File size: 118,435 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 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 | {
"paper_id": "P13-1029",
"header": {
"generated_with": "S2ORC 1.0.0",
"date_generated": "2023-01-19T09:32:46.268444Z"
},
"title": "Transfer Learning for Constituency-Based Grammars",
"authors": [
{
"first": "Yuan",
"middle": [],
"last": "Zhang",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "Massachusetts Institute of Technology",
"location": {}
},
"email": "yuanzh@csail.mit.edu"
},
{
"first": "Regina",
"middle": [],
"last": "Barzilay",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "Massachusetts Institute of Technology",
"location": {}
},
"email": "regina@csail.mit.edu"
},
{
"first": "Amir",
"middle": [],
"last": "Globerson",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "The Hebrew University",
"location": {}
},
"email": ""
}
],
"year": "",
"venue": null,
"identifiers": {},
"abstract": "In this paper, we consider the problem of cross-formalism transfer in parsing. We are interested in parsing constituencybased grammars such as HPSG and CCG using a small amount of data specific for the target formalism, and a large quantity of coarse CFG annotations from the Penn Treebank. While all of the target formalisms share a similar basic syntactic structure with Penn Treebank CFG, they also encode additional constraints and semantic features. To handle this apparent discrepancy, we design a probabilistic model that jointly generates CFG and target formalism parses. The model includes features of both parses, allowing transfer between the formalisms, while preserving parsing efficiency. We evaluate our approach on three constituency-based grammars-CCG, HPSG, and LFG, augmented with the Penn Treebank-1. Our experiments show that across all three formalisms, the target parsers significantly benefit from the coarse annotations. 1",
"pdf_parse": {
"paper_id": "P13-1029",
"_pdf_hash": "",
"abstract": [
{
"text": "In this paper, we consider the problem of cross-formalism transfer in parsing. We are interested in parsing constituencybased grammars such as HPSG and CCG using a small amount of data specific for the target formalism, and a large quantity of coarse CFG annotations from the Penn Treebank. While all of the target formalisms share a similar basic syntactic structure with Penn Treebank CFG, they also encode additional constraints and semantic features. To handle this apparent discrepancy, we design a probabilistic model that jointly generates CFG and target formalism parses. The model includes features of both parses, allowing transfer between the formalisms, while preserving parsing efficiency. We evaluate our approach on three constituency-based grammars-CCG, HPSG, and LFG, augmented with the Penn Treebank-1. Our experiments show that across all three formalisms, the target parsers significantly benefit from the coarse annotations. 1",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Abstract",
"sec_num": null
}
],
"body_text": [
{
"text": "Over the last several decades, linguists have introduced many different grammars for describing the syntax of natural languages. Moreover, the ongoing process of developing new formalisms is intrinsic to linguistic research. However, before these grammars can be used for statistical parsing, they require annotated sentences for training. The difficulty of obtaining such annotations is a key limiting factor that inhibits the effective use of these grammars.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "The standard solution to this bottleneck has relied on manually crafted transformation rules that map readily available syntactic annotations (e.g, the Penn Treebank) to the desired formalism. Designing these transformation rules is a major undertaking which requires multiple correction cycles and a deep understanding of the underlying grammar formalisms. In addition, designing these rules frequently requires external resources such as Wordnet, and even involves correction of the existing treebank. This effort has to be repeated for each new grammar formalism, each new annotation scheme and each new language.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "In this paper, we propose an alternative approach for parsing constituency-based grammars. Instead of using manually-crafted transformation rules, this approach relies on a small amount of annotations in the target formalism. Frequently, such annotations are available in linguistic texts that introduce the formalism. For instance, a textbook on HPSG (Pollard and Sag, 1994) illustrates grammatical constructions using about 600 examples. While these examples are informative, they are not sufficient for training. To compensate for the annotation sparsity, our approach utilizes coarsely annotated data readily available in large quantities. A natural candidate for such coarse annotations is context-free grammar (CFG) from the Penn Treebank, while the target formalism can be any constituency-based grammars, such as Combinatory Categorial Grammar (CCG) (Steedman, 2001) , Lexical Functional Grammar (LFG) (Bresnan, 1982) or Head-Driven Phrase Structure Grammar (HPSG) (Pollard and Sag, 1994) . All of these formalisms share a similar basic syntactic structure with Penn Treebank CFG. However, the target formalisms also encode additional constraints and semantic features. For instance, Penn Treebank annotations do not make an explicit distinction between complement and adjunct, while all the above grammars mark these roles explicitly. Moreover, even the identical syntactic information is encoded differently in these formalisms. An example of this phenomenon is the marking of subject. In LFG, this information is captured in the mapping equation, namely \u2191 SBJ =\u2193, while Penn Treebank represents it as a functional tag, such as NP-SBJ. Figure 1 shows derivations in the three target formalisms we consider, as well as a CFG derivation. We can see that the derivations of these formalisms share the same basic structure, while the formalism-specific information is mainly encoded in the lexical entries and node labels.",
"cite_spans": [
{
"start": 352,
"end": 375,
"text": "(Pollard and Sag, 1994)",
"ref_id": "BIBREF32"
},
{
"start": 858,
"end": 874,
"text": "(Steedman, 2001)",
"ref_id": null
},
{
"start": 910,
"end": 925,
"text": "(Bresnan, 1982)",
"ref_id": "BIBREF1"
},
{
"start": 973,
"end": 996,
"text": "(Pollard and Sag, 1994)",
"ref_id": "BIBREF32"
}
],
"ref_spans": [
{
"start": 1646,
"end": 1654,
"text": "Figure 1",
"ref_id": null
}
],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "To enable effective transfer the model has to identify shared structural components between the formalisms despite the apparent differences. Moreover, we do not assume parallel annotations. To this end, our model jointly parses the two corpora according to the corresponding annotations, enabling transfer via parameter sharing. In particular, we augment each target tree node with hidden variables that capture the connection to the coarse annotations. Specifically, each node in the target tree has two labels: an entry which is specific to the target formalism, and a latent label containing a value from the Penn Treebank tagset, such as NP (see Figure 2 ). This design enables us to represent three types of features: the target formalismspecific features, the coarse formalism features, and features that connect the two. This modeling approach makes it possible to perform transfer to a range of target formalisms, without manually drafting formalism-specific rules.",
"cite_spans": [],
"ref_spans": [
{
"start": 650,
"end": 658,
"text": "Figure 2",
"ref_id": null
}
],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "We evaluate our approach on three constituency-based grammars -CCG, HPSG, and LFG. As a source of coarse annotations, we use the Penn Treebank. 2 Our results clearly demonstrate that for all three formalisms, parsing accuracy can be improved by training with additional coarse annotations. For instance, the model trained on 500 HPSG sentences achieves labeled dependency F-score of 72.3%. Adding 15,000 Penn Treebank sentences during training leads to 78.5% labeled dependency F-score, an absolute improvement of 6.2%. To achieve similar performance in the absence of coarse annotations, the parser has to be trained on about 1,500 sentences, namely three times what is needed when using coarse annotations. Similar results are ",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "Our work belongs to a broader class of research on transfer learning in parsing. This area has garnered significant attention due to the expense associated with obtaining syntactic annotations. Transfer learning in parsing has been applied in different contexts, such as multilingual learning (Snyder et al., 2009; Hwa et al., 2005; McDonald et al., 2011; Jiang and Liu, 2009) , domain adaptation (McClosky et al., 2010; Dredze et al., 2007; Blitzer et al., 2006) , and crossformalism transfer (Hockenmaier and Steedman, 2002; Miyao et al., 2005; Cahill et al., 2002; Riezler et al., 2002; Chen and Shanker, 2005; Candito et al., 2010) . There have been several attempts to map annotations in coarse grammars like CFG to annotations in richer grammar, like HPSG, LFG, or CCG. Traditional approaches in this area typically rely on manually specified rules that encode the relation between the two formalisms. For instance, mappings may specify how to convert traces and functional tags in Penn Treebank to the f-structure in LFG (Cahill, 2004) . These conversion rules are typically utilized in two ways: (1) to create a new treebank which is consequently used to train a parser for the target formalism (Hockenmaier and Steedman, 2002; Clark and Curran, 2003; Miyao et al., 2005; Miyao and Tsujii, 2008) , (2) to translate the output of a CFG parser into the target formalism (Cahill et al., 2002) .",
"cite_spans": [
{
"start": 293,
"end": 314,
"text": "(Snyder et al., 2009;",
"ref_id": "BIBREF34"
},
{
"start": 315,
"end": 332,
"text": "Hwa et al., 2005;",
"ref_id": "BIBREF17"
},
{
"start": 333,
"end": 355,
"text": "McDonald et al., 2011;",
"ref_id": "BIBREF25"
},
{
"start": 356,
"end": 376,
"text": "Jiang and Liu, 2009)",
"ref_id": "BIBREF18"
},
{
"start": 397,
"end": 420,
"text": "(McClosky et al., 2010;",
"ref_id": "BIBREF22"
},
{
"start": 421,
"end": 441,
"text": "Dredze et al., 2007;",
"ref_id": "BIBREF13"
},
{
"start": 442,
"end": 463,
"text": "Blitzer et al., 2006)",
"ref_id": "BIBREF0"
},
{
"start": 494,
"end": 526,
"text": "(Hockenmaier and Steedman, 2002;",
"ref_id": "BIBREF15"
},
{
"start": 527,
"end": 546,
"text": "Miyao et al., 2005;",
"ref_id": "BIBREF27"
},
{
"start": 547,
"end": 567,
"text": "Cahill et al., 2002;",
"ref_id": "BIBREF2"
},
{
"start": 568,
"end": 589,
"text": "Riezler et al., 2002;",
"ref_id": "BIBREF33"
},
{
"start": 590,
"end": 613,
"text": "Chen and Shanker, 2005;",
"ref_id": "BIBREF8"
},
{
"start": 614,
"end": 635,
"text": "Candito et al., 2010)",
"ref_id": "BIBREF5"
},
{
"start": 1028,
"end": 1042,
"text": "(Cahill, 2004)",
"ref_id": "BIBREF4"
},
{
"start": 1203,
"end": 1235,
"text": "(Hockenmaier and Steedman, 2002;",
"ref_id": "BIBREF15"
},
{
"start": 1236,
"end": 1259,
"text": "Clark and Curran, 2003;",
"ref_id": "BIBREF9"
},
{
"start": 1260,
"end": 1279,
"text": "Miyao et al., 2005;",
"ref_id": "BIBREF27"
},
{
"start": 1280,
"end": 1303,
"text": "Miyao and Tsujii, 2008)",
"ref_id": "BIBREF26"
},
{
"start": 1376,
"end": 1397,
"text": "(Cahill et al., 2002)",
"ref_id": "BIBREF2"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Related Work",
"sec_num": "2"
},
{
"text": "The design of these rules is a major linguistic and computational undertaking, which requires multiple iterations over the data to increase coverage (Miyao et al., 2005; Oepen et al., 2004) . By nature, the mapping rules are formalism spe-cific and therefore not transferable. Moreover, frequently designing such mappings involves modification to the original annotations. For instance, Hockenmaier and Steedman (2002) made thousands of POS and constituent modifications to the Penn Treebank to facilitate transfer to CCG. More importantly, in some transfer scenarios, deterministic rules are not sufficient, due to the high ambiguity inherent in the mapping. Therefore, our work considers an alternative set-up for crossformalism transfer where a small amount of annotations in the target formalism is used as an alternative to using deterministic rules.",
"cite_spans": [
{
"start": 149,
"end": 169,
"text": "(Miyao et al., 2005;",
"ref_id": "BIBREF27"
},
{
"start": 170,
"end": 189,
"text": "Oepen et al., 2004)",
"ref_id": "BIBREF30"
},
{
"start": 387,
"end": 418,
"text": "Hockenmaier and Steedman (2002)",
"ref_id": "BIBREF15"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Related Work",
"sec_num": "2"
},
{
"text": "The limitation of deterministic transfer rules has been recognized in prior work (Riezler et al., 2002) . Their method uses a hand-crafted LFG parser to create a set of multiple parsing candidates for a given sentence. Using the partial mapping from CFG to LFG as the guidance, the resulting trees are ranked based on their consistency with the labeled LFG bracketing imported from CFG. In contrast to this method, we neither require a parser for the target formalism, nor manual rules for partial mapping. Consequently, our method can be applied to many different target grammar formalisms without significant engineering effort for each one. The utility of coarse-grained treebanks is determined by the degree of structural overlap with the target formalism.",
"cite_spans": [
{
"start": 81,
"end": 103,
"text": "(Riezler et al., 2002)",
"ref_id": "BIBREF33"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Related Work",
"sec_num": "2"
},
{
"text": "Recall that our goal is to learn how to parse the target formalisms while using two annotated sources: a small set of sentences annotated in the target formalism (e.g., CCG), and a large set of sentences with coarse annotations. For the latter, we use the CFG parses from the Penn Treebank. For simplicity we focus on the CCG formalism in what follows. We also generalize our model to other formalisms, as explained in Section 5.4.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "The Learning Problem",
"sec_num": "3"
},
{
"text": "Our notations are as follows: an input sentence is denoted by S. A CFG parse is denoted by y CF G and a CCG parse is denoted by y CCG . Clearly the set of possible values for y CF G and y CCG is determined by S and the grammar. The training set is a set of N sentences S 1 , . . . , S N with CFG parses y 1 CF G , . . . , y N CF G , and",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "The Learning Problem",
"sec_num": "3"
},
{
"text": "M sentencesS 1 , . . . ,S M with CCG parses y 1 CCG , . . . , y M CCG .",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "The Learning Problem",
"sec_num": "3"
},
{
"text": "It is important to note that we do not assume we have parallel data for CCG and CFG.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "The Learning Problem",
"sec_num": "3"
},
{
"text": "Our goal is to use such a corpus for learning Figure 2: Illustration of the joint CCG-CFG representation. The shadowed labels correspond to the CFG derivation yCF G, whereas the other labels correspond to the CCG derivation yCCG. Note that the two derivations share the same (binarized) tree structure. Also shown are features that are turned on for this joint derivation (see Section 6).",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "The Learning Problem",
"sec_num": "3"
},
{
"text": "how to generate CCG parses to unseen sentences.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "The Learning Problem",
"sec_num": "3"
},
{
"text": "The key idea behind our work is to learn a joint distribution over CCG and CFG parses. Such a distribution can be marginalized to obtain a distribution over CCG or CFG and is thus appropriate when the training data is not parallel, as it is in our setting.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "A Joint Model for Two Formalisms",
"sec_num": "4"
},
{
"text": "It is not immediately clear how to jointly model the CCG and CFG parses, which are structurally quite different. Furthermore, a joint distribution over these will become difficult to handle computationally if not constructed carefully. To address this difficulty, we make several simplifying assumptions. First, we assume that both parses are given in normal form, i.e., they correspond to binary derivation trees. CCG parses are already provided in this form in CCGBank. CFG parses in the Penn Treebank are not binary, and we therefore binarize them, as explained in Section 5.3.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "A Joint Model for Two Formalisms",
"sec_num": "4"
},
{
"text": "Second, we assume that any y CF G and y CCG jointly generated must share the same derivation tree structure. This makes sense. Since both formalisms are constituency-based, their trees are expected to describe the same constituents. We denote the set of valid CFG and CCG joint parses for sentence S by Y(S).",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "A Joint Model for Two Formalisms",
"sec_num": "4"
},
{
"text": "The above two simplifying assumptions make it easy to define joint features on the two parses, as explained in Section 6. The representation and features are illustrated in Figure 2 .",
"cite_spans": [],
"ref_spans": [
{
"start": 173,
"end": 181,
"text": "Figure 2",
"ref_id": null
}
],
"eq_spans": [],
"section": "A Joint Model for Two Formalisms",
"sec_num": "4"
},
{
"text": "We shall work within the discriminative framework, where given a sentence we model a distribution over parses. As is standard in such settings, the distribution will be log-linear in a set of features of these parses. Denoting y = (y CF G , y CCG ), we seek to model the distribution p(y|S) corresponding to the probability of generating a pair of parses (CFG and CCG) given a sentence. The distribution thus has the following form:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "A Joint Model for Two Formalisms",
"sec_num": "4"
},
{
"text": "p joint (y|S; \u03b8) = 1 Z(S; \u03b8) e f (y,S)\u2022\u03b8 . (1)",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "A Joint Model for Two Formalisms",
"sec_num": "4"
},
{
"text": "where \u03b8 is a vector of parameters to be learned from data, and f (y, S) is a feature vector. Z(S; \u03b8) is a normalization (partition) function normalized over y \u2208 Y(S) the set of valid joint parses. The feature vector contains three types of features: CFG specific, CCG specific and joint CFG-CCG. We denote these by f CF G , f CCG , f joint . These depend on y CCG , y CF G and y respectively. Accordingly, the parameter vector \u03b8 is a concatenation of \u03b8 CCG , \u03b8 CF G and \u03b8 joint .",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "A Joint Model for Two Formalisms",
"sec_num": "4"
},
{
"text": "As mentioned above, we can use Equation 1 to obtain distributions over y CCG and y CF G via marginalization. For the distribution over y CCG we do precisely this, namely use:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "A Joint Model for Two Formalisms",
"sec_num": "4"
},
{
"text": "p CCG (y CCG |S; \u03b8) = y CF G p joint (y|S; \u03b8) (2)",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "A Joint Model for Two Formalisms",
"sec_num": "4"
},
{
"text": "For the distribution over y CF G we could have marginalized p joint over y CCG . However, this computation is costly for each sentence, and has to be repeated for all the sentences. Instead, we assume that the distribution over y CF G is a loglinear model with parameters \u03b8 CF G (i.e., a subvector of \u03b8) , namely:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "A Joint Model for Two Formalisms",
"sec_num": "4"
},
{
"text": "p CF G (y CF G |S; \u03b8 CF G ) = e f CF G (y CF G ,S)\u2022\u03b8 CF G Z(S; \u03b8 CF G ) .",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "A Joint Model for Two Formalisms",
"sec_num": "4"
},
{
"text": "(3) Thus, we assume that both p joint and p CF G have the same dependence on the f CF G features.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "A Joint Model for Two Formalisms",
"sec_num": "4"
},
{
"text": "The Likelihood Objective: Given the models above, it is natural to use maximum likelihood to find the optimal parameters. To do this, we define the following regularized likelihood function:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "A Joint Model for Two Formalisms",
"sec_num": "4"
},
{
"text": "L(\u03b8) = N i=1 log p CF G (y i CF G |S i , \u03b8 CF G ) + M i=1 log p CCG (y i CCG |S i , \u03b8) \u2212 \u03bb 2 \u03b8 2 2",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "A Joint Model for Two Formalisms",
"sec_num": "4"
},
{
"text": "where p CCG and p CF G are defined in Equations 2 and 3 respectively. The last term is the l 2 -norm regularization. Our goal is then to find a \u03b8 that maximizes L(\u03b8).",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "A Joint Model for Two Formalisms",
"sec_num": "4"
},
{
"text": "Training Algorithm: For maximizing L(\u03b8) w.r.t. \u03b8 we use the limited-memory BFGS algorithm (Nocedal and Wright, 1999) . Calculating the gradient of L(\u03b8) requires evaluating the expected values of f (y, S) and f CF G under the distributions p joint and p CF G respectively. This can be done via the inside-outside algorithm. 3",
"cite_spans": [
{
"start": 90,
"end": 116,
"text": "(Nocedal and Wright, 1999)",
"ref_id": "BIBREF29"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "A Joint Model for Two Formalisms",
"sec_num": "4"
},
{
"text": "Parsing Using the Model: To parse a sentence S, we calculate the maximum probability assignment for p joint (y|S; \u03b8). 4 The result is both a CFG and a CCG parse. Here we will mostly be interested in the CCG parse. The joint parse with maximum probability is found using a standard CYK chart parsing algorithm. The chart construction will be explained in Section 5.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "A Joint Model for Two Formalisms",
"sec_num": "4"
},
{
"text": "This section introduces important implementation details, including supertagging, feature forest pruning and binarization methods. Finally, we explain how to generalize our model to other constituency-based formalisms.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Implementation",
"sec_num": "5"
},
{
"text": "When parsing a target formalism tree, one needs to associate each word with a lexical entry. However, since the number of candidates is typically more than one thousand, the size of the chart explodes. One effective way of reducing the number of candidates is via supertagging (Clark and Curran, 2007) . A supertagger is used for selecting a small set of lexical entry candidates for each word in the sentence. We use the tagger in (Clark and Curran, 2007) as a general suppertagger for all the grammars considered. The only difference is that we use different lexical entries in different grammars.",
"cite_spans": [
{
"start": 277,
"end": 301,
"text": "(Clark and Curran, 2007)",
"ref_id": "BIBREF10"
},
{
"start": 432,
"end": 456,
"text": "(Clark and Curran, 2007)",
"ref_id": "BIBREF10"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Supertagging",
"sec_num": "5.1"
},
{
"text": "In the BFGS algorithm (see Section 4), feature expectation is computed using the inside-outside algorithm. To perform this dynamic programming efficiently, we first need to build the packed chart, namely the feature forest (Miyao, 2006) to represent the exponential number of all possible tree structures. However, a common problem for lexicalized grammars is that the forest size is too large. In CFG, the forest is pruned according to the inside probability of a simple generative PCFG model and a prior (Collins, 2003) . The basic idea is to prune the trees with lower probability. For the target formalism, a common practice is to prune the forest using the supertagger (Clark and Curran, 2007; Miyao, 2006) . In our implementation, we applied all pruning techniques, because the forest is a combination of CFG and target grammar formalisms (e.g., CCG or HPSG).",
"cite_spans": [
{
"start": 223,
"end": 236,
"text": "(Miyao, 2006)",
"ref_id": "BIBREF28"
},
{
"start": 506,
"end": 521,
"text": "(Collins, 2003)",
"ref_id": "BIBREF12"
},
{
"start": 674,
"end": 698,
"text": "(Clark and Curran, 2007;",
"ref_id": "BIBREF10"
},
{
"start": 699,
"end": 711,
"text": "Miyao, 2006)",
"ref_id": "BIBREF28"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Feature Forest Pruning",
"sec_num": "5.2"
},
{
"text": "We assume that the derivation tree in the target formalism is in a normal form, which is indeed the case for the treebanks we consider. As mentioned in Section 4, we would also like to work with binarized CFG derivations, such that all trees are in normal form and it is easy to construct features that link the two (see Section 6).",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Binarization",
"sec_num": "5.3"
},
{
"text": "Since Penn Treebank trees are not binarized, we construct a simple procedure for binarizing them. The procedure is based on the available target formalism parses in the training corpus, which are binarized. We illustrate it with an example. In what follows, we describe derivations using the POS of the head words of the corresponding node in the tree. This makes it possible to transfer binarization rules between formalisms.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Binarization",
"sec_num": "5.3"
},
{
"text": "Suppose we want to learn the binarization rule of the following derivation in CFG:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Binarization",
"sec_num": "5.3"
},
{
"text": "EQUATION",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [
{
"start": 0,
"end": 8,
"text": "EQUATION",
"ref_id": "EQREF",
"raw_str": "NN \u2192 (DT JJ NN)",
"eq_num": "(4)"
}
],
"section": "Binarization",
"sec_num": "5.3"
},
{
"text": "We now look for binary derivations with these POS in the target formalism corpus, and take the most common binarization form. For example, we may find that the most common binarization to binarize the CFG derivation in Equation 4 is:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Binarization",
"sec_num": "5.3"
},
{
"text": "NN \u2192 (DT (JJ NN))",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Binarization",
"sec_num": "5.3"
},
{
"text": "If no (DT JJ NN) structure is observed in the CCG corpus, we first apply the binary branching on the children to the left of the head, and then on the children to the right of the head.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Binarization",
"sec_num": "5.3"
},
{
"text": "We also experiment with using fixed binarization rules such as left/right branching, instead of learning them. This results in a drop on the dependency F-score by about 5%.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Binarization",
"sec_num": "5.3"
},
{
"text": "We introduce our model in the context of CCG, but the model can easily be generalized to other constituency-based grammars, such as HPSG and LFG. In a derivation tree, the formalism-specific information is mainly encoded in the lexical entries and the applied grammar rules, rather than the tree structures. Therefore we only need to change the node labels and lexical entries to the languagespecific ones, while the framework of the model remains the same.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Implementation in Other Formalisms",
"sec_num": "5.4"
},
{
"text": "Feature functions in log-linear models are designed to capture the characteristics of each derivation in the tree. In our model, as mentioned in Section 1, the features are also defined to enable information transfer between coarse and rich formalisms. In this section, we first introduce how different types of feature templates are designed, and then show an example of how the features help transfer the syntactic structure information. Note that the same feature templates are used for all the target grammar formalisms.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Features",
"sec_num": "6"
},
{
"text": "Recall that our y contains both the CFG and CCG parses, and that these use the same derivation tree structure. Each feature will consider either the CFG derivation, the CCG derivation or these two derivations jointly.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Features",
"sec_num": "6"
},
{
"text": "The feature construction is similar to constructions used in previous work (Miyao, 2006) . The features are based on the atomic features listed in Table 1 . These will be used to construct f (y, S) as explained next.",
"cite_spans": [
{
"start": 75,
"end": 88,
"text": "(Miyao, 2006)",
"ref_id": "BIBREF28"
}
],
"ref_spans": [
{
"start": 147,
"end": 154,
"text": "Table 1",
"ref_id": "TABREF2"
}
],
"eq_spans": [],
"section": "Features",
"sec_num": "6"
},
{
"text": "hl lexical entries/CCG categories of the head word r grammar rules, i.e. HPSG schema, resulting CCG categories, LFG mapping equations sy CFG syntactic label of the node (e.g. NP, VP) d distance between the head words of the children c whether a comma exists between the head words of the children sp the span of the subtree rooted at the node hw surface form of the head word of the node hp part-of-speech of the head word pi part-of-speech of the i-th word in the sentence We define the following feature templates: f binary for binary derivations, f unary for unary derivations, and f root for the root nodes. These use the atomic features in Table 1 , resulting in the following templates: syr, spr, hwr, hpr, hlr, pst\u22121, pst\u22122, pen+1, pen+2 f unary = r, sy p , hw, hp, hl f root = sy, hw, hp, hl",
"cite_spans": [
{
"start": 693,
"end": 744,
"text": "syr, spr, hwr, hpr, hlr, pst\u22121, pst\u22122, pen+1, pen+2",
"ref_id": null
}
],
"ref_spans": [
{
"start": 645,
"end": 652,
"text": "Table 1",
"ref_id": "TABREF2"
}
],
"eq_spans": [],
"section": "Features",
"sec_num": "6"
},
{
"text": "f binary = r, syp, d, c sy l , sp l , hw l , hp l , hl l ,",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Features",
"sec_num": "6"
},
{
"text": "In the above we used the following notation: p, l, r denote the parent node and left/right child node, and st, en denote the starting and ending index of the constituent.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Features",
"sec_num": "6"
},
{
"text": "We also consider templates with subsets of the above features. The final list of binary feature templates is shown in Table 2 . It can be seen that some features depend only on the CFG derivations (i.e., those without r,hl), and are thus in f CF G (y, S). Others depend only on CCG derivations (i.e., those without sy), and are in f CCG (y, S). The rest depend on both CCG and CFG and are thus in f joint (y, S).",
"cite_spans": [],
"ref_spans": [
{
"start": 118,
"end": 125,
"text": "Table 2",
"ref_id": "TABREF3"
}
],
"eq_spans": [],
"section": "Features",
"sec_num": "6"
},
{
"text": "Note that after binarization, grandparent and sibling information becomes very important in encoding the structure. However, we limit the features to be designed locally in a derivation in order to run inside-outside efficiently. Therefore we use the preceding and succeeding POS tag information to approximate the grandparent and sibling information. Empirically, these features yield a significant improvement on the constituent accuracy. In order to apply the same feature templates to other target formalisms, we only need to assign the atomic features r and hl with the formalismspecific values. We do not need extra engineering work on redesigning the feature templates. Figure 3 gives an example in CCG of how features help transfer the syntactic information from Penn Treebank and learn the correspondence to the formalism-specific information. From the Penn Treebank CFG annotations, we can learn that the derivation VP\u2192(VP, NP) is common, as shown on the left of Figure 3 . In a CCG tree, this tendency will encourage the y CF G (latent) variables to take this CFG parse. Then weights on the f joint features will be learned to model the connection between the CFG and CCG labels. Moreover, the formalism-specific features f CCG can also encode the formalism-specific syntactic and semantic information. These three types of features work together to generate a tree skeleton and fill in the CFG and CCG labels. Datasets: As a source of coarse annotations, we use the Penn Treebank-1 (Marcus et al., 1993) . In addition, for CCG, HPSG and LFG, we rely on formalism-specific corpora developed in prior research (Hockenmaier and Steedman, 2002; Miyao et al., 2005; Cahill et al., 2002; King et al., 2003) . All of these corpora were derived via conversion of Penn Treebank to the target formalisms. In particular, our CCG and HPSG datasets were converted from the Penn Treebank based on hand- Figure 4 : Model performance with 500 target formalism trees and different numbers of CFG trees, evaluated using labeled/unlabeled dependency F-score and unlabeled PARSEVAL. crafted rules (Hockenmaier and Steedman, 2002; Miyao et al., 2005) . Table 3 shows which sections of the treebanks were used in training, testing and development for both formalisms. Our LFG training dataset was constructed in a similar fashion (Cahill et al., 2002) . However, we choose to use PARC700 as our LFG tesing and development datasets, following the previous work by (Kaplan et al., 2004) . It contains 700 manually annotated sentences that are randomly selected from Penn Treebank Section 23. The split of PARC700 follows the setting in (Kaplan et al., 2004 ). Since our model does not assume parallel data, we use distinct sentences in the source and target treebanks. Following previous work (Hockenmaier, 2003; Miyao and Tsujii, 2008) , we only consider sentences not exceeding 40 words, except on PARC700 where all sentences are used.",
"cite_spans": [
{
"start": 1483,
"end": 1515,
"text": "Treebank-1 (Marcus et al., 1993)",
"ref_id": null
},
{
"start": 1620,
"end": 1652,
"text": "(Hockenmaier and Steedman, 2002;",
"ref_id": "BIBREF15"
},
{
"start": 1653,
"end": 1672,
"text": "Miyao et al., 2005;",
"ref_id": "BIBREF27"
},
{
"start": 1673,
"end": 1693,
"text": "Cahill et al., 2002;",
"ref_id": "BIBREF2"
},
{
"start": 1694,
"end": 1712,
"text": "King et al., 2003)",
"ref_id": "BIBREF20"
},
{
"start": 2089,
"end": 2121,
"text": "(Hockenmaier and Steedman, 2002;",
"ref_id": "BIBREF15"
},
{
"start": 2122,
"end": 2141,
"text": "Miyao et al., 2005)",
"ref_id": "BIBREF27"
},
{
"start": 2320,
"end": 2341,
"text": "(Cahill et al., 2002)",
"ref_id": "BIBREF2"
},
{
"start": 2453,
"end": 2474,
"text": "(Kaplan et al., 2004)",
"ref_id": "BIBREF19"
},
{
"start": 2624,
"end": 2644,
"text": "(Kaplan et al., 2004",
"ref_id": "BIBREF19"
},
{
"start": 2781,
"end": 2800,
"text": "(Hockenmaier, 2003;",
"ref_id": "BIBREF16"
},
{
"start": 2801,
"end": 2824,
"text": "Miyao and Tsujii, 2008)",
"ref_id": "BIBREF26"
}
],
"ref_spans": [
{
"start": 677,
"end": 685,
"text": "Figure 3",
"ref_id": "FIGREF1"
},
{
"start": 973,
"end": 981,
"text": "Figure 3",
"ref_id": "FIGREF1"
},
{
"start": 1901,
"end": 1909,
"text": "Figure 4",
"ref_id": null
},
{
"start": 2144,
"end": 2151,
"text": "Table 3",
"ref_id": "TABREF5"
}
],
"eq_spans": [],
"section": "Features",
"sec_num": "6"
},
{
"text": "fCF G d, w l,r , hp l,r , sy p,l,r , d, w l,r , sy p,l,r , c, w l,r , hp l,r , sy p,l,r , c, w l,r , sy p,l,r , d, c, hp l,r , sy p,l,r , d, c, sy p,l,r , c, sp l,r , hp l,r , sy p,l,r , c, sp l,r , sy p,l,r , pst\u22121, sy p,l,r , pen+1, sy p,l,r , pst\u22121, pen+1, sy p,l,r , pst\u22121, pst\u22122, sy p,l,r , pen+1, pen+2, sy p,l,r , pst\u22121, pst\u22122, pen+1, pen+2, sy p,l,r , fCCG r, d, c, hw l,r , hp l,r , hl l,r , r, d, c, hw l,r , hp l,r r, d, c, hw l,r , hl l,r , r, c, sp l,r , hw l,r , hp l,r , hl l,r r, c, sp l,r , hw l,r , hp l,r , , r, c, sp l,r , hw l,r , hl l,r r, d, c, hp l,r , hl l,r , r, d, c, hp l,r , r, d, c, hl l,r r, c, hp l,r , hl l,",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Features",
"sec_num": "6"
},
{
"text": "Evaluation Metrics: We use two evaluation metrics. First, following previous work, we evaluate our method using the labeled and unlabeled predicate-argument dependency F-score. This metric is commonly used to measure parsing quality for the formalisms considered in this paper. The detailed definition of this measure as applied for each formalism is provided in (Clark and Curran, 2003; Miyao and Tsujii, 2008; . For CCG, we use the evaluation script from the C&C tools. 5 For HPSG, we evaluate all types of dependencies, including punctuations. For LFG, we consider the preds-only dependencies, which are the dependencies between pairs of words. Secondly, we also evaluate using unlabeled PARSEVAL, a standard measure for PCFG parsing (Petrov and Klein, 2007; Charniak and Johnson, 2005; Charniak, 2000; Collins, 1997) . The dependency F-score captures both the target-5 Available at http://svn.ask.it.usyd.edu.au/trac/candc/wiki grammar labels and tree-structural relations. The unlabeled PARSEVAL is used as an auxiliary measure that enables us to separate these two aspects by focusing on the structural relations exclusively.",
"cite_spans": [
{
"start": 363,
"end": 387,
"text": "(Clark and Curran, 2003;",
"ref_id": "BIBREF9"
},
{
"start": 388,
"end": 411,
"text": "Miyao and Tsujii, 2008;",
"ref_id": "BIBREF26"
},
{
"start": 737,
"end": 761,
"text": "(Petrov and Klein, 2007;",
"ref_id": "BIBREF31"
},
{
"start": 762,
"end": 789,
"text": "Charniak and Johnson, 2005;",
"ref_id": "BIBREF6"
},
{
"start": 790,
"end": 805,
"text": "Charniak, 2000;",
"ref_id": "BIBREF7"
},
{
"start": 806,
"end": 820,
"text": "Collins, 1997)",
"ref_id": "BIBREF11"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Evaluation Setup",
"sec_num": "7"
},
{
"text": "Training without CFG Data: To assess the impact of coarse data in the experiments below, we also consider the model trained only on formalism-specific annotations. When no CFG sentences are available, we assign all the CFG labels to a special value shared by all the nodes. In this set-up, the model reduces to a normal loglinear model for the target formalism.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Evaluation Setup",
"sec_num": "7"
},
{
"text": "Parameter Settings: During training, all the feature parameters \u03b8 are initialized to zero. The hyperparameters used in the model are tuned on the development sets. We noticed, however, that the resulting values are consistent across different formalisms. In particular, we set the l 2 -norm weight to \u03bb = 1.0, the supertagger threshold to \u03b2 = 0.01, and the PCFG pruning threshold to \u03b1 = 0.002.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Evaluation Setup",
"sec_num": "7"
},
{
"text": "To analyze the effectiveness of annotation transfer, we fix the number of annotated trees in the target formalism and vary the amount of coarse annotations available to the algorithm during training. In particular, we use 500 sentences with formalism-specific annotations, and vary the number of CFG trees from zero to 15,000.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Impact of Coarse Annotations on Target Formalism:",
"sec_num": null
},
{
"text": "As Figure 4 shows, CFG data boosts parsing accuracy for all the target formalisms. For instance, there is a gain of 6.2% in labeled dependency F-score for HPSG formalism when 15,000 CFG trees are used. Moreover, increasing the number of coarse annotations used in training leads to further improvement on different evaluation metrics. Tradeoff between Target and Coarse Annotations: We also assess the relative contribution of coarse annotations when the size of annotated training corpus in the target formalism varies. In this set of experiments, we fix the number of CFG trees to 15,000 and vary the number of target annotations from 500 to 4,000. Figure 5 shows the relative contribution of formalism-specific annotations compared to that of the coarse annotations. For instance, Figure 5a shows that the parsing performance achieved using 2,000 CCG sentences can be achieved using approximately 500 CCG sentences when coarse annotations are available for training. More generally, the result convincingly demonstrates that coarse annotations are helpful for all the sizes of formalism-specific training data. As expected, the improvement margin decreases when more formalism-specific data is used. Figure 5 also illustrates a slightly different characteristics of transfer performance between two evaluation metrics. Across all three grammars, we can observe that adding CFG data has a more pronounced effect on the PARSEVAL measure than the dependency F-score. This phenomenon can be explained as follows. The unlabeled PARSEVAL score (Figure 5d-f ) mainly relies on the coarse structural information. On the other hand, predicate-argument dependency Fscore (Figure 5a -c) also relies on the target grammar information. Because that our model only transfers structural information from the source treebank, the gains of PARSEVAL are expected to be larger than that of dependency F-score. Table 4 : The labeled/unlabeled dependency Fscore comparisons between our model and stateof-the-art parsers.",
"cite_spans": [],
"ref_spans": [
{
"start": 3,
"end": 11,
"text": "Figure 4",
"ref_id": null
},
{
"start": 651,
"end": 659,
"text": "Figure 5",
"ref_id": "FIGREF2"
},
{
"start": 784,
"end": 793,
"text": "Figure 5a",
"ref_id": "FIGREF2"
},
{
"start": 1203,
"end": 1211,
"text": "Figure 5",
"ref_id": "FIGREF2"
},
{
"start": 1541,
"end": 1553,
"text": "(Figure 5d-f",
"ref_id": "FIGREF2"
},
{
"start": 1664,
"end": 1674,
"text": "(Figure 5a",
"ref_id": "FIGREF2"
},
{
"start": 1894,
"end": 1901,
"text": "Table 4",
"ref_id": null
}
],
"eq_spans": [],
"section": "Impact of Coarse Annotations on Target Formalism:",
"sec_num": null
},
{
"text": "Comparison to State-of-the-art Parsers: We would also like to demonstrate that the above gains of our transfer model are achieved using an adequate formalism-specific parser. Since our model can be trained exclusively on formalismspecific data, we can compare it to state-of-theart formalism-specific parsers. For this experiment, we choose the C&C parser (Clark and Curran, 2003) for CCG, Enju parser (Miyao and Tsujii, 2008) for HPSG and pipeline automatic annotator with Charniak parser for LFG. For all three parsers, we use the implementation provided by the authors with the default parameter values. All the models are trained on either 1,000 or 15,000 sentences annotated with formalism-specific trees, thus evaluating their performances on small scale or large scale of data.",
"cite_spans": [
{
"start": 356,
"end": 380,
"text": "(Clark and Curran, 2003)",
"ref_id": "BIBREF9"
},
{
"start": 402,
"end": 426,
"text": "(Miyao and Tsujii, 2008)",
"ref_id": "BIBREF26"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Impact of Coarse Annotations on Target Formalism:",
"sec_num": null
},
{
"text": "As Table 4 shows, our model is competitive with all the baselines described above. It's not surprising that Cahill's model outperforms our loglinear model because it relies heavily on handcrafted rules optimized for the dataset.",
"cite_spans": [],
"ref_spans": [
{
"start": 3,
"end": 10,
"text": "Table 4",
"ref_id": null
}
],
"eq_spans": [],
"section": "Impact of Coarse Annotations on Target Formalism:",
"sec_num": null
},
{
"text": "Correspondence between CFG and Target Formalisms: Finally, we analyze highly weighted features. Table 5 shows such features for HPSG; similar patterns are also found for the other grammar formalisms. The first two features are formalism-specific ones, the first for HPSG and the second for CFG. They show that we correctly learn a frequent derivation in the target formalism and CFG. The third one shows an example of a connection between CFG and the target formalism. Our model correctly learns that a syntactic derivation with children VP and NP is very likely to be mapped to the derivation (head comp)\u2192 ([N V N] ",
"cite_spans": [
{
"start": 607,
"end": 615,
"text": "([N V N]",
"ref_id": null
}
],
"ref_spans": [
{
"start": 96,
"end": 103,
"text": "Table 5",
"ref_id": "TABREF9"
}
],
"eq_spans": [],
"section": "Impact of Coarse Annotations on Target Formalism:",
"sec_num": null
},
{
"text": "We present a method for cross-formalism transfer in parsing. Our model utilizes coarse syntactic annotations to supplement a small number of formalism-specific trees for training on constituency-based grammars. Our experimental results show that across a range of such formalisms, the model significantly benefits from the coarse annotations. ",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Conclusions",
"sec_num": "9"
},
{
"text": "The source code for the work is available at http://groups.csail.mit.edu/rbg/code/ grammar/acl2013.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
},
{
"text": "While the Penn Treebank-2 contains richer annotations, we decided to use the Penn Treebank-1 to demonstrate the feasibility of transfer from coarse annotations.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
},
{
"text": "To speed up the implementation, gradient computation is parallelized, using the Message Passing Interface package(Gropp et al., 1999).4 An alternative approach would be to marginalize over yCF G and maximize over yCCG. However, this is a harder computational problem.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
}
],
"back_matter": [
{
"text": "The authors acknowledge the support of the Army Research Office (grant 1130128-258552). We thank Yusuke Miyao, Ozlem Cetinoglu, Stephen Clark, Michael Auli and Yue Zhang for answering questions and sharing the codes of their work. We also thank the members of the MIT NLP group and the ACL reviewers for their suggestions and comments. Any opinions, findings, conclusions, or recommendations expressed in this paper are those of the authors, and do not necessarily reflect the views of the funding organizations.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Acknowledgments",
"sec_num": null
}
],
"bib_entries": {
"BIBREF0": {
"ref_id": "b0",
"title": "Domain adaptation with structural correspondence learning",
"authors": [
{
"first": "John",
"middle": [],
"last": "Blitzer",
"suffix": ""
},
{
"first": "Ryan",
"middle": [],
"last": "Mcdonald",
"suffix": ""
},
{
"first": "Fernando",
"middle": [],
"last": "Pereira",
"suffix": ""
}
],
"year": 2006,
"venue": "Proceedings of the 2006 Conference on Empirical Methods in Natural Language Processing",
"volume": "",
"issue": "",
"pages": "120--128",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "John Blitzer, Ryan McDonald, and Fernando Pereira. 2006. Domain adaptation with structural correspon- dence learning. In Proceedings of the 2006 Con- ference on Empirical Methods in Natural Language Processing, pages 120-128. Association for Com- putational Linguistics.",
"links": null
},
"BIBREF1": {
"ref_id": "b1",
"title": "The mental representation of grammatical relations",
"authors": [
{
"first": "Joan",
"middle": [],
"last": "Bresnan",
"suffix": ""
}
],
"year": 1982,
"venue": "",
"volume": "1",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Joan Bresnan. 1982. The mental representation of grammatical relations, volume 1. The MIT Press.",
"links": null
},
"BIBREF2": {
"ref_id": "b2",
"title": "Parsing with pcfgs and automatic f-structure annotation",
"authors": [
{
"first": "Aoife",
"middle": [],
"last": "Cahill",
"suffix": ""
},
{
"first": "Mairad",
"middle": [],
"last": "Mccarthy",
"suffix": ""
},
{
"first": "Josef",
"middle": [],
"last": "Van Genabith",
"suffix": ""
},
{
"first": "Andy",
"middle": [],
"last": "Way",
"suffix": ""
}
],
"year": 2002,
"venue": "Proceedings of the Seventh International Conference on LFG",
"volume": "",
"issue": "",
"pages": "76--95",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Aoife Cahill, Mairad McCarthy, Josef van Genabith, and Andy Way. 2002. Parsing with pcfgs and au- tomatic f-structure annotation. In Proceedings of the Seventh International Conference on LFG, pages 76-95. CSLI Publications.",
"links": null
},
"BIBREF3": {
"ref_id": "b3",
"title": "Long-distance dependency resolution in automatically acquired wide-coverage pcfg-based lfg approximations",
"authors": [
{
"first": "Aoife",
"middle": [],
"last": "Cahill",
"suffix": ""
},
{
"first": "Michael",
"middle": [],
"last": "Burke",
"suffix": ""
},
{
"first": "O'",
"middle": [],
"last": "Ruth",
"suffix": ""
},
{
"first": "Josef",
"middle": [],
"last": "Donovan",
"suffix": ""
},
{
"first": "Andy",
"middle": [],
"last": "Van Genabith",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Way",
"suffix": ""
}
],
"year": 2004,
"venue": "Proceedings of the 42nd Annual Meeting on Association for Computational Linguistics, page 319. Association for Computational Linguistics",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Aoife Cahill, Michael Burke, Ruth O'Donovan, Josef Van Genabith, and Andy Way. 2004. Long-distance dependency resolution in automatically acquired wide-coverage pcfg-based lfg approximations. In Proceedings of the 42nd Annual Meeting on Associ- ation for Computational Linguistics, page 319. As- sociation for Computational Linguistics.",
"links": null
},
"BIBREF4": {
"ref_id": "b4",
"title": "Parsing with Automatically Acquired, Wide-Coverage, Robust, Probabilistic LFG Approximation",
"authors": [
{
"first": "Aoife",
"middle": [],
"last": "Cahill",
"suffix": ""
}
],
"year": 2004,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Aoife Cahill. 2004. Parsing with Automatically Ac- quired, Wide-Coverage, Robust, Probabilistic LFG Approximation. Ph.D. thesis.",
"links": null
},
"BIBREF5": {
"ref_id": "b5",
"title": "Statistical french dependency parsing: treebank conversion and first results",
"authors": [
{
"first": "Marie",
"middle": [],
"last": "Candito",
"suffix": ""
},
{
"first": "Beno\u00eet",
"middle": [],
"last": "Crabb\u00e9",
"suffix": ""
},
{
"first": "Pascal",
"middle": [],
"last": "Denis",
"suffix": ""
}
],
"year": 2010,
"venue": "Proceedings of the Seventh International Conference on Language Resources and Evaluation (LREC 2010)",
"volume": "",
"issue": "",
"pages": "1840--1847",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Marie Candito, Beno\u00eet Crabb\u00e9, Pascal Denis, et al. 2010. Statistical french dependency parsing: tree- bank conversion and first results. In Proceed- ings of the Seventh International Conference on Language Resources and Evaluation (LREC 2010), pages 1840-1847.",
"links": null
},
"BIBREF6": {
"ref_id": "b6",
"title": "Coarseto-fine n-best parsing and maxent discriminative reranking",
"authors": [
{
"first": "Eugene",
"middle": [],
"last": "Charniak",
"suffix": ""
},
{
"first": "Mark",
"middle": [],
"last": "Johnson",
"suffix": ""
}
],
"year": 2005,
"venue": "Proceedings of the 43rd Annual Meeting on Association for Computational Linguistics",
"volume": "",
"issue": "",
"pages": "173--180",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Eugene Charniak and Mark Johnson. 2005. Coarse- to-fine n-best parsing and maxent discriminative reranking. In Proceedings of the 43rd Annual Meet- ing on Association for Computational Linguistics, pages 173-180. Association for Computational Lin- guistics.",
"links": null
},
"BIBREF7": {
"ref_id": "b7",
"title": "A maximum-entropyinspired parser",
"authors": [
{
"first": "Eugene",
"middle": [],
"last": "Charniak",
"suffix": ""
}
],
"year": 2000,
"venue": "Proceedings of the 1st North American chapter of the Association for Computational Linguistics conference",
"volume": "",
"issue": "",
"pages": "132--139",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Eugene Charniak. 2000. A maximum-entropy- inspired parser. In Proceedings of the 1st North American chapter of the Association for Computa- tional Linguistics conference, pages 132-139.",
"links": null
},
"BIBREF8": {
"ref_id": "b8",
"title": "Automated extraction of tags from the penn treebank. New developments in parsing technology",
"authors": [
{
"first": "John",
"middle": [],
"last": "Chen",
"suffix": ""
},
{
"first": "K",
"middle": [],
"last": "Vijay",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Shanker",
"suffix": ""
}
],
"year": 2005,
"venue": "",
"volume": "",
"issue": "",
"pages": "73--89",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "John Chen and Vijay K Shanker. 2005. Automated extraction of tags from the penn treebank. New de- velopments in parsing technology, pages 73-89.",
"links": null
},
"BIBREF9": {
"ref_id": "b9",
"title": "Log-linear models for wide-coverage ccg parsing",
"authors": [
{
"first": "Stephen",
"middle": [],
"last": "Clark",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "James R Curran",
"suffix": ""
}
],
"year": 2003,
"venue": "Proceedings of the 2003 conference on Empirical methods in natural language processing",
"volume": "",
"issue": "",
"pages": "97--104",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Stephen Clark and James R Curran. 2003. Log-linear models for wide-coverage ccg parsing. In Proceed- ings of the 2003 conference on Empirical methods in natural language processing, pages 97-104. As- sociation for Computational Linguistics.",
"links": null
},
"BIBREF10": {
"ref_id": "b10",
"title": "Widecoverage efficient statistical parsing with ccg and log-linear models",
"authors": [
{
"first": "Stephen",
"middle": [],
"last": "Clark",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "James R Curran",
"suffix": ""
}
],
"year": 2007,
"venue": "Computational Linguistics",
"volume": "33",
"issue": "4",
"pages": "493--552",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Stephen Clark and James R Curran. 2007. Wide- coverage efficient statistical parsing with ccg and log-linear models. Computational Linguistics, 33(4):493-552.",
"links": null
},
"BIBREF11": {
"ref_id": "b11",
"title": "Three generative, lexicalised models for statistical pprsing",
"authors": [
{
"first": "Michael",
"middle": [],
"last": "Collins",
"suffix": ""
}
],
"year": 1997,
"venue": "Proceedings of the eighth conference on European chapter of the Association for Computational Linguistics",
"volume": "",
"issue": "",
"pages": "16--23",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Michael Collins. 1997. Three generative, lexicalised models for statistical pprsing. In Proceedings of the eighth conference on European chapter of the Asso- ciation for Computational Linguistics, pages 16-23. Association for Computational Linguistics.",
"links": null
},
"BIBREF12": {
"ref_id": "b12",
"title": "Head-driven statistical models for natural language parsing",
"authors": [
{
"first": "Michael",
"middle": [],
"last": "Collins",
"suffix": ""
}
],
"year": 2003,
"venue": "Computational linguistics",
"volume": "29",
"issue": "4",
"pages": "589--637",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Michael Collins. 2003. Head-driven statistical mod- els for natural language parsing. Computational lin- guistics, 29(4):589-637.",
"links": null
},
"BIBREF13": {
"ref_id": "b13",
"title": "Frustratingly hard domain adaptation for dependency parsing",
"authors": [
{
"first": "Mark",
"middle": [],
"last": "Dredze",
"suffix": ""
},
{
"first": "John",
"middle": [],
"last": "Blitzer",
"suffix": ""
},
{
"first": "Partha",
"middle": [
"Pratim"
],
"last": "Talukdar",
"suffix": ""
},
{
"first": "Kuzman",
"middle": [],
"last": "Ganchev",
"suffix": ""
},
{
"first": "Joao",
"middle": [
"V"
],
"last": "Gra\u00e7a",
"suffix": ""
},
{
"first": "Fernando",
"middle": [],
"last": "Pereira",
"suffix": ""
}
],
"year": 2007,
"venue": "Proceedings of the CoNLL Shared Task Session of EMNLP-CoNLL",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Mark Dredze, John Blitzer, Partha Pratim Talukdar, Kuzman Ganchev, Joao V Gra\u00e7a, and Fernando Pereira. 2007. Frustratingly hard domain adap- tation for dependency parsing. In Proceedings of the CoNLL Shared Task Session of EMNLP-CoNLL, volume 2007.",
"links": null
},
"BIBREF14": {
"ref_id": "b14",
"title": "Using MPI: portable parallel programming with the message passing interface",
"authors": [
{
"first": "William",
"middle": [],
"last": "Gropp",
"suffix": ""
},
{
"first": "Ewing",
"middle": [],
"last": "Lusk",
"suffix": ""
},
{
"first": "Anthony",
"middle": [],
"last": "Skjellum",
"suffix": ""
}
],
"year": 1999,
"venue": "",
"volume": "1",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "William Gropp, Ewing Lusk, and Anthony Skjellum. 1999. Using MPI: portable parallel programming with the message passing interface, volume 1. MIT press.",
"links": null
},
"BIBREF15": {
"ref_id": "b15",
"title": "Acquiring compact lexicalized grammars from a cleaner treebank",
"authors": [
{
"first": "Julia",
"middle": [],
"last": "Hockenmaier",
"suffix": ""
},
{
"first": "Mark",
"middle": [],
"last": "Steedman",
"suffix": ""
}
],
"year": 2002,
"venue": "Proceedings of the Third LREC Conference",
"volume": "",
"issue": "",
"pages": "1974--1981",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Julia Hockenmaier and Mark Steedman. 2002. Acquiring compact lexicalized grammars from a cleaner treebank. In Proceedings of the Third LREC Conference, pages 1974-1981.",
"links": null
},
"BIBREF16": {
"ref_id": "b16",
"title": "Data and models for statistical parsing with combinatory categorial grammar",
"authors": [
{
"first": "Julia",
"middle": [],
"last": "Hockenmaier",
"suffix": ""
}
],
"year": 2003,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Julia Hockenmaier. 2003. Data and models for statis- tical parsing with combinatory categorial grammar.",
"links": null
},
"BIBREF17": {
"ref_id": "b17",
"title": "Breaking the resource bottleneck for multilingual parsing",
"authors": [
{
"first": "Rebecca",
"middle": [],
"last": "Hwa",
"suffix": ""
},
{
"first": "Philip",
"middle": [],
"last": "Resnik",
"suffix": ""
},
{
"first": "Amy",
"middle": [],
"last": "Weinberg",
"suffix": ""
},
{
"first": "; Dtic",
"middle": [],
"last": "Document",
"suffix": ""
}
],
"year": 2005,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Rebecca Hwa, Philip Resnik, and Amy Weinberg. 2005. Breaking the resource bottleneck for multi- lingual parsing. Technical report, DTIC Document.",
"links": null
},
"BIBREF18": {
"ref_id": "b18",
"title": "Automatic adaptation of annotation standards for dependency parsing: using projected treebank as source corpus",
"authors": [
{
"first": "Wenbin",
"middle": [],
"last": "Jiang",
"suffix": ""
},
{
"first": "Qun",
"middle": [],
"last": "Liu",
"suffix": ""
}
],
"year": 2009,
"venue": "Proceedings of the 11th International Conference on Parsing Technologies",
"volume": "",
"issue": "",
"pages": "25--28",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Wenbin Jiang and Qun Liu. 2009. Automatic adap- tation of annotation standards for dependency pars- ing: using projected treebank as source corpus. In Proceedings of the 11th International Conference on Parsing Technologies, pages 25-28. Association for Computational Linguistics.",
"links": null
},
"BIBREF19": {
"ref_id": "b19",
"title": "Speed and accuracy in shallow and deep stochastic parsing",
"authors": [
{
"first": "Ronald",
"middle": [
"M"
],
"last": "Kaplan",
"suffix": ""
},
{
"first": "Stefan",
"middle": [],
"last": "Riezler",
"suffix": ""
},
{
"first": "Tracy",
"middle": [
"H"
],
"last": "King",
"suffix": ""
},
{
"first": "John",
"middle": [
"T"
],
"last": "Maxwell",
"suffix": ""
},
{
"first": "Iii",
"middle": [],
"last": "",
"suffix": ""
},
{
"first": "Alexander",
"middle": [],
"last": "Vasserman",
"suffix": ""
},
{
"first": "Richard",
"middle": [],
"last": "Crouch",
"suffix": ""
}
],
"year": 2004,
"venue": "Proceedings of NAACL",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Ronald M. Kaplan, Stefan Riezler, Tracy H. King, John T. Maxwell III, Alexander Vasserman, and Richard Crouch. 2004. Speed and accuracy in shallow and deep stochastic parsing. In Proceedings of NAACL.",
"links": null
},
"BIBREF20": {
"ref_id": "b20",
"title": "The parc 700 dependency bank",
"authors": [
{
"first": "Tracy Holloway",
"middle": [],
"last": "King",
"suffix": ""
},
{
"first": "Richard",
"middle": [],
"last": "Crouch",
"suffix": ""
},
{
"first": "Stefan",
"middle": [],
"last": "Riezler",
"suffix": ""
},
{
"first": "Mary",
"middle": [],
"last": "Dalrymple",
"suffix": ""
},
{
"first": "Ronald M",
"middle": [],
"last": "Kaplan",
"suffix": ""
}
],
"year": 2003,
"venue": "Proceedings of the EACL03: 4th International Workshop on Linguistically Interpreted Corpora (LINC-03)",
"volume": "",
"issue": "",
"pages": "1--8",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Tracy Holloway King, Richard Crouch, Stefan Riezler, Mary Dalrymple, and Ronald M Kaplan. 2003. The parc 700 dependency bank. In Proceedings of the EACL03: 4th International Workshop on Linguisti- cally Interpreted Corpora (LINC-03), pages 1-8.",
"links": null
},
"BIBREF21": {
"ref_id": "b21",
"title": "Building a large annotated corpus of english: The penn treebank",
"authors": [
{
"first": "P",
"middle": [],
"last": "Mitchell",
"suffix": ""
},
{
"first": "Mary",
"middle": [
"Ann"
],
"last": "Marcus",
"suffix": ""
},
{
"first": "Beatrice",
"middle": [],
"last": "Marcinkiewicz",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Santorini",
"suffix": ""
}
],
"year": 1993,
"venue": "Computational linguistics",
"volume": "19",
"issue": "2",
"pages": "313--330",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Mitchell P Marcus, Mary Ann Marcinkiewicz, and Beatrice Santorini. 1993. Building a large anno- tated corpus of english: The penn treebank. Compu- tational linguistics, 19(2):313-330.",
"links": null
},
"BIBREF22": {
"ref_id": "b22",
"title": "Automatic domain adaptation for parsing",
"authors": [
{
"first": "David",
"middle": [],
"last": "Mcclosky",
"suffix": ""
},
{
"first": "Eugene",
"middle": [],
"last": "Charniak",
"suffix": ""
},
{
"first": "Mark",
"middle": [],
"last": "Johnson",
"suffix": ""
}
],
"year": 2010,
"venue": "Human Language Technologies: The",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "David McClosky, Eugene Charniak, and Mark John- son. 2010. Automatic domain adaptation for pars- ing. In Human Language Technologies: The 2010",
"links": null
},
"BIBREF23": {
"ref_id": "b23",
"title": "Annual Conference of the North American Chapter of the Association for Computational Linguistics",
"authors": [],
"year": null,
"venue": "",
"volume": "",
"issue": "",
"pages": "28--36",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Annual Conference of the North American Chap- ter of the Association for Computational Linguistics, pages 28-36. Association for Computational Lin- guistics.",
"links": null
},
"BIBREF24": {
"ref_id": "b24",
"title": "Multilingual dependency analysis with a twostage discriminative parser",
"authors": [
{
"first": "Ryan",
"middle": [],
"last": "Mcdonald",
"suffix": ""
},
{
"first": "Kevin",
"middle": [],
"last": "Lerman",
"suffix": ""
},
{
"first": "Fernando",
"middle": [],
"last": "Pereira",
"suffix": ""
}
],
"year": 2006,
"venue": "Proceedings of the Tenth Conference on Computational Natural Language Learning",
"volume": "",
"issue": "",
"pages": "216--220",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Ryan McDonald, Kevin Lerman, and Fernando Pereira. 2006. Multilingual dependency analysis with a two- stage discriminative parser. In Proceedings of the Tenth Conference on Computational Natural Lan- guage Learning, pages 216-220. Association for Computational Linguistics.",
"links": null
},
"BIBREF25": {
"ref_id": "b25",
"title": "Multi-source transfer of delexicalized dependency parsers",
"authors": [
{
"first": "Ryan",
"middle": [],
"last": "Mcdonald",
"suffix": ""
},
{
"first": "Slav",
"middle": [],
"last": "Petrov",
"suffix": ""
},
{
"first": "Keith",
"middle": [],
"last": "Hall",
"suffix": ""
}
],
"year": 2011,
"venue": "Proceedings of the Conference on Empirical Methods in Natural Language Processing",
"volume": "",
"issue": "",
"pages": "62--72",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Ryan McDonald, Slav Petrov, and Keith Hall. 2011. Multi-source transfer of delexicalized dependency parsers. In Proceedings of the Conference on Em- pirical Methods in Natural Language Processing, pages 62-72. Association for Computational Lin- guistics.",
"links": null
},
"BIBREF26": {
"ref_id": "b26",
"title": "Feature forest models for probabilistic hpsg parsing",
"authors": [
{
"first": "Yusuke",
"middle": [],
"last": "Miyao",
"suffix": ""
},
{
"first": "Jun'ichi",
"middle": [],
"last": "Tsujii",
"suffix": ""
}
],
"year": 2008,
"venue": "Computational Linguistics",
"volume": "34",
"issue": "1",
"pages": "35--80",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Yusuke Miyao and Jun'ichi Tsujii. 2008. Feature for- est models for probabilistic hpsg parsing. Computa- tional Linguistics, 34(1):35-80.",
"links": null
},
"BIBREF27": {
"ref_id": "b27",
"title": "Corpus-oriented grammar development for acquiring a head-driven phrase structure grammar from the penn treebank",
"authors": [
{
"first": "Yusuke",
"middle": [],
"last": "Miyao",
"suffix": ""
},
{
"first": "Takashi",
"middle": [],
"last": "Ninomiya",
"suffix": ""
},
{
"first": "Junichi",
"middle": [],
"last": "Tsujii",
"suffix": ""
}
],
"year": 2005,
"venue": "Natural Language Processing-IJCNLP 2004",
"volume": "",
"issue": "",
"pages": "684--693",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Yusuke Miyao, Takashi Ninomiya, and Junichi Tsu- jii. 2005. Corpus-oriented grammar development for acquiring a head-driven phrase structure gram- mar from the penn treebank. Natural Language Processing-IJCNLP 2004, pages 684-693.",
"links": null
},
"BIBREF28": {
"ref_id": "b28",
"title": "From Linguistic Theory to Syntactic Analysis: Corpus-Oriented Grammar Development and Feature Forest Model",
"authors": [
{
"first": "Yusuke",
"middle": [],
"last": "Miyao",
"suffix": ""
}
],
"year": 2006,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Yusuke Miyao. 2006. From Linguistic Theory to Syn- tactic Analysis: Corpus-Oriented Grammar Devel- opment and Feature Forest Model. Ph.D. thesis.",
"links": null
},
"BIBREF29": {
"ref_id": "b29",
"title": "Numerical optimization",
"authors": [
{
"first": "Jorge",
"middle": [],
"last": "Nocedal",
"suffix": ""
},
{
"first": "J",
"middle": [],
"last": "Stephen",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Wright",
"suffix": ""
}
],
"year": 1999,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Jorge Nocedal and Stephen J Wright. 1999. Numerical optimization. Springer verlag.",
"links": null
},
"BIBREF30": {
"ref_id": "b30",
"title": "Towards holistic grammar engineering and testing-grafting treebank maintenance into the grammar revision cycle",
"authors": [
{
"first": "Stephan",
"middle": [],
"last": "Oepen",
"suffix": ""
},
{
"first": "Dan",
"middle": [],
"last": "Flickinger",
"suffix": ""
},
{
"first": "Francis",
"middle": [],
"last": "Bond",
"suffix": ""
}
],
"year": 2004,
"venue": "Proceedings of the IJC-NLP workshop beyond shallow analysis",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Stephan Oepen, Dan Flickinger, and Francis Bond. 2004. Towards holistic grammar engineering and testing-grafting treebank maintenance into the grammar revision cycle. In Proceedings of the IJC- NLP workshop beyond shallow analysis. Citeseer.",
"links": null
},
"BIBREF31": {
"ref_id": "b31",
"title": "Improved inference for unlexicalized parsing",
"authors": [
{
"first": "Slav",
"middle": [],
"last": "Petrov",
"suffix": ""
},
{
"first": "Dan",
"middle": [],
"last": "Klein",
"suffix": ""
}
],
"year": 2007,
"venue": "Human Language Technologies 2007: The Conference of the North American Chapter of the Association for Computational Linguistics",
"volume": "",
"issue": "",
"pages": "404--411",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Slav Petrov and Dan Klein. 2007. Improved infer- ence for unlexicalized parsing. In Human Language Technologies 2007: The Conference of the North American Chapter of the Association for Computa- tional Linguistics, pages 404-411.",
"links": null
},
"BIBREF32": {
"ref_id": "b32",
"title": "Head-driven phrase structure grammar",
"authors": [
{
"first": "Carl",
"middle": [],
"last": "Pollard",
"suffix": ""
},
{
"first": "A",
"middle": [],
"last": "Ivan",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Sag",
"suffix": ""
}
],
"year": 1994,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Carl Pollard and Ivan A Sag. 1994. Head-driven phrase structure grammar. University of Chicago Press.",
"links": null
},
"BIBREF33": {
"ref_id": "b33",
"title": "Parsing the wall street journal using a lexical-functional grammar and discriminative estimation techniques",
"authors": [
{
"first": "Stefan",
"middle": [],
"last": "Riezler",
"suffix": ""
},
{
"first": "H",
"middle": [],
"last": "Tracy",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "King",
"suffix": ""
},
{
"first": "M",
"middle": [],
"last": "Ronald",
"suffix": ""
},
{
"first": "Richard",
"middle": [],
"last": "Kaplan",
"suffix": ""
},
{
"first": "John",
"middle": [
"T"
],
"last": "Crouch",
"suffix": ""
},
{
"first": "Iii",
"middle": [],
"last": "Maxwell",
"suffix": ""
},
{
"first": "Mark",
"middle": [],
"last": "Johnson",
"suffix": ""
}
],
"year": 2002,
"venue": "Proceedings of the 40th Annual Meeting on Association for Computational Linguistics",
"volume": "",
"issue": "",
"pages": "271--278",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Stefan Riezler, Tracy H King, Ronald M Kaplan, Richard Crouch, John T Maxwell III, and Mark Johnson. 2002. Parsing the wall street journal us- ing a lexical-functional grammar and discriminative estimation techniques. In Proceedings of the 40th Annual Meeting on Association for Computational Linguistics, pages 271-278. Association for Com- putational Linguistics.",
"links": null
},
"BIBREF34": {
"ref_id": "b34",
"title": "Unsupervised multilingual grammar induction",
"authors": [
{
"first": "Benjamin",
"middle": [],
"last": "Snyder",
"suffix": ""
},
{
"first": "Tahira",
"middle": [],
"last": "Naseem",
"suffix": ""
},
{
"first": "Regina",
"middle": [],
"last": "Barzilay",
"suffix": ""
}
],
"year": 2009,
"venue": "Proceedings of the Joint Conference of the 47th Annual Meeting of the ACL and the",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Benjamin Snyder, Tahira Naseem, and Regina Barzi- lay. 2009. Unsupervised multilingual grammar in- duction. In Proceedings of the Joint Conference of the 47th Annual Meeting of the ACL and the",
"links": null
}
},
"ref_entries": {
"FIGREF0": {
"uris": null,
"text": "r , r, c, hp l,r , r, c, hl l,r fjoint r, d, c, sy l,r , hl l,r , r, d, c, sy l,r r, c, sp l,r , sy l,r , hl l,r , r, c, sp l,r , sy l,r",
"type_str": "figure",
"num": null
},
"FIGREF1": {
"uris": null,
"text": "f CFG(y, S) :f CFG (y, S) : f CCG (y, S) : f joint (y, S) :Example of transfer between CFG and CCG formalisms.",
"type_str": "figure",
"num": null
},
"FIGREF2": {
"uris": null,
"text": "Model performance with different target formalism trees and zero or 15,000 CFG trees. The first row shows the results of labeled dependency F-score and the second row shows the results of unlabeled PARSEVAL.",
"type_str": "figure",
"num": null
},
"TABREF1": {
"html": null,
"type_str": "table",
"text": "\\NP (S[dcl]\\NP)/NP,NP joint feature on y CFG , y CCG VP, S[dcl]\\NP (VP, (S[dcl]\\NP)/NP), (NP, NP)",
"content": "<table><tr><td>VP S[dcl]\\NP</td><td>coarse feature on y CFG VP VP,NP</td></tr><tr><td/><td>formalism feature on y CCG</td></tr><tr><td/><td>S[dcl]</td></tr><tr><td>VP (S[dcl]\\NP)/NP</td><td>NP NP</td></tr><tr><td>eat</td><td>apples</td></tr></table>",
"num": null
},
"TABREF2": {
"html": null,
"type_str": "table",
"text": "Templates of atomic features.",
"content": "<table/>",
"num": null
},
"TABREF3": {
"html": null,
"type_str": "table",
"text": "Binary feature templates used in f (y, S). Unary and root features follow a similar pattern.",
"content": "<table/>",
"num": null
},
"TABREF5": {
"html": null,
"type_str": "table",
"text": "Training/Dev./Test split on WSJ sections and PARC700 for different grammar formalisms.",
"content": "<table/>",
"num": null
},
"TABREF8": {
"html": null,
"type_str": "table",
"text": ",[N.3sg]) in HPSG. \u2192 (sy l , hp l )(syr, hpr) Examples (VP)\u2192(VP,VB)(NP,NN) Joint features Template (r) \u2192 (hl l , sy l )(ler, syr)",
"content": "<table><tr><td colspan=\"2\">Feature type Features with high weight Template</td></tr><tr><td>Target formalism Coarse formalism</td><td>(r) \u2192 (hl l , hp l )(hlr, pr) Examples (head comp)\u2192 ([N V N],VB)([N.3sg],NN) Template (syp) Examples (head comp)\u2192 ([N V N],VP)([N.3sg],NP)</td></tr></table>",
"num": null
},
"TABREF9": {
"html": null,
"type_str": "table",
"text": "Example features with high weight.",
"content": "<table/>",
"num": null
},
"TABREF10": {
"html": null,
"type_str": "table",
"text": "4th International Joint Conference on Natural Language Processing of the AFNLP: Volume 1-Volume 1, pages 73-81. Association for Computational Linguistics. Mark Steedman. 2001. The syntactic process. MIT press. Yue Zhang, Stephen Clark, et al. 2011. Shift-reduce ccg parsing. In Proceedings of the 49th Meeting of the Association for Computational Linguistics, pages 683-692.",
"content": "<table/>",
"num": null
}
}
}
} |