": "Inoue",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "RIKEN Stony Brook University Ricoh Company, Ltd",
"location": {}
},
"email": ""
},
{
"first": "Shoichi",
"middle": [],
"last": "Naito",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "Tohoku University",
"location": {}
},
"email": ""
},
{
"first": "Kentaro",
"middle": [],
"last": "Inui",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "Tohoku University",
"location": {}
},
"email": "inui@tohoku.ac.jp"
}
],
"year": "",
"venue": null,
"identifiers": {},
"abstract": "Annotation of implicit reasoning (i.e., warrant) in arguments is a critical resource to train models in gaining deeper understanding and correct interpretation of arguments. However, warrants are usually annotated in unstructured form, having no restriction on their lexical structure which sometimes makes it difficult to interpret how warrants relate to any of the information given in claim and premise. Moreover, assessing and determining better warrants from the large variety of reasoning patterns of unstructured warrants becomes a formidable task. Therefore, in order to annotate warrants in a more interpretative and restrictive way, we propose two methodologies to annotate warrants in a semi-structured form. To the best of our knowledge, we are the first to show how such semi-structured warrants can be annotated on a large scale via crowdsourcing. We demonstrate through extensive quality evaluation that our methodologies enable collecting better quality warrants in comparison to unstructured annotations. To further facilitate research towards the task of explicating warrants in arguments, we release our materials publicly (i.e., crowdsourcing guidelines and collected warrants).",
"pdf_parse": {
"paper_id": "2021",
"_pdf_hash": "",
"abstract": [
{
"text": "Annotation of implicit reasoning (i.e., warrant) in arguments is a critical resource to train models in gaining deeper understanding and correct interpretation of arguments. However, warrants are usually annotated in unstructured form, having no restriction on their lexical structure which sometimes makes it difficult to interpret how warrants relate to any of the information given in claim and premise. Moreover, assessing and determining better warrants from the large variety of reasoning patterns of unstructured warrants becomes a formidable task. Therefore, in order to annotate warrants in a more interpretative and restrictive way, we propose two methodologies to annotate warrants in a semi-structured form. To the best of our knowledge, we are the first to show how such semi-structured warrants can be annotated on a large scale via crowdsourcing. We demonstrate through extensive quality evaluation that our methodologies enable collecting better quality warrants in comparison to unstructured annotations. To further facilitate research towards the task of explicating warrants in arguments, we release our materials publicly (i.e., crowdsourcing guidelines and collected warrants).",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Abstract",
"sec_num": null
}
],
"body_text": [
{
"text": "Implicit reasonings, commonly referred to as warrants (Toulmin, 1958) , have long been studied to understand the grounds on which a premise lends support to the claim (Freeman, 1992) . In other words, a warrant, when made explicit, clearly shows the inferential link between claim and premise (Pineau, 2013) . As depicted in Figure 1 , identification of such warrants by students has been shown to aid them in making better arguments (Erduran et al., 2004) , as well as improving their critical thinking skills (von der M\u00fchlen et al., 2019) and argument comprehension process (Hitchcock and Verheij, 2006) .",
"cite_spans": [
{
"start": 54,
"end": 69,
"text": "(Toulmin, 1958)",
"ref_id": "BIBREF18"
},
{
"start": 167,
"end": 182,
"text": "(Freeman, 1992)",
"ref_id": "BIBREF8"
},
{
"start": 293,
"end": 307,
"text": "(Pineau, 2013)",
"ref_id": "BIBREF16"
},
{
"start": 434,
"end": 456,
"text": "(Erduran et al., 2004)",
"ref_id": "BIBREF6"
},
{
"start": 511,
"end": 540,
"text": "(von der M\u00fchlen et al., 2019)",
"ref_id": "BIBREF19"
},
{
"start": 576,
"end": 605,
"text": "(Hitchcock and Verheij, 2006)",
"ref_id": "BIBREF11"
}
],
"ref_spans": [
{
"start": 325,
"end": 333,
"text": "Figure 1",
"ref_id": null
}
],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "While explication of warrants with assistance from teachers has been shown to be useful for improving students' argumentation skills, automating this explication process would not only help students to be less dependent on teachers, but it can also be beneficial for different downstream educational applications such as argument analysis (Becker et al., 2020) , enthymeme reconstruction (Razuvayevskaya and Teufel, 2017; Hulpus et al., 2019) and essay scoring (Williamson, 2013) . However, building an automated warrant explication system has been a challenge due to the difficulty of collecting warrants in a form that explicitly manifests the way a warrant relates the information between claim and premise. Generally, warrants are annotated in an unstructured (i.e., freetext) format which lays no restriction on its lexical structure (Boltu\u017ei\u0107 and \u0160najder, 2016) . As a result, sometimes the warrant consist of no information that overlaps with claim or premise which makes it difficult to understand how the warrant connects the claim to its premise. Furthermore, for a given argument, unstructured warrants can be framed in diverse ways that would have a wide variety of reasoning patterns (Kock, 2006 ) (given no restriction on the lexical structure), and identifying the correct ones from this large pool of warrants can be a preposterous task.",
"cite_spans": [
{
"start": 339,
"end": 360,
"text": "(Becker et al., 2020)",
"ref_id": "BIBREF3"
},
{
"start": 388,
"end": 421,
"text": "(Razuvayevskaya and Teufel, 2017;",
"ref_id": "BIBREF17"
},
{
"start": 422,
"end": 442,
"text": "Hulpus et al., 2019)",
"ref_id": "BIBREF13"
},
{
"start": 461,
"end": 479,
"text": "(Williamson, 2013)",
"ref_id": "BIBREF21"
},
{
"start": 839,
"end": 867,
"text": "(Boltu\u017ei\u0107 and \u0160najder, 2016)",
"ref_id": "BIBREF5"
},
{
"start": 1197,
"end": 1208,
"text": "(Kock, 2006",
"ref_id": "BIBREF14"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "In order to annotate warrants on a large-scale and in a way that overcomes the aforementioned challenges, we propose two novel warrant annotation methodologies: Pre-defined Keyword-based Warrant (PKW) and User-defined Keyword-based Warrant (UKW), which restrict a warrant's lexical structure to a semi-structured form. In contrast to approaches that crowdsource unstructured warrants, these methodologies explicate warrant by enforcing it to have the key information (i.e., keywords) from both claim and premise. The intuition behind our semi-structured approach is to restrict the structure of warrants to specific keyword-based Figure 1 : A typical example of warrant explication. Feedback provided by teachers or automated warrant explication system can help students identify correct warrants and leverage it in their revised argument to make the argument more reasonable.",
"cite_spans": [],
"ref_spans": [
{
"start": 630,
"end": 638,
"text": "Figure 1",
"ref_id": null
}
],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "reasoning patterns. We hypothesize that leveraging such keyword-based pattern could assist in collecting high quality warrants, where keywords are derived from the original argument (i.e., claim and premise). Our assumption follows the formal definition of warrants in the sense that warrants act as a inferential link between the contents of claim and its premise (Toulmin, 1958; Freeman, 1992) .",
"cite_spans": [
{
"start": 365,
"end": 380,
"text": "(Toulmin, 1958;",
"ref_id": "BIBREF18"
},
{
"start": 381,
"end": 395,
"text": "Freeman, 1992)",
"ref_id": "BIBREF8"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "Many previous work demonstrated different strategies of collecting high quality warrants (Becker et al., 2017; Habernal et al., 2018; Becker et al., 2020) , but did not apply any restriction on the structure of warrants to handle the variety of reasoning patterns in which warrants can be explicated. In contrast, our annotation methodologies are designed to restrict the reasoning pattern and ensure that warrant explicates the reasoning link between claim and premise.",
"cite_spans": [
{
"start": 89,
"end": 110,
"text": "(Becker et al., 2017;",
"ref_id": "BIBREF4"
},
{
"start": 111,
"end": 133,
"text": "Habernal et al., 2018;",
"ref_id": "BIBREF10"
},
{
"start": 134,
"end": 154,
"text": "Becker et al., 2020)",
"ref_id": "BIBREF3"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "In order to evaluate the warrants annotated via our proposed UKW and PKW methodologies, we devise specific guidelines to judge their quality through scoring. We also collect unstructured warrants (i.e., Natural Language Warrants (NLW)) and perform quality evaluation on them in order to compare with our annotated warrants. Our results suggest that in comparison to NLW, high quality warrants can be annotated via our proposed UKW methodology. To the best of our knowledge, this is the first study which targets large scale annotation of warrants in semi-structured form. To facilitate further research in warrant explication, we publish our crowdsourcing guidelines and the preliminary corpus of around 1700 warrants that are annotated via UKW methodology, covering over 600 arguments 1 .",
"cite_spans": [
{
"start": 786,
"end": 787,
"text": "1",
"ref_id": "BIBREF0"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "Explication of warrants in arguments has already been approached in many previous researches. In an initial attempt, Feng and Hirst (2011) proposed leveraging argumentation schemes (Walton et al., 2008) as a means to automatically reconstruct warrant, but did not approach the task due to the absence of training datasets. To overcome the unavailability of dataset, Boltu\u017ei\u0107 and \u0160najder (2016) leverage crowdsourcing and ask non-expert workers to annotate all possible variations of warrants for a given claim-premise pair. However, they concluded that the annotation of warrants varied both in number annotated per argument and in content due to no restrictions imposed on in the annotation process.",
"cite_spans": [
{
"start": 181,
"end": 202,
"text": "(Walton et al., 2008)",
"ref_id": "BIBREF20"
},
{
"start": 366,
"end": 393,
"text": "Boltu\u017ei\u0107 and \u0160najder (2016)",
"ref_id": "BIBREF5"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Related Work",
"sec_num": "2"
},
{
"text": "In order to overcome the prior difficulties, recent approaches leverage crowdsourcing to restrict the number of warrants collected per argument and either employ a step by step filtering process to weed out bad warrants (Habernal et al., 2018) or hire ex-Argument Warrant",
"cite_spans": [
{
"start": 220,
"end": 243,
"text": "(Habernal et al., 2018)",
"ref_id": "BIBREF10"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Related Work",
"sec_num": "2"
},
{
"text": "Claim: We should abolish zoos. Premise: Zoos are notorious for animal abuse.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "1.",
"sec_num": null
},
{
"text": "Abolishing zoos leads to animals being in their natural habitat which results in no animal abuse.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "1.",
"sec_num": null
},
{
"text": "Claim: We should ban whaling. Premise: Whaling is considered to be unacceptable cruelty towards animals.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "2.",
"sec_num": null
},
{
"text": "Banning whaling would stop the inhumane methods of stabbing whales which is unacceptable cruelty towards animals.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "2.",
"sec_num": null
},
{
"text": "Claim: We should introduce compulsory voting. Premise: Compulsory voting can help obtain better results during elections.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "3.",
"sec_num": null
},
{
"text": "Introducing compulsory voting leads to every citizen exercising the right to vote which can help obtain better results during elections. perts who iteratively converge to a single warrant annotation (Becker et al., 2017) . In an advanced attempt, Razuvayevskaya and Teufel (2017) explored whether it is feasible for human annotators to explicate warrants in arguments. They propose the idea of template-based warrant reconstruction using information from premise and claim, and employ experts to perform the task. In contrast, our annotation methodology is designed for expert as well as non-expert workers. Additionally, our annotation aims to restrict the warrant's structure to semi-structured format such that it comprises knowledge that is necessary to form an inferential link between the key contents of claim and premise. Furthermore, we do not restrict the number of warrants to one, but collect set of warrants per argument that fulfill the criteria of qualifying as a high quality warrant.",
"cite_spans": [
{
"start": 199,
"end": 220,
"text": "(Becker et al., 2017)",
"ref_id": "BIBREF4"
},
{
"start": 247,
"end": 279,
"text": "Razuvayevskaya and Teufel (2017)",
"ref_id": "BIBREF17"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "3.",
"sec_num": null
},
{
"text": "We define warrant that we want to annotate by characterizing its desired properties. Properties such as structure of warrant, quality in terms of how well a warrant links claim and premise, and feasibility of annotating warrant for an argument. In this work, we define this feasibility in terms of whether a warrant can be annotated for an argument, regardless of whether it is good or bad. We assume that explication of a warrant might not be possible if the argument is too good (i.e., warrant is explicated in the premise) or if the argument is too bad (i.e., no warrant can explicate the link between claim and premise).",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Warrant Desiderata",
"sec_num": "3"
},
{
"text": "Structure Warrants are implicit reasoning that logically link the contents of claim and premise (Toulmin, 1958; Freeman, 1992) . Therefore, we hypothesize that a warrant should have a structure that (a) comprises the key information given in claim and premise, and (b) explicates logical connection between the aforementioned key information with some implicit knowledge that is relevant to the argument. We define the warrants framed in such a way as semi-structured warrants. Examples of such semi-structured warrants are provided in Table 1 , where key information or keywords from claim and premise is linked with relevant implicit knowledge which all together forms a semi-structured warrant.",
"cite_spans": [
{
"start": 96,
"end": 111,
"text": "(Toulmin, 1958;",
"ref_id": "BIBREF18"
},
{
"start": 112,
"end": 126,
"text": "Freeman, 1992)",
"ref_id": "BIBREF8"
}
],
"ref_spans": [
{
"start": 536,
"end": 543,
"text": "Table 1",
"ref_id": "TABREF0"
}
],
"eq_spans": [],
"section": "Warrant Desiderata",
"sec_num": "3"
},
{
"text": "Feasibility Warrants may not be explicable for all the arguments. Specifically, for an argument with bad premise there may be no feasible way to explicate the logical link between claim and premise. For example, the warrant for the argument \"We should introduce multi-party system because it's the right thing to do\" is not feasible, since the argument is a fallacy (i.e. begging the question) where the premise: \"it's the right thing to do\" provides no adequate support to the claim: \"We should introduce multi-party system\". Similarly, for arguments with very good premise, it might not be necessary to explicate the warrant since the warrant might already be explicated in the premise. In contrast, as shown in Table 1 , for arguments with moderately good/bad premise, we assume that one can frame a warrant by leveraging argument relevant external knowledge. 2 .",
"cite_spans": [],
"ref_spans": [
{
"start": 714,
"end": 721,
"text": "Table 1",
"ref_id": "TABREF0"
}
],
"eq_spans": [],
"section": "Warrant Desiderata",
"sec_num": "3"
},
{
"text": "Quality A key factor distinguishing warrants from any other type of implicit knowledge or statement (e.g., commonsense knowledge) is their ability to justify the flow of reasoning between claim and premise. For example, in Table 1 , warrant (3) explicitly answers how introducing compulsory voting can help obtain better results in elections. These are the type of warrants we would like to annotate. Conversely, statements which do not serve this purpose cannot be qualified as a warrant. For example, given argument (3) from Table 1 , the statement \"Introducing compulsory voting enables people to freely choose their favourite candidate which results in encouraging better results during elections\" cannot be labeled as a warrant because it offers little to no help in bridging the implicit reasoning link between claim and premise.",
"cite_spans": [],
"ref_spans": [
{
"start": 223,
"end": 230,
"text": "Table 1",
"ref_id": "TABREF0"
},
{
"start": 527,
"end": 534,
"text": "Table 1",
"ref_id": "TABREF0"
}
],
"eq_spans": [],
"section": "Warrant Desiderata",
"sec_num": "3"
},
{
"text": "In this section, we discuss the development and design of our proposed semi-structured annotation methodologies. In particular, we consider two methodologies for annotating warrants: Predefined Keyword-based Warrant and User-defined Keyword-based Warrants.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Annotation Methodologies",
"sec_num": "4"
},
{
"text": "Pre-defined Keyword-based Warrants (PKW) In order to annotate semi-structured warrants that encompass information from claim and premise, we propose using keywords which encode the key information given in claim and premise. As shown in Table 1 , the purpose of these keywords is to create a semi-structured format for completing a warrant annotation such that keywords from claim form the initial (shown in red) and keywords from premise (shown in blue) form the latter part of the warrant. The keywords are linked by implicit knowledge that is necessary to connect the keywords in a meaningful way. Example annotation and task design of PKW annotation is shown in Figure 2 . For PKW methodology, the annotator is initially provided with keywords and is tasked to explain the flow of reasoning between them by writing implicit knowledge (i.e., hidden reasoning). In order to provide pre-defining keywords to the annotator, we employ spaCy (Honnibal et al., 2020) and automatically extract the key information from claim and premise by parsing the sentence into verb/noun phrases. For example, for claim \"We should ban whaling\", the verb phrase \"Banning whaling\" and for premise \"Whales are necessary for ecological sustainability of the oceans\", the noun phrase \"ecological sustainability of the oceans\" is extracted. To ensure the feasibility of framing a warrant with extracted keywords, we perform a manual check and if needed, make minimal changes to its tense or word-order.",
"cite_spans": [],
"ref_spans": [
{
"start": 237,
"end": 244,
"text": "Table 1",
"ref_id": "TABREF0"
},
{
"start": 666,
"end": 674,
"text": "Figure 2",
"ref_id": null
}
],
"eq_spans": [],
"section": "Annotation Methodologies",
"sec_num": "4"
},
{
"text": "User-defined Keyword-based Warrants (UKW) While PKW methodology introduces restrictions on warrant annotation via pre-defined keywords, they might be too restrictive or not provide sufficient flexibility for annotators to annotate the warrant. Moreover, automatically extracting keywords from claim and premise can be sometimes challenging due to varied syntactic structure of the argument. Therefore as an alternate approach, we ask the annotators to derive their own keywords from claim and premise. To do this, we provide detailed guidelines and concrete examples in our interface for annotators so that they can correctly understand the process of deriving keywords.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Annotation Methodologies",
"sec_num": "4"
},
{
"text": "The annotation design and example annotation is shown in Figure 3 , where for the given claim and premise, the keywords from claim can be a verb phrase: \"Introducing compulsory voting\", and keywords from premise can be verb phrase: \"obtain better results during elections\". To avoid annotations where annotator might write keywords with information from outside claim/premise, annotators were strictly advised not to use any external knowledge when writing the keywords, although the use of external knowledge was permitted for writing the hidden reasoning.",
"cite_spans": [],
"ref_spans": [
{
"start": 57,
"end": 65,
"text": "Figure 3",
"ref_id": "FIGREF0"
}
],
"eq_spans": [],
"section": "Annotation Methodologies",
"sec_num": "4"
},
{
"text": "Our goal is to establish a procedure for collecting semi-structured warrants and their annotations at large-scale. In order to collect such warrants from each methodology, we build a multi-step crowdsourcing process designed for encouraging annotator's creativity, while preventing biases in the annotations.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Warrant Collection Procedure",
"sec_num": "5"
},
{
"text": "In general, we break the warrant collection procedure into three steps of simple tasks: (i) Deriving keywords, (ii) Judging feasibility and (iii) Framing warrant. In addition, we implement several mechanisms for quality assurance and employ manual checks to ensure annotators understand and perform the final task correctly.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Warrant Collection Procedure",
"sec_num": "5"
},
{
"text": "(i) Deriving keywords In order to collect semistructured warrants, we require keywords derived from each of claim and premise. These keywords Figure 2 : The interface used for warrant annotations, along with an example of a annotator's annotation for Predefined Keyword-based Warrant. To avoid using complicated terminology, we used the terms stance, supporting statement and hidden reasoning to convey the notion of claim, premise and warrant respectively. act as the skeleton of the final annotated warrant. For example, as shown on the left side of Figure 3 , \"Introducing compulsory voting\" and \"obtain better results during elections\" are keywords derived from claim and premise respectively. To derive these keywords, annotators are instructed to strictly include only key information conveyed in their respective counterparts and no external information. For PKW methodology, keywords in this step are already derived as shown in Figure 2 . To ensure annotators understand the notion of keywords, we provide sufficient variety of examples for them to get used to this sub-task.",
"cite_spans": [],
"ref_spans": [
{
"start": 142,
"end": 150,
"text": "Figure 2",
"ref_id": null
},
{
"start": 552,
"end": 560,
"text": "Figure 3",
"ref_id": "FIGREF0"
},
{
"start": 937,
"end": 945,
"text": "Figure 2",
"ref_id": null
}
],
"eq_spans": [],
"section": "Warrant Collection Procedure",
"sec_num": "5"
},
{
"text": "(b) Judging feasibility Since warrants may not be explicable for all the arguments, i.e., if a premise is very bad or contrastingly very good, we explicitly ask annotators to judge the feasibility of writing a warrant by asking if they can complete the hidden reasoning. This step is rather tricky since annotators may be biased to answer \"No\" or \"Unsure\" (See Question in Figure 2 and 3) to avoid doing the task and finish the task quickly. To avoid this, we treat this step as bonus question and depending on majority response i.e., if majority of annotators believe a warrant can be explicated for the given argument, then the majority annotators get bonus. Similarly, if majority of annotators believe a warrant cannot be explicated for a given argument, then again majority annotators get bonus. We keep a high bonus for this step in order to compel anno-tators to do task as instructed instead of providing low quality response. This step helps us get a better judgement of feasibility of warrants and also identify annotators who are not doing the task properly.",
"cite_spans": [],
"ref_spans": [
{
"start": 373,
"end": 381,
"text": "Figure 2",
"ref_id": null
}
],
"eq_spans": [],
"section": "Warrant Collection Procedure",
"sec_num": "5"
},
{
"text": "(c) Framing Warrant The last step in warrant collection procedure is for the annotators to frame the hidden reasoning to complete the warrant. This step is the most challenging since it requires the annotator to be logical and use his background knowledge to complete warrant annotation. To complete this via PKW methodology, annotator's are restricted in terms of pre-defined keywords, while for UKW methodology the annotator can annotate the warrant by minimally changing the keywords as well as hidden reasoning.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Warrant Collection Procedure",
"sec_num": "5"
},
{
"text": "Auxiliary verification measures For each task, we hold preliminary qualification test that consists of several basic questions to judge the understanding of annotator's reasoning skills. Annotators who score more than a pre-defined threshold (\u2265 80%) are granted access to the main task. Our qualifications are open to annotators from major English speaking countries, namely USA, UK, New Zealand and Canada. Additionally, to address any ethical issues (Adda et al., 2011) raised by our task, we actively monitored multiple pilot tests to ensure annotators were satisfied with our task. Simultaneously, we corresponded directly to annotators that had questions/comments on our task. Annotators were paid in accordance with the minimum wage calculated by conducting many trials and based on average work-time. Additional bonus was paid to annotators that provided quality annotations.",
"cite_spans": [
{
"start": 452,
"end": 471,
"text": "(Adda et al., 2011)",
"ref_id": "BIBREF2"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Warrant Collection Procedure",
"sec_num": "5"
},
{
"text": "We choose Amazon Mechanical Turk (AMT) 3 as our crowdsourcing platform due to its success in previous argumentation mining tasks (Habernal et al., 2018) . As an initial step, we only allowed annotators who had \u2265 98% acceptance rate and \u2265 5,000 approved Human Intelligence Tasks (HITs). Each annotator was paid $0.30 for doing the task and workers who qualified for bonus were paid an additional $0.25 per HIT. For the task involving UKW annotations, the payment was increased to $0.40 for doing the task and $0.35 as bonus.",
"cite_spans": [
{
"start": 129,
"end": 152,
"text": "(Habernal et al., 2018)",
"ref_id": "BIBREF10"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Crowdsourcing",
"sec_num": "6"
},
{
"text": "Source data For the purpose of annotating warrants for a given set of arguments, we utilize a wellknown argumentation dataset, IBM-Rank-30K corpus (Gretz et al., 2019) , which already consists arguments in the form of claim and premise. The dataset contains around 15K crowd-sourced arguments covering 71 topics, annotated for supporting as well as opposing stance. 4 The arguments were collected with strict length limitations and accompanied by extensive quality control measures. Our inspection of arguments from IBM-Rank-30k revealed that a for large proportion of the arguments we can explicate warrants. To proceed with our warrant annotation, we selected a subset of three 3 well-known debatable topics: We should abolish zoos, We should ban whaling and We should introduce compulsory voting.",
"cite_spans": [
{
"start": 147,
"end": 167,
"text": "(Gretz et al., 2019)",
"ref_id": null
},
{
"start": 366,
"end": 367,
"text": "4",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Crowdsourcing",
"sec_num": "6"
},
{
"text": "Annotations In addition to warrants annotated via PKW and UKW methodologies, we also annotated unstructured warrants via crowdsourcing to compare the quality of annotated warrants across different methodologies. We followed similar crowdsourcing procedure as employed by Habernal et al. (2018) and refer to the warrants annotated via this process as Natural language Warrants (NLW). For each methodology, we annotate 40 unique claim and premise pairs, randomly chosen from the three pre-selected topics. Each argument was annotated by 5 annotators resulting in a total of 200 annotations per methodology and all warrants were limited to have a length between 60 and 200 characters excluding keywords.",
"cite_spans": [
{
"start": 271,
"end": 293,
"text": "Habernal et al. (2018)",
"ref_id": "BIBREF10"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Crowdsourcing",
"sec_num": "6"
},
{
"text": "Filtering For each methodology, it is possible to collect at most 200 warrants (given 40 arguments and 5 annotators). However, not all annotators Table 2 : Comparison between the different warrant crowdsourcing methodologies. Avg. corresponds to the average score given by both expert annotators.",
"cite_spans": [],
"ref_spans": [
{
"start": 146,
"end": 153,
"text": "Table 2",
"ref_id": null
}
],
"eq_spans": [],
"section": "Crowdsourcing",
"sec_num": "6"
},
{
"text": "Warrant is unrelated to the claim and its premise.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Score Explanation 0",
"sec_num": null
},
{
"text": "Warrant is related to the claim and premise but does not make the relationship between them easy to understand and/or strengthen the argument. In addition, the warrant may overlap or be a paraphrase of the premise.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "1",
"sec_num": null
},
{
"text": "The relationship between the claim and premise is easier to understand and/or strengthened because of the warrant. Table 3 : Guidelines used by our expert annotators for scoring the quality of warrants on a scale of 0-2.",
"cite_spans": [],
"ref_spans": [
{
"start": 115,
"end": 122,
"text": "Table 3",
"ref_id": null
}
],
"eq_spans": [],
"section": "2",
"sec_num": null
},
{
"text": "chose \"Yes\" when they were asked to judge if they can write a warrant. After filtering the negative responses at this step, we discover that, in total, annotators wrote 155, 101 and 65 warrants out of possible 200 warrants for Natural language Warrant, PKW and UKW methodologies respectively. We utilize these 321 collected warrants for further analysis.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "2",
"sec_num": null
},
{
"text": "In order to analyze the quality of the annotated warrants, we hired two annotators who are experts in the field of argumentation to score the crowdsourced warrants. To frame quality scoring guidelines for judging warrants, we ran several pilot tests and take expert advice to make our guidelines easier to interpret. Our final quality annotation guidelines are shown in Table 3 . The experts were asked to score a given warrant on a scale of (0-2), with 0 being the lowest and 2 being the highest. Both annotators were given 50 warrants randomly chosen from the pool of collected warrants for each methodology. Each topic was represented fairly in the quality annotation step with each topic having at least 15 warrants. In total, our experts annotated 150 warrants out of a total 321 annotated warrants. As shown in Table 2 , the agreement between both experts as judged by Krippendorff's alpha \u03b1 (Krippendorff, 2011 ) was found to be fairly good. We find that the average scores given by two expert annotators (Avg.) on a scale of (0-2) indicated userdefined and pre-defined methodologies with overall higher quality warrants as compared to natural language warrants. We also measure the combined average score given to the warrants for each topic to measure if the warrants belonging to one topic was of higher quality. We find that on average warrants annotated for Introducing compulsory voting were scored the lowest while for other topics was fairly high.",
"cite_spans": [
{
"start": 875,
"end": 917,
"text": "Krippendorff's alpha \u03b1 (Krippendorff, 2011",
"ref_id": null
}
],
"ref_spans": [
{
"start": 370,
"end": 377,
"text": "Table 3",
"ref_id": null
},
{
"start": 817,
"end": 824,
"text": "Table 2",
"ref_id": null
}
],
"eq_spans": [],
"section": "Results",
"sec_num": "7"
},
{
"text": "To further analyze the quality of warrants and the quality of the entire crowdsourcing process, we analyze sample of the warrants collected via each methodology and which were annotated by experts.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Qualitative Analysis",
"sec_num": "7.1"
},
{
"text": "In Table 4 , we can see that the warrants that were scored the highest by experts have fair amount of keyword overlap with the claim/premise. This follows from our initial motivation for using semistructured annotation methodology where we hypothesized that the inclusion of keyword information from the claim and premise can assist in framing better quality warrants. As shown in Table 4 , warrants UKW (1) and PKW (1) have similar keywords \"Banning whaling\" and \"ecological sustainability of the oceans\", hence the annotated implicit knowledge is also same. This indicates that our keyword-based PKW and UKW methodologies restrict the diverse ways in which warrants can be explicated. On the contrary, for NLW (1) the warrant is analogous to a general statement yet has 1. Claim: We should ban whaling. Premise: Whales are necessary for ecological sustainability of the oceans.",
"cite_spans": [
{
"start": 772,
"end": 774,
"text": "1.",
"ref_id": null
}
],
"ref_spans": [
{
"start": 3,
"end": 10,
"text": "Table 4",
"ref_id": null
},
{
"start": 381,
"end": 388,
"text": "Table 4",
"ref_id": null
}
],
"eq_spans": [],
"section": "Qualitative Analysis",
"sec_num": "7.1"
},
{
"text": "Marine life in the ocean cannot survive without ecological sustainability. PKW: Banning whaling would prevent the decreasing of whale population on which marine life thrives which will result in ecological sustainability of the oceans. UKW: Banning whaling prevents the decreasing of whale population which is necessary for ecological sustainability of the oceans.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "NLW:",
"sec_num": null
},
{
"text": "Claim: We should abolish zoos. Premise: We should abolish zoos to prevent the cruel confinement of wild animals.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "2.",
"sec_num": null
},
{
"text": "Animals in confinement suffer physically and emotionally. PKW: Abolishing zoos enables animals to live a more stimulating and fulfilling life in the wild which prevents the cruel confinement of wild animals. UKW: Zoos force many animals to live in prison-like environment with unhygienic conditions which is a cruel way of confining wild animals.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "NLW:",
"sec_num": null
},
{
"text": "Claim: We should introduce compulsory voting. Premise: Compulsory voting can help encourage better results during elections.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "3.",
"sec_num": null
},
{
"text": "Mandatory voting will reflect people's actual preferences in election results. PKW: Compulsory voting produces a winning candidate that is more accurately representative of all the voters which ensures better election results. UKW: Introducing compulsory voting stops one side from rigging the process by canvasing more people which results in better elections. Table 4 : Examples of warrants scored 2 by both experts. Majority of warrant belonging to UKW methodology were found to be of good quality. Table 5 : Examples of warrants scored 0 or 1 by both experts. Low scored warrants were mainly paraphrased (Scored 1) from the premise or did not relate to the given topic (Scored 0). similar implicit knowledge explicated in UKW (1) and PKW (1) warrants. This hints that NLW can also be of higher quality but its quality can vary to a larger extent due to no restrictions. In Table 5 , we can see that even though the warrants encode claim-premise information, the quality of the warrant can essentially be bad. For example, the warrants PLW (1) and UKW (1) explicate implicit knowledge which does not make the inferential link between claim and premise clear. Additionally, we note that while keyword-based methods restrict most warrants to a single sentence, natural language warrants often consist of shorter yet multiple sentences. Overall, we found that 23% of natural language warrants were composed of more than one sentences. Besides, such warrants were scored low and were often found to be paraphrased from the information in the premise or annotators rephrased previous premise in place of the warrant. Although this was mostly observed in natural language warrants, we found few such instances in our keyword-based collected warrants (See UKW (1) in Table 5 ). This suggests that while keyword-based methods assist in collecting warrants that explicate the inferential link between claim and premise, it still does not guarantee high quality warrant annotations and might require further adjustments.",
"cite_spans": [
{
"start": 730,
"end": 733,
"text": "(1)",
"ref_id": "BIBREF0"
},
{
"start": 742,
"end": 745,
"text": "(1)",
"ref_id": "BIBREF0"
}
],
"ref_spans": [
{
"start": 362,
"end": 369,
"text": "Table 4",
"ref_id": null
},
{
"start": 502,
"end": 509,
"text": "Table 5",
"ref_id": null
},
{
"start": 877,
"end": 884,
"text": "Table 5",
"ref_id": null
},
{
"start": 1763,
"end": 1770,
"text": "Table 5",
"ref_id": null
}
],
"eq_spans": [],
"section": "NLW:",
"sec_num": null
},
{
"text": "Based on our finding that user-defined keywordbased warrant methodology comparatively results in better warrants, we follow this method to collect a total of 1700 warrants across 3 topics, annotated for 600 claim-premise pairs. All warrants are limited to have a length between 60 and 200 characters. Since this is an ongoing work, we plan to make further analysis on our preliminary dataset in future.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Preliminary large-scale corpus",
"sec_num": "7.2"
},
{
"text": "In this work, we tackle the difficult task of explicating warrants in arguments and propose two novel methodologies to annotate warrants in semistructured format. We conduct extensive analysis and perform annotation study for determining the appropriate methodology for collecting warrants and show that user-defined keyword based approach produces the highest quality warrants as compared to pre-defined keyword-based warrant and natural language warrants. In future, we plan to extend the annotation of warrants for more diverse topics. Moreover, we plan to cover warrant annotations for claim-premise pairs with premises attacking the claim in addition to premises supporting the original claim. We would also like to test the usefulness of our annotations for constructing a model for automatic warrant explication which can be used to explicate warrant for any given argument. We believe that such a model can be useful in a pedagogical setting to perform downstream tasks such as argument analysis or giving constructive feedback to students.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Conclusion and Future work",
"sec_num": "8"
},
{
"text": "Our crowdsourcing guidelines and annotated warrants are publicly available at https://github.com/ cl-tohoku/ukw-warrants",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
},
{
"text": "The arguments shown inTable 1were already annotated with moderate quality scores in a larger study(Gretz et al., 2019)",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
},
{
"text": "www.mturk.com 4 In our work we only focus on arguments with supporting stance.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
}
],
"back_matter": [
{
"text": "This work was partially supported by JST CREST Grant Number JPMJCR20D2 and NEDO Grant Number J200001946. The authors would like to thank Paul Reisert, other members of the Tohoku NLP Lab, and the anonymous reviewers for their insightful feedback.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Acknowledgements",
"sec_num": null
}
],
"bib_entries": {
"BIBREF0": {
"ref_id": "b0",
"title": "NLW: Whales could soon die off completely. It is our duty to ban whaling. PKW: Banning whaling leads to whales not dying which has cause decrease in whale population over the years. UKW: Banning whaling is a harmful practice which results in decrease in whale population over the years. 2. Claim: We should abolish zoos. Premise:It is unfair to trap animals from their natural habitat and confine them to small spaces for human entertainment",
"authors": [],
"year": null,
"venue": "Claim: We should ban whaling. Premise: Whaling has led to a major decrease in whale populations over the years",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Claim: We should ban whaling. Premise: Whaling has led to a major decrease in whale populations over the years. NLW: Whales could soon die off completely. It is our duty to ban whaling. PKW: Banning whaling leads to whales not dying which has cause decrease in whale population over the years. UKW: Banning whaling is a harmful practice which results in decrease in whale population over the years. 2. Claim: We should abolish zoos. Premise:It is unfair to trap animals from their natural habitat and confine them to small spaces for human entertain- ment. NLW: Zoos keep animals captive where where they cannot run free and thrive.",
"links": null
},
"BIBREF1": {
"ref_id": "b1",
"title": "UKW: Abolishing zoos makes sure that animals are not being treated unfair and in small spaces. 3. Claim: We should introduce compulsory voting. Premise: Compulsive voting is a patriotic act that must be fully complied with. NLW: We enjoy the freedom and liberty enjoyed by expressing our opinions. We should take advantage of such compulsion. PKW: Introducing compulsory voting makes you feel that you belong to your country which is a patriotic act. UKW: Compulsory voting will force people to engage in politics and choose a",
"authors": [],
"year": null,
"venue": "PKW: Abolishing zoos is a bad practice because it is unfair to trap animals from their natural habitat and confine them to small spaces for human entertainment",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "PKW: Abolishing zoos is a bad practice because it is unfair to trap animals from their natural habitat and confine them to small spaces for human entertainment. UKW: Abolishing zoos makes sure that animals are not being treated unfair and in small spaces. 3. Claim: We should introduce compulsory voting. Premise: Compulsive voting is a patriotic act that must be fully complied with. NLW: We enjoy the freedom and liberty enjoyed by expressing our opinions. We should take advantage of such compulsion. PKW: Introducing compulsory voting makes you feel that you belong to your country which is a patriotic act. UKW: Compulsory voting will force people to engage in politics and choose a right leader that can lead their country in the future.",
"links": null
},
"BIBREF2": {
"ref_id": "b2",
"title": "Crowdsourcing for language resource development: Critical analysis of amazon mechanical turk overpowering use",
"authors": [
{
"first": "Gilles",
"middle": [],
"last": "Adda",
"suffix": ""
},
{
"first": "Beno\u00eet",
"middle": [],
"last": "Sagot",
"suffix": ""
},
{
"first": "Kar\u00ebn",
"middle": [],
"last": "Fort",
"suffix": ""
},
{
"first": "Joseph",
"middle": [],
"last": "",
"suffix": ""
}
],
"year": 2011,
"venue": "5th Language and Technology Conference",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Gilles Adda, Beno\u00eet Sagot, Kar\u00ebn Fort, and Joseph Mar- iani. 2011. Crowdsourcing for language resource de- velopment: Critical analysis of amazon mechanical turk overpowering use. In 5th Language and Tech- nology Conference.",
"links": null
},
"BIBREF3": {
"ref_id": "b3",
"title": "Explaining arguments with background knowledge",
"authors": [
{
"first": "Maria",
"middle": [],
"last": "Becker",
"suffix": ""
},
{
"first": "Ioana",
"middle": [],
"last": "Hulpu\u015f",
"suffix": ""
},
{
"first": "Juri",
"middle": [],
"last": "Opitz",
"suffix": ""
},
{
"first": "Debjit",
"middle": [],
"last": "Paul",
"suffix": ""
},
{
"first": "Jonathan",
"middle": [],
"last": "Kobbe",
"suffix": ""
},
{
"first": "Heiner",
"middle": [],
"last": "Stuckenschmidt",
"suffix": ""
},
{
"first": "Anette",
"middle": [],
"last": "Frank",
"suffix": ""
}
],
"year": 2020,
"venue": "Datenbank-Spektrum",
"volume": "20",
"issue": "2",
"pages": "131--141",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Maria Becker, Ioana Hulpu\u015f, Juri Opitz, Debjit Paul, Jonathan Kobbe, Heiner Stuckenschmidt, and Anette Frank. 2020. Explaining arguments with background knowledge. Datenbank-Spektrum, 20(2):131-141.",
"links": null
},
"BIBREF4": {
"ref_id": "b4",
"title": "Enriching argumentative texts with implicit knowledge",
"authors": [
{
"first": "Maria",
"middle": [],
"last": "Becker",
"suffix": ""
},
{
"first": "Michael",
"middle": [],
"last": "Staniek",
"suffix": ""
},
{
"first": "Vivi",
"middle": [],
"last": "Nastase",
"suffix": ""
},
{
"first": "Anette",
"middle": [],
"last": "Frank",
"suffix": ""
}
],
"year": 2017,
"venue": "International Conference on Applications of Natural Language to Information Systems",
"volume": "",
"issue": "",
"pages": "84--96",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Maria Becker, Michael Staniek, Vivi Nastase, and Anette Frank. 2017. Enriching argumentative texts with implicit knowledge. In International Confer- ence on Applications of Natural Language to Infor- mation Systems, pages 84-96. Springer.",
"links": null
},
"BIBREF5": {
"ref_id": "b5",
"title": "Fill the gap! analyzing implicit premises between claims from online debates",
"authors": [
{
"first": "Filip",
"middle": [],
"last": "Boltu\u017ei\u0107",
"suffix": ""
}
],
"year": 2016,
"venue": "Proceedings of the Third Workshop on Argument Mining (ArgMining2016)",
"volume": "",
"issue": "",
"pages": "124--133",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Filip Boltu\u017ei\u0107 and Jan \u0160najder. 2016. Fill the gap! an- alyzing implicit premises between claims from on- line debates. In Proceedings of the Third Work- shop on Argument Mining (ArgMining2016), pages 124-133, Berlin, Germany. Association for Compu- tational Linguistics.",
"links": null
},
"BIBREF6": {
"ref_id": "b6",
"title": "Tapping into argumentation: Developments in the application of toulmin's argument pattern for studying science discourse",
"authors": [
{
"first": "Sibel",
"middle": [],
"last": "Erduran",
"suffix": ""
},
{
"first": "Shirley",
"middle": [],
"last": "Simon",
"suffix": ""
},
{
"first": "Jonathan",
"middle": [],
"last": "Osborne",
"suffix": ""
}
],
"year": 2004,
"venue": "Science education",
"volume": "88",
"issue": "6",
"pages": "915--933",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Sibel Erduran, Shirley Simon, and Jonathan Osborne. 2004. Tapping into argumentation: Developments in the application of toulmin's argument pattern for studying science discourse. Science education, 88(6):915-933.",
"links": null
},
"BIBREF7": {
"ref_id": "b7",
"title": "Classifying arguments by scheme",
"authors": [
{
"first": "Vanessa",
"middle": [],
"last": "Wei Feng",
"suffix": ""
},
{
"first": "Graeme",
"middle": [],
"last": "Hirst",
"suffix": ""
}
],
"year": 2011,
"venue": "Proceedings of the 49th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies",
"volume": "",
"issue": "",
"pages": "987--996",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Vanessa Wei Feng and Graeme Hirst. 2011. Classify- ing arguments by scheme. In Proceedings of the 49th Annual Meeting of the Association for Com- putational Linguistics: Human Language Technolo- gies, pages 987-996, Portland, Oregon, USA. Asso- ciation for Computational Linguistics.",
"links": null
},
"BIBREF8": {
"ref_id": "b8",
"title": "Relevance, warrants, backing, inductive support",
"authors": [
{
"first": "",
"middle": [],
"last": "James B Freeman",
"suffix": ""
}
],
"year": 1992,
"venue": "Argumentation",
"volume": "6",
"issue": "2",
"pages": "219--275",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "James B Freeman. 1992. Relevance, warrants, backing, inductive support. Argumentation, 6(2):219-275.",
"links": null
},
"BIBREF9": {
"ref_id": "b9",
"title": "Ranit Aharonov, and Noam Slonim. 2019. A large-scale dataset for argument quality ranking: Construction and analysis",
"authors": [
{
"first": "Shai",
"middle": [],
"last": "Gretz",
"suffix": ""
},
{
"first": "Roni",
"middle": [],
"last": "Friedman",
"suffix": ""
},
{
"first": "Edo",
"middle": [],
"last": "Cohen-Karlik",
"suffix": ""
},
{
"first": "Assaf",
"middle": [],
"last": "Toledo",
"suffix": ""
},
{
"first": "Dan",
"middle": [],
"last": "Lahav",
"suffix": ""
}
],
"year": null,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {
"arXiv": [
"arXiv:1911.11408"
]
},
"num": null,
"urls": [],
"raw_text": "Shai Gretz, Roni Friedman, Edo Cohen-Karlik, As- saf Toledo, Dan Lahav, Ranit Aharonov, and Noam Slonim. 2019. A large-scale dataset for argument quality ranking: Construction and analysis. arXiv preprint arXiv:1911.11408.",
"links": null
},
"BIBREF10": {
"ref_id": "b10",
"title": "SemEval-2018 task 12: The argument reasoning comprehension task",
"authors": [
{
"first": "Ivan",
"middle": [],
"last": "Habernal",
"suffix": ""
},
{
"first": "Henning",
"middle": [],
"last": "Wachsmuth",
"suffix": ""
},
{
"first": "Iryna",
"middle": [],
"last": "Gurevych",
"suffix": ""
},
{
"first": "Benno",
"middle": [],
"last": "Stein",
"suffix": ""
}
],
"year": 2018,
"venue": "Proceedings of The 12th International Workshop on Semantic Evaluation",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Ivan Habernal, Henning Wachsmuth, Iryna Gurevych, and Benno Stein. 2018. SemEval-2018 task 12: The argument reasoning comprehension task. In Pro- ceedings of The 12th International Workshop on Se- mantic Evaluation, New Orleans, Louisiana. Associ- ation for Computational Linguistics.",
"links": null
},
"BIBREF11": {
"ref_id": "b11",
"title": "Arguing on the Toulmin model",
"authors": [
{
"first": "David",
"middle": [],
"last": "Hitchcock",
"suffix": ""
},
{
"first": "Bart",
"middle": [],
"last": "Verheij",
"suffix": ""
}
],
"year": 2006,
"venue": "",
"volume": "10",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "David Hitchcock and Bart Verheij. 2006. Arguing on the Toulmin model, volume 10. Springer.",
"links": null
},
"BIBREF12": {
"ref_id": "b12",
"title": "spaCy: Industrial-strength Natural Language Processing in Python",
"authors": [],
"year": null,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Matthew Honnibal, Ines Montani, Sofie Van Lan- deghem, and Adriane Boyd. 2020. spaCy: Industrial-strength Natural Language Processing in Python.",
"links": null
},
"BIBREF13": {
"ref_id": "b13",
"title": "Towards explaining natural language arguments with background knowledge",
"authors": [
{
"first": "Ioana",
"middle": [],
"last": "Hulpus",
"suffix": ""
},
{
"first": "Jonathan",
"middle": [],
"last": "Kobbe",
"suffix": ""
},
{
"first": "Christian",
"middle": [],
"last": "Meilicke",
"suffix": ""
},
{
"first": "Heiner",
"middle": [],
"last": "Stuckenschmidt",
"suffix": ""
},
{
"first": "Maria",
"middle": [],
"last": "Becker",
"suffix": ""
},
{
"first": "Juri",
"middle": [],
"last": "Opitz",
"suffix": ""
},
{
"first": "Vivi",
"middle": [],
"last": "Nastase",
"suffix": ""
},
{
"first": "Anette",
"middle": [],
"last": "Frank",
"suffix": ""
}
],
"year": 2019,
"venue": "PROFILES/SEMEX@ ISWC",
"volume": "",
"issue": "",
"pages": "62--77",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Ioana Hulpus, Jonathan Kobbe, Christian Meilicke, Heiner Stuckenschmidt, Maria Becker, Juri Opitz, Vivi Nastase, and Anette Frank. 2019. Towards explaining natural language arguments with back- ground knowledge. In PROFILES/SEMEX@ ISWC, pages 62-77.",
"links": null
},
"BIBREF14": {
"ref_id": "b14",
"title": "Multiple warrants in practical reasoning",
"authors": [
{
"first": "Christian",
"middle": [],
"last": "Kock",
"suffix": ""
}
],
"year": 2006,
"venue": "Arguing on the Toulmin model",
"volume": "",
"issue": "",
"pages": "247--259",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Christian Kock. 2006. Multiple warrants in practical reasoning. In Arguing on the Toulmin model, pages 247-259. Springer.",
"links": null
},
"BIBREF15": {
"ref_id": "b15",
"title": "Computing krippendorff's alpha-reliability",
"authors": [
{
"first": "Klaus",
"middle": [],
"last": "Krippendorff",
"suffix": ""
}
],
"year": 2011,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Klaus Krippendorff. 2011. Computing krippendorff's alpha-reliability.",
"links": null
},
"BIBREF16": {
"ref_id": "b16",
"title": "The abuses of argument: Understanding fallacies on toulmin's layout of argument",
"authors": [
{
"first": "Andrew",
"middle": [],
"last": "Pineau",
"suffix": ""
}
],
"year": 2013,
"venue": "Informal Logic",
"volume": "33",
"issue": "4",
"pages": "531--546",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Andrew Pineau. 2013. The abuses of argument: Under- standing fallacies on toulmin's layout of argument. Informal Logic, 33(4):531-546.",
"links": null
},
"BIBREF17": {
"ref_id": "b17",
"title": "Finding enthymemes in real-world texts: A feasibility study",
"authors": [
{
"first": "Olesya",
"middle": [],
"last": "Razuvayevskaya",
"suffix": ""
},
{
"first": "Simone",
"middle": [],
"last": "Teufel",
"suffix": ""
}
],
"year": 2017,
"venue": "Argument & computation",
"volume": "8",
"issue": "2",
"pages": "113--129",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Olesya Razuvayevskaya and Simone Teufel. 2017. Finding enthymemes in real-world texts: A feasibil- ity study. Argument & computation, 8(2):113-129.",
"links": null
},
"BIBREF18": {
"ref_id": "b18",
"title": "The use of argument",
"authors": [
{
"first": "",
"middle": [],
"last": "Stephen Edelston Toulmin",
"suffix": ""
}
],
"year": 1958,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Stephen Edelston Toulmin. 1958. The use of argument. Cambridge University Press.",
"links": null
},
"BIBREF19": {
"ref_id": "b19",
"title": "How to improve argumentation comprehension in university students: Experimental test of a training approach",
"authors": [
{
"first": "Tobias",
"middle": [],
"last": "Sarah Von Der M\u00fchlen",
"suffix": ""
},
{
"first": "Sebastian",
"middle": [],
"last": "Richter",
"suffix": ""
},
{
"first": "Kirsten",
"middle": [],
"last": "Schmid",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Berthold",
"suffix": ""
}
],
"year": 2019,
"venue": "Instructional Science",
"volume": "47",
"issue": "2",
"pages": "215--237",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Sarah von der M\u00fchlen, Tobias Richter, Sebastian Schmid, and Kirsten Berthold. 2019. How to im- prove argumentation comprehension in university students: Experimental test of a training approach. Instructional Science, 47(2):215-237.",
"links": null
},
"BIBREF20": {
"ref_id": "b20",
"title": "Argumentation schemes",
"authors": [
{
"first": "Douglas",
"middle": [],
"last": "Walton",
"suffix": ""
},
{
"first": "Christopher",
"middle": [],
"last": "Reed",
"suffix": ""
},
{
"first": "Fabrizio",
"middle": [],
"last": "Macagno",
"suffix": ""
}
],
"year": 2008,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Douglas Walton, Christopher Reed, and Fabrizio Macagno. 2008. Argumentation schemes. Cam- bridge University Press.",
"links": null
},
"BIBREF21": {
"ref_id": "b21",
"title": "Probable cause: Developing warrants for automated scoring of essays",
"authors": [
{
"first": "",
"middle": [],
"last": "David M Williamson",
"suffix": ""
}
],
"year": 2013,
"venue": "Handbook of automated essay evaluation",
"volume": "",
"issue": "",
"pages": "175--202",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "David M Williamson. 2013. Probable cause: De- veloping warrants for automated scoring of essays. In Handbook of automated essay evaluation, pages 175-202. Routledge.",
"links": null
}
},
"ref_entries": {
"FIGREF0": {
"uris": null,
"num": null,
"type_str": "figure",
"text": "The interface used for our warrant annotations, along with an example of a annotator's annotation for User-defined Keyword-based Warrant."
},
"TABREF0": {
"num": null,
"content": "<table/>",
"text": "Example warrants demonstrating the semi-structured form used for annotation.",
"html": null,
"type_str": "table"
}
}
}
} |