File size: 85,544 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 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 |
{
"paper_id": "2020",
"header": {
"generated_with": "S2ORC 1.0.0",
"date_generated": "2023-01-19T07:29:01.291048Z"
},
"title": "Generating Intelligible Plumitifs Descriptions: Use Case Application with Ethical Considerations",
"authors": [
{
"first": "David",
"middle": [],
"last": "Beauchemin",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "Universit\u00e9 Laval",
"location": {
"region": "Qu\u00e9bec",
"country": "Canada"
}
},
"email": "david.beauchemin.5@ulaval.ca"
},
{
"first": "Nicolas",
"middle": [],
"last": "Garneau",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "Universit\u00e9 Laval",
"location": {
"region": "Qu\u00e9bec",
"country": "Canada"
}
},
"email": "nicolas.garneau@ift.ulaval.ca"
},
{
"first": "Eve",
"middle": [],
"last": "Gaumond",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "Universit\u00e9 Laval",
"location": {
"region": "Qu\u00e9bec",
"country": "Canada"
}
},
"email": "eve.gaumond@observatoire-ia.ulaval.ca"
},
{
"first": "Pierre-Luc",
"middle": [],
"last": "D\u00e9ziel",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "Universit\u00e9 Laval",
"location": {
"region": "Qu\u00e9bec",
"country": "Canada"
}
},
"email": "pierre-luc.deziel@fd.ulaval.ca"
},
{
"first": "Richard",
"middle": [],
"last": "Khoury",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "Universit\u00e9 Laval",
"location": {
"region": "Qu\u00e9bec",
"country": "Canada"
}
},
"email": "richard.khoury@ift.ulaval.ca"
},
{
"first": "Luc",
"middle": [],
"last": "Lamontagne",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "Universit\u00e9 Laval",
"location": {
"region": "Qu\u00e9bec",
"country": "Canada"
}
},
"email": "luc.lamontagne@ift.ulaval.ca"
}
],
"year": "",
"venue": null,
"identifiers": {},
"abstract": "Plumitifs (dockets) were initially a tool for law clerks. Nowadays, they are used as summaries presenting all the steps of a judicial case. Information concerning parties' identity, jurisdiction in charge of administering the case, and some information relating to the nature and the course of the preceding are available through plumitifs. They are publicly accessible but barely understandable; they are written using abbreviations and referring to provisions from the Criminal Code of Canada, which makes them hard to reason about. In this paper, we propose a simple yet efficient multi-source language generation architecture that leverages both the plumitif and the Criminal Code's content to generate intelligible plumitifs descriptions. It goes without saying that ethical considerations rise with these sensitive documents made readable and available at scale, legitimate concerns that we address in this paper. This is, to the best of our knowledge, the first application of plumitifs descriptions generation made available for French speakers along with an ethical discussion about the topic.",
"pdf_parse": {
"paper_id": "2020",
"_pdf_hash": "",
"abstract": [
{
"text": "Plumitifs (dockets) were initially a tool for law clerks. Nowadays, they are used as summaries presenting all the steps of a judicial case. Information concerning parties' identity, jurisdiction in charge of administering the case, and some information relating to the nature and the course of the preceding are available through plumitifs. They are publicly accessible but barely understandable; they are written using abbreviations and referring to provisions from the Criminal Code of Canada, which makes them hard to reason about. In this paper, we propose a simple yet efficient multi-source language generation architecture that leverages both the plumitif and the Criminal Code's content to generate intelligible plumitifs descriptions. It goes without saying that ethical considerations rise with these sensitive documents made readable and available at scale, legitimate concerns that we address in this paper. This is, to the best of our knowledge, the first application of plumitifs descriptions generation made available for French speakers along with an ethical discussion about the topic.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Abstract",
"sec_num": null
}
],
"body_text": [
{
"text": "The right to access judicial information is a fundamental component of Canadian democracy and its judicial process (Vancouver Sun (Re), 2004; CBC. v Canada (A.G), 2011) 1 . This right has two main purposes. First, to enhance judicial accountability by providing opportunities to the public to scrutinize it and put forward criticisms of the judicial process (Sierra Club of Canada v Canada (Minister of Finance), 2002; CBC. v New Brunswick (A.G), [1996] ). Second, it has an educational purpose: by accessing judicial information, people acquire a better understanding of the court process (Edmonton Journal v Alberta (A.G), [1989] ). Given these purposes, the necessity to provide access to judicial information in an intelligible form cannot be ignored. Indeed, getting a copy of a document is not enough; people have to understand its contents. This is particularly crucial in a digital context since citizens face an overload of judicial information online (Eltis, 2011) . As a consequence, litigants have great difficulty in finding relevant information for their case online (Dionne, 2019) .",
"cite_spans": [
{
"start": 447,
"end": 453,
"text": "[1996]",
"ref_id": null
},
{
"start": 625,
"end": 631,
"text": "[1989]",
"ref_id": null
},
{
"start": 961,
"end": 974,
"text": "(Eltis, 2011)",
"ref_id": "BIBREF7"
},
{
"start": 1081,
"end": 1095,
"text": "(Dionne, 2019)",
"ref_id": "BIBREF6"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "Studies show that, in the province of Quebec, the plumitif (a public register where one can find an official trace of all the actions taken by the courts) lacks intelligibility (Tep et al., 2019) . Some users have called it \"non-sense\" for non-attorneys (Parada et al., 2020 ). Yet, the plumitif is necessary for every litigant as it provides information concerning the parties' identity, the jurisdiction responsible for administering cases, and information relating to the nature and the course of proceedings. In this work, we aim at leveraging both information extraction and natural language generation to increase the intelligibility of excerpts of the Court of Quebec's plumitif regarding criminal offenses under the Criminal Code of Canada (CCC) .",
"cite_spans": [
{
"start": 177,
"end": 195,
"text": "(Tep et al., 2019)",
"ref_id": "BIBREF28"
},
{
"start": 254,
"end": 274,
"text": "(Parada et al., 2020",
"ref_id": null
},
{
"start": 741,
"end": 753,
"text": "Canada (CCC)",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "Improving the comprehension of textual legal documents has been the subject of several studies in the past. For example, patent claims are long legal pieces of texts that contain complex sentences making it hard for a layperson to reason about. Sheremetyeva (2014) framed this problem into an automatic text simplification task while Farzindar et al. (2004) and Hachey and Grover (2006) proposed extractive summarization techniques to make them easier to understand. The plumitifs, while also lying in the \"legal texts\" family, take a completely different form; they are not written in a valid grammatical form, and contain many abbreviations and references to the CCC. This makes our use case application rather unique.",
"cite_spans": [
{
"start": 245,
"end": 264,
"text": "Sheremetyeva (2014)",
"ref_id": "BIBREF25"
},
{
"start": 334,
"end": 357,
"text": "Farzindar et al. (2004)",
"ref_id": "BIBREF8"
},
{
"start": 362,
"end": 386,
"text": "Hachey and Grover (2006)",
"ref_id": "BIBREF9"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "To handle this type of document, we have de-signed a description generation pipeline, divided into three steps. The first step consists of segmenting a plumitif into different parts. In the second step, we extract, for each part, the relevant information using a Named Entity Recognition (NER) model. For the final step, we generate sentences from the data extracted by the NER model. To this end, we use a template-filling approach to ensure there are no factual fallacies introduced in the generation, an essential concern in legal text generation. Moreover, we use a statistical language model in a controlled setting to augment the generation with vital contextual information, namely texts from the CCC, making our approach a hybrid generation model. Our contributions, in this work, are twofold; 1. We propose a simple yet robust data-to-text multi-source textual generation pipeline to make plumitifs easier to understand for the litigants (made available through a web application, see Appendix I); 2. We bring a discussion on the ethical considerations about privacy and discrimination that such an application may cause. We further describe our architecture, related work and methodology in Section 2 and evaluate its generation capabilities in Section 3. We bring important ethical considerations in Section 4 and open the discussion for future work in Section 5.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "Plumitifs are used as summaries presenting all the steps of a case heard by the court. In the context of criminal proceedings, they contain information about the plaintiff, the accused, different charges along with their associated penalty (if applicable). We present a plumitif example in Appendix A, Figure 2. Plumitifs are freely available in person at any courthouse and are also accessible on the Soci\u00e9t\u00e9 qu\u00e9b\u00e9coise d'information juridique (SOQUIJ) website 2 where they can be consulted for a fee. In this section, we detail our proposed architecture, which is broken down into three steps; segmenting the plumitif into parts, extracting relevant information from each part, and generating descriptions by also leveraging the CCC. We illustrate the whole architecture in Appendix B, Figure 4 and further detail each component in the following subsections.",
"cite_spans": [],
"ref_spans": [
{
"start": 302,
"end": 308,
"text": "Figure",
"ref_id": null
},
{
"start": 788,
"end": 796,
"text": "Figure 4",
"ref_id": null
}
],
"eq_spans": [],
"section": "Generating Intelligible Plumitifs Summaries",
"sec_num": "2"
},
{
"text": "2 https://soquij.qc.ca/",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Generating Intelligible Plumitifs Summaries",
"sec_num": "2"
},
{
"text": "We identify three parts in a plumitif ; the accused, the plaintiff, and the charges. Since the plumitif structure is pretty regular, it allows us to identify each one using simple heuristics based on the presence of specific strings (e.g. \"ACC.\" for \"accused\") with 100% accuracy. Splitting into parts simplifies the NER step since these models typically use a narrow contextual window of a few tokens on either side to make their prediction. It also provides more data points overall.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Segmenting the Plumitif",
"sec_num": "2.1"
},
{
"text": "As mentioned in Section 1, we frame the retrieval of the relevant entities in the plumitif as an information extraction problem. That is, given a raw part of the plumitif, a NER model extracts entities from the text to fill in a normalized view. We established nine types of entities that need to be extracted; Adresses (Addresses), Accusations et sp\u00e9cifications d'accusations (Charges and Charges Specifications), Dates, D\u00e9cisions (Decisions), Lois (Laws), Accusations, Organisations (Organizations), Personnes (Persons), Plaidoyer (Pleas) and Peines (Sentences). For the rest of the paper, we will use the French entities within the French templates and rules, and the English entities otherwise (i.e. in the text). We manually annotated 816 plumitifs from eight districts over the last five years, to cover as much variety as possible. These eight districts are the ones with the most cases for this date range. We train a NER model on the annotated dataset, which achieves, on average, a F1-Score of 0.965, thanks to the regularity in the form the plumitifs can take 3 .",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Extracting Relevant Information",
"sec_num": "2.2"
},
{
"text": "Once the relevant information is extracted and normalized, we use it in the third step of the pipeline, which consists of a data-to-text generation model, described in the following subsection.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Extracting Relevant Information",
"sec_num": "2.2"
},
{
"text": "Even though statistical and deep Natural Language Generation (NLG) has seen tremendous breakthroughs in recent years (Radford et al., 2018 Brown et al., 2020) , we decide not to strictly rely on this kind of Transformer model (Vaswani et al., 2017) for our use case. Several architectures (Ziegler et al., 2019; Keskar et al., 2019; Dathathri et al., 2020) attempt to control the generation of such pre-trained models by using conditioning elements that propose a specific stylistic or emotion for example. However, Brown et al. (2020) showed that one of the best neural language models to date (GPT-3) may generate non-factual utterances, often called hallucinations (Rohrbach et al., 2018; Rebuffel et al., 2020) , or even hide significant biases that may put the credibility of generation at stake.",
"cite_spans": [
{
"start": 117,
"end": 138,
"text": "(Radford et al., 2018",
"ref_id": "BIBREF20"
},
{
"start": 139,
"end": 158,
"text": "Brown et al., 2020)",
"ref_id": null
},
{
"start": 226,
"end": 248,
"text": "(Vaswani et al., 2017)",
"ref_id": null
},
{
"start": 289,
"end": 311,
"text": "(Ziegler et al., 2019;",
"ref_id": "BIBREF32"
},
{
"start": 312,
"end": 332,
"text": "Keskar et al., 2019;",
"ref_id": "BIBREF11"
},
{
"start": 333,
"end": 356,
"text": "Dathathri et al., 2020)",
"ref_id": "BIBREF3"
},
{
"start": 516,
"end": 535,
"text": "Brown et al. (2020)",
"ref_id": null
},
{
"start": 668,
"end": 691,
"text": "(Rohrbach et al., 2018;",
"ref_id": "BIBREF24"
},
{
"start": 692,
"end": 714,
"text": "Rebuffel et al., 2020)",
"ref_id": "BIBREF23"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Realisation of Plumitif Summaries",
"sec_num": "2.3"
},
{
"text": "Since we generate legal textual content that can be used in various sensitive applications (e.g. HR screening, (Parada et al., 2020 )), we can't afford to let a model \"statistically\" generate a non-factual decision (e.g. guilty but the accused is not) or a charge (e.g. something that the accused has not done). Thus, we prefer to sacrifice variability for control by using a template-filling approach. Puzikov and Gurevych (2018) showed that a template-based approach can be as good as a neural encoder-decoder model on generating restaurant descriptions from sets of key-value pairs. Deemter et al. 2005also argues that \"template-based approaches to the generation of language are not necessarily inferior to other [statistical] approaches as regards their maintainability, linguistic well-foundedness and quality of output\". This approach has been shown recently to perform well in different areas like weather reports (Ramos-Soto et al., 2015), financial analysis (Nesterenko, 2016) and soccer game reports (van der Lee et al., 2017) where they are used in production.",
"cite_spans": [
{
"start": 111,
"end": 131,
"text": "(Parada et al., 2020",
"ref_id": null
},
{
"start": 403,
"end": 430,
"text": "Puzikov and Gurevych (2018)",
"ref_id": "BIBREF19"
},
{
"start": 968,
"end": 986,
"text": "(Nesterenko, 2016)",
"ref_id": "BIBREF15"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Realisation of Plumitif Summaries",
"sec_num": "2.3"
},
{
"text": "In the same way Deemter et al. (2005) did, we manually deduce 66 patterns from a subset of the plumitifs to generate the description text using the extracted information from the model introduced in Section 2.2 4 . The generation rules (especially the sentence ones) have been written by a legal expert. Following the example in Figure 2 , with the corresponding extracted information about the accused and a really simple yet efficient rule, we can generate texts about the accused and the plaintiff, as illustrated in Appendix D.",
"cite_spans": [],
"ref_spans": [
{
"start": 329,
"end": 337,
"text": "Figure 2",
"ref_id": null
}
],
"eq_spans": [],
"section": "Template-Based, Data-to-Text Generation",
"sec_num": "2.3.1"
},
{
"text": "In the next subsection, we present how we combine the information extracted from the plumitif with a parsed version of the CCC 5 using a Masked Language Model.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Template-Based, Data-to-Text Generation",
"sec_num": "2.3.1"
},
{
"text": "The Criminal Code of Canada (CCC) is an act that contains most of the criminal law in Canada. It contains around 1,500 provisions (referred to with numbers) where each of them comprises paragraphs and subparagraphs. The plumitifs refers to provisions from the law using only the provision numbers, which provides little to no context to the litigants. Therefore, it is essential to extract the law's text from the Criminal Code when generating the plumitif 's summary. However, the CCC is only available in HTML or PDF format, making it hard to query it programmatically. Thus, we parsed the HTML version into the JSON format, which allows us to easily query for different articles, paragraphs and subparagraphs 6 . A plumitif may contain several charges. Each charge may refer to one or two provisions from the law. The first provision is most likely referring to the description of the law, where the title briefly summarizes the description. The second provision (if any) is usually there to specify the charge 7 .",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Leveraging the Criminal Code of Canada",
"sec_num": "2.3.2"
},
{
"text": "Given the following template (see Appendix G for a translated version); <Accus\u00e9> est accus\u00e9 <Article>.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Leveraging the Criminal Code of Canada",
"sec_num": "2.3.2"
},
{
"text": "we wish to insert the provision title syntactically. To this end, we propose to \"stitch\" the two pieces of the template using a Masked Language Model. We use the French pre-trained version of BERT (Devlin et al., 2019) , CamemBERT (Martin et al., 2020) , which has been trained on the French subset of OSCAR (Ortiz Su\u00e1rez et al., 2020), a huge multilingual corpus obtained by language classification and filtering of the Common Crawl corpus.",
"cite_spans": [
{
"start": 197,
"end": 218,
"text": "(Devlin et al., 2019)",
"ref_id": "BIBREF5"
},
{
"start": 231,
"end": 252,
"text": "(Martin et al., 2020)",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Leveraging the Criminal Code of Canada",
"sec_num": "2.3.2"
},
{
"text": "One of BERT's abilities is to predict randomly masked tokens in a sentence, usually referred to as a Cloze task in the literature (Taylor, 1953) . We specifically leverage this ability to our benefit, and let CamemBERT predict the proper preposition that should be inserted between the template and the charge's title (d\u00e9faut de se conformer\u00e0 une ordonnance here). The realisation of the previous template would then look like the following (Appendix G); John Doe est accus\u00e9 pour d\u00e9faut de se conformer\u00e0 une ordonnance.",
"cite_spans": [
{
"start": 130,
"end": 144,
"text": "(Taylor, 1953)",
"ref_id": "BIBREF27"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Leveraging the Criminal Code of Canada",
"sec_num": "2.3.2"
},
{
"text": "Using the 134 unique charges titles included in our dataset, we find that CamemBERT can predict the right preposition 84% of the time.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Leveraging the Criminal Code of Canada",
"sec_num": "2.3.2"
},
{
"text": "The generation of the pleas and decision text is simple since there are only a few possible situations, using 14 generation rules out of the 66 deduced. For the first, it is either guilty or not guilty. For the second, it is guilty, not guilty, or ten other technical situations such as \"arret\" (i.e. case where the court orders a stay of proceedings). In both cases, the mapping between the pleas and decision is one-toone with the associated generated text (i.e. a guilty decision can generate only one text). We illustrate this case in Appendix E.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Pleas, Decisions and Sentences",
"sec_num": "2.3.3"
},
{
"text": "On the other hand, generating Sentences is more complex. In our set of 66 deduced generation rules, 50 are used to generate the Sentences. This complexity is mostly due to the occurrences of different convictions in one Sentence, meaning that the mapping is one-to-many (i.e. a Sentence can have an unknown number of convictions). Given the Sentence's extracted convictions, we order them by types (i.e. the penalty inflicted of, fines and fees, community work, other convictions, probation and surcharge) and fill-in an \"on-the-fly merged generation template\" given the list of convictions. It is important to note that generation rules are not applied \"in cascade\" i.e. for a given list of convictions, there is one possible generation template. We illustrate the generation of the first Sentence's section in Appendix F.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Pleas, Decisions and Sentences",
"sec_num": "2.3.3"
},
{
"text": "Since our generation model mostly relies on rules, it is straightforward to evaluate its performance; we first need to make sure all the relevant information is fully extracted (NER step) and that it properly fills in the corresponding template (generation step).",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Evaluating the Realisation of the Summaries",
"sec_num": "3"
},
{
"text": "We thus quantify our model's performance in terms of \"Error Rate\" where a generation error is the lack of realizing a specific part (accused, plaintiff or list of charges paragraphs), instead of evaluating the textual generation. The counts are computed per text. Errors are split into two categories; Extractionbased Errors (EE) and Generation-based Errors (GE). For clarity, we display the Errors Rates by districts in Table 1 . In most cases, we find that a wrong extraction of the Plaintiff (due to the NER model) causes EE.",
"cite_spans": [],
"ref_spans": [
{
"start": 421,
"end": 428,
"text": "Table 1",
"ref_id": "TABREF1"
}
],
"eq_spans": [],
"section": "Evaluating the Realisation of the Summaries",
"sec_num": "3"
},
{
"text": "We can see that Granby and Sherbrooke have the highest EE rate; this is mostly due to the many different values an Organisation can take in these districts 8 . GE are mainly due to edge cases found in plumitifs which our rules do not cover. As we can see from the GE Rates in Table 1 , our generation rules commit most errors on the Montr\u00e9al, Sherbrooke and Gatineau districts. This is due to the numerous and diverse convictions these plumitifs hold. For example, a particular combination of convictions may not be associated to any generation rule. We illustrate this problem with an example in Figure 1 , where the Sentence comprises multiple convictions and are essentially edge cases about the duration. This highlights the need to have a better model at parsing and generating Sentences' paragraphs. Using a generative, sequence-to-sequence model, such as the one proposed by (Bahdanau et al., 2015) may be a better option, but we leave this study as future work. All in all, our model achieves low Error Rates (13% EE and 5% GE on average), allowing simple yet accurate textual generation of intelligible plumitifs. While these results are interesting, it raises some ethical concerns, that we discuss in the next section.",
"cite_spans": [
{
"start": 882,
"end": 905,
"text": "(Bahdanau et al., 2015)",
"ref_id": "BIBREF0"
}
],
"ref_spans": [
{
"start": 276,
"end": 283,
"text": "Table 1",
"ref_id": "TABREF1"
},
{
"start": 597,
"end": 605,
"text": "Figure 1",
"ref_id": null
}
],
"eq_spans": [],
"section": "Evaluating the Realisation of the Summaries",
"sec_num": "3"
},
{
"text": "There is some ethical considerations regarding our dataset's privacy that ought to be addressed. Plumitifs contain sensitive information such as the names, dates of birth, addresses and criminal backgrounds of accused people. The identity of judges, plaintiffs, clerks, and attorneys taking part in a criminal case are also found in the plumitifs. As explained in Section 1, all of this information must be publicly accessible. As long as this data is protected by practical obscurity 9 , the actual risks from public access of this information are limited (Vermeys, 2016) .",
"cite_spans": [
{
"start": 557,
"end": 572,
"text": "(Vermeys, 2016)",
"ref_id": "BIBREF30"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Ethical Considerations",
"sec_num": "4"
},
{
"text": "However, if this data was to be released in bulk to the scientific community, it would not be \"scattered [. . .] bits of information\" (US Department of Justice v. Reporters Committee for Freedom of the Press, 1989) that require time and resources to retrieve anymore. Information could be easily searched, aggregated or combined with information from other public sources. This poses a risk to the privacy of judicial stakeholders.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Ethical Considerations",
"sec_num": "4"
},
{
"text": "In this subsection, we explain why we decided not to release our data set publicly (raw or synthesized). To put it in straightforward terms: information collected in public records should not be \"up for grabs\". Its use can result in privacy violations. This is especially true in the digital context where aggregation, linkage and analytics are made easier (Martin and Nissenbaum, 2017) . There are several examples of privacy violations that occurred due to the malicious use of judicial information that was publicly accessible. For instance, more than 270 cases of identity theft have been linked to a security lapse in an American Municipal Court's website. (Bailey and Burkell, 2017) . The Office of the Privacy Commissioner of Canada had to intervene to end an extortion scheme relying on data available from the Canadian Legal Information Institute and SOQUIJ's websites (A.T. v Globe24h.com, 2017). United State's \"Public Access to Court Electronic Records\" system made the identity of some cooperating defendants and undercover agents publicly available, which contributed to the intimidation and harassment of witnesses in order to discourage them from testifying (Eltis, 2011). There have also been some docu-9 A term broadly used to explain that documents might be accessible to all in principle, but that the access is hindered by some obstacles such as fees to consult a document or the need to go physically to a location -as is the case for the plumitif.",
"cite_spans": [
{
"start": 357,
"end": 386,
"text": "(Martin and Nissenbaum, 2017)",
"ref_id": "BIBREF13"
},
{
"start": 662,
"end": 688,
"text": "(Bailey and Burkell, 2017)",
"ref_id": "BIBREF1"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Ethical Considerations",
"sec_num": "4"
},
{
"text": "mented cases of discrimination in the context of employment (Solove, 2002) and housing (Gichuru v Purewal and another, 2017) caused by judicial information available online. Moreover, academics have expressed significant concerns about the secondary use of judicial information for marketing purposes.",
"cite_spans": [
{
"start": 60,
"end": 74,
"text": "(Solove, 2002)",
"ref_id": "BIBREF26"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Ethical Considerations",
"sec_num": "4"
},
{
"text": "This is now prohibited by the Personal Information Protection and Electronic Documents Act, (Office of the Privacy Commissionner of Canada, 2014) , but (Bailey and Burkell, 2017) argues that this regulatory framework is not sufficient to prevent inappropriate uses of judicial data. Our team is currently working to develop a framework for the management of personal information contained in digital court records. However, for the moment, since the law provides no satisfactory solution, we chose not to release the dataset used to train our algorithm.",
"cite_spans": [
{
"start": 132,
"end": 145,
"text": "Canada, 2014)",
"ref_id": null
},
{
"start": 152,
"end": 178,
"text": "(Bailey and Burkell, 2017)",
"ref_id": "BIBREF1"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Ethical Considerations",
"sec_num": "4"
},
{
"text": "In this paper, we introduce a simple yet effective multi-source architecture able to generate digestible plumitifs for Canadian citizens. We also show that we are in a position to easily divulge who has been accused of what and the outcome of it, which raises some important ethical concerns. In the future, we plan to explore statistical natural language generation further by using case law, provide more diverse plumitifs descriptions and improve the generation of Sentences. Finally, we hope that our application will provide better insights to the community and give the right direction for the next applications of not only NLG, but Machine Learning in general, in the field of law.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Conclusion and Future Work",
"sec_num": "5"
},
{
"text": "Training details are available in Appendix C",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
},
{
"text": "We present a complete generation example in Appendix H based on the plumitif presented inFigure 2.5 https://laws-lois.justice.gc.ca/eng/ acts/c-46/",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
},
{
"text": "We were able to properly extract the 1518 provisions publicly release the JSON version of the French CCC here: https://bit.ly/3kiBdFd7 In this work, we do not leverage the second provision.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
},
{
"text": "This corroborates with the results of the NER model for the entity Organisation, in Section 2.2",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
}
],
"back_matter": [
{
"text": "We thank the reviewers for their insightful comments on our manuscript. This research was enabled in part by the support of the Natural Sciences and Engineering Research Council of Canada (NSERC) and the Social Sciences and Humanities Research Council (SSHR). Also, this research was made possible with the help of our partner, SOQUIJ.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Acknowledgements",
"sec_num": "6"
}
],
"bib_entries": {
"BIBREF0": {
"ref_id": "b0",
"title": "Neural machine translation by jointly learning to align and translate",
"authors": [
{
"first": "Dzmitry",
"middle": [],
"last": "Bahdanau",
"suffix": ""
},
{
"first": "Kyunghyun",
"middle": [],
"last": "Cho",
"suffix": ""
},
{
"first": "Yoshua",
"middle": [],
"last": "Bengio",
"suffix": ""
}
],
"year": 2015,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Dzmitry Bahdanau, Kyunghyun Cho, and Yoshua Bengio. 2015. Neural machine translation by jointly learning to align and translate. CoRR, abs/1409.0473.",
"links": null
},
"BIBREF1": {
"ref_id": "b1",
"title": "Privacy as Contextual Integrity",
"authors": [
{
"first": "Jane",
"middle": [],
"last": "Bailey",
"suffix": ""
},
{
"first": "Jacquelyn",
"middle": [],
"last": "Burkell",
"suffix": ""
}
],
"year": 2017,
"venue": "Ottawa Law Review",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Jane Bailey and Jacquelyn Burkell. 2017. Privacy as Contextual Integrity. Ottawa Law Review, 48.",
"links": null
},
"BIBREF3": {
"ref_id": "b3",
"title": "Plug and Play Language Models: A Simple Approach to Controlled Text Generation",
"authors": [
{
"first": "Sumanth",
"middle": [],
"last": "Dathathri",
"suffix": ""
},
{
"first": "Andrea",
"middle": [],
"last": "Madotto",
"suffix": ""
},
{
"first": "Janice",
"middle": [],
"last": "Lan",
"suffix": ""
},
{
"first": "Jane",
"middle": [],
"last": "Hung",
"suffix": ""
},
{
"first": "Eric",
"middle": [
"C"
],
"last": "Frank",
"suffix": ""
},
{
"first": "Piero",
"middle": [],
"last": "Molino",
"suffix": ""
},
{
"first": "Jason",
"middle": [],
"last": "Yosinski",
"suffix": ""
},
{
"first": "Rosanne",
"middle": [],
"last": "Liu",
"suffix": ""
}
],
"year": 2020,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Sumanth Dathathri, Andrea Madotto, Janice Lan, Jane Hung, Eric C. Frank, Piero Molino, Jason Yosinski, and Rosanne Liu. 2020. Plug and Play Language Models: A Simple Approach to Con- trolled Text Generation. ArXiv, 1912.02164.",
"links": null
},
"BIBREF4": {
"ref_id": "b4",
"title": "Real versus Template-Based Natural Language Generation: A False Opposition? Computational Linguistics",
"authors": [
{
"first": "Kees",
"middle": [],
"last": "Van Deemter",
"suffix": ""
},
{
"first": "Mari\u00ebt",
"middle": [],
"last": "Theune",
"suffix": ""
},
{
"first": "Emiel",
"middle": [],
"last": "Krahmer",
"suffix": ""
}
],
"year": 2005,
"venue": "",
"volume": "31",
"issue": "",
"pages": "15--24",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Kees Van Deemter, Mari\u00ebt Theune, and Emiel Krahmer. 2005. Real versus Template-Based Natural Language Generation: A False Opposi- tion? Computational Linguistics, 31:15-24.",
"links": null
},
"BIBREF5": {
"ref_id": "b5",
"title": "BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding",
"authors": [
{
"first": "Jacob",
"middle": [],
"last": "Devlin",
"suffix": ""
},
{
"first": "Ming-Wei",
"middle": [],
"last": "Chang",
"suffix": ""
},
{
"first": "Kenton",
"middle": [],
"last": "Lee",
"suffix": ""
},
{
"first": "Kristina",
"middle": [],
"last": "Toutanova",
"suffix": ""
}
],
"year": 2019,
"venue": "NAACL-HLT",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding. In NAACL-HLT.",
"links": null
},
"BIBREF6": {
"ref_id": "b6",
"title": "L'acc\u00e8s\u00e0 la justice en contexte num\u00e9rique : l'information juridique par et pour les justiciables sur les m\u00e9dias sociaux. Windsor Yearbook of Access to Justice",
"authors": [
{
"first": "Alexandra",
"middle": [],
"last": "Bahary",
"suffix": ""
},
{
"first": "Dionne",
"middle": [],
"last": "",
"suffix": ""
}
],
"year": 2019,
"venue": "",
"volume": "35",
"issue": "",
"pages": "337--362",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Alexandra Bahary Dionne. 2019. L'acc\u00e8s\u00e0 la jus- tice en contexte num\u00e9rique : l'information ju- ridique par et pour les justiciables sur les m\u00e9dias sociaux. Windsor Yearbook of Access to Justice, 35:337-362.",
"links": null
},
"BIBREF7": {
"ref_id": "b7",
"title": "The judicial system in the judicial age: revisiting the relashionship between privacy and accessibility in the cyber context. MLJ",
"authors": [
{
"first": "Karen",
"middle": [
"Eltis"
],
"last": "",
"suffix": ""
}
],
"year": 2011,
"venue": "",
"volume": "2",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Karen Eltis. 2011. The judicial system in the ju- dicial age: revisiting the relashionship between privacy and accessibility in the cyber context. MLJ, 2.",
"links": null
},
"BIBREF8": {
"ref_id": "b8",
"title": "R\u00e9sum\u00e9 de textes juridiques par identification de leur structure th\u00e9matique",
"authors": [
{
"first": "Atefeh",
"middle": [],
"last": "Farzindar",
"suffix": ""
},
{
"first": "Guy",
"middle": [],
"last": "Lapalme",
"suffix": ""
},
{
"first": "Jean-Pierre",
"middle": [],
"last": "Descl\u00e9s",
"suffix": ""
}
],
"year": 2004,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Atefeh Farzindar, Guy Lapalme, and Jean-Pierre Descl\u00e9s. 2004. R\u00e9sum\u00e9 de textes juridiques par identification de leur structure th\u00e9matique : R\u00e9sum\u00e9 automatique de textes.",
"links": null
},
"BIBREF9": {
"ref_id": "b9",
"title": "Extractive summarisation of legal texts",
"authors": [
{
"first": "Ben",
"middle": [],
"last": "Hachey",
"suffix": ""
},
{
"first": "Claire",
"middle": [],
"last": "Grover",
"suffix": ""
}
],
"year": 2006,
"venue": "Artificial Intelligence and Law",
"volume": "14",
"issue": "",
"pages": "305--345",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Ben Hachey and Claire Grover. 2006. Extractive summarisation of legal texts. Artificial Intelli- gence and Law, 14:305-345.",
"links": null
},
"BIBREF10": {
"ref_id": "b10",
"title": "Natural language understanding with Bloom embeddings, convolutional neural networks and incremental parsing",
"authors": [
{
"first": "Matthew",
"middle": [],
"last": "Honnibal",
"suffix": ""
},
{
"first": "Ines",
"middle": [],
"last": "Montani",
"suffix": ""
}
],
"year": 2017,
"venue": "",
"volume": "2",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Matthew Honnibal and Ines Montani. 2017. spaCy 2: Natural language understanding with Bloom embeddings, convolutional neural networks and incremental parsing.",
"links": null
},
"BIBREF11": {
"ref_id": "b11",
"title": "CTRL: A Conditional Transformer Language Model for Controllable Generation",
"authors": [
{
"first": "Bryan",
"middle": [],
"last": "Nitish Shirish Keskar",
"suffix": ""
},
{
"first": "Lav",
"middle": [
"R"
],
"last": "Mccann",
"suffix": ""
},
{
"first": "Caiming",
"middle": [],
"last": "Varshney",
"suffix": ""
},
{
"first": "Richard",
"middle": [],
"last": "Xiong",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Socher",
"suffix": ""
}
],
"year": 1909,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Nitish Shirish Keskar, Bryan McCann, Lav R. Varshney, Caiming Xiong, and Richard Socher. 2019. CTRL: A Conditional Transformer Lan- guage Model for Controllable Generation. ArXiv, 1909.05858.",
"links": null
},
"BIBREF12": {
"ref_id": "b12",
"title": "Pass: A dutch data-to-text system for soccer, targeted towards specific audiences",
"authors": [
{
"first": "Chris",
"middle": [],
"last": "Van Der Lee",
"suffix": ""
},
{
"first": "Emiel",
"middle": [],
"last": "Krahmer",
"suffix": ""
},
{
"first": "Sander",
"middle": [],
"last": "Wubben",
"suffix": ""
}
],
"year": 2017,
"venue": "INLG",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Chris van der Lee, Emiel Krahmer, and Sander Wubben. 2017. Pass: A dutch data-to-text sys- tem for soccer, targeted towards specific audi- ences. In INLG.",
"links": null
},
"BIBREF13": {
"ref_id": "b13",
"title": "Privacy Interests in Public Records: An Empirical Investigation",
"authors": [
{
"first": "Kristen",
"middle": [],
"last": "Martin",
"suffix": ""
},
{
"first": "Helen",
"middle": [],
"last": "Nissenbaum",
"suffix": ""
}
],
"year": 2017,
"venue": "Harvard Journal of Law & Technology",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Kristen Martin and Helen Nissenbaum. 2017. Pri- vacy Interests in Public Records: An Empirical Investigation. Harvard Journal of Law & Tech- nology, 31.",
"links": null
},
"BIBREF14": {
"ref_id": "b14",
"title": "Eric Villemonte de la Clergerie, Djam\u00e9 Seddah, and Beno\u00eet Sagot. 2020. CamemBERT: a Tasty French Language Model",
"authors": [
{
"first": "Louis",
"middle": [],
"last": "Martin",
"suffix": ""
},
{
"first": "Benjamin",
"middle": [],
"last": "Muller",
"suffix": ""
},
{
"first": "Pedro Javier Ortiz",
"middle": [],
"last": "Su\u00e1rez",
"suffix": ""
},
{
"first": "Yoann",
"middle": [],
"last": "Dupont",
"suffix": ""
},
{
"first": "Laurent",
"middle": [],
"last": "Romary",
"suffix": ""
}
],
"year": null,
"venue": "ACL",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Louis Martin, Benjamin Muller, Pedro Javier Or- tiz Su\u00e1rez, Yoann Dupont, Laurent Romary, Eric Villemonte de la Clergerie, Djam\u00e9 Seddah, and Beno\u00eet Sagot. 2020. CamemBERT: a Tasty French Language Model. In ACL.",
"links": null
},
"BIBREF15": {
"ref_id": "b15",
"title": "Building a system for stock news generation in russian",
"authors": [
{
"first": "Liubov",
"middle": [],
"last": "Nesterenko",
"suffix": ""
}
],
"year": 2016,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Liubov Nesterenko. 2016. Building a system for stock news generation in russian. In WebNLG.",
"links": null
},
"BIBREF16": {
"ref_id": "b16",
"title": "Office of the Privacy Commissionner of Canada",
"authors": [],
"year": 2014,
"venue": "Publicly available information -interpretation bulletin",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Office of the Privacy Commissionner of Canada. 2014. Publicly available information -interpre- tation bulletin.",
"links": null
},
"BIBREF17": {
"ref_id": "b17",
"title": "A monolingual approach to contextualized word embeddings for midresource languages",
"authors": [
{
"first": "Pedro Javier Ortiz",
"middle": [],
"last": "Su\u00e1rez",
"suffix": ""
},
{
"first": "Laurent",
"middle": [],
"last": "Romary",
"suffix": ""
},
{
"first": "Beno\u00eet",
"middle": [],
"last": "Sagot",
"suffix": ""
}
],
"year": 2020,
"venue": "ACL",
"volume": "",
"issue": "",
"pages": "1703--1714",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Pedro Javier Ortiz Su\u00e1rez, Laurent Romary, and Beno\u00eet Sagot. 2020. A monolingual approach to contextualized word embeddings for mid- resource languages. In ACL, pages 1703-1714.",
"links": null
},
"BIBREF18": {
"ref_id": "b18",
"title": "Digital Court Records : a Diversity of Uses. IJR",
"authors": [
{
"first": "Alexandra",
"middle": [],
"last": "Parada",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Sandrine Prom",
"suffix": ""
},
{
"first": "Florence",
"middle": [],
"last": "Tep",
"suffix": ""
},
{
"first": "Pierre",
"middle": [],
"last": "Millerand",
"suffix": ""
},
{
"first": "Anne-Marie",
"middle": [],
"last": "Noreau",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "San",
"suffix": ""
}
],
"year": null,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Alexandra Parada, Sandrine Prom Tep, Florence Millerand, Pierre Noreau, and Anne-Marie San- torineos. 2020. Digital Court Records : a Diver- sity of Uses. IJR, 9.",
"links": null
},
"BIBREF19": {
"ref_id": "b19",
"title": "E2e nlg challenge: Neural models vs. templates",
"authors": [
{
"first": "Yevgeniy",
"middle": [],
"last": "Puzikov",
"suffix": ""
},
{
"first": "Iryna",
"middle": [],
"last": "Gurevych",
"suffix": ""
}
],
"year": 2018,
"venue": "INLG",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Yevgeniy Puzikov and Iryna Gurevych. 2018. E2e nlg challenge: Neural models vs. templates. In INLG.",
"links": null
},
"BIBREF20": {
"ref_id": "b20",
"title": "Tim Salimans, and Ilya Sutskever",
"authors": [
{
"first": "Alec",
"middle": [],
"last": "Radford",
"suffix": ""
},
{
"first": "Karthik",
"middle": [],
"last": "Narasimhan",
"suffix": ""
}
],
"year": 2018,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Alec Radford, Karthik Narasimhan, Tim Salimans, and Ilya Sutskever. 2018. Improving language understanding by generative pre-training.",
"links": null
},
"BIBREF21": {
"ref_id": "b21",
"title": "Language models are unsupervised multitask learners",
"authors": [
{
"first": "Alec",
"middle": [],
"last": "Radford",
"suffix": ""
},
{
"first": "Jeffrey",
"middle": [],
"last": "Wu",
"suffix": ""
},
{
"first": "Rewon",
"middle": [],
"last": "Child",
"suffix": ""
},
{
"first": "David",
"middle": [],
"last": "Luan",
"suffix": ""
},
{
"first": "Dario",
"middle": [],
"last": "Amodei",
"suffix": ""
},
{
"first": "Ilya",
"middle": [],
"last": "Sutskever",
"suffix": ""
}
],
"year": 2019,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Alec Radford, Jeffrey Wu, Rewon Child, David Luan, Dario Amodei, and Ilya Sutskever. 2019. Language models are unsupervised multitask learners.",
"links": null
},
"BIBREF22": {
"ref_id": "b22",
"title": "Linguistic descriptions for automatic generation of textual short-term weather forecasts on real prediction data",
"authors": [
{
"first": "Alejandro",
"middle": [],
"last": "Ramos-Soto",
"suffix": ""
},
{
"first": "Alberto",
"middle": [],
"last": "Bugar\u00edn",
"suffix": ""
},
{
"first": "Sen\u00e9n",
"middle": [],
"last": "Barro",
"suffix": ""
},
{
"first": "Juan",
"middle": [],
"last": "Taboada",
"suffix": ""
}
],
"year": 2015,
"venue": "IEEE TFS",
"volume": "23",
"issue": "",
"pages": "44--57",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Alejandro Ramos-Soto, Alberto Bugar\u00edn, Sen\u00e9n Barro, and Juan Taboada. 2015. Linguistic de- scriptions for automatic generation of textual short-term weather forecasts on real prediction data. IEEE TFS, 23:44-57.",
"links": null
},
"BIBREF23": {
"ref_id": "b23",
"title": "Parenting via model-agnostic reinforcement learning to correct pathological behaviors in data-to-text generation",
"authors": [
{
"first": "Cl\u00e9ment",
"middle": [],
"last": "Rebuffel",
"suffix": ""
},
{
"first": "Laure",
"middle": [],
"last": "Soulier",
"suffix": ""
},
{
"first": "Geoffrey",
"middle": [],
"last": "Scoutheeten",
"suffix": ""
},
{
"first": "Patrick",
"middle": [],
"last": "Gallinari",
"suffix": ""
}
],
"year": 2020,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Cl\u00e9ment Rebuffel, Laure Soulier, Geoffrey Scoutheeten, and Patrick Gallinari. 2020. Parent- ing via model-agnostic reinforcement learning to correct pathological behaviors in data-to-text generation.",
"links": null
},
"BIBREF24": {
"ref_id": "b24",
"title": "Object hallucination in image captioning",
"authors": [
{
"first": "Anna",
"middle": [],
"last": "Rohrbach",
"suffix": ""
},
{
"first": "Lisa",
"middle": [
"Anne"
],
"last": "Hendricks",
"suffix": ""
},
{
"first": "Kaylee",
"middle": [],
"last": "Burns",
"suffix": ""
},
{
"first": "Trevor",
"middle": [],
"last": "Darrell",
"suffix": ""
},
{
"first": "Kate",
"middle": [],
"last": "Saenko",
"suffix": ""
}
],
"year": 2018,
"venue": "Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing",
"volume": "",
"issue": "",
"pages": "4035--4045",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Anna Rohrbach, Lisa Anne Hendricks, Kaylee Burns, Trevor Darrell, and Kate Saenko. 2018. Object hallucination in image captioning. In Pro- ceedings of the 2018 Conference on Empirical Methods in Natural Language Processing, pages 4035-4045, Brussels, Belgium. Association for Computational Linguistics.",
"links": null
},
"BIBREF25": {
"ref_id": "b25",
"title": "Automatic Text Simplification For Handling Intellectual Property (The Case of Multiple Patent Claims)",
"authors": [
{
"first": "Svetlana",
"middle": [],
"last": "Sheremetyeva",
"suffix": ""
}
],
"year": 2014,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Svetlana Sheremetyeva. 2014. Automatic Text Sim- plification For Handling Intellectual Property (The Case of Multiple Patent Claims).",
"links": null
},
"BIBREF26": {
"ref_id": "b26",
"title": "Access and Aggregation: Privacy, Public Records, and the Constitution",
"authors": [
{
"first": "J",
"middle": [],
"last": "Daniel",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Solove",
"suffix": ""
}
],
"year": 2002,
"venue": "Minnesota Law Review",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Daniel J. Solove. 2002. Access and Aggregation: Privacy, Public Records, and the Constitution. Minnesota Law Review, 86.",
"links": null
},
"BIBREF27": {
"ref_id": "b27",
"title": "Cloze Procedure\": A New Tool for Measuring Readability. JMCQ",
"authors": [
{
"first": "Wilson",
"middle": [
"L"
],
"last": "Taylor",
"suffix": ""
}
],
"year": 1953,
"venue": "",
"volume": "30",
"issue": "",
"pages": "415--433",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Wilson L. Taylor. 1953. \"Cloze Procedure\": A New Tool for Measuring Readability. JMCQ, 30:415 -433.",
"links": null
},
"BIBREF28": {
"ref_id": "b28",
"title": "Legal Information in Digital Form: the Challenge of Accessing Computerized Court Records. IJR",
"authors": [
{
"first": "Florence",
"middle": [],
"last": "Sandrine Prom Tep",
"suffix": ""
},
{
"first": "Alexandra",
"middle": [],
"last": "Millerand",
"suffix": ""
},
{
"first": "Alexandra",
"middle": [],
"last": "Parada",
"suffix": ""
},
{
"first": "Pierre",
"middle": [],
"last": "Bahary",
"suffix": ""
},
{
"first": "Anne-Marie",
"middle": [],
"last": "Noreau",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Santorineos",
"suffix": ""
}
],
"year": 2019,
"venue": "",
"volume": "8",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Sandrine Prom Tep, Florence Millerand, Alexandra Parada, Alexandra Bahary, Pierre Noreau, and Anne-Marie Santorineos. 2019. Legal Informa- tion in Digital Form: the Challenge of Accessing Computerized Court Records. IJR, 8.",
"links": null
},
"BIBREF30": {
"ref_id": "b30",
"title": "Privacy v. Transparency: How Remote Access to Court Records Forces Us to Re-examine Our Fundamental Values",
"authors": [
{
"first": "Nicolas",
"middle": [],
"last": "Vermeys",
"suffix": ""
}
],
"year": 2016,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Nicolas Vermeys. 2016. Privacy v. Transparency: How Remote Access to Court Records Forces Us to Re-examine Our Fundamental Values\",.",
"links": null
},
"BIBREF32": {
"ref_id": "b32",
"title": "Fine-Tuning Language Models from Human Preferences. ArXiv",
"authors": [
{
"first": "M",
"middle": [],
"last": "Daniel",
"suffix": ""
},
{
"first": "Nisan",
"middle": [],
"last": "Ziegler",
"suffix": ""
},
{
"first": "Jeffrey",
"middle": [],
"last": "Stiennon",
"suffix": ""
},
{
"first": "Tom",
"middle": [
"B"
],
"last": "Wu",
"suffix": ""
},
{
"first": "Alec",
"middle": [],
"last": "Brown",
"suffix": ""
},
{
"first": "Dario",
"middle": [],
"last": "Radford",
"suffix": ""
},
{
"first": "Paul",
"middle": [],
"last": "Amodei",
"suffix": ""
},
{
"first": "Geoffrey",
"middle": [],
"last": "Christiano",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Irving",
"suffix": ""
}
],
"year": 1909,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Daniel M. Ziegler, Nisan Stiennon, Jeffrey Wu, Tom B. Brown, Alec Radford, Dario Amodei, Paul Christiano, and Geoffrey Irving. 2019. Fine- Tuning Language Models from Human Prefer- ences. ArXiv, 1909.08593.",
"links": null
},
"BIBREF36": {
"ref_id": "b36",
"title": "Sierra Club of Canada v Canada (Minister of Finance)",
"authors": [],
"year": 2002,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Sierra Club of Canada v Canada (Minister of Fi- nance). 2002. SCC 41.",
"links": null
},
"BIBREF37": {
"ref_id": "b37",
"title": "Vancouver Sun (Re)",
"authors": [],
"year": 2004,
"venue": "SCC",
"volume": "43",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Vancouver Sun (Re). 2004. SCC 43.",
"links": null
}
},
"ref_entries": {
"TABREF1": {
"content": "<table><tr><td>PROBATION DE 2 ANS SURV. PROBATION DPAC:8.5MS/EMPR:6.5M TC 75 HS DEL 12 MS/SUIVI PROB 1 1/2 AN</td></tr><tr><td>Figure 1: Example of a complex Sentence containing a edge case about the duration of the different convic-tions. For this specific example, our model failed to generate a meaningful piece of text.</td></tr></table>",
"type_str": "table",
"html": null,
"text": "Error rates of the Extraction (EE) and Generation (GE) errors for each district.",
"num": null
}
}
}
} |