File size: 116,982 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 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 |
{
"paper_id": "2021",
"header": {
"generated_with": "S2ORC 1.0.0",
"date_generated": "2023-01-19T02:11:04.682896Z"
},
"title": "Few-Shot Learning of an Interleaved Text Summarization Model by Pretraining with Synthetic Data",
"authors": [
{
"first": "Sanjeev",
"middle": [
"Kumar"
],
"last": "Karn",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "LMU Munich",
"location": {}
},
"email": "skarn@cis.lmu.de"
},
{
"first": "Francine",
"middle": [],
"last": "Chen",
"suffix": "",
"affiliation": {},
"email": "francine@acm.orgyan-ying.chen@tri.global"
},
{
"first": "Yan-Ying",
"middle": [],
"last": "Chen",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "Toyota Research Institute",
"location": {
"addrLine": "California 4 Machine Intelligence",
"settlement": "Los Altos"
}
},
"email": ""
},
{
"first": "Ulli",
"middle": [],
"last": "Waltinger",
"suffix": "",
"affiliation": {},
"email": "ulli.waltinger@siemens.com"
},
{
"first": "Hinrich",
"middle": [],
"last": "Sch\u00fctze",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "LMU Munich",
"location": {}
},
"email": ""
}
],
"year": "",
"venue": null,
"identifiers": {},
"abstract": "Interleaved texts, where posts belonging to different threads occur in a sequence, commonly occur in online chat posts, so that it can be time-consuming to quickly obtain an overview of the discussions. Existing systems first disentangle the posts by threads and then extract summaries from those threads. A major issue with such systems is error propagation from the disentanglement component. While endto-end trainable summarization system could obviate explicit disentanglement, such systems require a large amount of labeled data. To address this, we propose to pretrain an endto-end trainable hierarchical encoder-decoder system using synthetic interleaved texts. We show that by fine-tuning on a real-world meeting dataset (AMI), such a system out-performs a traditional two-step system by 22%. We also compare against transformer models and observed that pretraining with synthetic data both the encoder and decoder outperforms the BertSumExtAbs transformer model which pretrains only the encoder on a large dataset.",
"pdf_parse": {
"paper_id": "2021",
"_pdf_hash": "",
"abstract": [
{
"text": "Interleaved texts, where posts belonging to different threads occur in a sequence, commonly occur in online chat posts, so that it can be time-consuming to quickly obtain an overview of the discussions. Existing systems first disentangle the posts by threads and then extract summaries from those threads. A major issue with such systems is error propagation from the disentanglement component. While endto-end trainable summarization system could obviate explicit disentanglement, such systems require a large amount of labeled data. To address this, we propose to pretrain an endto-end trainable hierarchical encoder-decoder system using synthetic interleaved texts. We show that by fine-tuning on a real-world meeting dataset (AMI), such a system out-performs a traditional two-step system by 22%. We also compare against transformer models and observed that pretraining with synthetic data both the encoder and decoder outperforms the BertSumExtAbs transformer model which pretrains only the encoder on a large dataset.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Abstract",
"sec_num": null
}
],
"body_text": [
{
"text": "Interleaved texts are increasingly common, occurring in social media conversations such as Slack and Stack Exchange, where posts belonging to different threads may be intermixed in the post sequence; see a meeting transcript from the AMI corpus (McCowan et al., 2005) in Table 1 . Due to the mixing, getting a quick sense of the different conversational threads is often difficult.",
"cite_spans": [
{
"start": 245,
"end": 267,
"text": "(McCowan et al., 2005)",
"ref_id": "BIBREF17"
}
],
"ref_spans": [
{
"start": 271,
"end": 278,
"text": "Table 1",
"ref_id": null
}
],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "In conversation disentanglement, interleaved posts are grouped by the thread. However, a reader still has to read all posts in each cluster of threads to get the gist. Shang et al. (2018) proposed a two-step system that takes an interleaved text as input and first disentangles the posts thread-wise by clustering, and then compresses the thread-wise posts to single-sentence summaries. However, disentanglement e.g., Wang and Oard (2009) , propagates error to the downstream summarization task. An end-to-end supervised summarization system that implicitly identifies the conversations would eliminate error propagation. However, labeling of interleaved texts is a difficult and expensive task Verberne et al., 2018) . AMI Utteracnes . . . Who is gonna do a PowerPoint presentation ? Think we all Huh. You will . . . . . . \u03be and uh the sender will send to the telly itself an infrared signal to tell it to switch on or switch. . . . . . \u03b6 so y so it's so it's so you got so that's something we should have a look into then i when desi when designing the ergonomics of see have a look . . . . . . \u03c8 ,the little tiny weeny batteries, all like special longlasting batteries. . .",
"cite_spans": [
{
"start": 168,
"end": 187,
"text": "Shang et al. (2018)",
"ref_id": "BIBREF22"
},
{
"start": 418,
"end": 438,
"text": "Wang and Oard (2009)",
"ref_id": "BIBREF26"
},
{
"start": 695,
"end": 717,
"text": "Verberne et al., 2018)",
"ref_id": "BIBREF25"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": ". . . Summary 1) the project manager had the team members re-introduce . . . 2) the industrial designer discussed the interior workings of a remote and the team discussed options for batteries and infra-red signals.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": ". . . 5) the marketing expert presented research on consumer preferences on remotes in general and on voice recognition and the team discussed the option to have an ergonomically designed remote.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": ". . . Table 1 : The top section shows AMI ASR transcripts and the bottom section shows human-written summaries. \u03be=150 th , \u03b6=522 th and \u03c8=570 th utterances. a) refer to the a th sentence in a multi-sentence summary.",
"cite_spans": [],
"ref_spans": [
{
"start": 6,
"end": 13,
"text": "Table 1",
"ref_id": null
}
],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "We propose a pretraining approach to tackle these issues. We synthesized a corpus of interleaved text-summary pairs out of a corpus of regular document-summary pairs and train an end-toend trainable encoder-decoder system. To generate the summary the model learns to infer (disentangle) the major topics in several threads. We show on synthetic and real-world data that the encoderdecoder system not only obviates a disentanglement component but also enhances performance. Thus, the summarization task acts as an auxiliary task for the disentanglement. Additionally, we show that fine-tuning of the encoder-decode system with the learned disentanglement representations on a real-world AMI dataset achieves a substantial increment in evaluation metrics despite a small number of labels.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "We also propose using a hierarchical attention in the encoder-decoder system with three levels of information from the interleaved text; posts, phrases, and words, rather than traditional two levels; post and word (Nallapati et al., 2017 (Nallapati et al., , 2016 Tan et al., 2017; Cheng and Lapata, 2016) .",
"cite_spans": [
{
"start": 214,
"end": 237,
"text": "(Nallapati et al., 2017",
"ref_id": "BIBREF18"
},
{
"start": 238,
"end": 263,
"text": "(Nallapati et al., , 2016",
"ref_id": "BIBREF19"
},
{
"start": 264,
"end": 281,
"text": "Tan et al., 2017;",
"ref_id": "BIBREF24"
},
{
"start": 282,
"end": 305,
"text": "Cheng and Lapata, 2016)",
"ref_id": "BIBREF5"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "The remaining paper is structured as follows. In Section 2, we discuss related work. In Section 3, we provide a detailed description of our hierarchical seq2seq model. In Section 4, we provide a detailed description on the synthetic data creation algorithm. In Section 5, we describe and discuss the experiments. And in Section 6, we present our conclusions. (2018) each designed a system that summarizes posts in multi-party conversations in order to provide readers with overview on the discussed matters. They broadly follow the same two-step approach: cluster the posts and then extract a summary from each cluster.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "There are two kinds of summarization: abstractive and extractive. In abstractive summarization, the model utilizes a corpus level vocabulary and generates novel sentences as the summary, while extractive models extract or rearrange the source words as the summary. Abstractive models based on neural sequence-to-sequence (seq2seq) (Rush et al., 2015) proved to generate summaries with higher ROUGE scores than the feature-based abstractive models. Li et al. (2015) proposed an encoder-decoder (auto-encoder) model that utilizes a hierarchy of networks: word-to-word followed by sentence-tosentence. Their model is better at capturing the underlying structure than a vanilla sequential encoderdecoder model (seq2seq). Krause et al. (2017) and Jing et al. (2018) showed multi-sentence captioning of an image through a hierarchical Recurrent Neural Network (RNN), topic-to-topic followed by word-to-word, is better than seq2seq. These works suggest a hierarchical decoder, thread-tothread followed by word-to-word, may intrinsically disentangle the posts, and therefore, generate more appropriate summaries.",
"cite_spans": [
{
"start": 331,
"end": 350,
"text": "(Rush et al., 2015)",
"ref_id": "BIBREF20"
},
{
"start": 448,
"end": 464,
"text": "Li et al. (2015)",
"ref_id": "BIBREF13"
},
{
"start": 717,
"end": 737,
"text": "Krause et al. (2017)",
"ref_id": "BIBREF11"
},
{
"start": 742,
"end": 760,
"text": "Jing et al. (2018)",
"ref_id": "BIBREF9"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Related Work",
"sec_num": "2"
},
{
"text": "Integration of attention into a seq2seq model (Bahdanau et al., 2014) led to further advancement of abstractive summarization (Nallapati et al., 2016; Chopra et al., 2016) . Nallapati et al. (2016) devised a hierarchical attention mechanism for a seq2seq model, where two levels of attention distributions over the source, i.e., sentence and word, are computed at every step of the word decoding. Based on the sentence attentions, the word attentions are rescaled. Our hierarchical attention is more intuitive, computes post(sentence)-level and phraselevel attentions for every new summary sentence, and is trained end-to-end.",
"cite_spans": [
{
"start": 46,
"end": 69,
"text": "(Bahdanau et al., 2014)",
"ref_id": "BIBREF2"
},
{
"start": 126,
"end": 150,
"text": "(Nallapati et al., 2016;",
"ref_id": "BIBREF19"
},
{
"start": 151,
"end": 171,
"text": "Chopra et al., 2016)",
"ref_id": "BIBREF6"
},
{
"start": 174,
"end": 197,
"text": "Nallapati et al. (2016)",
"ref_id": "BIBREF19"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Related Work",
"sec_num": "2"
},
{
"text": "Semi-supervised learning has recently gained popularity as it helps training parameters of large models without any training data. Researchers have pre-trained masked language models, in which only an encoder is used to reconstruct the text, e.g., BERT (Devlin et al., 2018) . Liu and Lapata (2019) used BERT as seq2seq encoder and showed improved performance on several abstractive summarization tasks. Similarly, researchers have published pre-trained seq2seq models using a different semi-supervised learning technique, where a seq2seq model is learned to reconstruct the original text, e.g., BART (Lewis et al., 2019) and MASS (Song et al., 2019) . In this work, we rely on transfer learning and demonstrate that by pretraining with appropriate interleaved text data, a seq2seq model readily transfers to a new domain with just a few examples.",
"cite_spans": [
{
"start": 253,
"end": 274,
"text": "(Devlin et al., 2018)",
"ref_id": "BIBREF8"
},
{
"start": 277,
"end": 298,
"text": "Liu and Lapata (2019)",
"ref_id": "BIBREF15"
},
{
"start": 601,
"end": 621,
"text": "(Lewis et al., 2019)",
"ref_id": "BIBREF12"
},
{
"start": 631,
"end": 650,
"text": "(Song et al., 2019)",
"ref_id": "BIBREF23"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Related Work",
"sec_num": "2"
},
{
"text": "Our hierarchical encoder (see Figure 1 left hand section) is based on Nallapati et al. (2017) , where word-to-word and post-to-post encoders are bidirectional LSTMs. The word-to-word BiLSTM encoder (E w2w ) runs over word embeddings of post P i and generates a set of hidden representations, h",
"cite_spans": [
{
"start": 70,
"end": 93,
"text": "Nallapati et al. (2017)",
"ref_id": "BIBREF18"
}
],
"ref_spans": [
{
"start": 30,
"end": 38,
"text": "Figure 1",
"ref_id": "FIGREF1"
}
],
"eq_spans": [],
"section": "Model",
"sec_num": "3"
},
{
"text": "E w2w i,0 , . . . , h E w2w i,p , of d dimensions.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Model",
"sec_num": "3"
},
{
"text": "The average pooled value of the word-to-word representations of post",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Model",
"sec_num": "3"
},
{
"text": "P i ( 1 p p j=0 h E w2w i,j",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Model",
"sec_num": "3"
},
{
"text": ") is input to the post-to-post BiLSTM encoder (E t2t ), which then generates a set of representations, h",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Model",
"sec_num": "3"
},
{
"text": "E p2p 0 , . . . , h E p2p n",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Model",
"sec_num": "3"
},
{
"text": ", corresponding to the posts. i.e., word-to-word (E w2w ) followed by post-to-post (E p2p ). On the right, summaries are generated hierarchically, thread-to-thread (D t2t ) followed by word-to-word (D t2t ).",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Model",
"sec_num": "3"
},
{
"text": "Overall, for a given channel C , output representations of word-to-word, W, and post-to-post, P, has n\u00d7p\u00d72d and n\u00d72d dimensions respectively. The hierarchical decoder has two uni-directional LSTM decoders, thread-to-thread and word-to-word (see right-hand side in Figure 1 ).",
"cite_spans": [],
"ref_spans": [
{
"start": 264,
"end": 272,
"text": "Figure 1",
"ref_id": "FIGREF1"
}
],
"eq_spans": [],
"section": "Model",
"sec_num": "3"
},
{
"text": "At step k of thread decoder (D t2t ), we compute elements of post-level attention as",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Model",
"sec_num": "3"
},
{
"text": "\u03b3 k i = \u03c3(attn \u03b3 (h D t2t k\u22121 , P i ) i \u2208 {1, .",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Model",
"sec_num": "3"
},
{
"text": ". . , n}, where attn \u03b3 aligns the current thread decoder state vector h D t2t k\u22121 to vectors of matrix P i . A phrase is a short sequences of words in a sentence/post. Phrases in interleaved texts are equivalent to visual patterns in images, and therefore, attending phrases are more relevant for thread recognition than attending posts. Thus, we have phrase-level attentions focusing on words in a channel and with a responsibility of disentangling threads. At step k of thread decoder, we also compute a sequence of attention weights,",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Model",
"sec_num": "3"
},
{
"text": "\u03b2 k = \u03b2 k 0,0 , . . . , \u03b2 k n,p , corresponding to the set of encoded word representations, h w2w 0,0 , . . . , h w2w n,p , as \u03b2 k i,j = \u03c3(attn \u03b2 (h D t2t k\u22121 , a i,j )) where a i,j = add(W i,j , P i ), i \u2208 {1, . . . , n}, j \u2208 {1, . . . , p}.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Model",
"sec_num": "3"
},
{
"text": "add aligns a post representation to its word representations and does element-wise addition, and attn \u03b2 maps the current thread decoder state h D t2t k\u22121 and vector a i,j to a scalar value. Then, we use the post-level attention, \u03b3 k , to rescale the sequence of attention weights \u03b2 k to obtain phrase-level atten-",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Model",
"sec_num": "3"
},
{
"text": "tions\u03b2 k as\u03b2 k i,j = \u03b2 k i,j * \u03b3 k i . A weighted representation of the words (crossed blue circle), n i=1 p j=1\u03b2 k i,j W",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Model",
"sec_num": "3"
},
{
"text": "ij , is used as an input to compute the next state of the thread-tothread decoder, D t2t . Additionally, we also use the last hidden state h D w2w k\u22121,q of the word-to-word decoder LSTM (D w2w ) of the previously generated summary sentence as the second input to D t2t .",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Model",
"sec_num": "3"
},
{
"text": "The motivation is to provide information about the previous sentence.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Model",
"sec_num": "3"
},
{
"text": "The current state h D t2t k is passed through a single layer feedforward network and a distribution over STOP=1 and CONTINUE=0 is computed:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Model",
"sec_num": "3"
},
{
"text": "p ST OP k = \u03c3(g h D t2t k )",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Model",
"sec_num": "3"
},
{
"text": ", where g is a feedforward network. In Figure 1 , the process is depicted by a yellow circle. The thread-to-thread decoder keeps decoding until p ST OP k is greater than 0.5.",
"cite_spans": [],
"ref_spans": [
{
"start": 39,
"end": 47,
"text": "Figure 1",
"ref_id": "FIGREF1"
}
],
"eq_spans": [],
"section": "Model",
"sec_num": "3"
},
{
"text": "Additionally, the new state h D t2t k and inputs to D t2t at that step are passed through a two-layer feedforward network, r, followed by a dropout layer to compute the thread representation s k . Given a thread representation s k , the word-toword decoder, a unidirectional attentional LSTM (D w2w ), generates a summary for the thread; see the right-hand side of Figure 1 . Our word-to-word decoder is based on Bahdanau et al. (2014) .",
"cite_spans": [
{
"start": 413,
"end": 435,
"text": "Bahdanau et al. (2014)",
"ref_id": "BIBREF2"
}
],
"ref_spans": [
{
"start": 365,
"end": 373,
"text": "Figure 1",
"ref_id": "FIGREF1"
}
],
"eq_spans": [],
"section": "Model",
"sec_num": "3"
},
{
"text": "At step l of word-to-word decoding of summary of thread k, we compute elements of word level attention, i.e., \u03b1 k,l i,\u2022 ; we refer to Bahdanau et al. (2014) for further details on it. However, we use phrase-level word attentions for rescaling the word level attention as\u03b1 k,l i,j = norm(\u03b2 k i,j \u00d7 \u03b1 k,l ij ), where norm (softmax) renormalizes the values. Thus, contrary to popular two-level hierarchical attention (Nallapati et al., 2016; Cheng and Lapata, 2016; Tan et al., 2017) , we have three levels of hierarchical attention and each with its responsibility and is coordinated through the rescaling operation. ",
"cite_spans": [
{
"start": 414,
"end": 438,
"text": "(Nallapati et al., 2016;",
"ref_id": "BIBREF19"
},
{
"start": 439,
"end": 462,
"text": "Cheng and Lapata, 2016;",
"ref_id": "BIBREF5"
},
{
"start": 463,
"end": 480,
"text": "Tan et al., 2017)",
"ref_id": "BIBREF24"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Model",
"sec_num": "3"
},
{
"text": "m k=1 q l=1 log p \u03b8 y k,l |w k,\u2022<l , W + \u03bb m k=1 y ST OP k log(p ST OP k )",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Model",
"sec_num": "3"
},
{
"text": ", where w k,0 , . . . , w k,q and y k,0 , . . . , y k,q are words in a ground-truth summary and D w2w generation respectively.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Model",
"sec_num": "3"
},
{
"text": "Obtaining labeled training data for interleaved conversation summarization is challenging. The available ones are either extractive Verberne et al. (2018) or too small Anguera et al., 2012) to train a neural model and thoroughly verify the architecture. To get around this issue, we synthesized a dataset by utilizing a corpus of conventional texts for which summaries are available. We created a corpus of interleaved texts from the abstracts and titles of articles from the PubMed corpus (Dernoncourt and Lee, 2017) . We chose PubMed abstracts as it has, in contrast to other corpora such as news articles or StackOverflow posts, a single-sentence summary that can only be comprehended out of a whole abstract. Further, the number of sentences more closely resembles that of a conversationalist in a conversation.",
"cite_spans": [
{
"start": 132,
"end": 154,
"text": "Verberne et al. (2018)",
"ref_id": "BIBREF25"
},
{
"start": 168,
"end": 189,
"text": "Anguera et al., 2012)",
"ref_id": "BIBREF1"
},
{
"start": 490,
"end": 517,
"text": "(Dernoncourt and Lee, 2017)",
"ref_id": "BIBREF7"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Synthetic Dataset",
"sec_num": "4"
},
{
"text": "Algorithm 1 Interleaving Algorithm 1:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Synthetic Dataset",
"sec_num": "4"
},
{
"text": "procedure INTERLEAVE(C, a, b, m, n) 2: O, Z \u2190 WINDOW(C, w, t), Array() 3: while O = \u2205 do 4: E, I , M , S \u2190 O.NEXT(), Array(), Array(), {} 5:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Synthetic Dataset",
"sec_num": "4"
},
{
"text": "r \u223c U(a, b) 6:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Synthetic Dataset",
"sec_num": "4"
},
{
"text": "for j = 1 to r do Selection 7:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Synthetic Dataset",
"sec_num": "4"
},
{
"text": "A, T \u2190 E[j] 8:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Synthetic Dataset",
"sec_num": "4"
},
{
"text": "q \u223c U(m, n) 9:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Synthetic Dataset",
"sec_num": "4"
},
{
"text": "I .ADD(A[1:q]) 10:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Synthetic Dataset",
"sec_num": "4"
},
{
"text": "M .ADD(T) 11:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Synthetic Dataset",
"sec_num": "4"
},
{
"text": "S \u2190 S \u222a {j\u00d7q} 12:\u00ce,M, l \u2190 Array(), Array(), |S| 13:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Synthetic Dataset",
"sec_num": "4"
},
{
"text": "for 1 to l do Interleaving 14:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Synthetic Dataset",
"sec_num": "4"
},
{
"text": "k \u2190 U(S) 15:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Synthetic Dataset",
"sec_num": "4"
},
{
"text": "P \u2190 REVERSE(I [k]",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Synthetic Dataset",
"sec_num": "4"
},
{
"text": ").POP() 16:\u00ce.ADD(P) 17:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Synthetic Dataset",
"sec_num": "4"
},
{
"text": "T \u2190 M [k] 18:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Synthetic Dataset",
"sec_num": "4"
},
{
"text": "if T \u2208M then: 19:M.ADD(T) 20:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Synthetic Dataset",
"sec_num": "4"
},
{
"text": "S \u2190 S \\ k 21: Z.ADD(\u00ce;M) 22:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Synthetic Dataset",
"sec_num": "4"
},
{
"text": "return Z",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Synthetic Dataset",
"sec_num": "4"
},
{
"text": "Random interleaving of the sentences from a small number of PubMed abstracts roughly resembles interleaved texts, and, correspondingly, interleaving of titles resembles its multi-sentence summary. We devised an algorithm for creating synthetic interleaved texts based on this idea.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Synthetic Dataset",
"sec_num": "4"
},
{
"text": "Interleave Algorithm: The Interleave Algorithm generates interleaved texts, each containing randomly interleaved sentences from a small number of abstracts, where the number is a random value within a specified range. The number of sentences used per abstract is also a random value within a specified range. Abstracts to be included in an interleaved text are first selected, then the selected abstracts are interleaved, and finally the interleaved texts together with a concatenation of the titles of the selected abstracts are returned.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Synthetic Dataset",
"sec_num": "4"
},
{
"text": "We first refer to Table 2 for terms and notations used in Algorithm. 1. INTERLEAVE takes a corpus of abstract-title pairs, C = A 1 ; T 1 , A 2 ; T 2 , . . . , A N ; T N , and returns a sequence of pairs of multi-thread interleaved texts and multi-sentence summaries, Z. Each interleaved text in the generated sequence will contain a number of threads ranging between a to b, where the number is randomly selected. Each thread, in turn, will contain a number of posts or sentences ranging between m and n, where this number is also randomly selected. The WINDOW function is given C, a desired window size, w, and step size, t, and it returns an iterator object, O, of size |N |\u2212w t +1.",
"cite_spans": [],
"ref_spans": [
{
"start": 18,
"end": 25,
"text": "Table 2",
"ref_id": "TABREF1"
}
],
"eq_spans": [],
"section": "Synthetic Dataset",
"sec_num": "4"
},
{
"text": "WINDOW helps to enlarge the interleaved corpus without redundancy as abstracts are randomly sampled out of an iterator element, E, and also new abstracts are always included in the next element through sliding. Similarly, sets of sentences are randomly sampled out of the selected abstracts. Thus, interleaved text-summary pairs in the corpus are different. The two parts of the INTERLEAVE algorithm, Selection and Interleaving, will be described next.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Synthetic Dataset",
"sec_num": "4"
},
{
"text": "Selection: U in step 5 determines the number of threads, r. Then, thread candidates for an interleaved text are chosen out of an iterator element E, a window size sequence of pairs of single-thread texts and single-sentence summaries. Next, post candidates for each selected thread, A, are chosen. U in step 8 determines the number of posts, q. Thread indices are repeated as many times as its posts and stored in a set, S.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Synthetic Dataset",
"sec_num": "4"
},
{
"text": "Interleaving: In every step in a loop of a size equivalent to the length of indices S, U randomly selects a thread index. REVERSE and POP in step 15 help in selecting a post, P , in the selected thread in a FIFO manner. The single-sentence summary, T, of the thread is added to the multi-sentence summary sequence,M, if it didn't exist previously.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Synthetic Dataset",
"sec_num": "4"
},
{
"text": "As an interleaved text-summary pair in the corpus has a thread size between a and b and post size per thread between m and n, the larger the difference between a and b and m and n in a corpus, the harder the disentangling and summarization task. So, we vary these parameters and create different synthetic corpora of varying difficulty for the experiments. Table 3 shows an example of a data instance from a Interleaved PubMed corpus compiled using a=2, b=5, m=2 and n=5.",
"cite_spans": [],
"ref_spans": [
{
"start": 357,
"end": 364,
"text": "Table 3",
"ref_id": "TABREF2"
}
],
"eq_spans": [],
"section": "Synthetic Dataset",
"sec_num": "4"
},
{
"text": "Parameters: For the word-to-word encoder, the steps are limited to 20, while the steps in the wordto-word decoder are limited to 15. The steps in the post-to-post encoder and thread-to-thread decoder depend on the corpus type, e.g., a Hard corpus compiled using a=2, b=5, m=2 and n=5 has 25 steps in post-to-post encoder, i.e., b\u00d7n (the maximum possible size of posts in an item in the corpus) and 5 steps in thread-to-thread decoder, i.e., b (the maximum possible threads in an item in the corpus). We initialized all weights, including word embeddings, with a random normal distribution with mean 0 and standard deviation 0.1. The embedding vectors and hidden states of the encoder and decoder in the models are set to dimension 100. Texts are lowercased. The vocabulary size is limited to 8000. We pad short sequences with a special token, P AD . We use Adam (Kingma and Ba, 2014) with an initial learning rate of .0001 and batch size of 64 for training. The training, evaluation and test sets in a Hard Interleaved PubMed corpus (a=2, b=5, m=2 and n=5) are of sizes of 170k, 4k and 4k respectively. We report ROUGE-1, ROUGE-2, and ROUGE-L as the quantitative evaluation of the models.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Experiments",
"sec_num": "5"
},
{
"text": "Upper-bound: In upper-bound experiments, we check the impact of disentanglement on the abstractive summarization models. In order to do this, we first evaluate the performance of a model when provided the ground-truth disentanglement (thread indices) information. We also evaluate the performance of models for either end-to-end or two-step summarization.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Experiments",
"sec_num": "5"
},
{
"text": "Ground-truth Disentangled: The ground-truth disentanglement information is used and posts of threads are disentangled and concatenated (posts are thread-wise sequentially arranged, i.e., noninterleaved). The first row in Table 4 shows performance of the hier2hier summarization model. Clearly, the model can easily detect a thread boundary in concatenated threads and perform very well, and therefore, sets an upper bound for the task.",
"cite_spans": [],
"ref_spans": [
{
"start": 221,
"end": 228,
"text": "Table 4",
"ref_id": "TABREF3"
}
],
"eq_spans": [],
"section": "Experiments",
"sec_num": "5"
},
{
"text": "No disentanglement: In real-world scenarios, i.e., with no disentanglement, Shang et al. (2018) 's unsupervised two-step system first disentangles/clusters the posts thread-wise and then compresses clusters to single-sentence summaries. While hier2hier is trained end-to-end, and therefore, generates multi-sentence summaries for a given interleaved text. Table 4 shows Shang et al. (2018) performs worse than hier2hier (compare rows 2 and 3), indicating that a hier2hier model trained on a sufficiently large dataset is better at summarization than the unsupervised sentence compression method, especially in fluency as indicated by an approximately 12 point increase in Rouge-2. Additionally, the hier2hier model trained on entangled texts achieves slightly lower performance to when it is trained on disentangled texts (compare rows 1 and 3), indicating that the disentanglement component can be avoided if summaries are available. The bottom section in Table 5 show an example of the model generations (shown in color). The top indexes of the phrase-level attention (\u03b2) is directly visualized in the table through the color coding matching the generation. This shows phrase level attention actually supports in learning to disentangle the interleaved texts.",
"cite_spans": [
{
"start": 76,
"end": 95,
"text": "Shang et al. (2018)",
"ref_id": "BIBREF22"
},
{
"start": 370,
"end": 389,
"text": "Shang et al. (2018)",
"ref_id": "BIBREF22"
}
],
"ref_spans": [
{
"start": 356,
"end": 363,
"text": "Table 4",
"ref_id": "TABREF3"
},
{
"start": 957,
"end": 964,
"text": "Table 5",
"ref_id": null
}
],
"eq_spans": [],
"section": "Experiments",
"sec_num": "5"
},
{
"text": "Transfer Learning: We utilize our interleaving algorithm and PubMed data to compile an interleaved corpus with a similar thread distribution as a corpus of real meetings, the AMI meeting corpus. AMI is a very small size corpus, so we have a train, eval and test split of 112, 10 and 20 respectively. Our analysis of the AMI corpus show that 90% of meetings have \u2264 12 summary sentences while 60% of meetings have \u2265 8 summary sentences, so we used 8 and 12 as the min (a) and max (b) number of threads respectively in the algorithm and create a synthetic corpus. We pretrain the hier2hier model for several iterations on the synthetic corpus, and then transfer and fine-tune the model on the AMI corpus with all parameters fixed except for the word-to-word decoder and hierarchical attention parameters. As PubMed and AMI are from Table 5 : An example of hier2hier generated summary sentences of a three thread interleaved text. Summaries are coloured differently and colors of attended phrases (\u03b2) in the text are identical to those of the generations. The table is best viewed in color. different domains, we use the byte pair encoding (BPE) (Sennrich et al., 2016) based subword dictionary. As shown in Table 6 , hier2hier readily transfers its disentangling knowledge, and therefore, obtains a boost in recall while maintaining its precision. The Li et al. (2019) system has the best ROUGE-1 scores, however their model is not directly comparable as unlike Shang et al. (2018) and our text-based model, it uses audio and video in addition to text.",
"cite_spans": [
{
"start": 1142,
"end": 1165,
"text": "(Sennrich et al., 2016)",
"ref_id": "BIBREF21"
},
{
"start": 1349,
"end": 1365,
"text": "Li et al. (2019)",
"ref_id": "BIBREF14"
},
{
"start": 1459,
"end": 1478,
"text": "Shang et al. (2018)",
"ref_id": "BIBREF22"
}
],
"ref_spans": [
{
"start": 829,
"end": 836,
"text": "Table 5",
"ref_id": null
},
{
"start": 1204,
"end": 1211,
"text": "Table 6",
"ref_id": null
}
],
"eq_spans": [],
"section": "Experiments",
"sec_num": "5"
},
{
"text": "Additionally, we also performed transfer learning experiments with models pre-trained for a dif-ferent number of iterations, and as seen in Figure 2 , hier2hier readily transfers its disentangling knowledge, and therefore, obtains a boost in recall while maintaining its precision. However, longer pretraining drives the model to generate shorter summaries similar to PubMed abstracts, and thereby, results in increasing precision and decreasing recall.",
"cite_spans": [],
"ref_spans": [
{
"start": 140,
"end": 148,
"text": "Figure 2",
"ref_id": "FIGREF4"
}
],
"eq_spans": [],
"section": "Experiments",
"sec_num": "5"
},
{
"text": "We also experimented with state of the art transformer-based seq2seq models, e.g., Bert-SumExtAbs (Liu and Lapata, 2019) and BART (Lewis et al., 2019) . BertSumExtAbs requires finetuning of the encoder and a de novo training of decoder while both encoder and decoder of BART are only fine-tuned. We use only AMI data for the de novo training and fine-tuning purpose, and the bottom two rows in Table 6 show the results from these models. 1 Although our hier2hier t-learn also only requires fine-tuning of the decoder and hierarchical attention, a highly-sophisticated semisupervised training of both the encoder and decoder of BART and larger model size (100x) yields better performance. However, for applications that have limited memory, as on some mobile devices, our model may be more desirable. Furthermore, despite a pre-trained encoder of BertSumExtAbs, a de novo training of a large size decoder with a tiny AMI data lead to over-fitting, and therefore, lower scores.",
"cite_spans": [
{
"start": 98,
"end": 120,
"text": "(Liu and Lapata, 2019)",
"ref_id": "BIBREF15"
},
{
"start": 130,
"end": 150,
"text": "(Lewis et al., 2019)",
"ref_id": "BIBREF12"
}
],
"ref_spans": [
{
"start": 394,
"end": 401,
"text": "Table 6",
"ref_id": null
}
],
"eq_spans": [],
"section": "Experiments",
"sec_num": "5"
},
{
"text": "Human Evaluation: We also performed a qualitative evaluation of our system using human judgments. Following Chen and Bansal (2018), we performed a comparative evaluation, where we provided six human judges (graduate students fluent in English) with meetings (\u2248 6000 words) and summaries from three sources, i.e., human reference, two-step baseline and hier2hier t-learn (here after referred to as the \"our model\"), and asked them to rate on a scale of 1 to 5 the two questions: 1) is the summary concise, fluent and grammatical (fluency) and 2) does the summary retain key information from the meeting (relevancy)?",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Experiments",
"sec_num": "5"
},
{
"text": "We sampled six meetings (each with three summaries corresponding to three sources), duplicated them, and then randomly sampled two dissimilar meetings and assigned them to each judge to annotate. For reference, an annotation sample would be an ASR transcript and human written summaries 1 Due to the small AMI data size, batch size and initial learning rate of BERTSumExt are set to 8 and 5e-4 respectively, batch size in BERTSumExtAbs is 16 and initial learning rates of BERT and transformer decoder in BERTSumExtAbs are 0.001 and 0.01 respectively. ROUGE-1 ROUGE-2 Model P R F1 P R F1 two-step (Shang et al., 2018) Table 6 : Rouge Scores for summary size 300 words on the AMI Corpus. t-learn=transfer-leaning. BART(base) = BART with 6 encoder and decoder layers and 140M parameters. Li et al. uses audio+video in addition to text and the transformer models (the bottom two rows) have lots of extra data for pre-training. . . . 5) the group discussed the shape of the device and decided to make the device easier. . .",
"cite_spans": [
{
"start": 596,
"end": 616,
"text": "(Shang et al., 2018)",
"ref_id": "BIBREF22"
}
],
"ref_spans": [
{
"start": 617,
"end": 624,
"text": "Table 6",
"ref_id": null
}
],
"eq_spans": [],
"section": "Experiments",
"sec_num": "5"
},
{
"text": ". . . two-step (Shang et al., 2018) summaries 1) marketing report uh we observed remote control users in a usability lab . . . 7) majority except for the forty five to fifty five year olds for some reason didnt want a voice act speech. . .",
"cite_spans": [
{
"start": 15,
"end": 35,
"text": "(Shang et al., 2018)",
"ref_id": "BIBREF22"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Experiments",
"sec_num": "5"
},
{
"text": ". . . 14) headed towards like a b a big yellow and black remote as far as maybe thats our next meeting that we discuss that . . . as in Table 1 and our model and Shang et al. (2018) summaries as in Table 7 . The judges were not shown the source of the summaries. The twelve ratings that we received are converted into two binary comparisons and are summarized in Table 8 . Our model summaries were often judged to be better than the Shang et al. (2018) system summaries in both fluency and relevancy. Gwet's AC1 and Brennan's and Prediger's kappa inter-rater agreement statistics show strong agreement for fluency. 2 However, compared to human summaries, our model summaries were similar in terms of fluency but were lower in terms of relevancy, with interrater statistics indicating fair strength of agreement. We also compared statistics of reference summaries against Our and Shang et al. (2018) model generated summaries of maximum 300 words. We observe our model generates approximately 145 words outputs, which is close to ground-truth human written summaries of size approximately 165 words. However, the Shang et al. (2018) system generates summaries of average 290 words. Further, the median number of threads (number of summaries) of our model, human written summaries and Shang et al. (2018) are 8, 8.5, 17, respectively. This indicates our model is learning to generate human-like summaries, while Shang et al. (2018) aims to distill words up to the permissible limit, and therefore, has high recall and very low precision; see Table 6 . Additionally, our model has twice the Shang et al. (2018) Rouge-2 values, which indicates high readability and was supported by human judges. Further, the difference in number of threads (summaries) between our model and reference are \u2264 3, 2, and 1 for 85%, 65%, and 40% of cases, respectively. This clearly indicates the strength of our hierarchical model in disentangling threads.",
"cite_spans": [
{
"start": 162,
"end": 181,
"text": "Shang et al. (2018)",
"ref_id": "BIBREF22"
},
{
"start": 879,
"end": 898,
"text": "Shang et al. (2018)",
"ref_id": "BIBREF22"
},
{
"start": 1112,
"end": 1131,
"text": "Shang et al. (2018)",
"ref_id": "BIBREF22"
},
{
"start": 1283,
"end": 1302,
"text": "Shang et al. (2018)",
"ref_id": "BIBREF22"
},
{
"start": 1410,
"end": 1429,
"text": "Shang et al. (2018)",
"ref_id": "BIBREF22"
}
],
"ref_spans": [
{
"start": 136,
"end": 143,
"text": "Table 1",
"ref_id": null
},
{
"start": 198,
"end": 205,
"text": "Table 7",
"ref_id": "TABREF5"
},
{
"start": 363,
"end": 370,
"text": "Table 8",
"ref_id": "TABREF7"
},
{
"start": 1540,
"end": 1547,
"text": "Table 6",
"ref_id": null
}
],
"eq_spans": [],
"section": "Experiments",
"sec_num": "5"
},
{
"text": "We investigated the use of an end-to-end hierarchical encoder-decoder model, hier2hier, with three levels of hierarchical attention for jointly summarizing and implicitly disentangling interleaved text. To train this model, we examined the use of pretraining using synthesized data and fine-tuning for adaptation to a new domain with limited labeled real-world data. On real-world AMI data, our fine-tuned end-to-end system outperforms a twostep system by 22% (Rouge-1). Experiments were also conducted against the transformer-based Bert-SumExtAbs and BART systems, which indicate that these transformer models can also summa-rize interleaved texts. Specifically our hier2hier model also outperformed the transformer-based BertSumExtAbs but not BART, which suggests that use of pretraining of both the decoder as well as the encoder is important, and also indicates the utility of our synthetic data.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Conclusion",
"sec_num": "6"
},
{
"text": "Gwet's AC(1) and Brennan and Prediger's Kappa adjust the impact of the empirical distributions over the chance agreement, and therefore, are better suited for cases where the proportion of agreements on one class differs from that of another.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
}
],
"back_matter": [],
"bib_entries": {
"BIBREF0": {
"ref_id": "b0",
"title": "Automatic label generation for news comment clusters",
"authors": [
{
"first": "Ahmet",
"middle": [],
"last": "Aker",
"suffix": ""
},
{
"first": "Monica",
"middle": [],
"last": "Paramita",
"suffix": ""
},
{
"first": "Emina",
"middle": [],
"last": "Kurtic",
"suffix": ""
},
{
"first": "Adam",
"middle": [],
"last": "Funk",
"suffix": ""
},
{
"first": "Emma",
"middle": [],
"last": "Barker",
"suffix": ""
},
{
"first": "Mark",
"middle": [],
"last": "Hepple",
"suffix": ""
},
{
"first": "Rob",
"middle": [],
"last": "Gaizauskas",
"suffix": ""
}
],
"year": 2016,
"venue": "Proceedings of the 9th International Natural Language Generation Conference",
"volume": "",
"issue": "",
"pages": "61--69",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Ahmet Aker, Monica Paramita, Emina Kurtic, Adam Funk, Emma Barker, Mark Hepple, and Rob Gaizauskas. 2016. Automatic label generation for news comment clusters. In Proceedings of the 9th International Natural Language Generation Confer- ence, pages 61-69.",
"links": null
},
"BIBREF1": {
"ref_id": "b1",
"title": "Speaker diarization: A review of recent research",
"authors": [
{
"first": "Xavier",
"middle": [],
"last": "Anguera",
"suffix": ""
},
{
"first": "Simon",
"middle": [],
"last": "Bozonnet",
"suffix": ""
},
{
"first": "Nicholas",
"middle": [],
"last": "Evans",
"suffix": ""
},
{
"first": "Corinne",
"middle": [],
"last": "Fredouille",
"suffix": ""
},
{
"first": "Gerald",
"middle": [],
"last": "Friedland",
"suffix": ""
},
{
"first": "Oriol",
"middle": [],
"last": "Vinyals",
"suffix": ""
}
],
"year": 2012,
"venue": "IEEE Transactions on Audio, Speech, and Language Processing",
"volume": "20",
"issue": "2",
"pages": "356--370",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Xavier Anguera, Simon Bozonnet, Nicholas Evans, Corinne Fredouille, Gerald Friedland, and Oriol Vinyals. 2012. Speaker diarization: A review of re- cent research. IEEE Transactions on Audio, Speech, and Language Processing, 20(2):356-370.",
"links": null
},
"BIBREF2": {
"ref_id": "b2",
"title": "Neural machine translation by jointly learning to align and translate",
"authors": [
{
"first": "Dzmitry",
"middle": [],
"last": "Bahdanau",
"suffix": ""
},
{
"first": "Kyunghyun",
"middle": [],
"last": "Cho",
"suffix": ""
},
{
"first": "Yoshua",
"middle": [],
"last": "Bengio",
"suffix": ""
}
],
"year": 2014,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Dzmitry Bahdanau, Kyunghyun Cho, and Yoshua Bengio. 2014. Neural machine translation by jointly learning to align and translate. CoRR, abs/1409.0473.",
"links": null
},
"BIBREF3": {
"ref_id": "b3",
"title": "The sensei annotated corpus: Human summaries of reader comment conversations in on-line news",
"authors": [
{
"first": "Emma",
"middle": [],
"last": "Barker",
"suffix": ""
},
{
"first": "Monica",
"middle": [
"Lestari"
],
"last": "Paramita",
"suffix": ""
},
{
"first": "Ahmet",
"middle": [],
"last": "Aker",
"suffix": ""
},
{
"first": "Emina",
"middle": [],
"last": "Kurtic",
"suffix": ""
},
{
"first": "Mark",
"middle": [],
"last": "Hepple",
"suffix": ""
},
{
"first": "Robert",
"middle": [],
"last": "Gaizauskas",
"suffix": ""
}
],
"year": 2016,
"venue": "Proceedings of the 17th annual meeting of the special interest group on discourse and dialogue",
"volume": "",
"issue": "",
"pages": "42--52",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Emma Barker, Monica Lestari Paramita, Ahmet Aker, Emina Kurtic, Mark Hepple, and Robert Gaizauskas. 2016. The sensei annotated corpus: Human sum- maries of reader comment conversations in on-line news. In Proceedings of the 17th annual meeting of the special interest group on discourse and dialogue, pages 42-52.",
"links": null
},
"BIBREF4": {
"ref_id": "b4",
"title": "Fast abstractive summarization with reinforce-selected sentence rewriting",
"authors": [
{
"first": "Yen-Chun",
"middle": [],
"last": "Chen",
"suffix": ""
},
{
"first": "Mohit",
"middle": [],
"last": "Bansal",
"suffix": ""
}
],
"year": 2018,
"venue": "Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics",
"volume": "1",
"issue": "",
"pages": "675--686",
"other_ids": {
"DOI": [
"10.18653/v1/P18-1063"
]
},
"num": null,
"urls": [],
"raw_text": "Yen-Chun Chen and Mohit Bansal. 2018. Fast abstrac- tive summarization with reinforce-selected sentence rewriting. In Proceedings of the 56th Annual Meet- ing of the Association for Computational Linguis- tics (Volume 1: Long Papers), pages 675-686, Mel- bourne, Australia. Association for Computational Linguistics.",
"links": null
},
"BIBREF5": {
"ref_id": "b5",
"title": "Neural summarization by extracting sentences and words",
"authors": [
{
"first": "Jianpeng",
"middle": [],
"last": "Cheng",
"suffix": ""
},
{
"first": "Mirella",
"middle": [],
"last": "Lapata",
"suffix": ""
}
],
"year": 2016,
"venue": "Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics",
"volume": "1",
"issue": "",
"pages": "484--494",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Jianpeng Cheng and Mirella Lapata. 2016. Neural sum- marization by extracting sentences and words. In Proceedings of the 54th Annual Meeting of the As- sociation for Computational Linguistics (Volume 1: Long Papers), pages 484-494.",
"links": null
},
"BIBREF6": {
"ref_id": "b6",
"title": "Abstractive sentence summarization with attentive recurrent neural networks",
"authors": [
{
"first": "Sumit",
"middle": [],
"last": "Chopra",
"suffix": ""
},
{
"first": "Michael",
"middle": [],
"last": "Auli",
"suffix": ""
},
{
"first": "Alexander",
"middle": [
"M"
],
"last": "Rush",
"suffix": ""
}
],
"year": 2016,
"venue": "The 2016 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies",
"volume": "",
"issue": "",
"pages": "93--98",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Sumit Chopra, Michael Auli, and Alexander M. Rush. 2016. Abstractive sentence summarization with at- tentive recurrent neural networks. In NAACL HLT 2016, The 2016 Conference of the North American Chapter of the Association for Computational Lin- guistics: Human Language Technologies, San Diego California, USA, June 12-17, 2016, pages 93-98. The Association for Computational Linguistics.",
"links": null
},
"BIBREF7": {
"ref_id": "b7",
"title": "Pubmed 200k rct: a dataset for sequential sentence classification in medical abstracts",
"authors": [
{
"first": "Franck",
"middle": [],
"last": "Dernoncourt",
"suffix": ""
},
{
"first": "Ji",
"middle": [
"Young"
],
"last": "Lee",
"suffix": ""
}
],
"year": 2017,
"venue": "Proceedings of the Eighth International Joint Conference on Natural Language Processing",
"volume": "2",
"issue": "",
"pages": "308--313",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Franck Dernoncourt and Ji Young Lee. 2017. Pubmed 200k rct: a dataset for sequential sentence classi- fication in medical abstracts. In Proceedings of the Eighth International Joint Conference on Natu- ral Language Processing (Volume 2: Short Papers), pages 308-313. Asian Federation of Natural Lan- guage Processing.",
"links": null
},
"BIBREF8": {
"ref_id": "b8",
"title": "Bert: Pre-training of deep bidirectional transformers for language understanding",
"authors": [
{
"first": "Jacob",
"middle": [],
"last": "Devlin",
"suffix": ""
},
{
"first": "Ming-Wei",
"middle": [],
"last": "Chang",
"suffix": ""
},
{
"first": "Kenton",
"middle": [],
"last": "Lee",
"suffix": ""
},
{
"first": "Kristina",
"middle": [],
"last": "Toutanova",
"suffix": ""
}
],
"year": 2018,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {
"arXiv": [
"arXiv:1810.04805"
]
},
"num": null,
"urls": [],
"raw_text": "Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2018. Bert: Pre-training of deep bidirectional transformers for language understand- ing. arXiv preprint arXiv:1810.04805.",
"links": null
},
"BIBREF9": {
"ref_id": "b9",
"title": "On the automatic generation of medical imaging reports",
"authors": [
{
"first": "Baoyu",
"middle": [],
"last": "Jing",
"suffix": ""
},
{
"first": "Pengtao",
"middle": [],
"last": "Xie",
"suffix": ""
},
{
"first": "Eric",
"middle": [],
"last": "Xing",
"suffix": ""
}
],
"year": 2018,
"venue": "Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics",
"volume": "1",
"issue": "",
"pages": "2577--2586",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Baoyu Jing, Pengtao Xie, and Eric Xing. 2018. On the automatic generation of medical imaging reports. In Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 2577-2586. Association for Computational Linguistics.",
"links": null
},
"BIBREF10": {
"ref_id": "b10",
"title": "Adam: A method for stochastic optimization",
"authors": [
{
"first": "P",
"middle": [],
"last": "Diederik",
"suffix": ""
},
{
"first": "Jimmy",
"middle": [],
"last": "Kingma",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Ba",
"suffix": ""
}
],
"year": 2014,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Diederik P. Kingma and Jimmy Ba. 2014. Adam: A method for stochastic optimization. CoRR, abs/1412.6980.",
"links": null
},
"BIBREF11": {
"ref_id": "b11",
"title": "A hierarchical approach for generating descriptive image paragraphs",
"authors": [
{
"first": "Jonathan",
"middle": [],
"last": "Krause",
"suffix": ""
},
{
"first": "Justin",
"middle": [],
"last": "Johnson",
"suffix": ""
},
{
"first": "Ranjay",
"middle": [],
"last": "Krishna",
"suffix": ""
},
{
"first": "Li",
"middle": [],
"last": "Fei-Fei",
"suffix": ""
}
],
"year": 2017,
"venue": "Computer Vision and Patterm Recognition (CVPR)",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Jonathan Krause, Justin Johnson, Ranjay Krishna, and Li Fei-Fei. 2017. A hierarchical approach for gen- erating descriptive image paragraphs. In Computer Vision and Patterm Recognition (CVPR).",
"links": null
},
"BIBREF12": {
"ref_id": "b12",
"title": "Bart: Denoising sequence-to-sequence pretraining for natural language generation, translation, and comprehension",
"authors": [
{
"first": "Mike",
"middle": [],
"last": "Lewis",
"suffix": ""
},
{
"first": "Yinhan",
"middle": [],
"last": "Liu",
"suffix": ""
},
{
"first": "Naman",
"middle": [],
"last": "Goyal ; Abdelrahman Mohamed",
"suffix": ""
},
{
"first": "Omer",
"middle": [],
"last": "Levy",
"suffix": ""
},
{
"first": "Veselin",
"middle": [],
"last": "Stoyanov",
"suffix": ""
},
{
"first": "Luke",
"middle": [],
"last": "Zettlemoyer",
"suffix": ""
}
],
"year": 2019,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {
"arXiv": [
"arXiv:1910.13461"
]
},
"num": null,
"urls": [],
"raw_text": "Mike Lewis, Yinhan Liu, Naman Goyal, Mar- jan Ghazvininejad, Abdelrahman Mohamed, Omer Levy, Veselin Stoyanov, and Luke Zettlemoyer. 2019. Bart: Denoising sequence-to-sequence pre- training for natural language generation, trans- lation, and comprehension. arXiv preprint arXiv:1910.13461.",
"links": null
},
"BIBREF13": {
"ref_id": "b13",
"title": "A hierarchical neural autoencoder for paragraphs and documents",
"authors": [
{
"first": "Jiwei",
"middle": [],
"last": "Li",
"suffix": ""
},
{
"first": "Thang",
"middle": [],
"last": "Luong",
"suffix": ""
},
{
"first": "Dan",
"middle": [],
"last": "Jurafsky",
"suffix": ""
}
],
"year": 2015,
"venue": "Proceedings of the 53rd Annual Meeting of the Association for Computational Linguistics and the 7th International Joint Conference on Natural Language Processing",
"volume": "1",
"issue": "",
"pages": "1106--1115",
"other_ids": {
"DOI": [
"10.3115/v1/P15-1107"
]
},
"num": null,
"urls": [],
"raw_text": "Jiwei Li, Thang Luong, and Dan Jurafsky. 2015. A hierarchical neural autoencoder for paragraphs and documents. In Proceedings of the 53rd Annual Meet- ing of the Association for Computational Linguis- tics and the 7th International Joint Conference on Natural Language Processing (Volume 1: Long Pa- pers), pages 1106-1115. Association for Computa- tional Linguistics.",
"links": null
},
"BIBREF14": {
"ref_id": "b14",
"title": "Keep meeting summaries on topic: Abstractive multi-modal meeting summarization",
"authors": [
{
"first": "Manling",
"middle": [],
"last": "Li",
"suffix": ""
},
{
"first": "Lingyu",
"middle": [],
"last": "Zhang",
"suffix": ""
},
{
"first": "Ji",
"middle": [],
"last": "Heng",
"suffix": ""
},
{
"first": "Richard",
"middle": [
"J"
],
"last": "Radke",
"suffix": ""
}
],
"year": 2019,
"venue": "Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics",
"volume": "",
"issue": "",
"pages": "2190--2196",
"other_ids": {
"DOI": [
"10.18653/v1/P19-1210"
]
},
"num": null,
"urls": [],
"raw_text": "Manling Li, Lingyu Zhang, Heng Ji, and Richard J. Radke. 2019. Keep meeting summaries on topic: Abstractive multi-modal meeting summarization. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages 2190-2196, Florence, Italy. Association for Compu- tational Linguistics.",
"links": null
},
"BIBREF15": {
"ref_id": "b15",
"title": "Text summarization with pretrained encoders",
"authors": [
{
"first": "Yang",
"middle": [],
"last": "Liu",
"suffix": ""
},
{
"first": "Mirella",
"middle": [],
"last": "Lapata",
"suffix": ""
}
],
"year": 2019,
"venue": "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP)",
"volume": "",
"issue": "",
"pages": "3730--3740",
"other_ids": {
"DOI": [
"10.18653/v1/D19-1387"
]
},
"num": null,
"urls": [],
"raw_text": "Yang Liu and Mirella Lapata. 2019. Text summariza- tion with pretrained encoders. In Proceedings of the 2019 Conference on Empirical Methods in Nat- ural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP), pages 3730-3740, Hong Kong, China. Association for Computational Linguistics.",
"links": null
},
"BIBREF16": {
"ref_id": "b16",
"title": "Topic-driven reader comments summarization",
"authors": [
{
"first": "Zongyang",
"middle": [],
"last": "Ma",
"suffix": ""
},
{
"first": "Aixin",
"middle": [],
"last": "Sun",
"suffix": ""
},
{
"first": "Quan",
"middle": [],
"last": "Yuan",
"suffix": ""
},
{
"first": "Gao",
"middle": [],
"last": "Cong",
"suffix": ""
}
],
"year": 2012,
"venue": "Proceedings of the 21st ACM international conference on Information and knowledge management",
"volume": "",
"issue": "",
"pages": "265--274",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Zongyang Ma, Aixin Sun, Quan Yuan, and Gao Cong. 2012. Topic-driven reader comments summariza- tion. In Proceedings of the 21st ACM international conference on Information and knowledge manage- ment, pages 265-274. ACM.",
"links": null
},
"BIBREF17": {
"ref_id": "b17",
"title": "The ami meeting corpus",
"authors": [
{
"first": "I",
"middle": [],
"last": "Mccowan",
"suffix": ""
},
{
"first": "J",
"middle": [],
"last": "Carletta",
"suffix": ""
},
{
"first": "W",
"middle": [],
"last": "Kraaij",
"suffix": ""
},
{
"first": "S",
"middle": [],
"last": "Ashby",
"suffix": ""
},
{
"first": "S",
"middle": [],
"last": "Bourban",
"suffix": ""
},
{
"first": "M",
"middle": [],
"last": "Flynn",
"suffix": ""
},
{
"first": "M",
"middle": [],
"last": "Guillemot",
"suffix": ""
},
{
"first": "T",
"middle": [],
"last": "Hain",
"suffix": ""
},
{
"first": "J",
"middle": [],
"last": "Kadlec",
"suffix": ""
},
{
"first": "V",
"middle": [],
"last": "Karaiskos",
"suffix": ""
},
{
"first": "M",
"middle": [],
"last": "Kronenthal",
"suffix": ""
},
{
"first": "G",
"middle": [],
"last": "Lathoud",
"suffix": ""
},
{
"first": "M",
"middle": [],
"last": "Lincoln",
"suffix": ""
},
{
"first": "A",
"middle": [],
"last": "Lisowska",
"suffix": ""
},
{
"first": "W",
"middle": [],
"last": "Post",
"suffix": ""
},
{
"first": "Dennis",
"middle": [],
"last": "Reidsma",
"suffix": ""
},
{
"first": "P",
"middle": [],
"last": "Wellner",
"suffix": ""
}
],
"year": 2005,
"venue": "Proceedings of Measuring Behavior 2005, 5th International Conference on Methods and Techniques in Behavioral Research",
"volume": "",
"issue": "",
"pages": "137--140",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "I. McCowan, J. Carletta, W. Kraaij, S. Ashby, S. Bour- ban, M. Flynn, M. Guillemot, T. Hain, J. Kadlec, V. Karaiskos, M. Kronenthal, G. Lathoud, M. Lin- coln, A. Lisowska, W. Post, Dennis Reidsma, and P. Wellner. 2005. The ami meeting corpus. In Pro- ceedings of Measuring Behavior 2005, 5th Interna- tional Conference on Methods and Techniques in Be- havioral Research, pages 137-140. Noldus Informa- tion Technology.",
"links": null
},
"BIBREF18": {
"ref_id": "b18",
"title": "Summarunner: A recurrent neural network based sequence model for extractive summarization of documents",
"authors": [
{
"first": "Ramesh",
"middle": [],
"last": "Nallapati",
"suffix": ""
},
{
"first": "Feifei",
"middle": [],
"last": "Zhai",
"suffix": ""
},
{
"first": "Bowen",
"middle": [],
"last": "Zhou",
"suffix": ""
}
],
"year": 2017,
"venue": "AAAI Conference on Artificial Intelligence",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Ramesh Nallapati, Feifei Zhai, and Bowen Zhou. 2017. Summarunner: A recurrent neural network based se- quence model for extractive summarization of doc- uments. In AAAI Conference on Artificial Intelli- gence.",
"links": null
},
"BIBREF19": {
"ref_id": "b19",
"title": "Abstractive text summarization using sequence-tosequence rnns and beyond",
"authors": [
{
"first": "Ramesh",
"middle": [],
"last": "Nallapati",
"suffix": ""
},
{
"first": "Bowen",
"middle": [],
"last": "Zhou",
"suffix": ""
},
{
"first": "C\u00edcero",
"middle": [],
"last": "Nogueira",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Santos",
"suffix": ""
},
{
"first": "Bing",
"middle": [],
"last": "Aglar G\u00fcl\u00e7ehre",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Xiang",
"suffix": ""
}
],
"year": 2016,
"venue": "Proceedings of the 20th SIGNLL Conference on Computational Natural Language Learning",
"volume": "",
"issue": "",
"pages": "280--290",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Ramesh Nallapati, Bowen Zhou, C\u00edcero Nogueira dos Santos, \u00c7 aglar G\u00fcl\u00e7ehre, and Bing Xiang. 2016. Abstractive text summarization using sequence-to- sequence rnns and beyond. In Proceedings of the 20th SIGNLL Conference on Computational Natural Language Learning, CoNLL 2016, Berlin, Germany, August 11-12, 2016, pages 280-290. ACL.",
"links": null
},
"BIBREF20": {
"ref_id": "b20",
"title": "A neural attention model for abstractive sentence summarization",
"authors": [
{
"first": "Alexander",
"middle": [
"M"
],
"last": "Rush",
"suffix": ""
},
{
"first": "Sumit",
"middle": [],
"last": "Chopra",
"suffix": ""
},
{
"first": "Jason",
"middle": [],
"last": "Weston",
"suffix": ""
}
],
"year": 2015,
"venue": "Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing",
"volume": "",
"issue": "",
"pages": "379--389",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Alexander M. Rush, Sumit Chopra, and Jason Weston. 2015. A neural attention model for abstractive sen- tence summarization. In Proceedings of the 2015 Conference on Empirical Methods in Natural Lan- guage Processing, EMNLP 2015, Lisbon, Portugal, September 17-21, 2015, pages 379-389. The Asso- ciation for Computational Linguistics.",
"links": null
},
"BIBREF21": {
"ref_id": "b21",
"title": "Neural machine translation of rare words with subword units",
"authors": [
{
"first": "Rico",
"middle": [],
"last": "Sennrich",
"suffix": ""
},
{
"first": "Barry",
"middle": [],
"last": "Haddow",
"suffix": ""
},
{
"first": "Alexandra",
"middle": [],
"last": "Birch",
"suffix": ""
}
],
"year": 2016,
"venue": "Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics",
"volume": "1",
"issue": "",
"pages": "1715--1725",
"other_ids": {
"DOI": [
"10.18653/v1/P16-1162"
]
},
"num": null,
"urls": [],
"raw_text": "Rico Sennrich, Barry Haddow, and Alexandra Birch. 2016. Neural machine translation of rare words with subword units. In Proceedings of the 54th An- nual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 1715- 1725, Berlin, Germany. Association for Computa- tional Linguistics.",
"links": null
},
"BIBREF22": {
"ref_id": "b22",
"title": "Unsupervised abstractive meeting summarization with multisentence compression and budgeted submodular maximization",
"authors": [
{
"first": "Guokan",
"middle": [],
"last": "Shang",
"suffix": ""
},
{
"first": "Wensi",
"middle": [],
"last": "Ding",
"suffix": ""
},
{
"first": "Zekun",
"middle": [],
"last": "Zhang",
"suffix": ""
},
{
"first": "Antoine",
"middle": [],
"last": "Tixier",
"suffix": ""
},
{
"first": "Polykarpos",
"middle": [],
"last": "Meladianos",
"suffix": ""
},
{
"first": "Michalis",
"middle": [],
"last": "Vazirgiannis",
"suffix": ""
},
{
"first": "Jean-Pierre",
"middle": [],
"last": "Lorr\u00e9",
"suffix": ""
}
],
"year": 2018,
"venue": "Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics",
"volume": "1",
"issue": "",
"pages": "664--674",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Guokan Shang, Wensi Ding, Zekun Zhang, An- toine Tixier, Polykarpos Meladianos, Michalis Vazir- giannis, and Jean-Pierre Lorr\u00e9. 2018. Unsuper- vised abstractive meeting summarization with multi- sentence compression and budgeted submodular maximization. In Proceedings of the 56th Annual Meeting of the Association for Computational Lin- guistics (Volume 1: Long Papers), pages 664-674. Association for Computational Linguistics.",
"links": null
},
"BIBREF23": {
"ref_id": "b23",
"title": "Mass: Masked sequence to sequence pre-training for language generation",
"authors": [
{
"first": "Kaitao",
"middle": [],
"last": "Song",
"suffix": ""
},
{
"first": "Xu",
"middle": [],
"last": "Tan",
"suffix": ""
},
{
"first": "Tao",
"middle": [],
"last": "Qin",
"suffix": ""
},
{
"first": "Jianfeng",
"middle": [],
"last": "Lu",
"suffix": ""
},
{
"first": "Tie-Yan",
"middle": [],
"last": "Liu",
"suffix": ""
}
],
"year": 2019,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {
"arXiv": [
"arXiv:1905.02450"
]
},
"num": null,
"urls": [],
"raw_text": "Kaitao Song, Xu Tan, Tao Qin, Jianfeng Lu, and Tie- Yan Liu. 2019. Mass: Masked sequence to sequence pre-training for language generation. arXiv preprint arXiv:1905.02450.",
"links": null
},
"BIBREF24": {
"ref_id": "b24",
"title": "From neural sentence summarization to headline generation: A coarse-to-fine approach",
"authors": [
{
"first": "Jiwei",
"middle": [],
"last": "Tan",
"suffix": ""
},
{
"first": "Xiaojun",
"middle": [],
"last": "Wan",
"suffix": ""
},
{
"first": "Jianguo",
"middle": [],
"last": "Xiao",
"suffix": ""
}
],
"year": 2017,
"venue": "IJCAI",
"volume": "",
"issue": "",
"pages": "4109--4115",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Jiwei Tan, Xiaojun Wan, and Jianguo Xiao. 2017. From neural sentence summarization to headline generation: A coarse-to-fine approach. In IJCAI, pages 4109-4115.",
"links": null
},
"BIBREF25": {
"ref_id": "b25",
"title": "Creating a reference data set for the summarization of discussion forum threads",
"authors": [
{
"first": "Suzan",
"middle": [],
"last": "Verberne",
"suffix": ""
},
{
"first": "Emiel",
"middle": [],
"last": "Krahmer",
"suffix": ""
},
{
"first": "Iris",
"middle": [],
"last": "Hendrickx",
"suffix": ""
},
{
"first": "Sander",
"middle": [],
"last": "Wubben",
"suffix": ""
},
{
"first": "Antal",
"middle": [],
"last": "Van Den",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Bosch",
"suffix": ""
}
],
"year": 2018,
"venue": "Language Resources and Evaluation",
"volume": "",
"issue": "",
"pages": "1--23",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Suzan Verberne, Emiel Krahmer, Iris Hendrickx, Sander Wubben, and Antal van Den Bosch. 2018. Creating a reference data set for the summarization of discussion forum threads. Language Resources and Evaluation, pages 1-23.",
"links": null
},
"BIBREF26": {
"ref_id": "b26",
"title": "Contextbased message expansion for disentanglement of interleaved text conversations",
"authors": [
{
"first": "Lidan",
"middle": [],
"last": "Wang",
"suffix": ""
},
{
"first": "W",
"middle": [],
"last": "Douglas",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Oard",
"suffix": ""
}
],
"year": 2009,
"venue": "Proceedings of human language technologies: The 2009 annual conference of the North American chapter of the association for computational linguistics",
"volume": "",
"issue": "",
"pages": "200--208",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Lidan Wang and Douglas W Oard. 2009. Context- based message expansion for disentanglement of in- terleaved text conversations. In Proceedings of hu- man language technologies: The 2009 annual con- ference of the North American chapter of the associ- ation for computational linguistics, pages 200-208. Association for Computational Linguistics.",
"links": null
}
},
"ref_entries": {
"FIGREF0": {
"uris": null,
"text": "Ma et al. (2012); Aker et al. (2016); Shang et al.",
"type_str": "figure",
"num": null
},
"FIGREF1": {
"uris": null,
"text": "Hierarchical encoder-decoder architecture. On the left, interleaved posts are encoded hierarchically,",
"type_str": "figure",
"num": null
},
"FIGREF2": {
"uris": null,
"text": "Interleaved text \u03c0 . . . conducted to evaluate the influence of excessive sweating during long-distance running on the urinary concentration of caffeine. . . \u03c9 . . . to assess the effect of a program of supervised fitness walking and patient education on functional status. . . . . . \u03c0 . . . 102 patients with a documented diagnosis of primary osteoarthritis of one or both knees participated. . . \u03c6 . . . examined the effects of intensity of training on ratings of perceived exertion (. . . . . . Summary \u03c0 caffeine in sport. influence of endurance exercise on the urinary caffeine concentration. \u03c9 supervised fitness walking in patients with osteoarthritis of the knee. a randomized , controlled trial. \u03c6 the effect of training intensity on ratings of perceived exertion.",
"type_str": "figure",
"num": null
},
"FIGREF3": {
"uris": null,
"text": "conducted to evaluate the influence of excessive sweating during long-distance running on the urinary concentration of caffeine. . . . . . to assess the effect of a program of supervised fitness walking and patient education on functional status. . . . . . . . . 102 patients with a documented diagnosis of primary osteoarthritis of one or both knees participated. . . . . . examined the effects of intensity of training on ratings of perceived exertion (. . . . . . Generation effect of excessive [UNK] during [UNK] running on the urinary concentration of caffeine . effect of a physical fitness walking on functional status , pain , and pain effects of intensity of training on perceived [UNK] in [UNK] athletes .",
"type_str": "figure",
"num": null
},
"FIGREF4": {
"uris": null,
"text": "ROUGE uni-and bi-gram precision (green) and recall (blue) of AMI fine-tuned hier2hier models with different numbers of pretraining iterations. Maximum words in a summary is 300. As a reference, solid horizontal lines show the scores of a model trained only on AMI.",
"type_str": "figure",
"num": null
},
"FIGREF5": {
"uris": null,
"text": "h2h t-learn summaries 1) the project manager opened the meeting and went over the minutes of the previous. . .. . . 3) the industrial designer discussed the interior workings of a remote and the team. . .",
"type_str": "figure",
"num": null
},
"TABREF1": {
"text": "We use lowercase italics for variables, uppercase italics for sets and sequences, math symbols for mathematical operations and uppercase words for methods.",
"content": "<table/>",
"html": null,
"num": null,
"type_str": "table"
},
"TABREF2": {
"text": "An example of a synthetic Interleaved text and summary pair compiled using PubMed corpus and Algorithm. 1. It includes three threads (abstracts) identifiable through superscribed symbols \u03c0, \u03c9, and \u03c6.",
"content": "<table><tr><td>Input Text</td><td>Model</td><td colspan=\"3\">Rouge-1 Rouge-2 Rouge-L</td></tr><tr><td colspan=\"2\">dis (upper bd) hier2hier</td><td>39.09</td><td>30.11</td><td>15.22</td></tr><tr><td colspan=\"2\">(Shang et al., 2018)</td><td>29.11</td><td>15.76</td><td>10.13</td></tr><tr><td>ent</td><td>hier2hier</td><td>37.11</td><td>27.97</td><td>14.26</td></tr></table>",
"html": null,
"num": null,
"type_str": "table"
},
"TABREF3": {
"text": "",
"content": "<table><tr><td>: Synthetic interleaved text summarization</td></tr><tr><td>performance (Rouge Recall-Scores) comparing models</td></tr><tr><td>when the threads are disentangled (top section, upper</td></tr><tr><td>bound) and when the threads are entangled (bottom sec-</td></tr><tr><td>tion, real-world) on an Interleaved PubMed Corpus. dis</td></tr><tr><td>= disentangled (ground-truth) and ent = entangled.</td></tr></table>",
"html": null,
"num": null,
"type_str": "table"
},
"TABREF4": {
"text": "34.61 41.84 37.37 6.92 8.29 7.45 hier2hier 46.30 38.17 41.30 14.84 12.23 13.13 hier2hier t-learn 47.68 44.37 45.56 16.02 14.98 15.35 (Li et al., 2019) --53.29 --13.51 BertSumExtAbs 55.95 36.21 43.24 18.35 12.16 14.39 BART(base) 42.17 59.19 49.03 16.52 23.05 19.15",
"content": "<table/>",
"html": null,
"num": null,
"type_str": "table"
},
"TABREF5": {
"text": "The top and bottom sections show our hierarchical and the Shang et al. (2018) system summaries respectively for ASR transcripts inTable 1. a) refer to the a th sentence in a multi-sentence summary.",
"content": "<table/>",
"html": null,
"num": null,
"type_str": "table"
},
"TABREF7": {
"text": "Comparative ratings by human judges of summaries on fluency and relevancy metrics. gwet ac1 and bp refer to Gwet's AC(1) and Brennan-Prediger Kappa coefficients respectively.",
"content": "<table/>",
"html": null,
"num": null,
"type_str": "table"
}
}
}
} |