File size: 122,963 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 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 | {
"paper_id": "2021",
"header": {
"generated_with": "S2ORC 1.0.0",
"date_generated": "2023-01-19T16:43:43.848869Z"
},
"title": "Is one head enough? Mention heads in coreference annotations compared with UD-style heads",
"authors": [
{
"first": "Anna",
"middle": [],
"last": "Nedoluzhko",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "Charles University",
"location": {}
},
"email": "nedoluzhko@ufal.mff.cuni.cz"
},
{
"first": "Michal",
"middle": [],
"last": "Nov\u00e1k",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "Charles University",
"location": {}
},
"email": "mnovak@ufal.mff.cuni.cz"
},
{
"first": "Martin",
"middle": [],
"last": "Popel",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "Charles University",
"location": {}
},
"email": "popel@ufal.mff.cuni.cz"
},
{
"first": "Zden\u011bk",
"middle": [],
"last": "\u017dabokrtsk\u00fd",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "Charles University",
"location": {}
},
"email": "zabokrtsky@ufal.mff.cuni.cz"
},
{
"first": "Daniel",
"middle": [],
"last": "Zeman",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "Charles University",
"location": {}
},
"email": "zeman@ufal.mff.cuni.cz"
}
],
"year": "",
"venue": null,
"identifiers": {},
"abstract": "We present an empirical study that compares mention heads as annotated manually in four coreference datasets (for Dutch, English, Polish, and Russian) on one hand, with heads induced from dependency trees parsed automatically, on the other hand. For parsing, we used UDPipe 2.6, a modern parser trained using the Universal Dependencies collection. We show that majority of mismatches (64%-94%) can be attributed to several classes of systematic differences in how the notion of head is treated in the respective data resources, while mismatches caused by parsing errors are relatively rare (4%-15%). Our conclusion is that consistency would be gained in (and across) coreference resources after migration to UD-style mention heads, without losing substantial information. This can be achieved with sufficient accuracy using modern dependency parsers even for coreference corpora that lack manual head annotation.",
"pdf_parse": {
"paper_id": "2021",
"_pdf_hash": "",
"abstract": [
{
"text": "We present an empirical study that compares mention heads as annotated manually in four coreference datasets (for Dutch, English, Polish, and Russian) on one hand, with heads induced from dependency trees parsed automatically, on the other hand. For parsing, we used UDPipe 2.6, a modern parser trained using the Universal Dependencies collection. We show that majority of mismatches (64%-94%) can be attributed to several classes of systematic differences in how the notion of head is treated in the respective data resources, while mismatches caused by parsing errors are relatively rare (4%-15%). Our conclusion is that consistency would be gained in (and across) coreference resources after migration to UD-style mention heads, without losing substantial information. This can be achieved with sufficient accuracy using modern dependency parsers even for coreference corpora that lack manual head annotation.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Abstract",
"sec_num": null
}
],
"body_text": [
{
"text": "Coreference is a relation between expressions in a text which refer to the same real-word entity or event; the referring expressions are called mentions. In most datasets annotated with coreference relations (see for a survey), a mention is represented simply by specifying the corresponding sequence of tokens (called a mention span), typically contiguous, mostly belonging to a single sentence.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "Naturally, a mention span can be analyzed syntactically. There is a vague consensus that some tokens, often delimited syntactically, carry more important information from the coreference resolution perspective than other tokens. The most crucial part is called a minimum span by some (as opposed to maximum span denoting the whole span; see e.g. Uryupina et al. (2020) , Hirschman and Chinchor (1998) ), or simply a head by others (Ogrodniczuk et al., 2013) , which is the term we adhere to. Identifying a mention's head is motivated not only linguistically, but also technically: with a long-span mention, there is a higher risk of annotation noise and requiring the exact match when evaluating span boundary prediction could be misleading. See e.g. Uryupina et al. (2020) , Elsner and Charniak (2010) , Peng et al. (2015) , or Wiseman et al. (2016) for more arguments on the importance of head for the task of coreference resolution.",
"cite_spans": [
{
"start": 346,
"end": 368,
"text": "Uryupina et al. (2020)",
"ref_id": "BIBREF25"
},
{
"start": 371,
"end": 400,
"text": "Hirschman and Chinchor (1998)",
"ref_id": "BIBREF7"
},
{
"start": 431,
"end": 457,
"text": "(Ogrodniczuk et al., 2013)",
"ref_id": "BIBREF14"
},
{
"start": 751,
"end": 773,
"text": "Uryupina et al. (2020)",
"ref_id": "BIBREF25"
},
{
"start": 776,
"end": 802,
"text": "Elsner and Charniak (2010)",
"ref_id": "BIBREF4"
},
{
"start": 805,
"end": 823,
"text": "Peng et al. (2015)",
"ref_id": "BIBREF18"
},
{
"start": 829,
"end": 850,
"text": "Wiseman et al. (2016)",
"ref_id": "BIBREF26"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "The notion of mention head in coreference annotations largely resembles the notion of head in dependency treebanks; however, with a few exceptions such as the Prague Dependency Treebank ) (PDT for short), the coreference and dependency-treebanking annotation efforts remain isolated to a surprising degree.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "In this paper we present a novel empirical study that compares manually annotated mention heads within coreference annotation projects with syntactic heads identified automatically by a modern parser trained on dependency treebanks from the Universal Dependencies (UD) collection (de Marneffe et al., 2021) . Our long-term motivation is based on the expectation that making the mention head notion convergent with heads induced from dependency structures following the UD guidelines could result in (a) improved annotation consistency in existing coreference datasets, and (b) more efficient and faster development of new coreference datasets (e.g. because of possible reuse of UD-related software tools), especially when it comes to extensions to multiple languages. However, in a shorter-term perspective, we should first try to explain the nature of differences between mention heads as annotated in existing coreference datasets on the one hand and UD-compliant heads of mentions on the other.",
"cite_spans": [
{
"start": 280,
"end": 306,
"text": "(de Marneffe et al., 2021)",
"ref_id": "BIBREF2"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "We make use of the CorefUD 0.2 collection, which contains 17 coreference datasets for 11 languages converted to a common annotation scheme . There is some notion of head used explicitly or implicitly in 13 out of the 17 datasets. However, we limit ourselves only to datasets in which mention heads are marked explicitly, and, at the same time, whose coreference annotations were created without using full-fledged hand-annotated syntactic structures (dependency or constituency). Thus, for example, the Prague Dependency Treebank dataset is excluded, since coreference and dependency annotations are tightly connected in it by design. The selection criterion leads to four resources: ARRAU (Uryupina et al., 2020) for English, COREA for Dutch (Hendrickx et al., 2008) , Polish Coreference Corpus (Ogrodniczuk et al., 2015; Ogrodniczuk et al., 2013) , and Russian Coreference Corpus (Toldova et al., 2014) . Datasets in CorefUD 0.2 have been parsed using the UDPipe 2 tool (Straka, 2018) with very recent parser models.",
"cite_spans": [
{
"start": 690,
"end": 713,
"text": "(Uryupina et al., 2020)",
"ref_id": "BIBREF25"
},
{
"start": 743,
"end": 767,
"text": "(Hendrickx et al., 2008)",
"ref_id": "BIBREF6"
},
{
"start": 796,
"end": 822,
"text": "(Ogrodniczuk et al., 2015;",
"ref_id": "BIBREF15"
},
{
"start": 823,
"end": 848,
"text": "Ogrodniczuk et al., 2013)",
"ref_id": "BIBREF14"
},
{
"start": 882,
"end": 904,
"text": "(Toldova et al., 2014)",
"ref_id": "BIBREF24"
},
{
"start": 972,
"end": 986,
"text": "(Straka, 2018)",
"ref_id": "BIBREF23"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "The rest of the paper is structured as follows. Section 2 summarizes different approaches to the notion of head, both from the syntactic and coreference perspectives. Section 3 gives basic information about the four coreference data resources included in our study. Section 4 describes our annotation of mentions selected from the four resources; we focused on mentions in which the mention head marked in the original coreference resource does not match the root of the mention in terms of automatically parsed UD tree. Section 5 analyzes and exemplifies types of such mismatches. Finally, Section 6 concludes.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "One can easily foresee that -in spite of the recent progress in parsing technology -there will be nonnegligible amount of head mismatches which are due to parsing errors; similarly, a non-zero amount of errors in manual annotation of mention heads can be expected too. However, we are interested in more principled sources of variability of the notion of head.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Head in dependency annotation schemes",
"sec_num": "2.1"
},
{
"text": "It was recognized by dependency-oriented scholars long time ago that multiple types of dependencies may be distinguished (especially syntactic and semantic ones), and that syntactic dependencies should not be confused with other types of relations, see e.g. a discussion on \"double dependency\" and \"mutual dependency\" in Mel'\u010duk and others (1988) . However, the trend in the current dependency treebanking in the last decade is inclined rather to maximize simplicity and robustness, with Universal Dependencies (de Marneffe et al., 2021) being the most prominent representative, rather than to design multilayered annotation schemes with strictly separated hypotactic and paratactic \"brackets\" (with the latter ones possibly interpreted as additional \"dimensions\" of dependency trees (Sgall, 1998) ) on each layer. This trend has a clear rationale especially if quick portability to multiple languages is one of the modern priorities, however, on the other hand, such formally simple structures are prone to various confusions concerning the notion of head.",
"cite_spans": [
{
"start": 340,
"end": 346,
"text": "(1988)",
"ref_id": null
},
{
"start": 784,
"end": 797,
"text": "(Sgall, 1998)",
"ref_id": "BIBREF22"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Head in dependency annotation schemes",
"sec_num": "2.1"
},
{
"text": "The fact that in some cases there is no unique obvious way for choosing a head of an expression, has been noticed many times as it has inevitable practical consequences in dependency-oriented projects. Above all, annotators' intuition concerning the dependency structure of sentences is insufficient for reaching reasonable annotation consistency, and thus artificial annotation rules must be introduced by convention. This can be illustrated by extensive annotation guidelines developed basically in every mature dependency treebanking project. We believe that most of the observed variability in the notion of head can be attributed to the following sources, as discussed in more detail the subsections below:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Head in dependency annotation schemes",
"sec_num": "2.1"
},
{
"text": "\u2022 opposite direction of syntactic and semantic dependencies (and other non-parallelisms),",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Head in dependency annotation schemes",
"sec_num": "2.1"
},
{
"text": "\u2022 representing functional words as nodes of their own,",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Head in dependency annotation schemes",
"sec_num": "2.1"
},
{
"text": "\u2022 representing paratactic relations within dependency trees,",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Head in dependency annotation schemes",
"sec_num": "2.1"
},
{
"text": "\u2022 no obvious head-dependent asymmetry in a syntactic constituent.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Head in dependency annotation schemes",
"sec_num": "2.1"
},
{
"text": "Several types of constructions are recognized in literature in which the direction of a syntactic dependency relation manifested by overt surface morphosyntactic means (such as agreement) is opposite to what is considered as semantic dependency; the syntactic and semantic heads are swapped, in other words.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Opposite direction of syntactic and semantic dependencies",
"sec_num": "2.1.1"
},
{
"text": "When designing treebank annotation guidelines, the authors either have to indicate whether syntactic or semantic dependencies are the preferred ones, or, alternatively, provide technical means for capturing both. The latter option can be illustrated by PDT, in which there are two separate dependency trees, one of them capturing surface syntax and the other one capturing deep syntax and semantics (to some extent). Similarly, the Enhanced representation in Universal Dependencies (Nivre et al., 2020, Section 3.4) adds extra edges to make explicit some semantically relevant relations that are otherwise implicit in basic dependencies.",
"cite_spans": [
{
"start": 482,
"end": 515,
"text": "(Nivre et al., 2020, Section 3.4)",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Opposite direction of syntactic and semantic dependencies",
"sec_num": "2.1.1"
},
{
"text": "If functional words have nodes of their own in a dependency representation, it can lead to problems related to head choice. A functional word is usually clearly associated with an autosemantic (meaningful) word, however, it is not clear which of them should be the head (more precisely, either choice can be justified with reasonable arguments, and one simply has to choose). Examples of such pairs are a preposition and a noun in a prepositional group, an auxiliary verb and an autosemantic verb in a complex verb form, or a determiner and a noun. For instance, if a prepositional group is considered, PDT surface syntax guidelines make the preposition the governor and put the noun below, while the two are connected the other way round in UD. If an auxiliary verb in a complex verb form bears congruent categories, then it becomes the governor in the PDT, while the autosemantic component of the complex verb form is the governor in most cases. Both PDT and UD annotation styles attach determiners below nouns being determined, but determiners are treated as governors of noun phrases in the Danish Dependency Treebank (Kromann et al., 2003) .",
"cite_spans": [
{
"start": 1122,
"end": 1144,
"text": "(Kromann et al., 2003)",
"ref_id": "BIBREF8"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Functional words",
"sec_num": "2.1.2"
},
{
"text": "A more complex example is that of expletives: in some cases insertion of expletive expressions (such as pronouns) is needed or preferred in a language, for instance if valency of a matrix clause verb requires a morphological case to be manifested with its argument, but the argument is a subordinating clause. Then, again, it is not clear whether the expletive pronoun or the subordinating clause head should be chosen as the head of it all.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Functional words",
"sec_num": "2.1.2"
},
{
"text": "In the case of parataxis, two syntactically connected expressions are in an equal relation with each other, instead of being subordinated one to the other. In other words, there is no head-dependent asymmetry. Typical examples are coordination and apposition constructions. Especially coordination has always been a nightmare for dependency grammarians, as it is very frequent and interferes in various ways with dependency relations. However, as long as we preserve the design decision that all we have for syntactic representation is nodes and edges, we have to encode paratactic constructions in this way too. There is a surprising number of different possible encodings for doing so, and a smaller, but even more surprising number of encodings that has been really used in existing treebanks, see Popel et al. (2013) for a survey. However, in most cases it boils down to either using coordination conjunction as the head node, or using one of the conjuncts as the head, selected in some canonical way.",
"cite_spans": [
{
"start": 801,
"end": 820,
"text": "Popel et al. (2013)",
"ref_id": "BIBREF19"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Paratactic structures",
"sec_num": "2.1.3"
},
{
"text": "Besides paratactic structures, there are also other types of expressions in which we perceive some internal structure and for which we do not possess intuition about what should be the head, but which are not paratactic either. A frequent example is a personal name consisting of a given name and a family name. UD has a dedicated relation type, flat, which is used in such exocentric constituents; the first word serves as the technical head, but there is no claim that it is a syntactically (or semantically) motivated head.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "No overt head-dependent asymmetry",
"sec_num": "2.1.4"
},
{
"text": "To summarize, head choice is far from obvious in various cases, which has both deeply linguistic and purely technical reasons; such situations can only be resolved unambiguously by adhering to artificial annotation rules.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "No overt head-dependent asymmetry",
"sec_num": "2.1.4"
},
{
"text": "For a better orientation, we suggest to classify language data resources containing coreference annotation tentatively as follows:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Head in coreference annotation schemes",
"sec_num": "2.2"
},
{
"text": "\u2022 head-agnostic approaches,",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Head in coreference annotation schemes",
"sec_num": "2.2"
},
{
"text": "\u2022 head-aware approaches,",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Head in coreference annotation schemes",
"sec_num": "2.2"
},
{
"text": "\u2022 head-centric approaches.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Head in coreference annotation schemes",
"sec_num": "2.2"
},
{
"text": "In head-agnostic approaches, a mention is considered the only meaningful unit that is needed for annotating coreference relations and no attempt to find its internal structure is made (at least not to our knowledge).",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Head-agnostic approaches.",
"sec_num": "2.2.1"
},
{
"text": "Examples of head-agnostic approaches are Potsdam Commentary Corpus (Bourgonje and Stede, 2020), the English-German parallel coreference corpus ParCorFull (Lapshinova-Koltunski et al., 2018) , and Lithuanian Coreference Corpus (\u017ditkus and Butkien\u0117, 2018).",
"cite_spans": [
{
"start": 154,
"end": 189,
"text": "(Lapshinova-Koltunski et al., 2018)",
"ref_id": "BIBREF9"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Head-agnostic approaches.",
"sec_num": "2.2.1"
},
{
"text": "In head-aware approaches, a mention delimited as a sequence of tokens is still the main entity, however, its internal structure is analyzed syntactically 1 (completely or partially) and/or its head is marked explicitly.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Head-aware approaches.",
"sec_num": "2.2.2"
},
{
"text": "Examples of head-aware approaches are Spanish and Catalan data contained in AnCora (Recasens and Mart\u00ed, 2010) and English data contained in ARRAU (Uryupina et al., 2020).",
"cite_spans": [
{
"start": 83,
"end": 109,
"text": "(Recasens and Mart\u00ed, 2010)",
"ref_id": "BIBREF21"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Head-aware approaches.",
"sec_num": "2.2.2"
},
{
"text": "In head-centric approaches, it is the head of a mention that is considered to be the argument of a coreference relation, while the exact span in terms of a token sequence is less important (or even left underspecified). Coreference datasets from the PDT family, in which coreference relations connect tectogrammatical (deep-syntactic) nodes and mention span is defined only implicitly, are examples of this approach.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Head-centric approaches.",
"sec_num": "2.2.3"
},
{
"text": "Examples of head-centric approaches are the Prague Dependency Treebank and the Prague Czech-English Dependency Treebank (Nedoluzhko et al., 2016) .",
"cite_spans": [
{
"start": 120,
"end": 145,
"text": "(Nedoluzhko et al., 2016)",
"ref_id": "BIBREF11"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Head-centric approaches.",
"sec_num": "2.2.3"
},
{
"text": "Our analysis is based on four datasets from CorefUD 0.2 2 whose original source corpora contain manual annotation of mentions: ARRAU, Polish Coreference Corpus, COREA, and Russian Coreference Corpus.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Coreference datasets with hand-annotated mention heads",
"sec_num": "3"
},
{
"text": "The ARRAU Corpus of Anaphoric Information (Uryupina et al., 2020) (further abbreviated as English-ARRAU) is a multi-genre corpus of English which provides large-scale annotations of a wide range of anaphoric phenomena. In English-ARRAU, the special attribute MIN (or minimal span) is manually annotated, similarly as it was once decided for MUC-7 (Hirschman and Chinchor, 1998) . This attribute corresponds to the head noun for non-proper nominal mentions, or to the entire proper name (for example, first name and surname) in case of multi-word named entities. It is not explicitly stated in the guidelines, if syntactic or semantic heads are preferred. According to the MUC-7 coreference task definition 3 , it maybe deduced that syntactic heads are preferred. However, this has not been stated explicitly for MUC-7 neither.",
"cite_spans": [
{
"start": 347,
"end": 377,
"text": "(Hirschman and Chinchor, 1998)",
"ref_id": "BIBREF7"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "ARRAU",
"sec_num": "3.1"
},
{
"text": "The Polish Coreference Corpus (Ogrodniczuk et al., 2013; Ogrodniczuk et al., 2015 ) (further abbreviated as Polish-PCC) is a corpus of Polish nominal coreference built upon the National Corpus of Polish. In Polish-PCC, semantic heads, i.e. the most important words from the point of view of the mention's sense, are annotated. The semantic head of a typical nominal group corresponds to the syntactic head but there are some exceptions. For example, in numeral groups like duzo pieniedzy 'a lot of money', or trzech z was 'three of you', the numeral is the syntactic head, and the noun is the semantic head and is annotated as such in Polish-PCC. The reason for such decision is the claim that coreference is a phenomenon on the level of semantics and discourse more than on the syntactic level. Thus, understanding the semantically central elements should help establish discourse links. Although not explicitly found in the guidelines, the head is understood semantically (an item with larger semantic weight is annotated as head) also in other types of constructions (od 1999 roku 'from the year 1999' with the numeral as a head, pan Ziolkowski 'Mr. Ziolkowski' with the surname as a head, etc.).",
"cite_spans": [
{
"start": 30,
"end": 56,
"text": "(Ogrodniczuk et al., 2013;",
"ref_id": "BIBREF14"
},
{
"start": 57,
"end": 81,
"text": "Ogrodniczuk et al., 2015",
"ref_id": "BIBREF15"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Polish Coreference Corpus",
"sec_num": "3.2"
},
{
"text": "The COREA coreference corpus (Hendrickx et al., 2008 ) (further abbreviated as Dutch-COREA) is a collection of written and transcribed oral texts in Dutch annotated for creating a coreference resolution system. Mentions are strings of text with specially distinguished heads which are defined as minimum strings representing semantic heads of the constituents. Nevertheless, rather than annotated from scratch, the semantic heads were acquired by manual post-editing of the heads obtained from syntactic representation of the underlying texts. For multi-word named entities, the head includes all words of the corresponding entity.",
"cite_spans": [
{
"start": 29,
"end": 52,
"text": "(Hendrickx et al., 2008",
"ref_id": "BIBREF6"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "COREA",
"sec_num": "3.3"
},
{
"text": "Russian Coreference Corpus (Toldova et al., 2014 ) (further abbreviated as Russian-RuCor) is annotated with anaphoric and coreferential relations between noun groups. Mentions are annotated as linear spans, with additionally distinguished heads. Similarly as for English-ARRAU and Dutch-COREA, heads are defined as one-word syntactic heads for common nouns and as sequences of words for multi-word proper nouns. For 'common noun + proper noun' constructions like the Pushkin street, the guidelines require the whole multi-word sequences to be annotated as heads, but in the annotated data, only one word is chosen as head (mostly the proper noun).",
"cite_spans": [
{
"start": 27,
"end": 48,
"text": "(Toldova et al., 2014",
"ref_id": "BIBREF24"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Russian Coreference Corpus",
"sec_num": "3.4"
},
{
"text": "The comparison of the guidelines for head annotation in the resources under analysis shows that there are differences in the following aspects:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Russian Coreference Corpus",
"sec_num": "3.4"
},
{
"text": "\u2022 Syntactic or semantic understanding of heads: Semantic heads are explicitly claimed to be annotated in Polish-PCC and partly in Dutch-COREA; in English-ARRAU and Russian-RuCor, there is no explicit claim about the syntactic nature of annotated heads but it may be deduced from the guidelines examples;",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Russian Coreference Corpus",
"sec_num": "3.4"
},
{
"text": "\u2022 Possibility to annotate multi-word entities as a head: Possible for multi-word named entities in English-ARRAU, Russian-RuCor and Dutch-COREA and not applied in Polish-PCC;",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Russian Coreference Corpus",
"sec_num": "3.4"
},
{
"text": "\u2022 Choice of the head in 'common noun + proper noun' constructions: the proper name in English-ARRAU, Dutch-COREA and Polish-PCC and both entities in Russian-RuCor; Table 1 : Statistics on mentions in the whole dataset. all is the total number of mentions in the train section of a given dataset. The other columns show percentage breakdown into mention types described in the first paragraph of Section 4. The types are detected automatically in a given order, so e.g. a non-catena mention with no annotated head is assigned the non-catena type (not missing head). The last column shows a percentage of multi-word catena mentions with a mismatch in annotated and syntactic head; a sample of 100 mentions of this type was annotated as shown in Table 2 .",
"cite_spans": [],
"ref_spans": [
{
"start": 164,
"end": 171,
"text": "Table 1",
"ref_id": null
},
{
"start": 743,
"end": 750,
"text": "Table 2",
"ref_id": null
}
],
"eq_spans": [],
"section": "Russian Coreference Corpus",
"sec_num": "3.4"
},
{
"text": "\u2022 Different technical conventions for apposition and coordination structures, special construction with dollar, percent, etc.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Russian Coreference Corpus",
"sec_num": "3.4"
},
{
"text": "Dependency trees in the four datasets under discussion have been obtained for CorefUD 0.2 using UDPipe 2 and models trained on UD 2.6, namely on English-GUM (Zeldes, 2017) , Polish-LFG (Patejuk and Przepi\u00f3rkowski, 2018) , Dutch-LassySmall (Bouma and van Noord, 2017) , and Russian-SynTagRus (Droganova et al., 2018) .",
"cite_spans": [
{
"start": 157,
"end": 171,
"text": "(Zeldes, 2017)",
"ref_id": "BIBREF27"
},
{
"start": 185,
"end": 219,
"text": "(Patejuk and Przepi\u00f3rkowski, 2018)",
"ref_id": "BIBREF17"
},
{
"start": 239,
"end": 266,
"text": "(Bouma and van Noord, 2017)",
"ref_id": "BIBREF0"
},
{
"start": 291,
"end": 315,
"text": "(Droganova et al., 2018)",
"ref_id": "BIBREF3"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Russian Coreference Corpus",
"sec_num": "3.4"
},
{
"text": "In our study, we focus on mentions, in which the head coming from the original annotation differs from the head of the mention with respect to the tree produced by automatic dependency parsing. 4 Consequently, all one-word mentions are excluded. In addition, we take into consideration only such mentions whose inner dependency structure forms a catena, i.e. a connected subgraph of a dependency tree (Osborne et al., 2012) . 5 Moreover, we focus only on mentions where at least one head was annotated. 6 We randomly sampled 100 such mentions from a train section of each of the four CorefUD datasets under analysis. The examples were examined and annotated by the authors of this work. As none of us is a speaker of Dutch, we utilized public machine-translation services in order to understand the example sentences.",
"cite_spans": [
{
"start": 194,
"end": 195,
"text": "4",
"ref_id": null
},
{
"start": 401,
"end": 423,
"text": "(Osborne et al., 2012)",
"ref_id": "BIBREF16"
},
{
"start": 426,
"end": 427,
"text": "5",
"ref_id": null
},
{
"start": 503,
"end": 504,
"text": "6",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Annotation of head mismatches",
"sec_num": "4"
},
{
"text": "During the annotation process, we settled upon the following categories of head mismatches:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Annotation of head mismatches",
"sec_num": "4"
},
{
"text": "\u2022 WRONG -we consider the mismatch to be an error.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Annotation of head mismatches",
"sec_num": "4"
},
{
"text": "-WRONGTREE -the automatically parsed UD tree is wrong -WRONGSPAN -wrong syntactic head caused by a wrong mention span, usually due to extra tokens. -WRONGHEAD -the manual annotation of head is wrong, i.e. it does not follow the original project annotation guidelines (or at least we were not able to find any guideline which would support such head annotation).",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Annotation of head mismatches",
"sec_num": "4"
},
{
"text": "CorefUD dataset COORD FLAT NUM OTHER TREE SPAN HEAD Dutch-COREA 25 31 11 7 7 7 16 English-ARRAU 1 44 14 13 4 0 25 Polish-PCC 11 21 23 9 15 1 13 Russian-RuCor 0 85 7 2 5 0 1 Table 2 : Result of our annotation of differences in annotated and syntactic heads in a sample of 100 mentions in each dataset. Disclaimer: Individual cases of WRONGHEAD may turn out to be cases of OK or vice versa. A deeper analysis of such cases is a subject of future studies.",
"cite_spans": [],
"ref_spans": [
{
"start": 52,
"end": 213,
"text": "Dutch-COREA 25 31 11 7 7 7 16 English-ARRAU 1 44 14 13 4 0 25 Polish-PCC 11 21 23 9 15 1 13 Russian-RuCor 0 85 7 2 5 0 1 Table 2",
"ref_id": null
}
],
"eq_spans": [],
"section": "OK WRONG",
"sec_num": null
},
{
"text": "\u2022 OK -the mismatch in head annotations is correct, as the respective guidelines do not agree on the head for a given phenomenon -OK-COORD -the first conjunct of a coordination is always marked as a head in UD. The original annotation marks the coordination conjunction or another conjunct as a head, instead. This is an example of the parataxis (see Section 2.1.3). -OK-FLAT -UD chooses the first token as head in flat structures (such as names, marked with deprel flat), appositions (marked with deprel appos) and lists, while the original dataset annotators decided to analyze it as a non-flat structure. This is an example of non-overt headdependent asymmetry as we describe it in Section 2.1.4, and parataxis for apposition (see Section 2.1.3). 7 -OK-NUM -the mismatch is caused by an opposite direction of syntactic and semantic dependencies (see Section 2.1.1). This most often includes numerals and containers (e.g. a group of people). -OK-OTHER -another subtype of OK. Table 2 summarizes the head mismatches annotation in the selected datasets. As we can see, there is a relatively low number of mismatches caused by wrong parsing. With a slightly larger number of such cases in Polish-PCC, there are just up to 7% of wrongly parsed annotated mentions in English-ARRAU, Dutch-COREA and Russian-RuCor. One of the reasons is that we included only multi-word catena structures into the analysis. 8 The remaining cases of wrong parsing are specific syntactic or derivation constructions, e.g. the deadjectival noun te\u015bciowa /mother-in-law/ in Example 1 9 from Polish which is falsely recognized as an adjective in UDPipe and thus gets a dependent position in the parsed tree. The surprisingly low overall number of parse errors can be justified by comparative simplicity of parsing of noun phrases (the majority of mentions are noun phrases).",
"cite_spans": [
{
"start": 1401,
"end": 1402,
"text": "8",
"ref_id": null
}
],
"ref_spans": [
{
"start": 977,
"end": 984,
"text": "Table 2",
"ref_id": null
}
],
"eq_spans": [],
"section": "OK WRONG",
"sec_num": null
},
{
"text": "( Another apparent general observation is a disproportion of incorrectly parsed or annotated sentences (WRONG labels) between Russian-RuCor (6%) and the other datasets (28-30%). This is likely a consequence of annotation mismatches in proper nouns that prevail in the selected sample (see Section 5.2).",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Analysis and discussion",
"sec_num": "5"
},
{
"text": "Our analysis reveals a number of mismatches (i) between syntactic heads generated by the UD parser and manually annotated heads in the datasets, but also inconsistencies (ii) across the datasets and (iii) within the annotated datasets themselves. The most typical categories of mismatches (OK-COORD, OK-FLAT and OK-NUM) and annotation inconsistencies are addressed in the following subsections.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Analysis and discussion",
"sec_num": "5"
},
{
"text": "The prevailing reason for mismatches of the OK-COORD type is that the coordination conjunction is annotated as the head of a coordination mention. In total it accounts for 73%.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Heads in coordinations",
"sec_num": "5.1"
},
{
"text": "In cases where a non-first conjunct is annotated as a mention head, the conjunct often comprises information that is shared among all conjuncts, e.g. in Example 2 from Dutch-COREA. ",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Heads in coordinations",
"sec_num": "5.1"
},
{
"text": "Constructions with proper names form a great deal of the OK-FLAT category. There are different annotation conventions for annotating heads in such constructions in UD (de Marneffe et al., 2021) and across the annotated datasets (see descriptions of the datasets in Section 3). Generally, whereas in phrases like President Carter the annotator more often chooses the proper noun as head (because it is referentially concrete), it is the first word (i.e. the general name President in our example) according to the UD convention (see Example 3 from English-ARRAU).",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Heads in expressions with proper names",
"sec_num": "5.2"
},
{
"text": "(3) Mr. Hastings was appointed to the federal bench by President Carter Interestingly, there are 85% such cases in the Russian sample, although the guidelines rather advise to label both expressions as a multi-word head. See the expression \u0436\u0443\u0440\u043d\u0430\u043b\u0430 Time /Time magazine/ in Example 4, where the annotated mention head is the proper name and the UD head is the common noun \u0436\u0443\u0440\u043d\u0430\u043b\u0430 /magazine/. This type of mismatches is also frequent in other datasets, see e.g. moja babcia Zofia /my grandma Zofia/ in Example 5 (Polish-PCC) or vitamine C in Example 6 (Dutch-COREA).",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Heads in expressions with proper names",
"sec_num": "5.2"
},
{
"text": "There is also a non-negligible number of inconsistencies in annotation of multi-word named entities within the datasets. Although, the guidelines require to mark the entire multi-word units as heads in all datasets except Polish-PCC, in some cases, only one more semantically significant word is annotated. See the annotation of only surname in the multi-word name Hillary Clinton in Example 7 from Russian-RuCor. ",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Heads in expressions with proper names",
"sec_num": "5.2"
},
{
"text": "Head mismatches in constructions with numerals grouped under the OK-NUM category may be further divided into the following subgroups.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Heads in expressions with numerals and quantifiers",
"sec_num": "5.3"
},
{
"text": "Cardinal numerals. Numeral mentions where a noun-like word is modified by a number (e.g. five cars) are typical cases of head mismatches. In most of them, the modified word is in fact a currency's name or symbol (e.g. $25 million, vijfhonderd zestig miljoen gulden /five hundred and sixty million guilders/, 90 \u043c\u043b\u0440\u0434 \u0440\u0443\u0431\u043b\u0435\u0439 /90 billion rubles/ ).",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Heads in expressions with numerals and quantifiers",
"sec_num": "5.3"
},
{
"text": "Heads are annotated inconsistently across datasets: numerals prevailingly serve as heads in English-ARRAU (Examples 8-9) and Polish-PCC (Example 10), while it is the modified words in Russian-RuCor (Example 13). Nevertheless, annotation of mention heads seems to be inconsistent also within some of the datasets. Let us look into mentions with a $ symbol as their syntactic head ($ mention) in English-ARRAU. Out of 727 such mentions scattered over 179 original documents, only in 43% of them the annotated head (minimal span) matches the syntactic head. Interestingly though, if an original document contains a matching $ mention, on average more than 92% of all $ mentions in the document are matching, too. The observed inconsistency thus occurs rather across than within original documents, suggesting that it is an artifact of the annotation workload having been distributed among multiple annotators on the document level.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Heads in expressions with numerals and quantifiers",
"sec_num": "5.3"
},
{
"text": "The mismatches between syntactic and annotated heads partly result also from inconsistencies in parses. However, we do not categorize them as parsing errors (WRONGTREE) since UDPipe models almost perfectly mimic the inconsistency that can be seen already across the manually annotated UD subcorpora they were trained on. While syntactic annotation of single-token numerals (e.g. five cars) seem to be identical across the languages, it differs considerably for multi-token numerals with large number names such as thousands, millions etc. (cf. Examples 8-13). Moreover, in Russian-SynTagRus the tree of multi-token numerals is shaped differently based on whether the word representing the large number name is in singular (Example 12) or plural (Example 13). The reasons for such mismatches may be twofold. First, these constructions are not clearly distinct from the structures like 'most of people, 'half of people' where most and half are syntactic heads. Another reason may be higher salience of the determiners in the given contexts.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Heads in expressions with numerals and quantifiers",
"sec_num": "5.3"
},
{
"text": "\u204e\u204e64 NUM nummod\u204e\u204e \u204e\u204eproc X obj <-------------------synt.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Heads in expressions with numerals and quantifiers",
"sec_num": "5.3"
},
{
"text": "Dates. Mismatches in dates seem to appear only in Polish-PCC. Years and months (if present) are consistently annotated as mention heads, as illustrated in Examples 20 and 21, respectively. Therefore, it should not be too difficult to obtain such mention heads using a rule-based transformation based on syntax. wystawa NOUN appos czynna ADJ amod do ADP case \u204e\u204e9 ADJ nmod <------------------synt. head\u204e\u204e \u204e\u204ekwietnia NOUN nmod:poss <=== annot. head\u204e\u204e \u204e\u204e2007 ADJ amod\u204e\u204e",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Heads in expressions with numerals and quantifiers",
"sec_num": "5.3"
},
{
"text": "We have provided a novel comparison of syntactic dependency structure on one hand, and annotation of coreferential mentions on the other hand. In particular, we focus on the notion of mention heads in coreference datasets where such a notion exists and it is not designed to be identical to the syntactic head. Nevertheless, we can compare mention heads with syntactic heads thanks to the CorefUD collection, which contains coreference corpora with dependency structures predicted by the UDPipe parser. We collected mention instances where the syntactic head did not match the designated mention head, then we manually examined a subset of such instances and analyzed the likely reasons for the difference. If we summarize our observations, the UD heads and manually annotated mention heads coincide in majority of multi-token mentions in all four studied datasets already now, while most differences can be attributed to one of the following reasons:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Conclusion",
"sec_num": "6"
},
{
"text": "\u2022 heads of a mention are different because of an error made by the UD parser, or because of an error made by an annotator; the amount of parsing errors is surprisingly low, likely due to relative simplicity of parsing of noun phrases (and will hopefully further fade out with progress in parsing technology),",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Conclusion",
"sec_num": "6"
},
{
"text": "\u2022 heads are selected using rather technical than linguistic rules in expressions such as named entities or coordination structures (in which linguistic intuitions for heads are weak); rule-based transformations could be used for translating UD convention to a coreference dataset convention or vice versa,",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Conclusion",
"sec_num": "6"
},
{
"text": "\u2022 semantic rather than syntactic heads are chosen in coreference annotations, e.g. in expressions with numerals; however, with an exception of some types of expressions (e.g. 'containers'), again a few rule-based patterns on the UD tree of a mention could be used to automatically identify the semantic head,",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Conclusion",
"sec_num": "6"
},
{
"text": "\u2022 in some cases, mention head annotations in coreference datasets bear information that seems intuitively semantically salient (such as contrast) and undeducible from UD syntax; however, such cases are rare and typically not supported by coreference annotation guidelines.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Conclusion",
"sec_num": "6"
},
{
"text": "Let us conclude by answering the question from the title. It seems that both inter-project and intraproject consistency would be gained and almost nothing would be lost if we start adhering to the UD notion of heads in mentions in coreference projects, instead of annotating coreference-specific heads. In addition, quality of mention heads derived from automatic UD parses based on modern parsing technology is quite high, which would further reduce potential benefits of manual annotation of mention heads in future coreference-oriented projects.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Conclusion",
"sec_num": "6"
},
{
"text": "An analysis whether or not coreference mentions do correspond to subtrees of UD trees can be found in, without a special attention paid to heads, though.2 http://hdl.handle.net/11234/1-4598",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
},
{
"text": "https://www-nlpir.nist.gov/related_projects/muc/proceedings/co_task.html",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
},
{
"text": "Multiple words could be annotated as heads (or minimal span) in the original annotation. In such cases, we focus on mentions where the syntactic head is not among the set of annotated heads.5 Note that catena differs from a subtree, which is a catena that spans the head and all its descendants. Non-catena mentions have multiple nodes that can be considered syntactic heads of the mention (i.e. their dependency parent is not part of the mention).6 SeeTable 1for statistics on the total count of mentions and their breakdown into the abovementioned types excluded from the annotation (one-word, non-catena, missing-head, same-head).",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
},
{
"text": "Even though appositions are paractic constructions, we rather included them in the OK-FLAT category. The reason is that they are closely related to hypotactic constructions such as president Trump, which are in fact treated as appositions in some of the datasets (e.g. Dutch-COREA).8 Note that Polish-PCC has the lowest percentage of head mismatches according to the last column inTable 1. Thus, in Polish-PCC we could expect 1.5% \u2022 15% = 0.23% mentions with head mismatch caused by wrong parsing in the whole dataset, while in Dutch-COREA it is twice as much: 6.6% \u2022 7% = 0.46%. 9 Examples in this work are presented in both glosses and trees. The first line of the gloss shows the original sentence / excerpt / phrase, optionally followed by its word-to-word translation and smooth translation to English. Nodes in the dependency tree show the word form, part-of-speech tag and dependency relation to the node's parent. While in gloss the annotated mention is typeset in bold, \u204e\u204etoken\u204e\u204e is used to mark each token of the mention in the tree. The annotated mention head and syntactic head given by the parser are labelled only in the tree.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
}
],
"back_matter": [
{
"text": "This work was supported by the Grants GA19-14534S and 20-16819X (LUSyD) of the Czech Science Foundation; LM2018101 (LINDAT/CLARIAH-CZ) of the Ministry of Education, Youth, and Sports of the Czech Republic; and EC/H2020/825303 (Bergamot) of the European Commision.We thank the three anonymous reviewers for their very insightful and useful comments.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Acknowledgements",
"sec_num": null
}
],
"bib_entries": {
"BIBREF0": {
"ref_id": "b0",
"title": "Increasing return on annotation investment: The automatic construction of a Universal Dependency treebank for Dutch",
"authors": [
{
"first": "Gosse",
"middle": [],
"last": "Bouma",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Gertjan Van Noord",
"suffix": ""
}
],
"year": 2017,
"venue": "Proceedings of the NoDaLiDa 2017 Workshop on Universal Dependencies (UDW 2017)",
"volume": "",
"issue": "",
"pages": "19--26",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Gosse Bouma and Gertjan van Noord. 2017. Increasing return on annotation investment: The automatic con- struction of a Universal Dependency treebank for Dutch. In Proceedings of the NoDaLiDa 2017 Workshop on Universal Dependencies (UDW 2017), pages 19-26, Gothenburg, Sweden, May. Association for Computational Linguistics.",
"links": null
},
"BIBREF1": {
"ref_id": "b1",
"title": "The Potsdam commentary corpus 2.2: Extending annotations for shallow discourse parsing",
"authors": [
{
"first": "Peter",
"middle": [],
"last": "Bourgonje",
"suffix": ""
},
{
"first": "Manfred",
"middle": [],
"last": "Stede",
"suffix": ""
}
],
"year": 2020,
"venue": "Proceedings of the 12th Language Resources and Evaluation Conference",
"volume": "",
"issue": "",
"pages": "1061--1066",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Peter Bourgonje and Manfred Stede. 2020. The Potsdam commentary corpus 2.2: Extending annotations for shallow discourse parsing. In Proceedings of the 12th Language Resources and Evaluation Conference, pages 1061-1066, Marseille, France, May. European Language Resources Association.",
"links": null
},
"BIBREF2": {
"ref_id": "b2",
"title": "Universal Dependencies",
"authors": [
{
"first": "Marie-Catherine",
"middle": [],
"last": "De Marneffe",
"suffix": ""
},
{
"first": "Christopher",
"middle": [],
"last": "Manning",
"suffix": ""
},
{
"first": "Joakim",
"middle": [],
"last": "Nivre",
"suffix": ""
},
{
"first": "Daniel",
"middle": [],
"last": "Zeman",
"suffix": ""
}
],
"year": 2021,
"venue": "Computational Linguistics",
"volume": "47",
"issue": "2",
"pages": "255--308",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Marie-Catherine de Marneffe, Christopher Manning, Joakim Nivre, and Daniel Zeman. 2021. Universal Depen- dencies. Computational Linguistics, 47(2):255-308.",
"links": null
},
"BIBREF3": {
"ref_id": "b3",
"title": "Data Conversion and Consistency of Monolingual Corpora: Russian UD Treebanks",
"authors": [
{
"first": "Kira",
"middle": [],
"last": "Droganova",
"suffix": ""
},
{
"first": "Olga",
"middle": [],
"last": "Lyashevskaya",
"suffix": ""
},
{
"first": "Daniel",
"middle": [],
"last": "Zeman",
"suffix": ""
}
],
"year": 2018,
"venue": "Proceedings of the 17th International Workshop on Treebanks and Linguistic Theories (TLT 2018)",
"volume": "",
"issue": "",
"pages": "53--66",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Kira Droganova, Olga Lyashevskaya, and Daniel Zeman. 2018. Data Conversion and Consistency of Monolin- gual Corpora: Russian UD Treebanks. In Proceedings of the 17th International Workshop on Treebanks and Linguistic Theories (TLT 2018), pages 53-66, Link\u00f6ping, Sweden. Link\u00f6ping University Electronic Press.",
"links": null
},
"BIBREF4": {
"ref_id": "b4",
"title": "The same-head heuristic for coreference",
"authors": [
{
"first": "Micha",
"middle": [],
"last": "Elsner",
"suffix": ""
},
{
"first": "Eugene",
"middle": [],
"last": "Charniak",
"suffix": ""
}
],
"year": 2010,
"venue": "Proceedings of the ACL 2010 Conference Short Papers",
"volume": "",
"issue": "",
"pages": "33--37",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Micha Elsner and Eugene Charniak. 2010. The same-head heuristic for coreference. In Proceedings of the ACL 2010 Conference Short Papers, pages 33-37.",
"links": null
},
"BIBREF5": {
"ref_id": "b5",
"title": "Prague Dependency Treebank -Consolidated 1.0",
"authors": [
{
"first": "Jan",
"middle": [],
"last": "Haji\u010d",
"suffix": ""
},
{
"first": "Eduard",
"middle": [],
"last": "Bej\u010dek",
"suffix": ""
},
{
"first": "Jaroslava",
"middle": [],
"last": "Hlav\u00e1\u010dov\u00e1",
"suffix": ""
},
{
"first": "Marie",
"middle": [],
"last": "Mikulov\u00e1",
"suffix": ""
},
{
"first": "Milan",
"middle": [],
"last": "Straka",
"suffix": ""
}
],
"year": 2020,
"venue": "Proceedings of the 12th International Conference on Language Resources and Evaluation (LREC 2020)",
"volume": "",
"issue": "",
"pages": "5208--5218",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Jan Haji\u010d, Eduard Bej\u010dek, Jaroslava Hlav\u00e1\u010dov\u00e1, Marie Mikulov\u00e1, Milan Straka, Jan \u0160t\u011bp\u00e1nek, and Barbora \u0160t\u011bp\u00e1nkov\u00e1. 2020. Prague Dependency Treebank -Consolidated 1.0. In Proceedings of the 12th Interna- tional Conference on Language Resources and Evaluation (LREC 2020), pages 5208-5218, Marseille, France. European Language Resources Association.",
"links": null
},
"BIBREF6": {
"ref_id": "b6",
"title": "A coreference corpus and resolution system for Dutch",
"authors": [
{
"first": "Iris",
"middle": [],
"last": "Hendrickx",
"suffix": ""
},
{
"first": "Gosse",
"middle": [],
"last": "Bouma",
"suffix": ""
},
{
"first": "Frederik",
"middle": [],
"last": "Coppens",
"suffix": ""
},
{
"first": "Walter",
"middle": [],
"last": "Daelemans",
"suffix": ""
},
{
"first": "Veronique",
"middle": [],
"last": "Hoste",
"suffix": ""
},
{
"first": "Geert",
"middle": [],
"last": "Kloosterman",
"suffix": ""
},
{
"first": "Anne-Marie",
"middle": [],
"last": "Mineur",
"suffix": ""
},
{
"first": "Joeri",
"middle": [],
"last": "Van Der",
"suffix": ""
},
{
"first": "Jean-Luc",
"middle": [],
"last": "Vloet",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Verschelde",
"suffix": ""
}
],
"year": 2008,
"venue": "Proceedings of the Sixth International Conference on Language Resources and Evaluation (LREC'08)",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Iris Hendrickx, Gosse Bouma, Frederik Coppens, Walter Daelemans, Veronique Hoste, Geert Kloosterman, Anne- Marie Mineur, Joeri Van Der Vloet, and Jean-Luc Verschelde. 2008. A coreference corpus and resolution system for Dutch. In Proceedings of the Sixth International Conference on Language Resources and Evaluation (LREC'08), Marrakech, Morocco, May. European Language Resources Association (ELRA).",
"links": null
},
"BIBREF7": {
"ref_id": "b7",
"title": "Appendix F: MUC-7 coreference task definition (version 3.0)",
"authors": [
{
"first": "Lynette",
"middle": [],
"last": "Hirschman",
"suffix": ""
},
{
"first": "Nancy",
"middle": [],
"last": "Chinchor",
"suffix": ""
}
],
"year": 1998,
"venue": "Seventh Message Understanding Conference",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Lynette Hirschman and Nancy Chinchor. 1998. Appendix F: MUC-7 coreference task definition (version 3.0). In Seventh Message Understanding Conference (MUC-7): Proceedings of a Conference Held in Fairfax, Virginia, April 29 -May 1, 1998.",
"links": null
},
"BIBREF8": {
"ref_id": "b8",
"title": "Danish Dependency Treebank",
"authors": [
{
"first": "Matthias",
"middle": [
"T"
],
"last": "Kromann",
"suffix": ""
},
{
"first": "Line",
"middle": [],
"last": "Mikkelsen",
"suffix": ""
},
{
"first": "Stine",
"middle": [],
"last": "Kern Lynge",
"suffix": ""
}
],
"year": 2003,
"venue": "Proceedings of the Second Workshop on Treebanks and Linguistic Theories",
"volume": "",
"issue": "",
"pages": "217--220",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Matthias T. Kromann, Line Mikkelsen, and Stine Kern Lynge. 2003. Danish Dependency Treebank. In Proceed- ings of the Second Workshop on Treebanks and Linguistic Theories, pages 217-220.",
"links": null
},
"BIBREF9": {
"ref_id": "b9",
"title": "ParCorFull: A parallel corpus annotated with full coreference",
"authors": [
{
"first": "Ekaterina",
"middle": [],
"last": "Lapshinova-Koltunski",
"suffix": ""
},
{
"first": "Christian",
"middle": [],
"last": "Hardmeier",
"suffix": ""
},
{
"first": "Pauline",
"middle": [],
"last": "Krielke",
"suffix": ""
}
],
"year": 2018,
"venue": "LINDAT/CLARIAH-CZ digital library at the Institute of Formal and Applied Linguistics (\u00daFAL), Faculty of Mathematics and Physics",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Ekaterina Lapshinova-Koltunski, Christian Hardmeier, and Pauline Krielke. 2018. ParCorFull: A parallel corpus annotated with full coreference. LINDAT/CLARIAH-CZ digital library at the Institute of Formal and Applied Linguistics (\u00daFAL), Faculty of Mathematics and Physics, Charles University.",
"links": null
},
"BIBREF10": {
"ref_id": "b10",
"title": "Dependency syntax: theory and practice",
"authors": [
{
"first": "Igor",
"middle": [],
"last": "",
"suffix": ""
}
],
"year": 1988,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Igor Aleksandrovi\u010d Mel'\u010duk et al. 1988. Dependency syntax: theory and practice. SUNY press.",
"links": null
},
"BIBREF11": {
"ref_id": "b11",
"title": "Coreference in Prague Czech-English Dependency Treebank",
"authors": [
{
"first": "Anna",
"middle": [],
"last": "Nedoluzhko",
"suffix": ""
},
{
"first": "Michal",
"middle": [],
"last": "Nov\u00e1k",
"suffix": ""
},
{
"first": "Silvie",
"middle": [],
"last": "Cinkov\u00e1",
"suffix": ""
},
{
"first": "Marie",
"middle": [],
"last": "Mikulov\u00e1",
"suffix": ""
},
{
"first": "Ji\u0159\u00ed",
"middle": [],
"last": "M\u00edrovsk\u00fd",
"suffix": ""
}
],
"year": 2016,
"venue": "Proceedings of the Tenth International Conference on Language Resources and Evaluation (LREC'16)",
"volume": "",
"issue": "",
"pages": "169--176",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Anna Nedoluzhko, Michal Nov\u00e1k, Silvie Cinkov\u00e1, Marie Mikulov\u00e1, and Ji\u0159\u00ed M\u00edrovsk\u00fd. 2016. Coreference in Prague Czech-English Dependency Treebank. In Proceedings of the Tenth International Conference on Lan- guage Resources and Evaluation (LREC'16), pages 169-176, Portoro\u017e, Slovenia, May. European Language Resources Association (ELRA).",
"links": null
},
"BIBREF12": {
"ref_id": "b12",
"title": "Coreference meets Universal Dependencies -a pilot experiment on harmonizing coreference datasets for 11 languages",
"authors": [
{
"first": "Anna",
"middle": [],
"last": "Nedoluzhko",
"suffix": ""
},
{
"first": "Michal",
"middle": [],
"last": "Nov\u00e1k",
"suffix": ""
},
{
"first": "Martin",
"middle": [],
"last": "Popel",
"suffix": ""
},
{
"first": "Zden\u011bk",
"middle": [],
"last": "\u017dabokrtsk\u00fd",
"suffix": ""
},
{
"first": "Daniel",
"middle": [],
"last": "Zeman",
"suffix": ""
}
],
"year": 2021,
"venue": "\u00daFAL MFF UK",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Anna Nedoluzhko, Michal Nov\u00e1k, Martin Popel, Zden\u011bk \u017dabokrtsk\u00fd, and Daniel Zeman. 2021. Coreference meets Universal Dependencies -a pilot experiment on harmonizing coreference datasets for 11 languages. Tech- nical Report 66, \u00daFAL MFF UK, Praha, Czechia.",
"links": null
},
"BIBREF13": {
"ref_id": "b13",
"title": "Universal Dependencies v2: An evergrowing multilingual treebank collection",
"authors": [
{
"first": "Joakim",
"middle": [],
"last": "Nivre",
"suffix": ""
},
{
"first": "Marie-Catherine",
"middle": [],
"last": "De Marneffe",
"suffix": ""
},
{
"first": "Filip",
"middle": [],
"last": "Ginter",
"suffix": ""
},
{
"first": "Jan",
"middle": [],
"last": "Haji\u010d",
"suffix": ""
},
{
"first": "Christopher",
"middle": [],
"last": "Manning",
"suffix": ""
},
{
"first": "Sampo",
"middle": [],
"last": "Pyysalo",
"suffix": ""
},
{
"first": "Sebastian",
"middle": [],
"last": "Schuster",
"suffix": ""
},
{
"first": "Francis",
"middle": [],
"last": "Tyers",
"suffix": ""
},
{
"first": "Daniel",
"middle": [],
"last": "Zeman",
"suffix": ""
}
],
"year": 2020,
"venue": "Proceedings of the 12th International Conference on Language Resources and Evaluation (LREC 2020)",
"volume": "",
"issue": "",
"pages": "4034--4043",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Joakim Nivre, Marie-Catherine de Marneffe, Filip Ginter, Jan Haji\u010d, Christopher Manning, Sampo Pyysalo, Se- bastian Schuster, Francis Tyers, and Daniel Zeman. 2020. Universal Dependencies v2: An evergrowing multi- lingual treebank collection. In Proceedings of the 12th International Conference on Language Resources and Evaluation (LREC 2020), pages 4034-4043, Marseille, France. European Language Resources Association.",
"links": null
},
"BIBREF14": {
"ref_id": "b14",
"title": "Polish coreference corpus",
"authors": [
{
"first": "Maciej",
"middle": [],
"last": "Ogrodniczuk",
"suffix": ""
},
{
"first": "Katarzyna",
"middle": [],
"last": "Glowi\u0144ska",
"suffix": ""
},
{
"first": "Mateusz",
"middle": [],
"last": "Kope\u0107",
"suffix": ""
},
{
"first": "Agata",
"middle": [],
"last": "Savary",
"suffix": ""
},
{
"first": "Magdalena",
"middle": [],
"last": "Zawis\u0142awska",
"suffix": ""
}
],
"year": 2013,
"venue": "Human Language Technology. Challenges for Computer Science and Linguistics -6th Language and Technology Conference, LTC 2013",
"volume": "9561",
"issue": "",
"pages": "215--226",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Maciej Ogrodniczuk, Katarzyna Glowi\u0144ska, Mateusz Kope\u0107, Agata Savary, and Magdalena Zawis\u0142awska. 2013. Polish coreference corpus. In Human Language Technology. Challenges for Computer Science and Linguistics -6th Language and Technology Conference, LTC 2013, Pozna\u0144, Poland, December 7-9, 2013. Revised Selected Papers, volume 9561 of Lecture Notes in Computer Science, pages 215-226. Springer.",
"links": null
},
"BIBREF15": {
"ref_id": "b15",
"title": "Coreference in Polish: Annotation, Resolution and Evaluation",
"authors": [
{
"first": "Maciej",
"middle": [],
"last": "Ogrodniczuk",
"suffix": ""
},
{
"first": "Katarzyna",
"middle": [],
"last": "G\u0142owi\u0144ska",
"suffix": ""
},
{
"first": "Mateusz",
"middle": [],
"last": "Kope\u0107",
"suffix": ""
},
{
"first": "Agata",
"middle": [],
"last": "Savary",
"suffix": ""
},
{
"first": "Magdalena",
"middle": [],
"last": "Zawis\u0142awska",
"suffix": ""
}
],
"year": 2015,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Maciej Ogrodniczuk, Katarzyna G\u0142owi\u0144ska, Mateusz Kope\u0107, Agata Savary, and Magdalena Zawis\u0142awska. 2015. Coreference in Polish: Annotation, Resolution and Evaluation. Walter De Gruyter.",
"links": null
},
"BIBREF16": {
"ref_id": "b16",
"title": "Catenae: Introducing a novel unit of syntactic analysis",
"authors": [
{
"first": "Timothy",
"middle": [],
"last": "Osborne",
"suffix": ""
},
{
"first": "Michael",
"middle": [],
"last": "Putnam",
"suffix": ""
},
{
"first": "Thomas",
"middle": [],
"last": "Gro\u00df",
"suffix": ""
}
],
"year": 2012,
"venue": "Syntax",
"volume": "15",
"issue": "4",
"pages": "354--396",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Timothy Osborne, Michael Putnam, and Thomas Gro\u00df. 2012. Catenae: Introducing a novel unit of syntactic analysis. Syntax, 15(4):354-396.",
"links": null
},
"BIBREF17": {
"ref_id": "b17",
"title": "From Lexical Functional Grammar to Enhanced Universal Dependencies: Linguistically informed treebanks of Polish",
"authors": [
{
"first": "Agnieszka",
"middle": [],
"last": "Patejuk",
"suffix": ""
},
{
"first": "Adam",
"middle": [],
"last": "Przepi\u00f3rkowski",
"suffix": ""
}
],
"year": 2018,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Agnieszka Patejuk and Adam Przepi\u00f3rkowski. 2018. From Lexical Functional Grammar to Enhanced Universal Dependencies: Linguistically informed treebanks of Polish. Institute of Computer Science, Polish Academy of Sciences, Warsaw.",
"links": null
},
"BIBREF18": {
"ref_id": "b18",
"title": "A joint framework for coreference resolution and mention head detection",
"authors": [
{
"first": "Haoruo",
"middle": [],
"last": "Peng",
"suffix": ""
},
{
"first": "Kai-Wei",
"middle": [],
"last": "Chang",
"suffix": ""
},
{
"first": "Dan",
"middle": [],
"last": "Roth",
"suffix": ""
}
],
"year": 2015,
"venue": "Proceedings of the Nineteenth Conference on Computational Natural Language Learning",
"volume": "",
"issue": "",
"pages": "12--21",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Haoruo Peng, Kai-Wei Chang, and Dan Roth. 2015. A joint framework for coreference resolution and mention head detection. In Proceedings of the Nineteenth Conference on Computational Natural Language Learning, pages 12-21.",
"links": null
},
"BIBREF19": {
"ref_id": "b19",
"title": "Coordination structures in dependency treebanks",
"authors": [
{
"first": "Martin",
"middle": [],
"last": "Popel",
"suffix": ""
},
{
"first": "David",
"middle": [],
"last": "Mare\u010dek",
"suffix": ""
},
{
"first": "Jan",
"middle": [],
"last": "\u0160t\u011bp\u00e1nek",
"suffix": ""
},
{
"first": "Daniel",
"middle": [],
"last": "Zeman",
"suffix": ""
},
{
"first": "Zden\u011bk",
"middle": [],
"last": "\u017dabokrtsk\u00fd",
"suffix": ""
}
],
"year": 2013,
"venue": "Proceedings of the 51st Annual Meeting of the Association for Computational Linguistics",
"volume": "",
"issue": "",
"pages": "517--527",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Martin Popel, David Mare\u010dek, Jan \u0160t\u011bp\u00e1nek, Daniel Zeman, and Zden\u011bk \u017dabokrtsk\u00fd. 2013. Coordination struc- tures in dependency treebanks. In Proceedings of the 51st Annual Meeting of the Association for Computational Linguistics, pages 517-527, Sofija, Bulgaria. Association for Computational Linguistics.",
"links": null
},
"BIBREF20": {
"ref_id": "b20",
"title": "Do UD Trees Match Mention Spans in Coreference Annotations?",
"authors": [
{
"first": "Martin",
"middle": [],
"last": "Popel",
"suffix": ""
},
{
"first": "Zden\u011bk",
"middle": [],
"last": "\u017dabokrtsk\u00fd",
"suffix": ""
},
{
"first": "Anna",
"middle": [],
"last": "Nedoluzhko",
"suffix": ""
},
{
"first": "Michal",
"middle": [],
"last": "Nov\u00e1k",
"suffix": ""
},
{
"first": "Daniel",
"middle": [],
"last": "Zeman",
"suffix": ""
}
],
"year": 2021,
"venue": "Findings of EMNLP 2021. Association for Computational Linguistics",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Martin Popel, Zden\u011bk \u017dabokrtsk\u00fd, Anna Nedoluzhko, Michal Nov\u00e1k, and Daniel Zeman. 2021. Do UD Trees Match Mention Spans in Coreference Annotations? In Findings of EMNLP 2021. Association for Computa- tional Linguistics.",
"links": null
},
"BIBREF21": {
"ref_id": "b21",
"title": "AnCora-CO: Coreferentially Annotated Corpora for Spanish and Catalan",
"authors": [
{
"first": "Marta",
"middle": [],
"last": "Recasens",
"suffix": ""
},
{
"first": "M",
"middle": [],
"last": "Ant\u00f2nia Mart\u00ed",
"suffix": ""
}
],
"year": 2010,
"venue": "Lang. Resour. Eval",
"volume": "44",
"issue": "4",
"pages": "315--345",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Marta Recasens and M. Ant\u00f2nia Mart\u00ed. 2010. AnCora-CO: Coreferentially Annotated Corpora for Spanish and Catalan. Lang. Resour. Eval., 44(4):315-345, December.",
"links": null
},
"BIBREF22": {
"ref_id": "b22",
"title": "Teorie valence a jej\u00ed form\u00e1ln\u00ed zpracov\u00e1n\u00ed. Slovo a slovesnost",
"authors": [
{
"first": "Petr",
"middle": [],
"last": "Sgall",
"suffix": ""
}
],
"year": 1998,
"venue": "",
"volume": "59",
"issue": "",
"pages": "15--29",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Petr Sgall. 1998. Teorie valence a jej\u00ed form\u00e1ln\u00ed zpracov\u00e1n\u00ed. Slovo a slovesnost, 59(1):15-29.",
"links": null
},
"BIBREF23": {
"ref_id": "b23",
"title": "UDPipe 2.0 prototype at CoNLL 2018 UD shared task",
"authors": [
{
"first": "Milan",
"middle": [],
"last": "Straka",
"suffix": ""
}
],
"year": 2018,
"venue": "Proceedings of the CoNLL 2018 Shared Task: Multilingual Parsing from Raw Text to Universal Dependencies",
"volume": "",
"issue": "",
"pages": "197--207",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Milan Straka. 2018. UDPipe 2.0 prototype at CoNLL 2018 UD shared task. In Proceedings of the CoNLL 2018 Shared Task: Multilingual Parsing from Raw Text to Universal Dependencies, pages 197-207, Brussels, Belgium, October. Association for Computational Linguistics.",
"links": null
},
"BIBREF24": {
"ref_id": "b24",
"title": "Evaluating Anaphora and Coreference Resolution for Russian",
"authors": [
{
"first": "Svetlana",
"middle": [],
"last": "Toldova",
"suffix": ""
},
{
"first": "Anna",
"middle": [],
"last": "Roytberg",
"suffix": ""
},
{
"first": "Alina",
"middle": [],
"last": "Ladygina",
"suffix": ""
},
{
"first": "Maria",
"middle": [],
"last": "Vasilyeva",
"suffix": ""
},
{
"first": "Ilya",
"middle": [],
"last": "Azerkovich",
"suffix": ""
},
{
"first": "Matvei",
"middle": [],
"last": "Kurzukov",
"suffix": ""
},
{
"first": "G",
"middle": [],
"last": "Sim",
"suffix": ""
},
{
"first": "D",
"middle": [
"V"
],
"last": "Gorshkov",
"suffix": ""
},
{
"first": "A",
"middle": [],
"last": "Ivanova",
"suffix": ""
},
{
"first": "Anna",
"middle": [],
"last": "Nedoluzhko",
"suffix": ""
},
{
"first": "Yulia",
"middle": [],
"last": "Grishina",
"suffix": ""
}
],
"year": 2014,
"venue": "Komp'juternaja lingvistika i intellektual'nye tehnologii. Po materialam ezhegodnoj Mezhdunarodnoj konferencii Dialog",
"volume": "",
"issue": "",
"pages": "681--695",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Svetlana Toldova, Anna Roytberg, Alina Ladygina, Maria Vasilyeva, Ilya Azerkovich, Matvei Kurzukov, G. Sim, D.V. Gorshkov, A. Ivanova, Anna Nedoluzhko, and Yulia Grishina. 2014. Evaluating Anaphora and Corefer- ence Resolution for Russian. In Komp'juternaja lingvistika i intellektual'nye tehnologii. Po materialam ezhe- godnoj Mezhdunarodnoj konferencii Dialog, pages 681-695.",
"links": null
},
"BIBREF25": {
"ref_id": "b25",
"title": "Annotating a broad range of anaphoric phenomena, in a variety of genres: the ARRAU Corpus",
"authors": [
{
"first": "Olga",
"middle": [],
"last": "Uryupina",
"suffix": ""
},
{
"first": "Ron",
"middle": [],
"last": "Artstein",
"suffix": ""
},
{
"first": "Antonella",
"middle": [],
"last": "Bristot",
"suffix": ""
},
{
"first": "Federica",
"middle": [],
"last": "Cavicchio",
"suffix": ""
},
{
"first": "Francesca",
"middle": [],
"last": "Delogu",
"suffix": ""
},
{
"first": "J",
"middle": [],
"last": "Kepa",
"suffix": ""
},
{
"first": "Massimo",
"middle": [],
"last": "Rodriguez",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Poesio",
"suffix": ""
}
],
"year": 2020,
"venue": "Natural Language Engineering",
"volume": "26",
"issue": "1",
"pages": "95--128",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Olga Uryupina, Ron Artstein, Antonella Bristot, Federica Cavicchio, Francesca Delogu, Kepa J. Rodriguez, and Massimo Poesio. 2020. Annotating a broad range of anaphoric phenomena, in a variety of genres: the ARRAU Corpus. Natural Language Engineering, 26(1):95-128, January.",
"links": null
},
"BIBREF26": {
"ref_id": "b26",
"title": "Learning Global Features for Coreference Resolution",
"authors": [
{
"first": "Sam",
"middle": [],
"last": "Wiseman",
"suffix": ""
},
{
"first": "Alexander",
"middle": [
"M"
],
"last": "Rush",
"suffix": ""
},
{
"first": "Stuart",
"middle": [
"M"
],
"last": "Shieber",
"suffix": ""
}
],
"year": 2016,
"venue": "Proceedings of the 2016 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies",
"volume": "",
"issue": "",
"pages": "994--1004",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Sam Wiseman, Alexander M. Rush, and Stuart M. Shieber. 2016. Learning Global Features for Coreference Reso- lution. In Proceedings of the 2016 Conference of the North American Chapter of the Association for Computa- tional Linguistics: Human Language Technologies, pages 994-1004, San Diego, California, June. Association for Computational Linguistics.",
"links": null
},
"BIBREF27": {
"ref_id": "b27",
"title": "The GUM Corpus: Creating Multilayer Resources in the Classroom",
"authors": [
{
"first": "Amir",
"middle": [],
"last": "Zeldes",
"suffix": ""
}
],
"year": 2017,
"venue": "Language Resources and Evaluation",
"volume": "51",
"issue": "3",
"pages": "581--612",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Amir Zeldes. 2017. The GUM Corpus: Creating Multilayer Resources in the Classroom. Language Resources and Evaluation, 51(3):581-612, September.",
"links": null
},
"BIBREF28": {
"ref_id": "b28",
"title": "Coreference annotation scheme and corpus for Lithuanian language",
"authors": [
{
"first": "Voldemaras",
"middle": [],
"last": "\u017ditkus",
"suffix": ""
},
{
"first": "Rita",
"middle": [],
"last": "Butkien\u0117",
"suffix": ""
}
],
"year": 2018,
"venue": "Fifth International Conference on Social Networks Analysis, Management and Security",
"volume": "",
"issue": "",
"pages": "243--250",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Voldemaras \u017ditkus and Rita Butkien\u0117. 2018. Coreference annotation scheme and corpus for Lithuanian language. In Fifth International Conference on Social Networks Analysis, Management and Security, SNAMS 2018, Valen- cia, Spain, October 15-18, 2018, pages 243-250. IEEE.",
"links": null
}
},
"ref_entries": {
"FIGREF0": {
"text": "or virus carriers' \u204e\u204egezonde VERB amod\u204e\u204e \u204e\u204ebacterie-X obl <-------------synt. head\u204e\u204e \u204e\u204eof CCONJ cc\u204e\u204e \u204e\u204evirusdragers NOUN conj <==== annot. head\u204e\u204e",
"type_str": "figure",
"uris": null,
"num": null
},
"FIGREF1": {
"text": ", a new contract was signed by Mr. Marian's wife ' w ADP case \u204e\u204e1998 ADJ amod <====== annot. head\u204e\u204e \u204e\u204eroku NOUN obl <--------synt. head\u204e\u204e now\u0105 is open until April 9, 2007'",
"type_str": "figure",
"uris": null,
"num": null
},
"TABREF1": {
"num": null,
"type_str": "table",
"text": "He remembered Dr. Goldmann's mother-in-law's claws.'",
"html": null,
"content": "<table><tr><td>1) Przypomnia\u0142</td><td>sobie</td><td>pazury</td><td>te\u015bciowej</td><td>doktora</td><td>Goldmanna.</td></tr><tr><td>He remembered</td><td>himself</td><td>claws</td><td>of mother-in-law</td><td>of dr.</td><td>Goldmann.</td></tr><tr><td colspan=\"3\">'Przypomnia\u0142 VERB root</td><td/><td/><td/></tr><tr><td colspan=\"2\">sobie PRON iobj</td><td/><td/><td/><td/></tr><tr><td colspan=\"2\">pazury NOUN obj</td><td/><td/><td/><td/></tr><tr><td colspan=\"6\">\u204e\u204ete\u015bciowej ADJ amod <======= annot. head\u204e\u204e</td></tr><tr><td colspan=\"5\">\u204e\u204edoktora NOUN nmod:poss <-----synt. head\u204e\u204e</td><td/></tr><tr><td colspan=\"4\">\u204e\u204eGoldmanna PROPN flat\u204e\u204e</td><td/><td/></tr><tr><td colspan=\"2\">. PUNCT punct</td><td/><td/><td/><td/></tr></table>"
},
"TABREF4": {
"num": null,
"type_str": "table",
"text": "Syntactically governing numerals and containers. In constructions with governing numerals (e.g. one of the candidates, all of this) and so-called 'containers' (e.g. group of tourists), UDPipe systematically marks the numerals or containers as heads. On the other hand, manual annotation often chooses their syntactic dependent members as more important, putting the emphasis on the semantic point of view. Nevertheless, Examples 14-15 from Polish-PCC and Examples 16-17 from Russian-RuCor show that the manual annotation of mention heads in constructions with containers and governing numerals, respectively, is not systematic. Although we admit there may be another aspect (e.g. semantic salience) that convinced the annotators to label heads in these examples differently, it is neither obvious nor described in the guidelines.",
"html": null,
"content": "<table><tr><td colspan=\"4\">(8) 3.5 million ounces</td><td/><td colspan=\"2\">(9) $25 million</td></tr><tr><td/><td colspan=\"4\">\u204e\u204e3.5 NUM compound\u204e\u204e</td><td/><td colspan=\"2\">\u204e\u204e$ SYM appos <-------------synt. head\u204e\u204e</td></tr><tr><td/><td colspan=\"4\">\u204e\u204emillion NUM nummod\u204e\u204e</td><td/><td colspan=\"2\">\u204e\u204e25 NUM compound\u204e\u204e</td></tr><tr><td colspan=\"5\">\u204e\u204eounces NOUN root <===== annot./synt. head\u204e\u204e</td><td/><td colspan=\"2\">\u204e\u204emillion NUM nummod <==== annot. head\u204e\u204e</td></tr><tr><td>(10) 40</td><td colspan=\"2\">milion\u00f3w</td><td>z\u0142otych</td><td/><td colspan=\"2\">(11) vijfhonderd</td><td>zestig</td><td>miljoen</td><td>gulden</td></tr><tr><td>40</td><td>million</td><td/><td>z\u0142oty</td><td/><td colspan=\"2\">five hundred</td><td>sixty</td><td>million</td><td>guilders</td></tr><tr><td/><td colspan=\"4\">\u204e\u204e40 NUM nummod\u204e\u204e</td><td/><td colspan=\"2\">\u204e\u204evijfhonderd NUM nummod\u204e\u204e</td></tr><tr><td/><td colspan=\"4\">\u204e\u204emilion\u00f3w NOUN flat <==== annot. head\u204e\u204e</td><td/><td colspan=\"2\">\u204e\u204ezestig NUM nummod\u204e\u204e</td></tr><tr><td colspan=\"5\">\u204e\u204ez\u0142otych NOUN nmod:poss <----synt. head\u204e\u204e</td><td/><td colspan=\"2\">\u204e\u204emiljoen NOUN nmod <-----synt. head\u204e\u204e</td></tr><tr><td/><td/><td/><td/><td/><td/><td colspan=\"2\">\u204e\u204egulden NOUN nmod <==== annot. head\u204e\u204e</td></tr><tr><td>(12) \u0434\u0432\u0435</td><td colspan=\"2\">\u0442\u044b\u0441\u044f\u0447\u0438</td><td/><td>\u043f\u0440\u0435\u0434\u043b\u043e\u0436\u0435\u043d\u0438\u0439</td><td>(13) 5</td><td>\u0442\u044b\u0441\u044f\u0447</td><td>\u0432\u043e\u0435\u043d\u043d\u044b\u0445</td></tr><tr><td>two</td><td colspan=\"3\">thousand.GEN.SG</td><td>sentences</td><td>5</td><td colspan=\"2\">thousand.GEN.PL</td><td>soldiers</td></tr><tr><td/><td colspan=\"4\">**\u0434\u0432\u0435 NUM nummod:gov**</td><td/><td colspan=\"2\">**5 NUM nummod**</td></tr><tr><td/><td colspan=\"4\">**\u0442\u044b\u0441\u044f\u0447\u0438 NUM nummod:gov**</td><td/><td colspan=\"2\">**\u0442\u044b\u0441\u044f\u0447 NOUN nsubj:pass <--synt. head**</td></tr><tr><td colspan=\"5\">**\u043f\u0440\u0435\u0434\u043b\u043e\u0436\u0435\u043d\u0438\u0439 NOUN obl <----synt. head**</td><td/><td colspan=\"2\">**\u0432\u043e\u0435\u043d\u043d\u044b\u0445 NOUN nmod <==== annot. head**</td></tr><tr><td>(14) 64</td><td>proc.</td><td colspan=\"2\">przemys\u0142u</td><td>chemicznego</td><td/><td/></tr><tr><td>64</td><td>perc.</td><td colspan=\"2\">industry</td><td>chemical</td><td/><td/></tr><tr><td colspan=\"4\">'64% chemical industry'</td><td/><td/><td/></tr></table>"
},
"TABREF5": {
"num": null,
"type_str": "table",
"text": "Quantifiers as determiners. Interestingly, we find quite a lot of cases of quantifiers in the syntactic position of determiners (some, most, each, half and even no). They are heads neither from the syntactic nor the semantic point of view. However, in some cases they are marked as heads in manual annotations, e.g. in most analysts in Example 18, half the total in Example 19, some investors, each bond, no trading (all from English-ARRAU) and in \u043d\u0435\u0441\u043a\u043e\u043b\u044c\u043a\u043e \u0441\u0435\u0440\u044b\u0445 \u043f\u044f\u0442\u0435\u043d /some grey spots/ from Russian-RuCor.(18) Most analysts had expected a sharper decline after the steep rise more than half the trust's total of $268 million",
"html": null,
"content": "<table><tr><td/><td colspan=\"3\">\u204e\u204eMost ADJ amod <======== annot. head\u204e\u204e</td><td colspan=\"4\">(19) more ADJ advmod</td><td/></tr><tr><td colspan=\"4\">\u204e\u204eanalysts NOUN nsubj <----synt. head\u204e\u204e</td><td/><td/><td colspan=\"2\">than ADP fixed</td><td/></tr><tr><td colspan=\"2\">had AUX aux</td><td/><td/><td/><td colspan=\"5\">\u204e\u204ehalf PRON det:predet <==== annot. head\u204e\u204e</td></tr><tr><td colspan=\"2\">expected VERB root</td><td/><td/><td/><td colspan=\"3\">\u204e\u204ethe DET det\u204e\u204e</td><td/></tr><tr><td/><td>a DET det</td><td/><td/><td/><td colspan=\"3\">\u204e\u204etrust NOUN nmod:poss\u204e\u204e</td><td/></tr><tr><td/><td colspan=\"2\">sharper ADJ amod</td><td/><td/><td colspan=\"3\">\u204e\u204e's PART case\u204e\u204e</td><td/></tr><tr><td colspan=\"2\">decline NOUN obj</td><td/><td/><td/><td colspan=\"5\">\u204e\u204etotal NOUN appos <----------synt. head\u204e\u204e</td></tr><tr><td/><td>after ADP case</td><td/><td/><td/><td colspan=\"3\">\u204e\u204eof ADP case\u204e\u204e</td><td/></tr><tr><td/><td>the DET det</td><td/><td/><td/><td colspan=\"2\">\u204e\u204e$ SYM nmod\u204e\u204e</td><td/><td/></tr><tr><td/><td>steep ADJ amod</td><td/><td/><td/><td/><td colspan=\"2\">\u204e\u204e268 NUM compound\u204e\u204e</td><td/></tr><tr><td colspan=\"2\">rise NOUN obl</td><td/><td/><td/><td colspan=\"3\">\u204e\u204emillion NUM nummod\u204e\u204e</td><td/></tr><tr><td/><td/><td/><td/><td>(15) 3</td><td>proc.</td><td>kupowanego</td><td>towaru</td><td/></tr><tr><td/><td/><td/><td/><td>3</td><td>perc.</td><td>purchased</td><td>goods</td><td/></tr><tr><td/><td/><td/><td/><td colspan=\"3\">'3% purchased goods'</td><td/><td/></tr><tr><td/><td/><td/><td/><td/><td colspan=\"4\">\u204e\u204e3 NUM nummod <====== annot. head\u204e\u204e</td></tr><tr><td/><td/><td/><td>head\u204e\u204e</td><td/><td colspan=\"4\">\u204e\u204eproc X nmod:poss <-----synt. head\u204e\u204e</td></tr><tr><td colspan=\"3\">\u204e\u204e. PUNCT punct\u204e\u204e</td><td/><td/><td colspan=\"3\">\u204e\u204e. PUNCT punct\u204e\u204e</td><td/></tr><tr><td colspan=\"4\">\u204e\u204eprzemys\u0142u NOUN nmod:poss <=== annot. head\u204e\u204e</td><td/><td colspan=\"3\">\u204e\u204ekupowanego ADJ acl\u204e\u204e</td><td/></tr><tr><td/><td colspan=\"3\">\u204e\u204echemicznego ADJ amod\u204e\u204e</td><td/><td colspan=\"3\">\u204e\u204etowaru NOUN nmod:poss\u204e\u204e</td><td/></tr><tr><td>(16) \u0433\u0440\u0443\u043f\u043f\u0430</td><td>\u0430\u043a\u0442\u0438\u0432\u0438\u0441\u0442\u043e\u0432</td><td>\u0437\u0430\u043d\u044f\u043b\u0430\u0441\u044c</td><td>\u0441\u0442\u0440\u043e\u0438\u0442\u0435\u043b\u044c\u0441\u0442\u0432\u043e\u043c</td><td colspan=\"2\">(17) \u0433\u0440\u0443\u043f\u043f\u0430</td><td>\u0443\u0447\u0451\u043d\u044b\u0445</td><td>\u043f\u043b\u0430\u043d\u0438\u0440\u043e\u0432\u0430\u043b\u0430</td><td>\u043f\u0440\u043e\u0432\u0435\u0441\u0442\u0438</td><td>\u043d\u0430\u0431\u043b\u044e\u0434\u0435\u043d\u0438\u044f</td></tr><tr><td>group</td><td>activists.GEN</td><td>took up</td><td>construction</td><td colspan=\"2\">group</td><td>scientists.GEN</td><td>planned</td><td>to conduct</td><td>observations</td></tr><tr><td colspan=\"2\">\u043a\u0430\u0442\u0430\u043f\u0443\u043b\u044c\u0442\u044b catapult.GEN</td><td/><td/><td colspan=\"5\">'a group of scientists planned to conduct observations'</td></tr><tr><td colspan=\"4\">'a group of activists took up the construction of the catapult'</td><td/><td colspan=\"5\">**\u0433\u0440\u0443\u043f\u043f\u0430 NOUN nsubj <=== annot./synt. head**</td></tr><tr><td/><td/><td/><td/><td/><td colspan=\"3\">**\u0443\u0447\u0451\u043d\u044b\u0445 NOUN nmod</td><td/></tr><tr><td colspan=\"4\">**\u0433\u0440\u0443\u043f\u043f\u0430 NOUN nsubj <--------synt. head** **\u0430\u043a\u0442\u0438\u0432\u0438\u0441\u0442\u043e\u0432 NOUN nmod <=== annot. head** \u0437\u0430\u043d\u044f\u043b\u0430\u0441\u044c VERB root</td><td/><td colspan=\"3\">\u0437\u0430\u043d\u044f\u043b\u0430\u0441\u044c VERB parataxis \u043f\u0440\u043e\u0432\u0435\u0441\u0442\u0438 VERB xcomp \u043d\u0430\u0431\u043b\u044e\u0434\u0435\u043d\u0438\u044f NOUN obj</td><td/></tr><tr><td colspan=\"3\">\u0441\u0442\u0440\u043e\u0438\u0442\u0435\u043b\u044c\u0441\u0442\u0432\u043e\u043c NOUN obl</td><td/><td/><td/><td/><td/><td/></tr><tr><td/><td colspan=\"2\">\u043a\u0430\u0442\u0430\u043f\u0443\u043b\u044c\u0442\u044b NOUN nmod</td><td/><td/><td/><td/><td/><td/></tr></table>"
}
}
}
} |