File size: 114,442 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 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 |
{
"paper_id": "O03-4004",
"header": {
"generated_with": "S2ORC 1.0.0",
"date_generated": "2023-01-19T08:01:31.667495Z"
},
"title": "N-grams for English and Chinese",
"authors": [
{
"first": "",
"middle": [],
"last": "Le Quan Ha",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "Queen's University Belfast",
"location": {
"postCode": "BT7 1NN",
"settlement": "Belfast",
"region": "Northern Ireland, UK"
}
},
"email": ""
},
{
"first": "E",
"middle": [
"I"
],
"last": "Sicilia-Garcia",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "Queen's University Belfast",
"location": {
"postCode": "BT7 1NN",
"settlement": "Belfast",
"region": "Northern Ireland, UK"
}
},
"email": ""
},
{
"first": "Ji",
"middle": [],
"last": "Ming",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "Queen's University Belfast",
"location": {
"postCode": "BT7 1NN",
"settlement": "Belfast",
"region": "Northern Ireland, UK"
}
},
"email": "j.ming@qub.ac.uk"
},
{
"first": "F",
"middle": [
"J"
],
"last": "Smith",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "Queen's University Belfast",
"location": {
"postCode": "BT7 1NN",
"settlement": "Belfast",
"region": "Northern Ireland, UK"
}
},
"email": "fj.smith@qub.ac.uk"
}
],
"year": "",
"venue": null,
"identifiers": {},
"abstract": "It is shown that for a large corpus, Zipf 's law for both words in English and characters in Chinese does not hold for all ranks. The frequency falls below the frequency predicted by Zipf's law for English words for rank greater than about 5,000 and for Chinese characters for rank greater than about 1,000. However, when single words or characters are combined together with n-gram words or characters in one list and put in order of frequency, the frequency of tokens in the combined list follows Zipf's law approximately with the slope close to-1 on a loglog plot for all n-grams, down to the lowest frequencies in both languages. This behaviour is also found for English 2-byte and 3-byte word fragments. It only happens when all n-grams are used, including semantically incomplete n-grams. Previous theories do not predict this behaviour, possibly because conditional probabilities of tokens have not been properly represented.",
"pdf_parse": {
"paper_id": "O03-4004",
"_pdf_hash": "",
"abstract": [
{
"text": "It is shown that for a large corpus, Zipf 's law for both words in English and characters in Chinese does not hold for all ranks. The frequency falls below the frequency predicted by Zipf's law for English words for rank greater than about 5,000 and for Chinese characters for rank greater than about 1,000. However, when single words or characters are combined together with n-gram words or characters in one list and put in order of frequency, the frequency of tokens in the combined list follows Zipf's law approximately with the slope close to-1 on a loglog plot for all n-grams, down to the lowest frequencies in both languages. This behaviour is also found for English 2-byte and 3-byte word fragments. It only happens when all n-grams are used, including semantically incomplete n-grams. Previous theories do not predict this behaviour, possibly because conditional probabilities of tokens have not been properly represented.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Abstract",
"sec_num": null
}
],
"body_text": [
{
"text": "The law discovered empirically by [Zipf 1949] for word tokens in a corpus states that if f is the frequency of a word in the corpus and r is the rank, then:",
"cite_spans": [
{
"start": 34,
"end": 45,
"text": "[Zipf 1949]",
"ref_id": "BIBREF42"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Zipf's law",
"sec_num": "1.1"
},
{
"text": "EQUATION",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [
{
"start": 0,
"end": 8,
"text": "EQUATION",
"ref_id": "EQREF",
"raw_str": "r k f =",
"eq_num": "(1)"
}
],
"section": "Zipf's law",
"sec_num": "1.1"
},
{
"text": "where k is a constant for the corpus. When log(f) is drawn against log(r) in a graph (which is called a Zipf curve), a straight line is obtained with a slope of -1. An example with a small corpus of 250,000 tokens made up of paragraphs chosen at random from the Brown corpus of American English [Francis and Kucera 1964] is given in Figure 1 ; in this the tokens do not include punctuation marks and numbers. Typographical errors, if any, will appear in the hapax legomenon. Zipf's discovery was followed by a large body of literature, reviewed in a series of papers edited by [Guiter and Arapov 1982] . Notable among these are papers by [Mandelbrot 1953 [Mandelbrot , 1954 [Mandelbrot , 1959 , [Miller 1954 [Miller , 1957 [Miller , 1958 , [Simon 1955 [Simon , 1960 , [Sichel 1975 [Sichel , 1986 , [Carroll 1967 [Carroll , 1969 , [Baayen 1991] , [Chitashvili 1983 [Chitashvili , 1989 and [Orlov 1983] . It continues to stimulate interest today [Samuelson 1996] ; [Baayen 2001] ; [Hatzigeorgiu, Mikros and Carayannis 2001] ; [Montermurro 2001] ; [Ferrer and Sol\u00e9 2002] and, for example, it has been recently applied to citations [Silagadze 1997] , to biological species-abundance [Sichel 1997 ] and to DNA sequences [Yonezawa and Motohasi 1999] ; [Li 2001 ].",
"cite_spans": [
{
"start": 295,
"end": 320,
"text": "[Francis and Kucera 1964]",
"ref_id": "BIBREF9"
},
{
"start": 577,
"end": 601,
"text": "[Guiter and Arapov 1982]",
"ref_id": "BIBREF10"
},
{
"start": 638,
"end": 654,
"text": "[Mandelbrot 1953",
"ref_id": "BIBREF14"
},
{
"start": 655,
"end": 673,
"text": "[Mandelbrot , 1954",
"ref_id": "BIBREF15"
},
{
"start": 674,
"end": 692,
"text": "[Mandelbrot , 1959",
"ref_id": "BIBREF16"
},
{
"start": 695,
"end": 707,
"text": "[Miller 1954",
"ref_id": "BIBREF19"
},
{
"start": 708,
"end": 722,
"text": "[Miller , 1957",
"ref_id": "BIBREF20"
},
{
"start": 723,
"end": 737,
"text": "[Miller , 1958",
"ref_id": "BIBREF21"
},
{
"start": 740,
"end": 751,
"text": "[Simon 1955",
"ref_id": "BIBREF33"
},
{
"start": 752,
"end": 765,
"text": "[Simon , 1960",
"ref_id": "BIBREF34"
},
{
"start": 768,
"end": 780,
"text": "[Sichel 1975",
"ref_id": "BIBREF29"
},
{
"start": 781,
"end": 795,
"text": "[Sichel , 1986",
"ref_id": "BIBREF30"
},
{
"start": 798,
"end": 811,
"text": "[Carroll 1967",
"ref_id": null
},
{
"start": 812,
"end": 827,
"text": "[Carroll , 1969",
"ref_id": "BIBREF4"
},
{
"start": 830,
"end": 843,
"text": "[Baayen 1991]",
"ref_id": "BIBREF0"
},
{
"start": 846,
"end": 863,
"text": "[Chitashvili 1983",
"ref_id": "BIBREF25"
},
{
"start": 864,
"end": 883,
"text": "[Chitashvili , 1989",
"ref_id": null
},
{
"start": 888,
"end": 900,
"text": "[Orlov 1983]",
"ref_id": "BIBREF25"
},
{
"start": 944,
"end": 960,
"text": "[Samuelson 1996]",
"ref_id": "BIBREF28"
},
{
"start": 963,
"end": 976,
"text": "[Baayen 2001]",
"ref_id": "BIBREF1"
},
{
"start": 979,
"end": 1021,
"text": "[Hatzigeorgiu, Mikros and Carayannis 2001]",
"ref_id": "BIBREF11"
},
{
"start": 1024,
"end": 1042,
"text": "[Montermurro 2001]",
"ref_id": null
},
{
"start": 1045,
"end": 1067,
"text": "[Ferrer and Sol\u00e9 2002]",
"ref_id": "BIBREF8"
},
{
"start": 1128,
"end": 1144,
"text": "[Silagadze 1997]",
"ref_id": "BIBREF32"
},
{
"start": 1179,
"end": 1191,
"text": "[Sichel 1997",
"ref_id": "BIBREF31"
},
{
"start": 1215,
"end": 1243,
"text": "[Yonezawa and Motohasi 1999]",
"ref_id": "BIBREF40"
},
{
"start": 1246,
"end": 1254,
"text": "[Li 2001",
"ref_id": "BIBREF13"
}
],
"ref_spans": [
{
"start": 333,
"end": 341,
"text": "Figure 1",
"ref_id": "FIGREF0"
}
],
"eq_spans": [],
"section": "Zipf's law",
"sec_num": "1.1"
},
{
"text": "Zipf discovered the law by analysing manually the frequencies of words in the novel \"Ulysses\" by James Joyce. It contains a vocabulary of 29,899 different word types associated with 260,430 word tokens.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Zipf's law",
"sec_num": "1.1"
},
{
"text": "Following its discovery in 1949, several experiments aided by the appearance of the computer in the 1960's, confirmed that the law was correct for the small corpora that could be processed at that time. The slope of the curve was found to vary slightly from -1 for some corpora; also the frequencies for the highest ranked words sometimes deviated from the straight line, which suggested several modifications of the law, and in particular one derived theoretically by [Mandelbrot 1953 ] with the form:",
"cite_spans": [
{
"start": 469,
"end": 485,
"text": "[Mandelbrot 1953",
"ref_id": "BIBREF14"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Theoretical developments:",
"sec_num": "1.2"
},
{
"text": "EQUATION",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [
{
"start": 0,
"end": 8,
"text": "EQUATION",
"ref_id": "EQREF",
"raw_str": "\u03b2 \u03b1 )( + = r k f",
"eq_num": "(2)"
}
],
"section": "Theoretical developments:",
"sec_num": "1.2"
},
{
"text": "where \u03b1 and \u03b2 are constants for the corpus being analysed. However, generally the constants \u03b1 and \u03b2 were found to be only small varying deviations from the original law by Zipf.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Theoretical developments:",
"sec_num": "1.2"
},
{
"text": "Exceptions include legal texts which have smaller slopes (\u22480.9) showing that lawyers use more word types than other people! [Smith and Devine 1985] .",
"cite_spans": [
{
"start": 124,
"end": 147,
"text": "[Smith and Devine 1985]",
"ref_id": "BIBREF37"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Theoretical developments:",
"sec_num": "1.2"
},
{
"text": "A number of theoretical explanations for Zipf's law had been derived, many reviewed by [Fedorowicz 1982] ; notably are those due to [Mandelbrot 1954 [Mandelbrot , 1957 , [Miller 1954 [Miller , 1958 , [Simon 1955] , [Booth 1967] , and [Sichel 1975 [Sichel , 1986 ]. Simon's derivation was controversial and a correspondence in the scientific press developed between Mandelbrot and Simon on the validity of this derivation (1959) (1960) (1961) ; the dispute was not resolved by the time Zipf curves for larger corpora were beginning to be computed. ",
"cite_spans": [
{
"start": 87,
"end": 104,
"text": "[Fedorowicz 1982]",
"ref_id": "BIBREF7"
},
{
"start": 132,
"end": 148,
"text": "[Mandelbrot 1954",
"ref_id": "BIBREF15"
},
{
"start": 149,
"end": 167,
"text": "[Mandelbrot , 1957",
"ref_id": null
},
{
"start": 170,
"end": 182,
"text": "[Miller 1954",
"ref_id": "BIBREF19"
},
{
"start": 183,
"end": 197,
"text": "[Miller , 1958",
"ref_id": "BIBREF21"
},
{
"start": 200,
"end": 212,
"text": "[Simon 1955]",
"ref_id": "BIBREF33"
},
{
"start": 215,
"end": 227,
"text": "[Booth 1967]",
"ref_id": "BIBREF3"
},
{
"start": 234,
"end": 246,
"text": "[Sichel 1975",
"ref_id": "BIBREF29"
},
{
"start": 247,
"end": 261,
"text": "[Sichel , 1986",
"ref_id": "BIBREF30"
},
{
"start": 421,
"end": 427,
"text": "(1959)",
"ref_id": null
},
{
"start": 428,
"end": 434,
"text": "(1960)",
"ref_id": null
},
{
"start": 435,
"end": 441,
"text": "(1961)",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Theoretical developments:",
"sec_num": "1.2"
},
{
"text": "The processing of larger corpora with 1 million words or more was facilitated by the development of PC's in the 1980's. When Zipf curves for these corpora were drawn, they were found to drop below the Zipf straight line with slope of -1 at the bottom of the curve, for rank greater than about 5,000. This is illustrated in Figure 2 , which shows the Zipf curve for the whole of the Brown corpus (1 million words), again excluding punctuations and numbers.",
"cite_spans": [],
"ref_spans": [
{
"start": 323,
"end": 331,
"text": "Figure 2",
"ref_id": null
}
],
"eq_spans": [],
"section": "Figure 2 Zipf curve for the unigrams extracted from the 1 million words of the Brown corpus showing that the Zipf curve falls below the line with slope -1 for rank > 5,000.",
"sec_num": null
},
{
"text": "This deviation from Zipf 's law was interpreted for single-author texts to represent the limited numbers of words in each author's diction. But we see in Figure 2 that a deviation also occurs for a multi-author corpus covering a wide range of domains such as the Brown corpus; so the drop in the curve is not likely to be only due to the limited number of words.",
"cite_spans": [],
"ref_spans": [
{
"start": 154,
"end": 162,
"text": "Figure 2",
"ref_id": null
}
],
"eq_spans": [],
"section": "Figure 2 Zipf curve for the unigrams extracted from the 1 million words of the Brown corpus showing that the Zipf curve falls below the line with slope -1 for rank > 5,000.",
"sec_num": null
},
{
"text": "We are going to explore the above deviation from Zipf's law for large corpora in two languages: Chinese and English. We begin with English.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Zipf curves for large English corpora",
"sec_num": "2."
},
{
"text": "The English corpora used in our experiments are the full text of articles appearing in the Wall Street Journal [Paul and Baker 1992 ] for 1987 , 1988 , 1989 , with sizes approximately 19 million, 16 million and 6 million tokens respectively. The Zipf curves for the 3 corpora are shown in Figure 3 .",
"cite_spans": [
{
"start": 111,
"end": 131,
"text": "[Paul and Baker 1992",
"ref_id": "BIBREF27"
},
{
"start": 132,
"end": 142,
"text": "] for 1987",
"ref_id": null
},
{
"start": 143,
"end": 149,
"text": ", 1988",
"ref_id": null
},
{
"start": 150,
"end": 156,
"text": ", 1989",
"ref_id": null
}
],
"ref_spans": [
{
"start": 289,
"end": 297,
"text": "Figure 3",
"ref_id": null
}
],
"eq_spans": [],
"section": "Single words",
"sec_num": "2.1"
},
{
"text": "For pre-execution of this corpus, numbers were written as words, e.g. 23 became \"twenty three\" and punctuation marks were excluded. The characters \"=\", \"#\", \"~\", \"<\", \">\", \"|\", \"+\", \"-\", \"^\", \"*\", \"@\", \"/\" and \"\\\", etc. were also ignored. ",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Single words",
"sec_num": "2.1"
},
{
"text": "Language is not made of individual words, each with its own separate piece of information, but consists of sequences of words, made up of individual words and of phrases of 2, 3 or more words together called n-grams. So it is interesting to measure the frequencies of n-grams and draw the corresponding Zipf curves.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "n-Grams",
"sec_num": "2.2"
},
{
"text": "To do this we allowed n-grams to overlap. For example, for the sentence: \"The cat sat on the mat\", there are four trigrams: (1) \"the cat sat\", (2) \"cat sat on\", (3) \"sat on the\" and 4\"on the mat\". So semantically incomplete n-grams such as \"cat sat on\" are included in our study. No n-gram crossed over a punctuation mark. So a fullstop, comma, colon, etc. always ends an n-gram and a new n-gram starts after the punctuation. Thus the sentence \"Three blind mice, see how they run\" has only three trigrams \"three blind mice\", \"see how they\" and \"how they run\".",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "n-Grams",
"sec_num": "2.2"
},
{
"text": "For each value of n between 2 and 5, we thus computed the frequencies of all n-grams in each corpus and put them in rank order as we had done for the words. This enabled us to draw the Zipf curves for 2-, 3-, 4-and 5-grams which are shown along with the single word curves in Figure 4 , Figure 5 and Figure 6 for the three corpora. These curves are similar to the first Zipf curves drawn for n-grams by [Smith and Devine 1985] ; but these earlier curves were for a much smaller corpus. ",
"cite_spans": [
{
"start": 403,
"end": 426,
"text": "[Smith and Devine 1985]",
"ref_id": "BIBREF37"
}
],
"ref_spans": [
{
"start": 276,
"end": 284,
"text": "Figure 4",
"ref_id": null
},
{
"start": 287,
"end": 295,
"text": "Figure 5",
"ref_id": null
},
{
"start": 300,
"end": 308,
"text": "Figure 6",
"ref_id": null
}
],
"eq_spans": [],
"section": "n-Grams",
"sec_num": "2.2"
},
{
"text": "The n-gram Zipf curves do not follow straight lines but curve gently downwards. The average slope decreases from about 0.66 for the bigrams to about 0.59 for the 5-grams. ",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Figure 4 Zipf curves for the WSJ87 corpus",
"sec_num": null
},
{
"text": "First for WSJ87, the crossing point between the unigram and bigram curves is at rank 2,943 and for the unigram and trigram curves, it is at rank 8,497. For WSJ88, these crossing points are similar, at rank 2,913 and at rank 8,404, and for WSJ89, they are at rank 2,908 and 7,960. So the unigram curves cross the bigram curves when the rank \u2245 3,000 in all 3 cases, and for the unigram and trigram curves, they cross at rank \u2245 8,000.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Figure 6 Zipf curves for the WSJ89 corpus",
"sec_num": null
},
{
"text": "The ten most common words, bigrams and trigrams in the combined WSJ corpus of 40 million words are listed in Table 1 . ",
"cite_spans": [],
"ref_spans": [
{
"start": 109,
"end": 116,
"text": "Table 1",
"ref_id": "TABREF1"
}
],
"eq_spans": [],
"section": "Figure 6 Zipf curves for the WSJ89 corpus",
"sec_num": null
},
{
"text": "The size of the hapax legomena (tokens with frequency 1) for the n-grams rises rapidly with n as shown in Table 2a , but it can not rise above the number of tokens; so the rate of increase has slowed when n = 5 since almost all tokens are in the hapax legomena. The hapax dis legomena (tokens with frequency 2) is much smaller and reaches a maximum for trigrams from all 3 corpora (see Table 2b ) because almost all of the tokens have frequency 1, leaving a smaller number with frequency 2 when n = 4 and 5. ",
"cite_spans": [],
"ref_spans": [
{
"start": 106,
"end": 114,
"text": "Table 2a",
"ref_id": "TABREF2"
},
{
"start": 386,
"end": 394,
"text": "Table 2b",
"ref_id": "TABREF3"
}
],
"eq_spans": [],
"section": "Hapax legomena and dis legomena",
"sec_num": "2.3"
},
{
"text": "It can be argued that most of the n-grams in the hapax legomena or hapax dis legomena are not meaningful, since they are semantically incomplete. Certainly that meaning may be incomplete and they need the words on either side of them to realise their full meaning. But then it can be argued that this is true of every n-gram (and indeed for every word). So we take the view that every n-gram taken from a natural language text produced by humans has meaning, though often incomplete.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "The nature of n-grams",
"sec_num": "2.4"
},
{
"text": "However, Miller's monkey typing on a word typewriter would produce mainly meaningless n-grams, e.g. \"the the the\", as well as those others which have meaning by accident. The number of possible n-grams which the monkey can type is huge. For example, for the WSJ87 corpus there are more than 10 15 possible trigrams of which less than 7 million produced by humans appear in the Hapax legomenon for the corpus.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "The nature of n-grams",
"sec_num": "2.4"
},
{
"text": "Whatever one's views on the meaning of some of these incomplete n-grams, we report in this paper on the Zipf curves for all n-grams in a corpus. A later paper will include discussion on the equivalent curves for semantically complete phrases.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "The nature of n-grams",
"sec_num": "2.4"
},
{
"text": "One of our reasons for including all n-grams is that statistical language modellers have been using n-grams, similar to the ones we have defined, which include semantically incomplete n-grams, with great success in modelling language over the last 20 years [Jelinek and Mercer 1985] ; [O'Boyle, Owens and Smith 1994] ; [Ney 1999 ].",
"cite_spans": [
{
"start": 257,
"end": 282,
"text": "[Jelinek and Mercer 1985]",
"ref_id": "BIBREF12"
},
{
"start": 285,
"end": 316,
"text": "[O'Boyle, Owens and Smith 1994]",
"ref_id": null
},
{
"start": 319,
"end": 328,
"text": "[Ney 1999",
"ref_id": "BIBREF23"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "The nature of n-grams",
"sec_num": "2.4"
},
{
"text": "In Chinese, compound words can be created, made up of two or more characters. However, it is not always easy to automatically segment a written sentence in Chinese into compound words as these are not separated by spaces as in English. Nevertheless, the extraction of a word sequence from a Chinese document has been the subject of study by many authors [Zhu 1981 ]; [Chen and Shi 1992] ; [Bates, Chen, Li, Opie and Tzeng 1993] ; [Packard 2000] ; [Sproat 2002 ]; [Tsai and Hsu 2002] ; who reference other papers.",
"cite_spans": [
{
"start": 354,
"end": 363,
"text": "[Zhu 1981",
"ref_id": "BIBREF41"
},
{
"start": 367,
"end": 386,
"text": "[Chen and Shi 1992]",
"ref_id": "BIBREF6"
},
{
"start": 389,
"end": 427,
"text": "[Bates, Chen, Li, Opie and Tzeng 1993]",
"ref_id": "BIBREF2"
},
{
"start": 430,
"end": 444,
"text": "[Packard 2000]",
"ref_id": "BIBREF26"
},
{
"start": 447,
"end": 459,
"text": "[Sproat 2002",
"ref_id": "BIBREF38"
},
{
"start": 463,
"end": 482,
"text": "[Tsai and Hsu 2002]",
"ref_id": "BIBREF39"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Zipf Curves for Chinese Corpora",
"sec_num": "3."
},
{
"text": "Unfortunately, there is still ambiguity in the process of compound word extraction. For example, the following string of characters can be broken into the words: Therefore, it is difficult to write a computer program to extract the correct word sequence, and for a corpus of 250 million syllables, it is impossible to do by hand. So we proceeded as follows: first of all, we used a 50,000 word-syllable dictionary (which can be found at http://www.euroasiasoftware.com/), but the extraction of the words from the text is still partly ambiguous. When a sequence of syllables was found that matched a word in the dictionary, it was usually accepted as a word. When an ambiguity occurs, e.g. \u66b4\u98ce\u9aa4\u96e8 which can be one word hurricane, or two bi-syllable words: \u66b4\u98ce \u9aa4\u96e8 storm shower, then the longer word was accepted \u66b4\u98ce\u9aa4\u96e8 hurricane. Similarly, \u767e\u4e07\u5bcc\u7fc1 millionaire is accepted as one word instead of the three words \u767e\u4e07 \u5bcc \u7fc1 million(s) rich elder.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Zipf Curves for Chinese Corpora",
"sec_num": "3."
},
{
"text": "\u5317\u4eac (Beijing) \u57ce (city) \u91cc (in) \u4ea4\u901a (traffic) \u7e41\u5fd9 (busy) (",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Zipf Curves for Chinese Corpora",
"sec_num": "3."
},
{
"text": "Although the whole corpus could not be checked manually, the higher frequency n-grams can be checked, for example the following 6-gram has been broken into the pattern: \u57c3\u53ca \u603b \u7edf \u7a46 \u5df4 \u62c9 \u514b rather than the pattern \u57c3\u53ca \u603b\u7edf \u7a46\u5df4\u62c9\u514b Egyptian president Mubarak.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Zipf Curves for Chinese Corpora",
"sec_num": "3."
},
{
"text": "This occurs 1,865 times and could be corrected for all 1,865 occurrences in one step all over the corpus. Another example is the 7-gram: \u963f \u8054 \u914b \u4e4c \u5179 \u522b \u514b occurring 7 times which should be the 2 names \u963f\u8054\u914b \u4e4c\u5179\u522b\u514b Alanqiu Wuzibieke to be correct. Because of the multiple occurrence of n-grams all of which can be corrected by one change, this speeded-up the manual process considerably. Checking all of the high frequency n-grams took more than 2 months work; after this, a check on a test text of 3,117 tokens was found to have 82 errors (2.6%) by an independent native speaker (other than the authors), which we took as acceptable. (The corpus can be made available on request to q.le@qub.ac.uk or fj.smith@qub.ac.uk).",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Zipf Curves for Chinese Corpora",
"sec_num": "3."
},
{
"text": "Two corpora were used in our experiments: the TREC corpus and the Mandarin Daily News corpus. Both are from the Linguistic Data Consortium 1 .",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Zipf Curves for Chinese Corpora",
"sec_num": "3."
},
{
"text": "There is a small overlap between the Chinese TREC corpus and the Mandarin News corpus (less than 10% of the smaller TREC corpus). This overlap could have been removed, but it was not, to retain the full size of both corpora in the analysis. The effect of overlap will be small.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Zipf Curves for Chinese Corpora",
"sec_num": "3."
},
{
"text": "The TREC Corpus was obtained from the full articles in the People's Daily Newspaper from 01/1991 to 12/1993 and from the Xinhua News Agency from 04/1994 to 09/1995.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "TREC Corpus (compound words)",
"sec_num": "3.1"
},
{
"text": "The Zipf curves for the TREC compound words are shown in Figure 7 . Note that the unigram curve is different from the curve for English, first with a slope less than 1 then falling rapidly after a rank of about 1,000. ",
"cite_spans": [],
"ref_spans": [
{
"start": 57,
"end": 65,
"text": "Figure 7",
"ref_id": null
}
],
"eq_spans": [],
"section": "TREC Corpus (compound words)",
"sec_num": "3.1"
},
{
"text": "The crossing-point between compound word unigrams and bigrams is at rank: 4,999, and between the unigram and trigram curves at rank: 8,589, similar to English.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Figure 7 Zipf curves for Mandarin compound words from TREC",
"sec_num": null
},
{
"text": "The second corpus is the Mandarin News corpus, obtained from the People's Daily Newspaper from 1991 to 1996 (125 million syllables); from the Xinhua News Agency from 1994 to 1996 (25 million syllables); and from transcripts from China Radio International broadcast from 1994 to 1996 (100 million syllables), altogether over 250 million syllables.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Mandarin News corpus (compound words)",
"sec_num": "3.2"
},
{
"text": "The Zipf curves for the Mandarin News compound words are drawn in Figure 8 and look like those for the TREC corpus. The rapid fall in the curve after rank 10,000 is due to the restricted word dictionary of 50,000 word types used in the experiment. The ten highest frequency Mandarin unigrams, bigrams and trigrams from the Mandarin News are in Table 3 and Table 4 . ",
"cite_spans": [],
"ref_spans": [
{
"start": 66,
"end": 74,
"text": "Figure 8",
"ref_id": null
},
{
"start": 344,
"end": 351,
"text": "Table 3",
"ref_id": "TABREF5"
},
{
"start": 356,
"end": 363,
"text": "Table 4",
"ref_id": "TABREF6"
}
],
"eq_spans": [],
"section": "Mandarin News corpus (compound words)",
"sec_num": "3.2"
},
{
"text": "The crossing-point between compound word unigrams and bigrams is at rank: 5,544 and between unigrams and trigrams at rank: 9,577 similar to previous values for TREC and English. So these appear to be invariants of language, not just of English. Figure 9 falls very rapidly after rank about 300. It is similar to previous curves, one for a smaller Chinese corpus of 2 million tokens by [Clark, Lua and McCallum 1986] and one for 10 million tokens by [Sproat 2002 ]. The Zipf curves for syllable n-grams for the TREC corpus are also shown in Figure 9 . ",
"cite_spans": [
{
"start": 385,
"end": 415,
"text": "[Clark, Lua and McCallum 1986]",
"ref_id": "BIBREF5"
},
{
"start": 449,
"end": 461,
"text": "[Sproat 2002",
"ref_id": "BIBREF38"
}
],
"ref_spans": [
{
"start": 245,
"end": 253,
"text": "Figure 9",
"ref_id": null
},
{
"start": 540,
"end": 548,
"text": "Figure 9",
"ref_id": null
}
],
"eq_spans": [],
"section": "Figure 8 Zipf curves for the Mandarin News corpus (compound words)",
"sec_num": null
},
{
"text": "Except for the unigrams, the shapes of the other TREC syllable n-gram Zipf curves are similar to but not quite the same as those for compound words. In particular the syllable bigram curve for Chinese is more curved than the word curve because there are more highfrequency syllable bigrams than word bigrams. The crossing points between the syllable unigram curve and the bigram and trigram curves are at rank: 1,224 and 1,920, respectively, very different from compound words.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Figure 9 Zipf curves for syllables from the TREC Mandarin corpus",
"sec_num": null
},
{
"text": "The number of syllable-types (i.e. unigrams) in the Mandarin News corpus is 6,800, similar to the TREC corpus. The Zipf curves and crossing points are also similar as shown in Figure 10 . ",
"cite_spans": [],
"ref_spans": [
{
"start": 176,
"end": 185,
"text": "Figure 10",
"ref_id": "FIGREF0"
}
],
"eq_spans": [],
"section": "Figure 9 Zipf curves for syllables from the TREC Mandarin corpus",
"sec_num": null
},
{
"text": "The hapax legomena and dis legomena for the Chinese corpora Zipf curves are shown in Table 5a and 5b. Their behaviour as n increases is similar to the English corpora. ",
"cite_spans": [],
"ref_spans": [
{
"start": 85,
"end": 93,
"text": "Table 5a",
"ref_id": "TABREF7"
}
],
"eq_spans": [],
"section": "Figure 10 Zipf curves for syllables from the Mandarin News corpus",
"sec_num": null
},
{
"text": "Following a suggestion by a reviewer of this paper, we built the Zipf curves on English 2-byte and 3-byte substrings to compare them with the Chinese syllable results.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "English byte substring",
"sec_num": "4.2"
},
{
"text": "From the WSJ88 corpus, we built a corpus of the first 2 million tokens. Then we took 2byte and 3-byte moving windows on this corpus ignoring spaces and stopping the 2-bytes or 3-bytes at punctuation marks. As predicted by the reviewer, the results in Figure 11 and Figure 12 show that the Zipf curve for 3-byte substrings looks particularly similar to the Chinese syllable curves. ",
"cite_spans": [],
"ref_spans": [
{
"start": 251,
"end": 260,
"text": "Figure 11",
"ref_id": "FIGREF0"
},
{
"start": 265,
"end": 274,
"text": "Figure 12",
"ref_id": "FIGREF0"
}
],
"eq_spans": [],
"section": "English byte substring",
"sec_num": "4.2"
},
{
"text": "Note that the number of 2-byte and 3-byte types in these curves equal 673 and 10,548, compared with the maximum possible numbers 26 2 = 676 and 26 3 = 17,576. ",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Figure 12 Zipf curves for English 3-byte substrings",
"sec_num": null
},
{
"text": "The Zipf curves for unigrams for the combined WSJ corpus, the Mandarin News word corpus, the Mandarin News syllable corpus and 3-byte English corpus are compared in Figure 13 . ",
"cite_spans": [],
"ref_spans": [
{
"start": 165,
"end": 174,
"text": "Figure 13",
"ref_id": "FIGREF0"
}
],
"eq_spans": [],
"section": "Figure 13 Comparison of Zipf curves for unigrams",
"sec_num": null
},
{
"text": "These Figures show two things as n increases. First, the curves straighten out for high n.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Figure 17 Comparison of Zipf curves for 5-grams",
"sec_num": null
},
{
"text": "Secondly, the number of hapax legomena becomes very large, often larger than one would expect from the last 10 steps of the rank-frequency step function. This is exactly the pattern one gets when Markov models are used to generate data sets [Baayen 1991 [Baayen , 2001 ].",
"cite_spans": [
{
"start": 241,
"end": 253,
"text": "[Baayen 1991",
"ref_id": "BIBREF0"
},
{
"start": 254,
"end": 268,
"text": "[Baayen , 2001",
"ref_id": "BIBREF1"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Figure 17 Comparison of Zipf curves for 5-grams",
"sec_num": null
},
{
"text": "The theoretical justifications for Zipf's law by Mandelbrot, Miller, Simon and others were based on single word tokens and they worked quite well for small corpora, but none of them could predict the drop in the Zipf curve below Zipf 's law for English and Chinese when the rank is greater than 5,000 word types. In the case of Chinese syllables, Zipf 's law could not hold for rank greater than about 100, but when these syllables are combined into compound words then Zipf 's law is valid for a wider range, up to about rank 1,000. Therefore, by combining Chinese syllables into larger units, Zipf 's law was extended from rank 100 to rank 1,000. This led us to combine all syllable n-grams, to see if the law could be extended to even higher rank and to combine word n-grams in Chinese and English for the same purpose.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Combined n-grams",
"sec_num": "6."
},
{
"text": "We therefore put all unigrams and n-grams together with their frequencies into one large file, sorted on frequency and put in rank order as previously. The resulting combined Zipf curve is shown with the unigram curve for English words for the combined WSJ corpus in Figure 18 and for the Chinese syllables for Mandarin News in Figure 19 . ",
"cite_spans": [],
"ref_spans": [
{
"start": 267,
"end": 276,
"text": "Figure 18",
"ref_id": "FIGREF0"
},
{
"start": 328,
"end": 337,
"text": "Figure 19",
"ref_id": "FIGREF0"
}
],
"eq_spans": [],
"section": "Combined n-grams",
"sec_num": "6."
},
{
"text": "This shows the remarkable result that as the unigram curve drops away from Zipf 's slope of -1, the shortfall is made up almost exactly by the n-grams in both cases, even though those shortfalls are very different in the two cases. So when all n-grams are combined together, including unigrams, Zipf's law is found to be approximately correct with a slope close to -1 for all ranks. If semantically incomplete n-grams had been excluded from this analysis, this result would not have been obtained.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Figure 19 The unigram and combined curves for the Mandarin News syllable corpus",
"sec_num": null
},
{
"text": "The resulting Zipf curves for the combined n-grams from all of the corpora are shown in ",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Figure 19 The unigram and combined curves for the Mandarin News syllable corpus",
"sec_num": null
},
{
"text": "This shows that the 6 combined Zipf curves are all approximately straight lines with slopes close to -1 for all ranks > 1,000. For ranks < 1,000, the unigram curves dominate and are not so straight. As in Figure 18 and 19, the n-grams (n \u2265 2) almost exactly make up for the deviation of the unigram Zipf curve from Zipf 's law for the six very different unigram curves. So the results in Figure 20 are a new confirmation of Zipf 's original law in an extended form.",
"cite_spans": [],
"ref_spans": [
{
"start": 205,
"end": 214,
"text": "Figure 18",
"ref_id": "FIGREF0"
},
{
"start": 388,
"end": 397,
"text": "Figure 20",
"ref_id": null
}
],
"eq_spans": [],
"section": "Figure 20 Combined Zipf curves for both of the languages",
"sec_num": null
},
{
"text": "This paper reports on the results of some experiments conducted on Zipf curves for English and Chinese corpora. It was confirmed that Zipf curves on a log-log graph for single word unigram distributions for both languages fall below the straight line with slope -1 as predicted by Zipf 's law. The deviation from Zipf 's law occurs at a rank close to rank = 5,000, for the 3 corpora in English and 2 corpora in Chinese. This rank (5,000) is also the rank near which the unigram and bigram Zipf curves cross for all 5 corpora.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Summary and Conclusions",
"sec_num": "7."
},
{
"text": "The more significant result was the discovery that when the frequency distribution of words is combined with the distributions of all 2-, 3-, 4-and 5-grams, the combined Zipf curve approximately obeys Zipf 's law for all ranks and frequencies for both languages. This effectively extends Zipf 's law, with the higher n-grams almost exactly making up for the falloff in the Zipf curve for words. Furthermore, this extended form of Zipf 's law also holds for the syllables of Chinese (as well as for 2-byte and 3-byte word fragments in English), even though the distribution of syllable unigrams is very different from the distribution for words.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Summary and Conclusions",
"sec_num": "7."
},
{
"text": "This paper does not explain why Zipf 's law in an extended form is valid for large corpora or what this result means. This must be left for further experiments and other researchers. However, preliminary results, not yet complete, for other languages suggest that these results are universal for all languages. We also know that they do not hold for all artificial distributions of words, because some experiments with computer generated artificial distributions did not yield an extended Zipf curve, (with a random distribution, and with Zipf distributions for words with slopes \u03b2 = 2 and \u03b2 = 0.5).",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Summary and Conclusions",
"sec_num": "7."
},
{
"text": "The earlier derivations of Zipf's law due to Mandelbrot, Miller, Simon and others fail to predict the fall-off in the Zipf curve from about rank 5,000 and to predict the extended form of Zipf's law for the combined n-gram curves. We believe that this is because these derivations do not properly take account of the fact that each token is part of a sequence and its information is dependent on a conditional probability, conditional on the words or characters around it; this can be approximated in terms of the frequency of n-grams [O 'Boyle, Owens and Smith 1994 ].",
"cite_spans": [
{
"start": 534,
"end": 565,
"text": "[O 'Boyle, Owens and Smith 1994",
"ref_id": "BIBREF24"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Summary and Conclusions",
"sec_num": "7."
},
{
"text": "http://www.ldc.upenn.edu/",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
},
{
"text": "This is how Chinese people read and write the year for example 1993 as \"one nine nine three\"; therefore we elminated numbers but kept the written form.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
}
],
"back_matter": [
{
"text": "The authors would like to express their appreciation to reviewers of this paper whose comments and suggestions made a great improvement to the paper and to Dr Xiaoyu Qiao for her contribution of testing and standardising the Chinese morphology.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Acknowledgement",
"sec_num": null
}
],
"bib_entries": {
"BIBREF0": {
"ref_id": "b0",
"title": "A Stochastic Process for Word Frequency Distributions",
"authors": [
{
"first": "H",
"middle": [],
"last": "Baayen",
"suffix": ""
}
],
"year": 1991,
"venue": "Proceedings of the 29th Annual Meeting of the Association for Computational Linguistics (ACL-29)",
"volume": "",
"issue": "",
"pages": "271--278",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Baayen, H. \"A Stochastic Process for Word Frequency Distributions\", In Proceedings of the 29th Annual Meeting of the Association for Computational Linguistics (ACL-29), Berkeley, California, USA, 1991, pp. 271-278.",
"links": null
},
"BIBREF1": {
"ref_id": "b1",
"title": "Word Frequency Distributions",
"authors": [
{
"first": "H",
"middle": [],
"last": "Baayen",
"suffix": ""
}
],
"year": 2001,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Baayen, H. \"Word Frequency Distributions\", Kluwer Academic Publishers, 2001.",
"links": null
},
"BIBREF2": {
"ref_id": "b2",
"title": "Where is the boundary between compounds and phrases in Chinese ? A reply to",
"authors": [
{
"first": "E",
"middle": [],
"last": "Bates",
"suffix": ""
},
{
"first": "S",
"middle": [],
"last": "Chen",
"suffix": ""
},
{
"first": "P",
"middle": [],
"last": "Li",
"suffix": ""
},
{
"first": "M",
"middle": [],
"last": "Opie",
"suffix": ""
},
{
"first": "O",
"middle": [],
"last": "Tzeng",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Zhou",
"suffix": ""
}
],
"year": 1993,
"venue": "Brain and Language",
"volume": "45",
"issue": "",
"pages": "94--107",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Bates, E., Chen, S., Li, P., Opie, M. and Tzeng, O. \"Where is the boundary between compounds and phrases in Chinese ? A reply to Zhou et al.\", Brain and Language, 45, 1993, pp. 94-107.",
"links": null
},
"BIBREF3": {
"ref_id": "b3",
"title": "A Law of Occurrences for Words of Low Frequency",
"authors": [
{
"first": "A",
"middle": [
"D"
],
"last": "Booth",
"suffix": ""
}
],
"year": 1967,
"venue": "Information and Control",
"volume": "10",
"issue": "4",
"pages": "386--393",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Booth, A. D. \"A Law of Occurrences for Words of Low Frequency\", Information and Control, Vol. 10, No. 4, April 1967, pp. 386-393.",
"links": null
},
"BIBREF4": {
"ref_id": "b4",
"title": "A Rationale for an Asymptotic Lognormal Form of Word Frequency Distributions",
"authors": [
{
"first": "J",
"middle": [
"B"
],
"last": "Carroll",
"suffix": ""
}
],
"year": 1969,
"venue": "Research Bulletin --Educational Testing Service",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Carroll, J. B. \"A Rationale for an Asymptotic Lognormal Form of Word Frequency Distributions\", Research Bulletin --Educational Testing Service, Princeton, November 1969.",
"links": null
},
"BIBREF5": {
"ref_id": "b5",
"title": "Using Zipf's Law to Analyse the Rank Frequency Distribution of Elements in Chinese Text",
"authors": [
{
"first": "J",
"middle": [
"L"
],
"last": "Clark",
"suffix": ""
},
{
"first": "K",
"middle": [
"T"
],
"last": "Lua",
"suffix": ""
},
{
"first": "J",
"middle": [],
"last": "Mccallum",
"suffix": ""
}
],
"year": 1986,
"venue": "Proceedings of International Conference on Chinese Computing",
"volume": "",
"issue": "",
"pages": "321--324",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Clark, J. L., Lua, K. T. and McCallum, J. (1986). \"Using Zipf's Law to Analyse the Rank Frequency Distribution of Elements in Chinese Text\", In Proceedings of International Conference on Chinese Computing, Singapore, August 1986, pp. 321-324.",
"links": null
},
"BIBREF6": {
"ref_id": "b6",
"title": "On the feeding relation between syntax and morphology: Evidence from Chinese V-N compounds",
"authors": [
{
"first": "S",
"middle": [],
"last": "Chen",
"suffix": ""
},
{
"first": "D-X",
"middle": [],
"last": "Shi",
"suffix": ""
}
],
"year": 1992,
"venue": "Proceedings of the Third International Symposium on Chinese Languages and Linguistics",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Chen, S., and Shi, D-X, \"On the feeding relation between syntax and morphology: Evidence from Chinese V-N compounds\", In Proceedings of the Third International Symposium on Chinese Languages and Linguistics, Taiwan: Chinghwa University, 1992.",
"links": null
},
"BIBREF7": {
"ref_id": "b7",
"title": "A Zipfian Model of an Automatic Bibliographic System: an Application to MEDLINE",
"authors": [
{
"first": "J",
"middle": [],
"last": "Fedorowicz",
"suffix": ""
}
],
"year": 1982,
"venue": "Journal of American Society of Information Science",
"volume": "33",
"issue": "",
"pages": "223--232",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Fedorowicz, J. \"A Zipfian Model of an Automatic Bibliographic System: an Application to MEDLINE\", Journal of American Society of Information Science, Vol. 33, 1982, pp. 223-232.",
"links": null
},
"BIBREF8": {
"ref_id": "b8",
"title": "Two Regimes in the Frequency of Words and the Origin of Complex Lexicons",
"authors": [
{
"first": "R",
"middle": [],
"last": "Ferrer I Cancho",
"suffix": ""
},
{
"first": "R",
"middle": [
"V"
],
"last": "Sol\u00e9",
"suffix": ""
}
],
"year": 2002,
"venue": "Journal of Quantitative Linguistics",
"volume": "8",
"issue": "3",
"pages": "165--173",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Ferrer i Cancho, R., Sol\u00e9, R. V., \"Two Regimes in the Frequency of Words and the Origin of Complex Lexicons\", Journal of Quantitative Linguistics, Vol. 8, No. 3, 2002, pp. 165 - 173.",
"links": null
},
"BIBREF9": {
"ref_id": "b9",
"title": "Manual of Information to Accompany A Standard Corpus of Present-Day Edited American English, for use with Digital Computers",
"authors": [
{
"first": "W",
"middle": [
"N"
],
"last": "Francis",
"suffix": ""
},
{
"first": "H",
"middle": [],
"last": "Kucera",
"suffix": ""
}
],
"year": 1964,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Francis, W. N. and Kucera, H. \"Manual of Information to Accompany A Standard Corpus of Present-Day Edited American English, for use with Digital Computers\", Department of Linguistics, Brown University, Providence, Rhode Island, 1964.",
"links": null
},
"BIBREF10": {
"ref_id": "b10",
"title": "Studies on Zipf 's Law",
"authors": [
{
"first": "H",
"middle": [],
"last": "Guiter",
"suffix": ""
},
{
"first": "M",
"middle": [],
"last": "Arapov",
"suffix": ""
}
],
"year": 1982,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Guiter, H. and Arapov, M., editors. \"Studies on Zipf 's Law\", Brochmeyer, Bochum, 1982.",
"links": null
},
"BIBREF11": {
"ref_id": "b11",
"title": "Word Length, Word Frequencies and Zipf's Law in the Greek Language",
"authors": [
{
"first": "N",
"middle": [],
"last": "Hatzigeorgiu",
"suffix": ""
},
{
"first": "G",
"middle": [],
"last": "Mikros",
"suffix": ""
},
{
"first": "G",
"middle": [],
"last": "Carayannis",
"suffix": ""
}
],
"year": 2001,
"venue": "Journal of Quantitative Linguistics",
"volume": "8",
"issue": "3",
"pages": "175--185",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Hatzigeorgiu, N., Mikros, G., and Carayannis, G., \"Word Length, Word Frequencies and Zipf's Law in the Greek Language\", Journal of Quantitative Linguistics, Vol. 8, No. 3, 2001, pp. 175 -185.",
"links": null
},
"BIBREF12": {
"ref_id": "b12",
"title": "Probability distribution estimation from sparse data",
"authors": [
{
"first": "F",
"middle": [],
"last": "Jelinek",
"suffix": ""
},
{
"first": "R",
"middle": [
"L"
],
"last": "Mercer",
"suffix": ""
}
],
"year": 1985,
"venue": "IBM Technical Disclosure Bulletin",
"volume": "28",
"issue": "6",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Jelinek, F., Mercer, R. L. \"Probability distribution estimation from sparse data\", IBM Technical Disclosure Bulletin, Vol. 28, No. 6, November 1985.",
"links": null
},
"BIBREF13": {
"ref_id": "b13",
"title": "Zipf's Law in Importance of Genes for Cancer Classification Using Microarray Data",
"authors": [
{
"first": "W",
"middle": [],
"last": "Li",
"suffix": ""
}
],
"year": 2001,
"venue": "Laboratory of Statistical Genetics",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Li, W. \"Zipf's Law in Importance of Genes for Cancer Classification Using Microarray Data \", Laboratory of Statistical Genetics, Rockefeller University, New York, 2001.",
"links": null
},
"BIBREF14": {
"ref_id": "b14",
"title": "An Information Theory of the Statistical Structure of Language",
"authors": [
{
"first": "B",
"middle": [],
"last": "Mandelbrot",
"suffix": ""
}
],
"year": 1953,
"venue": "Communication Theory",
"volume": "",
"issue": "",
"pages": "486--502",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Mandelbrot, B. \"An Information Theory of the Statistical Structure of Language\", Communication Theory, edited by Willis Jackson, New York: Academic Press, 1953, pp. 486-502.",
"links": null
},
"BIBREF15": {
"ref_id": "b15",
"title": "Simple Games of Strategy Occurring in Communication through Natural Languages",
"authors": [
{
"first": "B",
"middle": [],
"last": "Mandelbrot",
"suffix": ""
}
],
"year": 1954,
"venue": "Transactions of the IRE Professional Group on Information Theory",
"volume": "3",
"issue": "",
"pages": "124--137",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Mandelbrot, B. \"Simple Games of Strategy Occurring in Communication through Natural Languages\", Transactions of the IRE Professional Group on Information Theory , Vol. 3, 1954, pp. 124-137.",
"links": null
},
"BIBREF16": {
"ref_id": "b16",
"title": "A note on a class of skew distribution function analysis and critique of a paper by H. A. Simon",
"authors": [
{
"first": "B",
"middle": [],
"last": "Mandelbrot",
"suffix": ""
}
],
"year": 1959,
"venue": "Information and Control",
"volume": "2",
"issue": "",
"pages": "90--99",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Mandelbrot, B. \"A note on a class of skew distribution function analysis and critique of a paper by H. A. Simon\", Information and Control, Vol. 2, 1959, pp. 90-99.",
"links": null
},
"BIBREF17": {
"ref_id": "b17",
"title": "Final note on a class of skew distribution functions: analysis and critique of a model due to H. A. Simon",
"authors": [
{
"first": "B",
"middle": [],
"last": "Mandelbrot",
"suffix": ""
}
],
"year": 1961,
"venue": "Information and Control",
"volume": "4",
"issue": "",
"pages": "198--216",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Mandelbrot, B. \"Final note on a class of skew distribution functions: analysis and critique of a model due to H. A. Simon\", Information and Control, Vol. 4, 1961, pp. 198-216.",
"links": null
},
"BIBREF18": {
"ref_id": "b18",
"title": "Post Scriptum to 'final note",
"authors": [
{
"first": "B",
"middle": [],
"last": "Mandelbrot",
"suffix": ""
}
],
"year": 1961,
"venue": "Information and Control",
"volume": "4",
"issue": "",
"pages": "300--304",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Mandelbrot, B. B. \"Post Scriptum to 'final note'\", Information and Control, Vol. 4, 1961, pp. 300-304.",
"links": null
},
"BIBREF19": {
"ref_id": "b19",
"title": "Communication",
"authors": [
{
"first": "G",
"middle": [
"A"
],
"last": "Miller",
"suffix": ""
}
],
"year": 1954,
"venue": "Annual Review of Psychology",
"volume": "5",
"issue": "",
"pages": "401--420",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Miller, G. A. \"Communication\", Annual Review of Psychology, 5, 1954, pp. 401-420.",
"links": null
},
"BIBREF20": {
"ref_id": "b20",
"title": "Some effects of intermittent silence",
"authors": [
{
"first": "G",
"middle": [
"A"
],
"last": "Miller",
"suffix": ""
}
],
"year": 1957,
"venue": "The American Journal of Psychology",
"volume": "52",
"issue": "",
"pages": "311--314",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Miller, G. A. \"Some effects of intermittent silence\", The American Journal of Psychology, 52, 1957, pp. 311-314.",
"links": null
},
"BIBREF21": {
"ref_id": "b21",
"title": "Length-Frequency Statistics for Written English",
"authors": [
{
"first": "G",
"middle": [
"A"
],
"last": "Miller",
"suffix": ""
},
{
"first": "E",
"middle": [
"B"
],
"last": "Newman",
"suffix": ""
},
{
"first": "E",
"middle": [
"A"
],
"last": "Friedman",
"suffix": ""
}
],
"year": 1958,
"venue": "Information and control",
"volume": "1",
"issue": "",
"pages": "370--389",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Miller, G. A., Newman, E. B. and Friedman, E. A. \"Length-Frequency Statistics for Written English\", Information and control, Vol. 1, 1958, pp. 370-389.",
"links": null
},
"BIBREF22": {
"ref_id": "b22",
"title": "Beyond the Zipf-Mandelbrot Law in Quantitative Linguistics",
"authors": [
{
"first": "M",
"middle": [],
"last": "Montemurro",
"suffix": ""
}
],
"year": 2001,
"venue": "Physica A: Statistical Mechanics and its Applications",
"volume": "300",
"issue": "",
"pages": "567--578",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Montemurro, M. \"Beyond the Zipf-Mandelbrot Law in Quantitative Linguistics\", Physica A: Statistical Mechanics and its Applications, Vol. 300, Issues 3-4, November 2001, pp. 567-578.",
"links": null
},
"BIBREF23": {
"ref_id": "b23",
"title": "The Use of the Maximum Likelihood Criterion in Language Modelling",
"authors": [
{
"first": "H",
"middle": [],
"last": "Ney",
"suffix": ""
}
],
"year": 1999,
"venue": "Computational Models of Speech Pattern Processing",
"volume": "",
"issue": "",
"pages": "259--279",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Ney, H. \"The Use of the Maximum Likelihood Criterion in Language Modelling\", In K. Ponting (*ed.): Computational Models of Speech Pattern Processing, Springer, Berlin, Germany, 1999, pp. 259-279.",
"links": null
},
"BIBREF24": {
"ref_id": "b24",
"title": "A weighted average n-gram model of natural language",
"authors": [
{
"first": "P",
"middle": [],
"last": "O 'boyle",
"suffix": ""
},
{
"first": "M",
"middle": [],
"last": "Owens",
"suffix": ""
},
{
"first": "F",
"middle": [
"J"
],
"last": "Smith",
"suffix": ""
}
],
"year": 1994,
"venue": "Computer Speech and Language",
"volume": "8",
"issue": "",
"pages": "337--349",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "O 'Boyle, P., Owens, M. and Smith, F. J. \"A weighted average n-gram model of natural language\", Computer Speech and Language, Vol. 8, 1994, pp. 337-349.",
"links": null
},
"BIBREF25": {
"ref_id": "b25",
"title": "Generalized Z-distribution generating the well-known 'rank-distributions",
"authors": [
{
"first": "J",
"middle": [
"K"
],
"last": "Orlov",
"suffix": ""
},
{
"first": "R",
"middle": [
"Y"
],
"last": "Chitashvili",
"suffix": ""
}
],
"year": 1983,
"venue": "Bulletin of the Academy of Sciences",
"volume": "110",
"issue": "2",
"pages": "269--272",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Orlov, J. K. and Chitashvili, R. Y. \"Generalized Z-distribution generating the well-known 'rank-distributions' \", Bulletin of the Academy of Sciences, Georgia, 110.2, 1983, pp. 269-272.",
"links": null
},
"BIBREF26": {
"ref_id": "b26",
"title": "The Morphology of Chinese A Linguistic and Cognitive Approach",
"authors": [
{
"first": "J",
"middle": [
"L"
],
"last": "Packard",
"suffix": ""
}
],
"year": 2000,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Packard, J. L., \"The Morphology of Chinese A Linguistic and Cognitive Approach\", Cambridge University Press, 2000, UK.",
"links": null
},
"BIBREF27": {
"ref_id": "b27",
"title": "The Design for the Wall Street Journal-based CSR Corpus",
"authors": [
{
"first": "D",
"middle": [
"B"
],
"last": "Paul",
"suffix": ""
},
{
"first": "J",
"middle": [],
"last": "Baker",
"suffix": ""
}
],
"year": 1992,
"venue": "Proceedings of International Conference on Spoken Language Processing (ICSLP)",
"volume": "",
"issue": "",
"pages": "899--902",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Paul, D. B. and Baker, J. M. \"The Design for the Wall Street Journal-based CSR Corpus\", In Proceedings of International Conference on Spoken Language Processing (ICSLP) , Banff, Alberta, Canada, October 1992, pp. 899-902.",
"links": null
},
"BIBREF28": {
"ref_id": "b28",
"title": "Relating Turing's Formula and Zipf's Law",
"authors": [
{
"first": "C",
"middle": [],
"last": "Samuelson",
"suffix": ""
}
],
"year": 1996,
"venue": "Proceedings of the 4 th Workshop on Very Large Corpora",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Samuelson, C. \"Relating Turing's Formula and Zipf's Law\", In Proceedings of the 4 th Workshop on Very Large Corpora, Copenhagen, Denmark, 1996.",
"links": null
},
"BIBREF29": {
"ref_id": "b29",
"title": "On a Distribution Law for Word Frequencies",
"authors": [
{
"first": "H",
"middle": [],
"last": "Sichel",
"suffix": ""
}
],
"year": 1975,
"venue": "Journal of the American Statistical Association",
"volume": "70",
"issue": "",
"pages": "542--547",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Sichel, H. S. \"On a Distribution Law for Word Frequencies\", Journal of the American Statistical Association, 70, 1975, pp. 542-547.",
"links": null
},
"BIBREF30": {
"ref_id": "b30",
"title": "Word Frequency Distributions and Type-Token Characteristics",
"authors": [
{
"first": "H",
"middle": [],
"last": "Sichel",
"suffix": ""
}
],
"year": 1986,
"venue": "Mathematical Scientist",
"volume": "11",
"issue": "",
"pages": "45--72",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Sichel, H. S. \"Word Frequency Distributions and Type-Token Characteristics\", Mathematical Scientist, 11, 1986, pp. 45-72.",
"links": null
},
"BIBREF31": {
"ref_id": "b31",
"title": "Modelling Species-Abundance Frequencies and Species-Individual Functions with the Generalized Inverse Gaussion-Poisson Distribution",
"authors": [
{
"first": "H",
"middle": [
"S"
],
"last": "Sichel",
"suffix": ""
}
],
"year": 1997,
"venue": "South African Statistical Journal",
"volume": "31",
"issue": "",
"pages": "13--37",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Sichel, H. S. \"Modelling Species-Abundance Frequencies and Species-Individual Functions with the Generalized Inverse Gaussion-Poisson Distribution\", South African Statistical Journal, 31, 1997, pp. 13-37.",
"links": null
},
"BIBREF32": {
"ref_id": "b32",
"title": "Citations and the Zipf-Mandelbrot Law",
"authors": [
{
"first": "Z",
"middle": [],
"last": "Silagadze",
"suffix": ""
}
],
"year": 1997,
"venue": "Complex Systems",
"volume": "11",
"issue": "6",
"pages": "487--499",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Silagadze, Z. K. \"Citations and the Zipf-Mandelbrot Law\", Complex Systems, Vol. 11, No. 6, 1997, pp. 487-499.",
"links": null
},
"BIBREF33": {
"ref_id": "b33",
"title": "On a Class of Skew Distribution Functions",
"authors": [
{
"first": "H",
"middle": [
"A"
],
"last": "Simon",
"suffix": ""
}
],
"year": 1955,
"venue": "Biometrika",
"volume": "42",
"issue": "",
"pages": "425--440",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Simon, H. A. \"On a Class of Skew Distribution Functions\", Biometrika, Vol. 42, 1955, pp. 425-440.",
"links": null
},
"BIBREF34": {
"ref_id": "b34",
"title": "Some Further Notes on a Class of Skew Distribution Functions",
"authors": [
{
"first": "H",
"middle": [
"A"
],
"last": "Simon",
"suffix": ""
}
],
"year": 1960,
"venue": "Information and Control",
"volume": "3",
"issue": "",
"pages": "80--88",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Simon, H. A. \"Some Further Notes on a Class of Skew Distribution Functions\", Information and Control, Vol. 3, 1960, pp. 80-88.",
"links": null
},
"BIBREF35": {
"ref_id": "b35",
"title": "Reply to 'final note' by Benoit Mandelbrot",
"authors": [
{
"first": "H",
"middle": [
"A"
],
"last": "Simon",
"suffix": ""
}
],
"year": 1961,
"venue": "Information and Control",
"volume": "4",
"issue": "",
"pages": "217--223",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Simon, H. A. \"Reply to 'final note' by Benoit Mandelbrot\", Information and Control, Vol. 4, 1961, pp. 217-223.",
"links": null
},
"BIBREF36": {
"ref_id": "b36",
"title": "Reply to Dr. Mandelbrot's post Scriptum",
"authors": [
{
"first": "H",
"middle": [
"A"
],
"last": "Simon",
"suffix": ""
}
],
"year": 1961,
"venue": "Information and Control",
"volume": "4",
"issue": "",
"pages": "305--308",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Simon, H. A. \"Reply to Dr. Mandelbrot's post Scriptum\", Information and Control, Vol. 4, 1961, pp. 305-308.",
"links": null
},
"BIBREF37": {
"ref_id": "b37",
"title": "Storing and Retrieving Word Phrases",
"authors": [
{
"first": "F",
"middle": [
"J"
],
"last": "Smith",
"suffix": ""
},
{
"first": "K",
"middle": [],
"last": "Devine",
"suffix": ""
}
],
"year": 1985,
"venue": "Information Processing and Management",
"volume": "21",
"issue": "",
"pages": "215--224",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Smith, F. J. and Devine, K. \"Storing and Retrieving Word Phrases\", Information Processing and Management, Vol. 21, No. 3, 1985, pp. 215-224.",
"links": null
},
"BIBREF38": {
"ref_id": "b38",
"title": "Corpus-Based methods in Chinese Morphology",
"authors": [
{
"first": "R",
"middle": [],
"last": "Sproat",
"suffix": ""
}
],
"year": 2002,
"venue": "Tutorial of the 19 th International Conference on Computational Linguistics (COLING)",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Sproat, R., \"Corpus-Based methods in Chinese Morphology\", Tutorial of the 19 th International Conference on Computational Linguistics (COLING), Taipei, Taiwan, August 2002.",
"links": null
},
"BIBREF39": {
"ref_id": "b39",
"title": "Applying an NVEF Word-Pair Identifier to the Chinese Syllable-to-Word Conversion Problem",
"authors": [
{
"first": "J-L",
"middle": [],
"last": "Tsai",
"suffix": ""
},
{
"first": "W-L",
"middle": [],
"last": "Hsu",
"suffix": ""
}
],
"year": 2002,
"venue": "Proceedings of the 19 th International Conference on Computational Linguistics (COLING)",
"volume": "",
"issue": "",
"pages": "1016--1022",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Tsai, J-L., Hsu, W-L. \"Applying an NVEF Word-Pair Identifier to the Chinese Syllable-to- Word Conversion Problem\", In Proceedings of the 19 th International Conference on Computational Linguistics (COLING), Taipei, Taiwan, August 2002, pp. 1016-1022.",
"links": null
},
"BIBREF40": {
"ref_id": "b40",
"title": "Zipf-Scaling Description in the DNA Sequence",
"authors": [
{
"first": "Y",
"middle": [],
"last": "Yonezawa",
"suffix": ""
},
{
"first": "H",
"middle": [],
"last": "Motohasi",
"suffix": ""
}
],
"year": 1999,
"venue": "Proceedings of the 10 th Workshop on Genome Informatics",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Yonezawa, Y. and Motohasi, H. \"Zipf-Scaling Description in the DNA Sequence\", In Proceedings of the 10 th Workshop on Genome Informatics, Japan, December 1999.",
"links": null
},
"BIBREF41": {
"ref_id": "b41",
"title": "Yufa Jiangyi (Chinese Syntax)",
"authors": [
{
"first": "D",
"middle": [
"X"
],
"last": "Zhu",
"suffix": ""
}
],
"year": 1981,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Zhu, D. X. \"Yufa Jiangyi (Chinese Syntax)\", Shanghai: The Commercial Publisher, China, 1981.",
"links": null
},
"BIBREF42": {
"ref_id": "b42",
"title": "Human Behaviour and the Principle of Least Effort",
"authors": [
{
"first": "G",
"middle": [
"K"
],
"last": "Zipf",
"suffix": ""
}
],
"year": 1949,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Zipf, G. K. \"Human Behaviour and the Principle of Least Effort\", Reading, MA: Addison- Wesley Publishing Co., 1949.",
"links": null
}
},
"ref_entries": {
"FIGREF0": {
"text": "Zipf curve for the unigrams extracted from a 250,000-word tokens corpus.",
"type_str": "figure",
"uris": null,
"num": null
},
"FIGREF5": {
"text": "Figure 5 Zipf curves for the WSJ88 corpus",
"type_str": "figure",
"uris": null,
"num": null
},
"FIGREF10": {
"text": "Figure 11 Zipf curves for English 2-byte substrings",
"type_str": "figure",
"uris": null,
"num": null
},
"FIGREF12": {
"text": "Figure 14 Comparison of Zipf curves for bigrams",
"type_str": "figure",
"uris": null,
"num": null
},
"FIGREF13": {
"text": "Figure 18 The unigram and combined curves for the combined WSJ corpus",
"type_str": "figure",
"uris": null,
"num": null
},
"FIGREF14": {
"text": "Figure 20.",
"type_str": "figure",
"uris": null,
"num": null
},
"TABREF1": {
"text": "",
"type_str": "table",
"num": null,
"content": "<table><tr><td>Unigrams</td><td/><td/><td>Bigrams</td><td/><td>Trigrams</td></tr><tr><td colspan=\"3\">Frequency Token Frequency</td><td>Token</td><td>Frequency</td><td>Token</td></tr><tr><td>2,057,968</td><td>the</td><td>217,427</td><td>of the</td><td>42,030</td><td>the U. S.</td></tr><tr><td>973,650</td><td>of</td><td>173,797</td><td>in the</td><td>27,260</td><td>in nineteen eighty</td></tr><tr><td>940,525</td><td>to</td><td>110,291</td><td>million dollars</td><td>24,165</td><td>cents a share</td></tr><tr><td>853,342</td><td>a</td><td>89,184</td><td>U. S.</td><td>18,233</td><td>nineteen eighty six</td></tr><tr><td>825,489</td><td>and</td><td>83,799</td><td>nineteen eighty</td><td>16,786</td><td>nineteen eighty seven</td></tr><tr><td>711,462</td><td>in</td><td>76,187</td><td>for the</td><td>15,316</td><td>five million dollars</td></tr><tr><td>368,012</td><td>that</td><td>72,312</td><td>to the</td><td>14,943</td><td>million dollars or</td></tr><tr><td>362,771</td><td>for</td><td>65,565</td><td>on the</td><td>14,517</td><td>million dollars in</td></tr><tr><td>298,646</td><td>one</td><td>63,838</td><td>one hundred</td><td>12,327</td><td>in New York</td></tr><tr><td>281,190</td><td>is</td><td>55,014</td><td>that the</td><td>11,981</td><td>a year earlier</td></tr></table>",
"html": null
},
"TABREF2": {
"text": "",
"type_str": "table",
"num": null,
"content": "<table><tr><td/><td>Corpus</td><td>WSJ87</td><td>WSJ88</td><td>WSJ89</td></tr><tr><td colspan=\"2\">No of Tokens</td><td>18,790,794</td><td>15,757,051</td><td>5,946,585</td></tr><tr><td colspan=\"2\">No of Types</td><td>114,581</td><td>108,522</td><td>71,837</td></tr><tr><td>Hapax</td><td>Unigram</td><td>38,853</td><td>36,945</td><td>25,162</td></tr><tr><td>legomena</td><td>Bigram</td><td>1,786,290</td><td>1,620,385</td><td>851,542</td></tr><tr><td/><td>Trigram</td><td>6,601,243</td><td>5,799,257</td><td>2,598,509</td></tr><tr><td/><td>4-gram</td><td>10,635,310</td><td>9,137,402</td><td>3,736,880</td></tr><tr><td/><td>5-gram</td><td>12,493,656</td><td>10,612,036</td><td>4,376,741</td></tr></table>",
"html": null
},
"TABREF3": {
"text": "",
"type_str": "table",
"num": null,
"content": "<table><tr><td colspan=\"2\">Corpus</td><td>WSJ87</td><td>WSJ88</td><td>WSJ89</td></tr><tr><td colspan=\"2\">No of Tokens</td><td>18,790,794</td><td>15,757,051</td><td>5,946,585</td></tr><tr><td colspan=\"2\">No of Types</td><td>114,581</td><td>108,522</td><td>71,837</td></tr><tr><td>Hapax</td><td>Unigram</td><td>14,855</td><td>14,431</td><td>9,861</td></tr><tr><td>dis-</td><td>Bigram</td><td>349,205</td><td>314,496</td><td>155,068</td></tr><tr><td>-legomena</td><td>Trigram</td><td>742,771</td><td>632,372</td><td>251,435</td></tr><tr><td/><td>4-gram</td><td>670,106</td><td>546,951</td><td>190,947</td></tr><tr><td/><td>5-gram</td><td>485,487</td><td>389,113</td><td>130,544</td></tr></table>",
"html": null
},
"TABREF5": {
"text": "",
"type_str": "table",
"num": null,
"content": "<table><tr><td>Rank</td><td/><td>Unigrams</td><td/><td/><td>Bigrams</td><td/></tr><tr><td/><td>Freq</td><td>Token</td><td>Meaning</td><td>Freq</td><td>Token</td><td>Meaning</td></tr><tr><td>1</td><td>7,356,017</td><td>\u7684</td><td>of</td><td>114,910</td><td>\u65e5 \u7535</td><td>daily news</td></tr><tr><td>2</td><td>1,825,758</td><td>\u5728</td><td>in / at</td><td>92,259</td><td>\u8fd9 \u4e00</td><td>this one</td></tr><tr><td>3</td><td>1,515,473</td><td>\u548c</td><td>and</td><td>82,705</td><td>\u8fd9 \u662f</td><td>this is</td></tr><tr><td>4</td><td>1,502,098</td><td>\u4e86</td><td>perfective</td><td>81,930</td><td>\u4e2d\u56fd \u7684</td><td>of China</td></tr><tr><td/><td/><td/><td>marker</td><td/><td/><td/></tr><tr><td>5</td><td>1,331,433</td><td>\u662f</td><td>yes / right</td><td>79,390</td><td>\u7684 \u53d1\u5c55</td><td>of development</td></tr></table>",
"html": null
},
"TABREF6": {
"text": "Because of the difficulty in extracting the compound words in Chinese, we decided to draw Zipf curves for the syllables for both Chinese corpora. TREC has 19,546,872 syllable tokens but only 6,300 syllable types, so it is not surprising that the Zipf curve for syllable unigrams in Chinese in",
"type_str": "table",
"num": null,
"content": "<table><tr><td>Rank</td><td/><td>Trigrams</td><td/></tr><tr><td/><td>Freq</td><td>Token</td><td>Meaning</td></tr><tr><td>1</td><td>60,214</td><td>\u56fd\u9645 \u5e7f\u64ad \u7535\u53f0</td><td>international broadcast station</td></tr><tr><td>2</td><td>60,057</td><td>\u4e2d\u56fd \u56fd\u9645 \u5e7f\u64ad</td><td>China international broadcast</td></tr><tr><td>3</td><td>35,584</td><td>\u4e00 \u4e5d \u4e5d</td><td>one nine nine 2</td></tr><tr><td>4</td><td>28,589</td><td>\u636e \u4e2d\u56fd \u56fd\u9645</td><td>according to China international</td></tr><tr><td>5</td><td>28,240</td><td>\u5e7f\u64ad \u7535\u53f0 \u62a5\u5bfc</td><td>broadcast station report</td></tr><tr><td>6</td><td>26,240</td><td>\u5b66\u5386 \u6536\u542c \u8bed\u8a00</td><td>degree listen (to) language</td></tr><tr><td>7</td><td>26,232</td><td>\u5e74\u9f84 \u5b66\u5386 \u6536\u542c</td><td>age degree listen (to)</td></tr><tr><td>8</td><td>26,203</td><td>\u6536\u542c \u8bed\u8a00 \u5907\u6ce8</td><td>listen (to) language remarks/notes</td></tr><tr><td>9</td><td>26,154</td><td>\u804c\u4e1a \u5e74\u9f84 \u5b66\u5386</td><td>profession age degree</td></tr><tr><td>10</td><td>26,081</td><td>\u4f20\u771f \u5355\u4f4d \u804c\u4e1a</td><td>fax department profession</td></tr></table>",
"html": null
},
"TABREF7": {
"text": "",
"type_str": "table",
"num": null,
"content": "<table><tr><td colspan=\"2\">Corpus</td><td>TREC</td><td>TREC compound</td><td>Mandarin News</td><td>Mandarin News</td></tr><tr><td/><td/><td>syllables</td><td>words</td><td>syllables</td><td>compound words</td></tr><tr><td colspan=\"2\">No of Tokens</td><td>19,720,320</td><td>13,467,443</td><td>223,222,788</td><td>153,942,010</td></tr><tr><td colspan=\"2\">No of Types</td><td>6,356</td><td>20,587</td><td>6,891</td><td>29,688</td></tr><tr><td>Hapax</td><td>Unigram</td><td>676</td><td>2,642</td><td>259</td><td>4,192</td></tr><tr><td>legomena</td><td>Bigram</td><td>351,691</td><td>1,013,276</td><td>667,966</td><td>2,671,406</td></tr><tr><td/><td>Trigram</td><td>2,447,451</td><td>4,009,020</td><td>8,462,775</td><td>17,794,466</td></tr><tr><td/><td>4-gram</td><td>5,309,654</td><td>5,661,530</td><td>23,812,934</td><td>30,885,192</td></tr><tr><td/><td>5-gram</td><td>7,279,824</td><td>5,875,696</td><td>37,348,300</td><td>34,617,579</td></tr></table>",
"html": null
},
"TABREF8": {
"text": "",
"type_str": "table",
"num": null,
"content": "<table><tr><td>dis-</td><td>Bigram</td><td>131,619</td><td>287,225</td><td>294,634</td><td>1,001,809</td></tr><tr><td>legomena</td><td>Trigram</td><td>565,069</td><td>624,895</td><td>2,591,830</td><td>4,549,748</td></tr><tr><td/><td>4-gram</td><td>834,227</td><td>553,965</td><td>5,806,633</td><td>6,168,487</td></tr><tr><td/><td>5-gram</td><td>840,276</td><td>417,656</td><td>7,874,536</td><td>6,057,974</td></tr><tr><td colspan=\"2\">Corpus</td><td>TREC</td><td>TREC compound</td><td>Mandarin News</td><td>Mandarin News</td></tr><tr><td/><td/><td>syllables</td><td>words</td><td>syllables</td><td>compound words</td></tr><tr><td colspan=\"2\">No of Tokens</td><td>19,720,320</td><td>13,467,443</td><td>223,222,788</td><td>153,942,010</td></tr><tr><td colspan=\"2\">No of Types</td><td>6,356</td><td>20,587</td><td>6,891</td><td>29,688</td></tr><tr><td>Hapax</td><td>Unigram</td><td>347</td><td>1,260</td><td>155</td><td>1,701</td></tr></table>",
"html": null
}
}
}
} |