comprehend.d.ts 127 KB
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
import {Request} from '../lib/request';
import {Response} from '../lib/response';
import {AWSError} from '../lib/error';
import {Service} from '../lib/service';
import {ServiceConfigurationOptions} from '../lib/service';
import {ConfigBase as Config} from '../lib/config';
interface Blob {}
declare class Comprehend extends Service {
  /**
   * Constructs a service object. This object has one method for each API operation.
   */
  constructor(options?: Comprehend.Types.ClientConfiguration)
  config: Config & Comprehend.Types.ClientConfiguration;
  /**
   * Determines the dominant language of the input text for a batch of documents. For a list of languages that Amazon Comprehend can detect, see Amazon Comprehend Supported Languages. 
   */
  batchDetectDominantLanguage(params: Comprehend.Types.BatchDetectDominantLanguageRequest, callback?: (err: AWSError, data: Comprehend.Types.BatchDetectDominantLanguageResponse) => void): Request<Comprehend.Types.BatchDetectDominantLanguageResponse, AWSError>;
  /**
   * Determines the dominant language of the input text for a batch of documents. For a list of languages that Amazon Comprehend can detect, see Amazon Comprehend Supported Languages. 
   */
  batchDetectDominantLanguage(callback?: (err: AWSError, data: Comprehend.Types.BatchDetectDominantLanguageResponse) => void): Request<Comprehend.Types.BatchDetectDominantLanguageResponse, AWSError>;
  /**
   * Inspects the text of a batch of documents for named entities and returns information about them. For more information about named entities, see how-entities 
   */
  batchDetectEntities(params: Comprehend.Types.BatchDetectEntitiesRequest, callback?: (err: AWSError, data: Comprehend.Types.BatchDetectEntitiesResponse) => void): Request<Comprehend.Types.BatchDetectEntitiesResponse, AWSError>;
  /**
   * Inspects the text of a batch of documents for named entities and returns information about them. For more information about named entities, see how-entities 
   */
  batchDetectEntities(callback?: (err: AWSError, data: Comprehend.Types.BatchDetectEntitiesResponse) => void): Request<Comprehend.Types.BatchDetectEntitiesResponse, AWSError>;
  /**
   * Detects the key noun phrases found in a batch of documents.
   */
  batchDetectKeyPhrases(params: Comprehend.Types.BatchDetectKeyPhrasesRequest, callback?: (err: AWSError, data: Comprehend.Types.BatchDetectKeyPhrasesResponse) => void): Request<Comprehend.Types.BatchDetectKeyPhrasesResponse, AWSError>;
  /**
   * Detects the key noun phrases found in a batch of documents.
   */
  batchDetectKeyPhrases(callback?: (err: AWSError, data: Comprehend.Types.BatchDetectKeyPhrasesResponse) => void): Request<Comprehend.Types.BatchDetectKeyPhrasesResponse, AWSError>;
  /**
   * Inspects a batch of documents and returns an inference of the prevailing sentiment, POSITIVE, NEUTRAL, MIXED, or NEGATIVE, in each one.
   */
  batchDetectSentiment(params: Comprehend.Types.BatchDetectSentimentRequest, callback?: (err: AWSError, data: Comprehend.Types.BatchDetectSentimentResponse) => void): Request<Comprehend.Types.BatchDetectSentimentResponse, AWSError>;
  /**
   * Inspects a batch of documents and returns an inference of the prevailing sentiment, POSITIVE, NEUTRAL, MIXED, or NEGATIVE, in each one.
   */
  batchDetectSentiment(callback?: (err: AWSError, data: Comprehend.Types.BatchDetectSentimentResponse) => void): Request<Comprehend.Types.BatchDetectSentimentResponse, AWSError>;
  /**
   * Inspects the text of a batch of documents for the syntax and part of speech of the words in the document and returns information about them. For more information, see how-syntax.
   */
  batchDetectSyntax(params: Comprehend.Types.BatchDetectSyntaxRequest, callback?: (err: AWSError, data: Comprehend.Types.BatchDetectSyntaxResponse) => void): Request<Comprehend.Types.BatchDetectSyntaxResponse, AWSError>;
  /**
   * Inspects the text of a batch of documents for the syntax and part of speech of the words in the document and returns information about them. For more information, see how-syntax.
   */
  batchDetectSyntax(callback?: (err: AWSError, data: Comprehend.Types.BatchDetectSyntaxResponse) => void): Request<Comprehend.Types.BatchDetectSyntaxResponse, AWSError>;
  /**
   * Creates a new document classifier that you can use to categorize documents. To create a classifier you provide a set of training documents that labeled with the categories that you want to use. After the classifier is trained you can use it to categorize a set of labeled documents into the categories. For more information, see how-document-classification.
   */
  createDocumentClassifier(params: Comprehend.Types.CreateDocumentClassifierRequest, callback?: (err: AWSError, data: Comprehend.Types.CreateDocumentClassifierResponse) => void): Request<Comprehend.Types.CreateDocumentClassifierResponse, AWSError>;
  /**
   * Creates a new document classifier that you can use to categorize documents. To create a classifier you provide a set of training documents that labeled with the categories that you want to use. After the classifier is trained you can use it to categorize a set of labeled documents into the categories. For more information, see how-document-classification.
   */
  createDocumentClassifier(callback?: (err: AWSError, data: Comprehend.Types.CreateDocumentClassifierResponse) => void): Request<Comprehend.Types.CreateDocumentClassifierResponse, AWSError>;
  /**
   * Creates an entity recognizer using submitted files. After your CreateEntityRecognizer request is submitted, you can check job status using the API. 
   */
  createEntityRecognizer(params: Comprehend.Types.CreateEntityRecognizerRequest, callback?: (err: AWSError, data: Comprehend.Types.CreateEntityRecognizerResponse) => void): Request<Comprehend.Types.CreateEntityRecognizerResponse, AWSError>;
  /**
   * Creates an entity recognizer using submitted files. After your CreateEntityRecognizer request is submitted, you can check job status using the API. 
   */
  createEntityRecognizer(callback?: (err: AWSError, data: Comprehend.Types.CreateEntityRecognizerResponse) => void): Request<Comprehend.Types.CreateEntityRecognizerResponse, AWSError>;
  /**
   * Deletes a previously created document classifier Only those classifiers that are in terminated states (IN_ERROR, TRAINED) will be deleted. If an active inference job is using the model, a ResourceInUseException will be returned. This is an asynchronous action that puts the classifier into a DELETING state, and it is then removed by a background job. Once removed, the classifier disappears from your account and is no longer available for use. 
   */
  deleteDocumentClassifier(params: Comprehend.Types.DeleteDocumentClassifierRequest, callback?: (err: AWSError, data: Comprehend.Types.DeleteDocumentClassifierResponse) => void): Request<Comprehend.Types.DeleteDocumentClassifierResponse, AWSError>;
  /**
   * Deletes a previously created document classifier Only those classifiers that are in terminated states (IN_ERROR, TRAINED) will be deleted. If an active inference job is using the model, a ResourceInUseException will be returned. This is an asynchronous action that puts the classifier into a DELETING state, and it is then removed by a background job. Once removed, the classifier disappears from your account and is no longer available for use. 
   */
  deleteDocumentClassifier(callback?: (err: AWSError, data: Comprehend.Types.DeleteDocumentClassifierResponse) => void): Request<Comprehend.Types.DeleteDocumentClassifierResponse, AWSError>;
  /**
   * Deletes an entity recognizer. Only those recognizers that are in terminated states (IN_ERROR, TRAINED) will be deleted. If an active inference job is using the model, a ResourceInUseException will be returned. This is an asynchronous action that puts the recognizer into a DELETING state, and it is then removed by a background job. Once removed, the recognizer disappears from your account and is no longer available for use. 
   */
  deleteEntityRecognizer(params: Comprehend.Types.DeleteEntityRecognizerRequest, callback?: (err: AWSError, data: Comprehend.Types.DeleteEntityRecognizerResponse) => void): Request<Comprehend.Types.DeleteEntityRecognizerResponse, AWSError>;
  /**
   * Deletes an entity recognizer. Only those recognizers that are in terminated states (IN_ERROR, TRAINED) will be deleted. If an active inference job is using the model, a ResourceInUseException will be returned. This is an asynchronous action that puts the recognizer into a DELETING state, and it is then removed by a background job. Once removed, the recognizer disappears from your account and is no longer available for use. 
   */
  deleteEntityRecognizer(callback?: (err: AWSError, data: Comprehend.Types.DeleteEntityRecognizerResponse) => void): Request<Comprehend.Types.DeleteEntityRecognizerResponse, AWSError>;
  /**
   * Gets the properties associated with a document classification job. Use this operation to get the status of a classification job.
   */
  describeDocumentClassificationJob(params: Comprehend.Types.DescribeDocumentClassificationJobRequest, callback?: (err: AWSError, data: Comprehend.Types.DescribeDocumentClassificationJobResponse) => void): Request<Comprehend.Types.DescribeDocumentClassificationJobResponse, AWSError>;
  /**
   * Gets the properties associated with a document classification job. Use this operation to get the status of a classification job.
   */
  describeDocumentClassificationJob(callback?: (err: AWSError, data: Comprehend.Types.DescribeDocumentClassificationJobResponse) => void): Request<Comprehend.Types.DescribeDocumentClassificationJobResponse, AWSError>;
  /**
   * Gets the properties associated with a document classifier.
   */
  describeDocumentClassifier(params: Comprehend.Types.DescribeDocumentClassifierRequest, callback?: (err: AWSError, data: Comprehend.Types.DescribeDocumentClassifierResponse) => void): Request<Comprehend.Types.DescribeDocumentClassifierResponse, AWSError>;
  /**
   * Gets the properties associated with a document classifier.
   */
  describeDocumentClassifier(callback?: (err: AWSError, data: Comprehend.Types.DescribeDocumentClassifierResponse) => void): Request<Comprehend.Types.DescribeDocumentClassifierResponse, AWSError>;
  /**
   * Gets the properties associated with a dominant language detection job. Use this operation to get the status of a detection job.
   */
  describeDominantLanguageDetectionJob(params: Comprehend.Types.DescribeDominantLanguageDetectionJobRequest, callback?: (err: AWSError, data: Comprehend.Types.DescribeDominantLanguageDetectionJobResponse) => void): Request<Comprehend.Types.DescribeDominantLanguageDetectionJobResponse, AWSError>;
  /**
   * Gets the properties associated with a dominant language detection job. Use this operation to get the status of a detection job.
   */
  describeDominantLanguageDetectionJob(callback?: (err: AWSError, data: Comprehend.Types.DescribeDominantLanguageDetectionJobResponse) => void): Request<Comprehend.Types.DescribeDominantLanguageDetectionJobResponse, AWSError>;
  /**
   * Gets the properties associated with an entities detection job. Use this operation to get the status of a detection job.
   */
  describeEntitiesDetectionJob(params: Comprehend.Types.DescribeEntitiesDetectionJobRequest, callback?: (err: AWSError, data: Comprehend.Types.DescribeEntitiesDetectionJobResponse) => void): Request<Comprehend.Types.DescribeEntitiesDetectionJobResponse, AWSError>;
  /**
   * Gets the properties associated with an entities detection job. Use this operation to get the status of a detection job.
   */
  describeEntitiesDetectionJob(callback?: (err: AWSError, data: Comprehend.Types.DescribeEntitiesDetectionJobResponse) => void): Request<Comprehend.Types.DescribeEntitiesDetectionJobResponse, AWSError>;
  /**
   * Provides details about an entity recognizer including status, S3 buckets containing training data, recognizer metadata, metrics, and so on.
   */
  describeEntityRecognizer(params: Comprehend.Types.DescribeEntityRecognizerRequest, callback?: (err: AWSError, data: Comprehend.Types.DescribeEntityRecognizerResponse) => void): Request<Comprehend.Types.DescribeEntityRecognizerResponse, AWSError>;
  /**
   * Provides details about an entity recognizer including status, S3 buckets containing training data, recognizer metadata, metrics, and so on.
   */
  describeEntityRecognizer(callback?: (err: AWSError, data: Comprehend.Types.DescribeEntityRecognizerResponse) => void): Request<Comprehend.Types.DescribeEntityRecognizerResponse, AWSError>;
  /**
   * Gets the properties associated with a key phrases detection job. Use this operation to get the status of a detection job.
   */
  describeKeyPhrasesDetectionJob(params: Comprehend.Types.DescribeKeyPhrasesDetectionJobRequest, callback?: (err: AWSError, data: Comprehend.Types.DescribeKeyPhrasesDetectionJobResponse) => void): Request<Comprehend.Types.DescribeKeyPhrasesDetectionJobResponse, AWSError>;
  /**
   * Gets the properties associated with a key phrases detection job. Use this operation to get the status of a detection job.
   */
  describeKeyPhrasesDetectionJob(callback?: (err: AWSError, data: Comprehend.Types.DescribeKeyPhrasesDetectionJobResponse) => void): Request<Comprehend.Types.DescribeKeyPhrasesDetectionJobResponse, AWSError>;
  /**
   * Gets the properties associated with a sentiment detection job. Use this operation to get the status of a detection job.
   */
  describeSentimentDetectionJob(params: Comprehend.Types.DescribeSentimentDetectionJobRequest, callback?: (err: AWSError, data: Comprehend.Types.DescribeSentimentDetectionJobResponse) => void): Request<Comprehend.Types.DescribeSentimentDetectionJobResponse, AWSError>;
  /**
   * Gets the properties associated with a sentiment detection job. Use this operation to get the status of a detection job.
   */
  describeSentimentDetectionJob(callback?: (err: AWSError, data: Comprehend.Types.DescribeSentimentDetectionJobResponse) => void): Request<Comprehend.Types.DescribeSentimentDetectionJobResponse, AWSError>;
  /**
   * Gets the properties associated with a topic detection job. Use this operation to get the status of a detection job.
   */
  describeTopicsDetectionJob(params: Comprehend.Types.DescribeTopicsDetectionJobRequest, callback?: (err: AWSError, data: Comprehend.Types.DescribeTopicsDetectionJobResponse) => void): Request<Comprehend.Types.DescribeTopicsDetectionJobResponse, AWSError>;
  /**
   * Gets the properties associated with a topic detection job. Use this operation to get the status of a detection job.
   */
  describeTopicsDetectionJob(callback?: (err: AWSError, data: Comprehend.Types.DescribeTopicsDetectionJobResponse) => void): Request<Comprehend.Types.DescribeTopicsDetectionJobResponse, AWSError>;
  /**
   * Determines the dominant language of the input text. For a list of languages that Amazon Comprehend can detect, see Amazon Comprehend Supported Languages. 
   */
  detectDominantLanguage(params: Comprehend.Types.DetectDominantLanguageRequest, callback?: (err: AWSError, data: Comprehend.Types.DetectDominantLanguageResponse) => void): Request<Comprehend.Types.DetectDominantLanguageResponse, AWSError>;
  /**
   * Determines the dominant language of the input text. For a list of languages that Amazon Comprehend can detect, see Amazon Comprehend Supported Languages. 
   */
  detectDominantLanguage(callback?: (err: AWSError, data: Comprehend.Types.DetectDominantLanguageResponse) => void): Request<Comprehend.Types.DetectDominantLanguageResponse, AWSError>;
  /**
   * Inspects text for named entities, and returns information about them. For more information, about named entities, see how-entities. 
   */
  detectEntities(params: Comprehend.Types.DetectEntitiesRequest, callback?: (err: AWSError, data: Comprehend.Types.DetectEntitiesResponse) => void): Request<Comprehend.Types.DetectEntitiesResponse, AWSError>;
  /**
   * Inspects text for named entities, and returns information about them. For more information, about named entities, see how-entities. 
   */
  detectEntities(callback?: (err: AWSError, data: Comprehend.Types.DetectEntitiesResponse) => void): Request<Comprehend.Types.DetectEntitiesResponse, AWSError>;
  /**
   * Detects the key noun phrases found in the text. 
   */
  detectKeyPhrases(params: Comprehend.Types.DetectKeyPhrasesRequest, callback?: (err: AWSError, data: Comprehend.Types.DetectKeyPhrasesResponse) => void): Request<Comprehend.Types.DetectKeyPhrasesResponse, AWSError>;
  /**
   * Detects the key noun phrases found in the text. 
   */
  detectKeyPhrases(callback?: (err: AWSError, data: Comprehend.Types.DetectKeyPhrasesResponse) => void): Request<Comprehend.Types.DetectKeyPhrasesResponse, AWSError>;
  /**
   * Inspects text and returns an inference of the prevailing sentiment (POSITIVE, NEUTRAL, MIXED, or NEGATIVE). 
   */
  detectSentiment(params: Comprehend.Types.DetectSentimentRequest, callback?: (err: AWSError, data: Comprehend.Types.DetectSentimentResponse) => void): Request<Comprehend.Types.DetectSentimentResponse, AWSError>;
  /**
   * Inspects text and returns an inference of the prevailing sentiment (POSITIVE, NEUTRAL, MIXED, or NEGATIVE). 
   */
  detectSentiment(callback?: (err: AWSError, data: Comprehend.Types.DetectSentimentResponse) => void): Request<Comprehend.Types.DetectSentimentResponse, AWSError>;
  /**
   * Inspects text for syntax and the part of speech of words in the document. For more information, how-syntax.
   */
  detectSyntax(params: Comprehend.Types.DetectSyntaxRequest, callback?: (err: AWSError, data: Comprehend.Types.DetectSyntaxResponse) => void): Request<Comprehend.Types.DetectSyntaxResponse, AWSError>;
  /**
   * Inspects text for syntax and the part of speech of words in the document. For more information, how-syntax.
   */
  detectSyntax(callback?: (err: AWSError, data: Comprehend.Types.DetectSyntaxResponse) => void): Request<Comprehend.Types.DetectSyntaxResponse, AWSError>;
  /**
   * Gets a list of the documentation classification jobs that you have submitted.
   */
  listDocumentClassificationJobs(params: Comprehend.Types.ListDocumentClassificationJobsRequest, callback?: (err: AWSError, data: Comprehend.Types.ListDocumentClassificationJobsResponse) => void): Request<Comprehend.Types.ListDocumentClassificationJobsResponse, AWSError>;
  /**
   * Gets a list of the documentation classification jobs that you have submitted.
   */
  listDocumentClassificationJobs(callback?: (err: AWSError, data: Comprehend.Types.ListDocumentClassificationJobsResponse) => void): Request<Comprehend.Types.ListDocumentClassificationJobsResponse, AWSError>;
  /**
   * Gets a list of the document classifiers that you have created.
   */
  listDocumentClassifiers(params: Comprehend.Types.ListDocumentClassifiersRequest, callback?: (err: AWSError, data: Comprehend.Types.ListDocumentClassifiersResponse) => void): Request<Comprehend.Types.ListDocumentClassifiersResponse, AWSError>;
  /**
   * Gets a list of the document classifiers that you have created.
   */
  listDocumentClassifiers(callback?: (err: AWSError, data: Comprehend.Types.ListDocumentClassifiersResponse) => void): Request<Comprehend.Types.ListDocumentClassifiersResponse, AWSError>;
  /**
   * Gets a list of the dominant language detection jobs that you have submitted.
   */
  listDominantLanguageDetectionJobs(params: Comprehend.Types.ListDominantLanguageDetectionJobsRequest, callback?: (err: AWSError, data: Comprehend.Types.ListDominantLanguageDetectionJobsResponse) => void): Request<Comprehend.Types.ListDominantLanguageDetectionJobsResponse, AWSError>;
  /**
   * Gets a list of the dominant language detection jobs that you have submitted.
   */
  listDominantLanguageDetectionJobs(callback?: (err: AWSError, data: Comprehend.Types.ListDominantLanguageDetectionJobsResponse) => void): Request<Comprehend.Types.ListDominantLanguageDetectionJobsResponse, AWSError>;
  /**
   * Gets a list of the entity detection jobs that you have submitted.
   */
  listEntitiesDetectionJobs(params: Comprehend.Types.ListEntitiesDetectionJobsRequest, callback?: (err: AWSError, data: Comprehend.Types.ListEntitiesDetectionJobsResponse) => void): Request<Comprehend.Types.ListEntitiesDetectionJobsResponse, AWSError>;
  /**
   * Gets a list of the entity detection jobs that you have submitted.
   */
  listEntitiesDetectionJobs(callback?: (err: AWSError, data: Comprehend.Types.ListEntitiesDetectionJobsResponse) => void): Request<Comprehend.Types.ListEntitiesDetectionJobsResponse, AWSError>;
  /**
   * Gets a list of the properties of all entity recognizers that you created, including recognizers currently in training. Allows you to filter the list of recognizers based on criteria such as status and submission time. This call returns up to 500 entity recognizers in the list, with a default number of 100 recognizers in the list. The results of this list are not in any particular order. Please get the list and sort locally if needed.
   */
  listEntityRecognizers(params: Comprehend.Types.ListEntityRecognizersRequest, callback?: (err: AWSError, data: Comprehend.Types.ListEntityRecognizersResponse) => void): Request<Comprehend.Types.ListEntityRecognizersResponse, AWSError>;
  /**
   * Gets a list of the properties of all entity recognizers that you created, including recognizers currently in training. Allows you to filter the list of recognizers based on criteria such as status and submission time. This call returns up to 500 entity recognizers in the list, with a default number of 100 recognizers in the list. The results of this list are not in any particular order. Please get the list and sort locally if needed.
   */
  listEntityRecognizers(callback?: (err: AWSError, data: Comprehend.Types.ListEntityRecognizersResponse) => void): Request<Comprehend.Types.ListEntityRecognizersResponse, AWSError>;
  /**
   * Get a list of key phrase detection jobs that you have submitted.
   */
  listKeyPhrasesDetectionJobs(params: Comprehend.Types.ListKeyPhrasesDetectionJobsRequest, callback?: (err: AWSError, data: Comprehend.Types.ListKeyPhrasesDetectionJobsResponse) => void): Request<Comprehend.Types.ListKeyPhrasesDetectionJobsResponse, AWSError>;
  /**
   * Get a list of key phrase detection jobs that you have submitted.
   */
  listKeyPhrasesDetectionJobs(callback?: (err: AWSError, data: Comprehend.Types.ListKeyPhrasesDetectionJobsResponse) => void): Request<Comprehend.Types.ListKeyPhrasesDetectionJobsResponse, AWSError>;
  /**
   * Gets a list of sentiment detection jobs that you have submitted.
   */
  listSentimentDetectionJobs(params: Comprehend.Types.ListSentimentDetectionJobsRequest, callback?: (err: AWSError, data: Comprehend.Types.ListSentimentDetectionJobsResponse) => void): Request<Comprehend.Types.ListSentimentDetectionJobsResponse, AWSError>;
  /**
   * Gets a list of sentiment detection jobs that you have submitted.
   */
  listSentimentDetectionJobs(callback?: (err: AWSError, data: Comprehend.Types.ListSentimentDetectionJobsResponse) => void): Request<Comprehend.Types.ListSentimentDetectionJobsResponse, AWSError>;
  /**
   * Lists all tags associated with a given Amazon Comprehend resource. 
   */
  listTagsForResource(params: Comprehend.Types.ListTagsForResourceRequest, callback?: (err: AWSError, data: Comprehend.Types.ListTagsForResourceResponse) => void): Request<Comprehend.Types.ListTagsForResourceResponse, AWSError>;
  /**
   * Lists all tags associated with a given Amazon Comprehend resource. 
   */
  listTagsForResource(callback?: (err: AWSError, data: Comprehend.Types.ListTagsForResourceResponse) => void): Request<Comprehend.Types.ListTagsForResourceResponse, AWSError>;
  /**
   * Gets a list of the topic detection jobs that you have submitted.
   */
  listTopicsDetectionJobs(params: Comprehend.Types.ListTopicsDetectionJobsRequest, callback?: (err: AWSError, data: Comprehend.Types.ListTopicsDetectionJobsResponse) => void): Request<Comprehend.Types.ListTopicsDetectionJobsResponse, AWSError>;
  /**
   * Gets a list of the topic detection jobs that you have submitted.
   */
  listTopicsDetectionJobs(callback?: (err: AWSError, data: Comprehend.Types.ListTopicsDetectionJobsResponse) => void): Request<Comprehend.Types.ListTopicsDetectionJobsResponse, AWSError>;
  /**
   * Starts an asynchronous document classification job. Use the operation to track the progress of the job.
   */
  startDocumentClassificationJob(params: Comprehend.Types.StartDocumentClassificationJobRequest, callback?: (err: AWSError, data: Comprehend.Types.StartDocumentClassificationJobResponse) => void): Request<Comprehend.Types.StartDocumentClassificationJobResponse, AWSError>;
  /**
   * Starts an asynchronous document classification job. Use the operation to track the progress of the job.
   */
  startDocumentClassificationJob(callback?: (err: AWSError, data: Comprehend.Types.StartDocumentClassificationJobResponse) => void): Request<Comprehend.Types.StartDocumentClassificationJobResponse, AWSError>;
  /**
   * Starts an asynchronous dominant language detection job for a collection of documents. Use the operation to track the status of a job.
   */
  startDominantLanguageDetectionJob(params: Comprehend.Types.StartDominantLanguageDetectionJobRequest, callback?: (err: AWSError, data: Comprehend.Types.StartDominantLanguageDetectionJobResponse) => void): Request<Comprehend.Types.StartDominantLanguageDetectionJobResponse, AWSError>;
  /**
   * Starts an asynchronous dominant language detection job for a collection of documents. Use the operation to track the status of a job.
   */
  startDominantLanguageDetectionJob(callback?: (err: AWSError, data: Comprehend.Types.StartDominantLanguageDetectionJobResponse) => void): Request<Comprehend.Types.StartDominantLanguageDetectionJobResponse, AWSError>;
  /**
   * Starts an asynchronous entity detection job for a collection of documents. Use the operation to track the status of a job. This API can be used for either standard entity detection or custom entity recognition. In order to be used for custom entity recognition, the optional EntityRecognizerArn must be used in order to provide access to the recognizer being used to detect the custom entity.
   */
  startEntitiesDetectionJob(params: Comprehend.Types.StartEntitiesDetectionJobRequest, callback?: (err: AWSError, data: Comprehend.Types.StartEntitiesDetectionJobResponse) => void): Request<Comprehend.Types.StartEntitiesDetectionJobResponse, AWSError>;
  /**
   * Starts an asynchronous entity detection job for a collection of documents. Use the operation to track the status of a job. This API can be used for either standard entity detection or custom entity recognition. In order to be used for custom entity recognition, the optional EntityRecognizerArn must be used in order to provide access to the recognizer being used to detect the custom entity.
   */
  startEntitiesDetectionJob(callback?: (err: AWSError, data: Comprehend.Types.StartEntitiesDetectionJobResponse) => void): Request<Comprehend.Types.StartEntitiesDetectionJobResponse, AWSError>;
  /**
   * Starts an asynchronous key phrase detection job for a collection of documents. Use the operation to track the status of a job.
   */
  startKeyPhrasesDetectionJob(params: Comprehend.Types.StartKeyPhrasesDetectionJobRequest, callback?: (err: AWSError, data: Comprehend.Types.StartKeyPhrasesDetectionJobResponse) => void): Request<Comprehend.Types.StartKeyPhrasesDetectionJobResponse, AWSError>;
  /**
   * Starts an asynchronous key phrase detection job for a collection of documents. Use the operation to track the status of a job.
   */
  startKeyPhrasesDetectionJob(callback?: (err: AWSError, data: Comprehend.Types.StartKeyPhrasesDetectionJobResponse) => void): Request<Comprehend.Types.StartKeyPhrasesDetectionJobResponse, AWSError>;
  /**
   * Starts an asynchronous sentiment detection job for a collection of documents. use the operation to track the status of a job.
   */
  startSentimentDetectionJob(params: Comprehend.Types.StartSentimentDetectionJobRequest, callback?: (err: AWSError, data: Comprehend.Types.StartSentimentDetectionJobResponse) => void): Request<Comprehend.Types.StartSentimentDetectionJobResponse, AWSError>;
  /**
   * Starts an asynchronous sentiment detection job for a collection of documents. use the operation to track the status of a job.
   */
  startSentimentDetectionJob(callback?: (err: AWSError, data: Comprehend.Types.StartSentimentDetectionJobResponse) => void): Request<Comprehend.Types.StartSentimentDetectionJobResponse, AWSError>;
  /**
   * Starts an asynchronous topic detection job. Use the DescribeTopicDetectionJob operation to track the status of a job.
   */
  startTopicsDetectionJob(params: Comprehend.Types.StartTopicsDetectionJobRequest, callback?: (err: AWSError, data: Comprehend.Types.StartTopicsDetectionJobResponse) => void): Request<Comprehend.Types.StartTopicsDetectionJobResponse, AWSError>;
  /**
   * Starts an asynchronous topic detection job. Use the DescribeTopicDetectionJob operation to track the status of a job.
   */
  startTopicsDetectionJob(callback?: (err: AWSError, data: Comprehend.Types.StartTopicsDetectionJobResponse) => void): Request<Comprehend.Types.StartTopicsDetectionJobResponse, AWSError>;
  /**
   * Stops a dominant language detection job in progress. If the job state is IN_PROGRESS the job is marked for termination and put into the STOP_REQUESTED state. If the job completes before it can be stopped, it is put into the COMPLETED state; otherwise the job is stopped and put into the STOPPED state. If the job is in the COMPLETED or FAILED state when you call the StopDominantLanguageDetectionJob operation, the operation returns a 400 Internal Request Exception.  When a job is stopped, any documents already processed are written to the output location.
   */
  stopDominantLanguageDetectionJob(params: Comprehend.Types.StopDominantLanguageDetectionJobRequest, callback?: (err: AWSError, data: Comprehend.Types.StopDominantLanguageDetectionJobResponse) => void): Request<Comprehend.Types.StopDominantLanguageDetectionJobResponse, AWSError>;
  /**
   * Stops a dominant language detection job in progress. If the job state is IN_PROGRESS the job is marked for termination and put into the STOP_REQUESTED state. If the job completes before it can be stopped, it is put into the COMPLETED state; otherwise the job is stopped and put into the STOPPED state. If the job is in the COMPLETED or FAILED state when you call the StopDominantLanguageDetectionJob operation, the operation returns a 400 Internal Request Exception.  When a job is stopped, any documents already processed are written to the output location.
   */
  stopDominantLanguageDetectionJob(callback?: (err: AWSError, data: Comprehend.Types.StopDominantLanguageDetectionJobResponse) => void): Request<Comprehend.Types.StopDominantLanguageDetectionJobResponse, AWSError>;
  /**
   * Stops an entities detection job in progress. If the job state is IN_PROGRESS the job is marked for termination and put into the STOP_REQUESTED state. If the job completes before it can be stopped, it is put into the COMPLETED state; otherwise the job is stopped and put into the STOPPED state. If the job is in the COMPLETED or FAILED state when you call the StopDominantLanguageDetectionJob operation, the operation returns a 400 Internal Request Exception.  When a job is stopped, any documents already processed are written to the output location.
   */
  stopEntitiesDetectionJob(params: Comprehend.Types.StopEntitiesDetectionJobRequest, callback?: (err: AWSError, data: Comprehend.Types.StopEntitiesDetectionJobResponse) => void): Request<Comprehend.Types.StopEntitiesDetectionJobResponse, AWSError>;
  /**
   * Stops an entities detection job in progress. If the job state is IN_PROGRESS the job is marked for termination and put into the STOP_REQUESTED state. If the job completes before it can be stopped, it is put into the COMPLETED state; otherwise the job is stopped and put into the STOPPED state. If the job is in the COMPLETED or FAILED state when you call the StopDominantLanguageDetectionJob operation, the operation returns a 400 Internal Request Exception.  When a job is stopped, any documents already processed are written to the output location.
   */
  stopEntitiesDetectionJob(callback?: (err: AWSError, data: Comprehend.Types.StopEntitiesDetectionJobResponse) => void): Request<Comprehend.Types.StopEntitiesDetectionJobResponse, AWSError>;
  /**
   * Stops a key phrases detection job in progress. If the job state is IN_PROGRESS the job is marked for termination and put into the STOP_REQUESTED state. If the job completes before it can be stopped, it is put into the COMPLETED state; otherwise the job is stopped and put into the STOPPED state. If the job is in the COMPLETED or FAILED state when you call the StopDominantLanguageDetectionJob operation, the operation returns a 400 Internal Request Exception.  When a job is stopped, any documents already processed are written to the output location.
   */
  stopKeyPhrasesDetectionJob(params: Comprehend.Types.StopKeyPhrasesDetectionJobRequest, callback?: (err: AWSError, data: Comprehend.Types.StopKeyPhrasesDetectionJobResponse) => void): Request<Comprehend.Types.StopKeyPhrasesDetectionJobResponse, AWSError>;
  /**
   * Stops a key phrases detection job in progress. If the job state is IN_PROGRESS the job is marked for termination and put into the STOP_REQUESTED state. If the job completes before it can be stopped, it is put into the COMPLETED state; otherwise the job is stopped and put into the STOPPED state. If the job is in the COMPLETED or FAILED state when you call the StopDominantLanguageDetectionJob operation, the operation returns a 400 Internal Request Exception.  When a job is stopped, any documents already processed are written to the output location.
   */
  stopKeyPhrasesDetectionJob(callback?: (err: AWSError, data: Comprehend.Types.StopKeyPhrasesDetectionJobResponse) => void): Request<Comprehend.Types.StopKeyPhrasesDetectionJobResponse, AWSError>;
  /**
   * Stops a sentiment detection job in progress. If the job state is IN_PROGRESS the job is marked for termination and put into the STOP_REQUESTED state. If the job completes before it can be stopped, it is put into the COMPLETED state; otherwise the job is be stopped and put into the STOPPED state. If the job is in the COMPLETED or FAILED state when you call the StopDominantLanguageDetectionJob operation, the operation returns a 400 Internal Request Exception.  When a job is stopped, any documents already processed are written to the output location.
   */
  stopSentimentDetectionJob(params: Comprehend.Types.StopSentimentDetectionJobRequest, callback?: (err: AWSError, data: Comprehend.Types.StopSentimentDetectionJobResponse) => void): Request<Comprehend.Types.StopSentimentDetectionJobResponse, AWSError>;
  /**
   * Stops a sentiment detection job in progress. If the job state is IN_PROGRESS the job is marked for termination and put into the STOP_REQUESTED state. If the job completes before it can be stopped, it is put into the COMPLETED state; otherwise the job is be stopped and put into the STOPPED state. If the job is in the COMPLETED or FAILED state when you call the StopDominantLanguageDetectionJob operation, the operation returns a 400 Internal Request Exception.  When a job is stopped, any documents already processed are written to the output location.
   */
  stopSentimentDetectionJob(callback?: (err: AWSError, data: Comprehend.Types.StopSentimentDetectionJobResponse) => void): Request<Comprehend.Types.StopSentimentDetectionJobResponse, AWSError>;
  /**
   * Stops a document classifier training job while in progress. If the training job state is TRAINING, the job is marked for termination and put into the STOP_REQUESTED state. If the training job completes before it can be stopped, it is put into the TRAINED; otherwise the training job is stopped and put into the STOPPED state and the service sends back an HTTP 200 response with an empty HTTP body. 
   */
  stopTrainingDocumentClassifier(params: Comprehend.Types.StopTrainingDocumentClassifierRequest, callback?: (err: AWSError, data: Comprehend.Types.StopTrainingDocumentClassifierResponse) => void): Request<Comprehend.Types.StopTrainingDocumentClassifierResponse, AWSError>;
  /**
   * Stops a document classifier training job while in progress. If the training job state is TRAINING, the job is marked for termination and put into the STOP_REQUESTED state. If the training job completes before it can be stopped, it is put into the TRAINED; otherwise the training job is stopped and put into the STOPPED state and the service sends back an HTTP 200 response with an empty HTTP body. 
   */
  stopTrainingDocumentClassifier(callback?: (err: AWSError, data: Comprehend.Types.StopTrainingDocumentClassifierResponse) => void): Request<Comprehend.Types.StopTrainingDocumentClassifierResponse, AWSError>;
  /**
   * Stops an entity recognizer training job while in progress. If the training job state is TRAINING, the job is marked for termination and put into the STOP_REQUESTED state. If the training job completes before it can be stopped, it is put into the TRAINED; otherwise the training job is stopped and putted into the STOPPED state and the service sends back an HTTP 200 response with an empty HTTP body.
   */
  stopTrainingEntityRecognizer(params: Comprehend.Types.StopTrainingEntityRecognizerRequest, callback?: (err: AWSError, data: Comprehend.Types.StopTrainingEntityRecognizerResponse) => void): Request<Comprehend.Types.StopTrainingEntityRecognizerResponse, AWSError>;
  /**
   * Stops an entity recognizer training job while in progress. If the training job state is TRAINING, the job is marked for termination and put into the STOP_REQUESTED state. If the training job completes before it can be stopped, it is put into the TRAINED; otherwise the training job is stopped and putted into the STOPPED state and the service sends back an HTTP 200 response with an empty HTTP body.
   */
  stopTrainingEntityRecognizer(callback?: (err: AWSError, data: Comprehend.Types.StopTrainingEntityRecognizerResponse) => void): Request<Comprehend.Types.StopTrainingEntityRecognizerResponse, AWSError>;
  /**
   * Associates a specific tag with an Amazon Comprehend resource. A tag is a key-value pair that adds as a metadata to a resource used by Amazon Comprehend. For example, a tag with "Sales" as the key might be added to a resource to indicate its use by the sales department. 
   */
  tagResource(params: Comprehend.Types.TagResourceRequest, callback?: (err: AWSError, data: Comprehend.Types.TagResourceResponse) => void): Request<Comprehend.Types.TagResourceResponse, AWSError>;
  /**
   * Associates a specific tag with an Amazon Comprehend resource. A tag is a key-value pair that adds as a metadata to a resource used by Amazon Comprehend. For example, a tag with "Sales" as the key might be added to a resource to indicate its use by the sales department. 
   */
  tagResource(callback?: (err: AWSError, data: Comprehend.Types.TagResourceResponse) => void): Request<Comprehend.Types.TagResourceResponse, AWSError>;
  /**
   * Removes a specific tag associated with an Amazon Comprehend resource. 
   */
  untagResource(params: Comprehend.Types.UntagResourceRequest, callback?: (err: AWSError, data: Comprehend.Types.UntagResourceResponse) => void): Request<Comprehend.Types.UntagResourceResponse, AWSError>;
  /**
   * Removes a specific tag associated with an Amazon Comprehend resource. 
   */
  untagResource(callback?: (err: AWSError, data: Comprehend.Types.UntagResourceResponse) => void): Request<Comprehend.Types.UntagResourceResponse, AWSError>;
}
declare namespace Comprehend {
  export type AnyLengthString = string;
  export interface BatchDetectDominantLanguageItemResult {
    /**
     * The zero-based index of the document in the input list.
     */
    Index?: Integer;
    /**
     * One or more DominantLanguage objects describing the dominant languages in the document.
     */
    Languages?: ListOfDominantLanguages;
  }
  export interface BatchDetectDominantLanguageRequest {
    /**
     * A list containing the text of the input documents. The list can contain a maximum of 25 documents. Each document should contain at least 20 characters and must contain fewer than 5,000 bytes of UTF-8 encoded characters.
     */
    TextList: StringList;
  }
  export interface BatchDetectDominantLanguageResponse {
    /**
     * A list of objects containing the results of the operation. The results are sorted in ascending order by the Index field and match the order of the documents in the input list. If all of the documents contain an error, the ResultList is empty.
     */
    ResultList: ListOfDetectDominantLanguageResult;
    /**
     * A list containing one object for each document that contained an error. The results are sorted in ascending order by the Index field and match the order of the documents in the input list. If there are no errors in the batch, the ErrorList is empty.
     */
    ErrorList: BatchItemErrorList;
  }
  export interface BatchDetectEntitiesItemResult {
    /**
     * The zero-based index of the document in the input list.
     */
    Index?: Integer;
    /**
     * One or more Entity objects, one for each entity detected in the document.
     */
    Entities?: ListOfEntities;
  }
  export interface BatchDetectEntitiesRequest {
    /**
     * A list containing the text of the input documents. The list can contain a maximum of 25 documents. Each document must contain fewer than 5,000 bytes of UTF-8 encoded characters.
     */
    TextList: StringList;
    /**
     * The language of the input documents. You can specify any of the primary languages supported by Amazon Comprehend: German ("de"), English ("en"), Spanish ("es"), French ("fr"), Italian ("it"), or Portuguese ("pt"). All documents must be in the same language.
     */
    LanguageCode: LanguageCode;
  }
  export interface BatchDetectEntitiesResponse {
    /**
     * A list of objects containing the results of the operation. The results are sorted in ascending order by the Index field and match the order of the documents in the input list. If all of the documents contain an error, the ResultList is empty.
     */
    ResultList: ListOfDetectEntitiesResult;
    /**
     * A list containing one object for each document that contained an error. The results are sorted in ascending order by the Index field and match the order of the documents in the input list. If there are no errors in the batch, the ErrorList is empty.
     */
    ErrorList: BatchItemErrorList;
  }
  export interface BatchDetectKeyPhrasesItemResult {
    /**
     * The zero-based index of the document in the input list.
     */
    Index?: Integer;
    /**
     * One or more KeyPhrase objects, one for each key phrase detected in the document.
     */
    KeyPhrases?: ListOfKeyPhrases;
  }
  export interface BatchDetectKeyPhrasesRequest {
    /**
     * A list containing the text of the input documents. The list can contain a maximum of 25 documents. Each document must contain fewer that 5,000 bytes of UTF-8 encoded characters.
     */
    TextList: StringList;
    /**
     * The language of the input documents. You can specify any of the primary languages supported by Amazon Comprehend: German ("de"), English ("en"), Spanish ("es"), French ("fr"), Italian ("it"), or Portuguese ("pt"). All documents must be in the same language.
     */
    LanguageCode: LanguageCode;
  }
  export interface BatchDetectKeyPhrasesResponse {
    /**
     * A list of objects containing the results of the operation. The results are sorted in ascending order by the Index field and match the order of the documents in the input list. If all of the documents contain an error, the ResultList is empty.
     */
    ResultList: ListOfDetectKeyPhrasesResult;
    /**
     * A list containing one object for each document that contained an error. The results are sorted in ascending order by the Index field and match the order of the documents in the input list. If there are no errors in the batch, the ErrorList is empty.
     */
    ErrorList: BatchItemErrorList;
  }
  export interface BatchDetectSentimentItemResult {
    /**
     * The zero-based index of the document in the input list.
     */
    Index?: Integer;
    /**
     * The sentiment detected in the document.
     */
    Sentiment?: SentimentType;
    /**
     * The level of confidence that Amazon Comprehend has in the accuracy of its sentiment detection.
     */
    SentimentScore?: SentimentScore;
  }
  export interface BatchDetectSentimentRequest {
    /**
     * A list containing the text of the input documents. The list can contain a maximum of 25 documents. Each document must contain fewer that 5,000 bytes of UTF-8 encoded characters.
     */
    TextList: StringList;
    /**
     * The language of the input documents. You can specify any of the primary languages supported by Amazon Comprehend: German ("de"), English ("en"), Spanish ("es"), French ("fr"), Italian ("it"), or Portuguese ("pt"). All documents must be in the same language.
     */
    LanguageCode: LanguageCode;
  }
  export interface BatchDetectSentimentResponse {
    /**
     * A list of objects containing the results of the operation. The results are sorted in ascending order by the Index field and match the order of the documents in the input list. If all of the documents contain an error, the ResultList is empty.
     */
    ResultList: ListOfDetectSentimentResult;
    /**
     * A list containing one object for each document that contained an error. The results are sorted in ascending order by the Index field and match the order of the documents in the input list. If there are no errors in the batch, the ErrorList is empty.
     */
    ErrorList: BatchItemErrorList;
  }
  export interface BatchDetectSyntaxItemResult {
    /**
     * The zero-based index of the document in the input list.
     */
    Index?: Integer;
    /**
     * The syntax tokens for the words in the document, one token for each word.
     */
    SyntaxTokens?: ListOfSyntaxTokens;
  }
  export interface BatchDetectSyntaxRequest {
    /**
     * A list containing the text of the input documents. The list can contain a maximum of 25 documents. Each document must contain fewer that 5,000 bytes of UTF-8 encoded characters.
     */
    TextList: StringList;
    /**
     * The language of the input documents. You can specify any of the primary languages supported by Amazon Comprehend: German ("de"), English ("en"), Spanish ("es"), French ("fr"), Italian ("it"), or Portuguese ("pt"). All documents must be in the same language.
     */
    LanguageCode: SyntaxLanguageCode;
  }
  export interface BatchDetectSyntaxResponse {
    /**
     * A list of objects containing the results of the operation. The results are sorted in ascending order by the Index field and match the order of the documents in the input list. If all of the documents contain an error, the ResultList is empty.
     */
    ResultList: ListOfDetectSyntaxResult;
    /**
     * A list containing one object for each document that contained an error. The results are sorted in ascending order by the Index field and match the order of the documents in the input list. If there are no errors in the batch, the ErrorList is empty.
     */
    ErrorList: BatchItemErrorList;
  }
  export interface BatchItemError {
    /**
     * The zero-based index of the document in the input list.
     */
    Index?: Integer;
    /**
     * The numeric error code of the error.
     */
    ErrorCode?: String;
    /**
     * A text description of the error.
     */
    ErrorMessage?: String;
  }
  export type BatchItemErrorList = BatchItemError[];
  export interface ClassifierEvaluationMetrics {
    /**
     * The fraction of the labels that were correct recognized. It is computed by dividing the number of labels in the test documents that were correctly recognized by the total number of labels in the test documents.
     */
    Accuracy?: Double;
    /**
     * A measure of the usefulness of the classifier results in the test data. High precision means that the classifier returned substantially more relevant results than irrelevant ones.
     */
    Precision?: Double;
    /**
     * A measure of how complete the classifier results are for the test data. High recall means that the classifier returned most of the relevant results. 
     */
    Recall?: Double;
    /**
     * A measure of how accurate the classifier results are for the test data. It is derived from the Precision and Recall values. The F1Score is the harmonic average of the two scores. The highest score is 1, and the worst score is 0. 
     */
    F1Score?: Double;
  }
  export interface ClassifierMetadata {
    /**
     * The number of labels in the input data. 
     */
    NumberOfLabels?: Integer;
    /**
     * The number of documents in the input data that were used to train the classifier. Typically this is 80 to 90 percent of the input documents.
     */
    NumberOfTrainedDocuments?: Integer;
    /**
     * The number of documents in the input data that were used to test the classifier. Typically this is 10 to 20 percent of the input documents.
     */
    NumberOfTestDocuments?: Integer;
    /**
     *  Describes the result metrics for the test data associated with an documentation classifier.
     */
    EvaluationMetrics?: ClassifierEvaluationMetrics;
  }
  export type ClientRequestTokenString = string;
  export type ComprehendArn = string;
  export type ComprehendArnName = string;
  export interface CreateDocumentClassifierRequest {
    /**
     * The name of the document classifier.
     */
    DocumentClassifierName: ComprehendArnName;
    /**
     * The Amazon Resource Name (ARN) of the AWS Identity and Management (IAM) role that grants Amazon Comprehend read access to your input data.
     */
    DataAccessRoleArn: IamRoleArn;
    /**
     * Tags to be associated with the document classifier being created. A tag is a key-value pair that adds as a metadata to a resource used by Amazon Comprehend. For example, a tag with "Sales" as the key might be added to a resource to indicate its use by the sales department. 
     */
    Tags?: TagList;
    /**
     * Specifies the format and location of the input data for the job.
     */
    InputDataConfig: DocumentClassifierInputDataConfig;
    /**
     * Enables the addition of output results configuration parameters for custom classifier jobs.
     */
    OutputDataConfig?: DocumentClassifierOutputDataConfig;
    /**
     * A unique identifier for the request. If you don't set the client request token, Amazon Comprehend generates one.
     */
    ClientRequestToken?: ClientRequestTokenString;
    /**
     * The language of the input documents. You can specify any of the primary languages supported by Amazon Comprehend: German ("de"), English ("en"), Spanish ("es"), French ("fr"), Italian ("it"), or Portuguese ("pt"). All documents must be in the same language.
     */
    LanguageCode: LanguageCode;
    /**
     * ID for the AWS Key Management Service (KMS) key that Amazon Comprehend uses to encrypt data on the storage volume attached to the ML compute instance(s) that process the analysis job. The VolumeKmsKeyId can be either of the following formats:   KMS Key ID: "1234abcd-12ab-34cd-56ef-1234567890ab"    Amazon Resource Name (ARN) of a KMS Key: "arn:aws:kms:us-west-2:111122223333:key/1234abcd-12ab-34cd-56ef-1234567890ab"   
     */
    VolumeKmsKeyId?: KmsKeyId;
    /**
     * Configuration parameters for an optional private Virtual Private Cloud (VPC) containing the resources you are using for your custom classifier. For more information, see Amazon VPC. 
     */
    VpcConfig?: VpcConfig;
  }
  export interface CreateDocumentClassifierResponse {
    /**
     * The Amazon Resource Name (ARN) that identifies the document classifier.
     */
    DocumentClassifierArn?: DocumentClassifierArn;
  }
  export interface CreateEntityRecognizerRequest {
    /**
     * The name given to the newly created recognizer. Recognizer names can be a maximum of 256 characters. Alphanumeric characters, hyphens (-) and underscores (_) are allowed. The name must be unique in the account/region.
     */
    RecognizerName: ComprehendArnName;
    /**
     * The Amazon Resource Name (ARN) of the AWS Identity and Management (IAM) role that grants Amazon Comprehend read access to your input data.
     */
    DataAccessRoleArn: IamRoleArn;
    /**
     * Tags to be associated with the entity recognizer being created. A tag is a key-value pair that adds as a metadata to a resource used by Amazon Comprehend. For example, a tag with "Sales" as the key might be added to a resource to indicate its use by the sales department. 
     */
    Tags?: TagList;
    /**
     * Specifies the format and location of the input data. The S3 bucket containing the input data must be located in the same region as the entity recognizer being created. 
     */
    InputDataConfig: EntityRecognizerInputDataConfig;
    /**
     *  A unique identifier for the request. If you don't set the client request token, Amazon Comprehend generates one.
     */
    ClientRequestToken?: ClientRequestTokenString;
    /**
     *  The language of the input documents. All documents must be in the same language. Only English ("en") is currently supported. 
     */
    LanguageCode: LanguageCode;
    /**
     * ID for the AWS Key Management Service (KMS) key that Amazon Comprehend uses to encrypt data on the storage volume attached to the ML compute instance(s) that process the analysis job. The VolumeKmsKeyId can be either of the following formats:   KMS Key ID: "1234abcd-12ab-34cd-56ef-1234567890ab"    Amazon Resource Name (ARN) of a KMS Key: "arn:aws:kms:us-west-2:111122223333:key/1234abcd-12ab-34cd-56ef-1234567890ab"   
     */
    VolumeKmsKeyId?: KmsKeyId;
    /**
     * Configuration parameters for an optional private Virtual Private Cloud (VPC) containing the resources you are using for your custom entity recognizer. For more information, see Amazon VPC. 
     */
    VpcConfig?: VpcConfig;
  }
  export interface CreateEntityRecognizerResponse {
    /**
     * The Amazon Resource Name (ARN) that identifies the entity recognizer.
     */
    EntityRecognizerArn?: EntityRecognizerArn;
  }
  export interface DeleteDocumentClassifierRequest {
    /**
     * The Amazon Resource Name (ARN) that identifies the document classifier. 
     */
    DocumentClassifierArn: DocumentClassifierArn;
  }
  export interface DeleteDocumentClassifierResponse {
  }
  export interface DeleteEntityRecognizerRequest {
    /**
     * The Amazon Resource Name (ARN) that identifies the entity recognizer.
     */
    EntityRecognizerArn: EntityRecognizerArn;
  }
  export interface DeleteEntityRecognizerResponse {
  }
  export interface DescribeDocumentClassificationJobRequest {
    /**
     * The identifier that Amazon Comprehend generated for the job. The operation returns this identifier in its response.
     */
    JobId: JobId;
  }
  export interface DescribeDocumentClassificationJobResponse {
    /**
     * An object that describes the properties associated with the document classification job.
     */
    DocumentClassificationJobProperties?: DocumentClassificationJobProperties;
  }
  export interface DescribeDocumentClassifierRequest {
    /**
     * The Amazon Resource Name (ARN) that identifies the document classifier. The operation returns this identifier in its response.
     */
    DocumentClassifierArn: DocumentClassifierArn;
  }
  export interface DescribeDocumentClassifierResponse {
    /**
     * An object that contains the properties associated with a document classifier.
     */
    DocumentClassifierProperties?: DocumentClassifierProperties;
  }
  export interface DescribeDominantLanguageDetectionJobRequest {
    /**
     * The identifier that Amazon Comprehend generated for the job. The operation returns this identifier in its response.
     */
    JobId: JobId;
  }
  export interface DescribeDominantLanguageDetectionJobResponse {
    /**
     * An object that contains the properties associated with a dominant language detection job.
     */
    DominantLanguageDetectionJobProperties?: DominantLanguageDetectionJobProperties;
  }
  export interface DescribeEntitiesDetectionJobRequest {
    /**
     * The identifier that Amazon Comprehend generated for the job. The operation returns this identifier in its response.
     */
    JobId: JobId;
  }
  export interface DescribeEntitiesDetectionJobResponse {
    /**
     * An object that contains the properties associated with an entities detection job.
     */
    EntitiesDetectionJobProperties?: EntitiesDetectionJobProperties;
  }
  export interface DescribeEntityRecognizerRequest {
    /**
     * The Amazon Resource Name (ARN) that identifies the entity recognizer.
     */
    EntityRecognizerArn: EntityRecognizerArn;
  }
  export interface DescribeEntityRecognizerResponse {
    /**
     * Describes information associated with an entity recognizer.
     */
    EntityRecognizerProperties?: EntityRecognizerProperties;
  }
  export interface DescribeKeyPhrasesDetectionJobRequest {
    /**
     * The identifier that Amazon Comprehend generated for the job. The operation returns this identifier in its response.
     */
    JobId: JobId;
  }
  export interface DescribeKeyPhrasesDetectionJobResponse {
    /**
     * An object that contains the properties associated with a key phrases detection job. 
     */
    KeyPhrasesDetectionJobProperties?: KeyPhrasesDetectionJobProperties;
  }
  export interface DescribeSentimentDetectionJobRequest {
    /**
     * The identifier that Amazon Comprehend generated for the job. The operation returns this identifier in its response.
     */
    JobId: JobId;
  }
  export interface DescribeSentimentDetectionJobResponse {
    /**
     * An object that contains the properties associated with a sentiment detection job.
     */
    SentimentDetectionJobProperties?: SentimentDetectionJobProperties;
  }
  export interface DescribeTopicsDetectionJobRequest {
    /**
     * The identifier assigned by the user to the detection job.
     */
    JobId: JobId;
  }
  export interface DescribeTopicsDetectionJobResponse {
    /**
     * The list of properties for the requested job.
     */
    TopicsDetectionJobProperties?: TopicsDetectionJobProperties;
  }
  export interface DetectDominantLanguageRequest {
    /**
     * A UTF-8 text string. Each string should contain at least 20 characters and must contain fewer that 5,000 bytes of UTF-8 encoded characters.
     */
    Text: String;
  }
  export interface DetectDominantLanguageResponse {
    /**
     * The languages that Amazon Comprehend detected in the input text. For each language, the response returns the RFC 5646 language code and the level of confidence that Amazon Comprehend has in the accuracy of its inference. For more information about RFC 5646, see Tags for Identifying Languages on the IETF Tools web site.
     */
    Languages?: ListOfDominantLanguages;
  }
  export interface DetectEntitiesRequest {
    /**
     * A UTF-8 text string. Each string must contain fewer that 5,000 bytes of UTF-8 encoded characters.
     */
    Text: String;
    /**
     * The language of the input documents. You can specify any of the primary languages supported by Amazon Comprehend: German ("de"), English ("en"), Spanish ("es"), French ("fr"), Italian ("it"), or Portuguese ("pt"). All documents must be in the same language.
     */
    LanguageCode: LanguageCode;
  }
  export interface DetectEntitiesResponse {
    /**
     * A collection of entities identified in the input text. For each entity, the response provides the entity text, entity type, where the entity text begins and ends, and the level of confidence that Amazon Comprehend has in the detection. For a list of entity types, see how-entities. 
     */
    Entities?: ListOfEntities;
  }
  export interface DetectKeyPhrasesRequest {
    /**
     * A UTF-8 text string. Each string must contain fewer that 5,000 bytes of UTF-8 encoded characters.
     */
    Text: String;
    /**
     * The language of the input documents. You can specify any of the primary languages supported by Amazon Comprehend: German ("de"), English ("en"), Spanish ("es"), French ("fr"), Italian ("it"), or Portuguese ("pt"). All documents must be in the same language.
     */
    LanguageCode: LanguageCode;
  }
  export interface DetectKeyPhrasesResponse {
    /**
     * A collection of key phrases that Amazon Comprehend identified in the input text. For each key phrase, the response provides the text of the key phrase, where the key phrase begins and ends, and the level of confidence that Amazon Comprehend has in the accuracy of the detection. 
     */
    KeyPhrases?: ListOfKeyPhrases;
  }
  export interface DetectSentimentRequest {
    /**
     * A UTF-8 text string. Each string must contain fewer that 5,000 bytes of UTF-8 encoded characters.
     */
    Text: String;
    /**
     * The language of the input documents. You can specify any of the primary languages supported by Amazon Comprehend: German ("de"), English ("en"), Spanish ("es"), French ("fr"), Italian ("it"), or Portuguese ("pt"). All documents must be in the same language.
     */
    LanguageCode: LanguageCode;
  }
  export interface DetectSentimentResponse {
    /**
     * The inferred sentiment that Amazon Comprehend has the highest level of confidence in.
     */
    Sentiment?: SentimentType;
    /**
     * An object that lists the sentiments, and their corresponding confidence levels.
     */
    SentimentScore?: SentimentScore;
  }
  export interface DetectSyntaxRequest {
    /**
     * A UTF-8 string. Each string must contain fewer that 5,000 bytes of UTF encoded characters.
     */
    Text: String;
    /**
     * The language code of the input documents. You can specify any of the primary languages supported by Amazon Comprehend: German ("de"), English ("en"), Spanish ("es"), French ("fr"), Italian ("it"), or Portuguese ("pt").
     */
    LanguageCode: SyntaxLanguageCode;
  }
  export interface DetectSyntaxResponse {
    /**
     * A collection of syntax tokens describing the text. For each token, the response provides the text, the token type, where the text begins and ends, and the level of confidence that Amazon Comprehend has that the token is correct. For a list of token types, see how-syntax.
     */
    SyntaxTokens?: ListOfSyntaxTokens;
  }
  export interface DocumentClassificationJobFilter {
    /**
     * Filters on the name of the job.
     */
    JobName?: JobName;
    /**
     * Filters the list based on job status. Returns only jobs with the specified status.
     */
    JobStatus?: JobStatus;
    /**
     * Filters the list of jobs based on the time that the job was submitted for processing. Returns only jobs submitted after the specified time. Jobs are returned in ascending order, oldest to newest.
     */
    SubmitTimeBefore?: Timestamp;
    /**
     * Filters the list of jobs based on the time that the job was submitted for processing. Returns only jobs submitted before the specified time. Jobs are returned in descending order, newest to oldest.
     */
    SubmitTimeAfter?: Timestamp;
  }
  export interface DocumentClassificationJobProperties {
    /**
     * The identifier assigned to the document classification job.
     */
    JobId?: JobId;
    /**
     * The name that you assigned to the document classification job.
     */
    JobName?: JobName;
    /**
     * The current status of the document classification job. If the status is FAILED, the Message field shows the reason for the failure.
     */
    JobStatus?: JobStatus;
    /**
     * A description of the status of the job.
     */
    Message?: AnyLengthString;
    /**
     * The time that the document classification job was submitted for processing.
     */
    SubmitTime?: Timestamp;
    /**
     * The time that the document classification job completed.
     */
    EndTime?: Timestamp;
    /**
     * The Amazon Resource Name (ARN) that identifies the document classifier. 
     */
    DocumentClassifierArn?: DocumentClassifierArn;
    /**
     * The input data configuration that you supplied when you created the document classification job.
     */
    InputDataConfig?: InputDataConfig;
    /**
     * The output data configuration that you supplied when you created the document classification job.
     */
    OutputDataConfig?: OutputDataConfig;
    /**
     * The Amazon Resource Name (ARN) of the AWS identity and Access Management (IAM) role that grants Amazon Comprehend read access to your input data.
     */
    DataAccessRoleArn?: IamRoleArn;
    /**
     * ID for the AWS Key Management Service (KMS) key that Amazon Comprehend uses to encrypt data on the storage volume attached to the ML compute instance(s) that process the analysis job. The VolumeKmsKeyId can be either of the following formats:   KMS Key ID: "1234abcd-12ab-34cd-56ef-1234567890ab"    Amazon Resource Name (ARN) of a KMS Key: "arn:aws:kms:us-west-2:111122223333:key/1234abcd-12ab-34cd-56ef-1234567890ab"   
     */
    VolumeKmsKeyId?: KmsKeyId;
    /**
     *  Configuration parameters for a private Virtual Private Cloud (VPC) containing the resources you are using for your document classification job. For more information, see Amazon VPC. 
     */
    VpcConfig?: VpcConfig;
  }
  export type DocumentClassificationJobPropertiesList = DocumentClassificationJobProperties[];
  export type DocumentClassifierArn = string;
  export interface DocumentClassifierFilter {
    /**
     * Filters the list of classifiers based on status. 
     */
    Status?: ModelStatus;
    /**
     * Filters the list of classifiers based on the time that the classifier was submitted for processing. Returns only classifiers submitted before the specified time. Classifiers are returned in ascending order, oldest to newest.
     */
    SubmitTimeBefore?: Timestamp;
    /**
     * Filters the list of classifiers based on the time that the classifier was submitted for processing. Returns only classifiers submitted after the specified time. Classifiers are returned in descending order, newest to oldest.
     */
    SubmitTimeAfter?: Timestamp;
  }
  export interface DocumentClassifierInputDataConfig {
    /**
     * The Amazon S3 URI for the input data. The S3 bucket must be in the same region as the API endpoint that you are calling. The URI can point to a single input file or it can provide the prefix for a collection of input files. For example, if you use the URI S3://bucketName/prefix, if the prefix is a single file, Amazon Comprehend uses that file as input. If more than one file begins with the prefix, Amazon Comprehend uses all of them as input.
     */
    S3Uri: S3Uri;
  }
  export interface DocumentClassifierOutputDataConfig {
    /**
     * When you use the OutputDataConfig object while creating a custom classifier, you specify the Amazon S3 location where you want to write the confusion matrix. The URI must be in the same region as the API endpoint that you are calling. The location is used as the prefix for the actual location of this output file. When the custom classifier job is finished, the service creates the output file in a directory specific to the job. The S3Uri field contains the location of the output file, called output.tar.gz. It is a compressed archive that contains the confusion matrix.
     */
    S3Uri?: S3Uri;
    /**
     * ID for the AWS Key Management Service (KMS) key that Amazon Comprehend uses to encrypt the output results from an analysis job. The KmsKeyId can be one of the following formats:   KMS Key ID: "1234abcd-12ab-34cd-56ef-1234567890ab"    Amazon Resource Name (ARN) of a KMS Key: "arn:aws:kms:us-west-2:111122223333:key/1234abcd-12ab-34cd-56ef-1234567890ab"    KMS Key Alias: "alias/ExampleAlias"    ARN of a KMS Key Alias: "arn:aws:kms:us-west-2:111122223333:alias/ExampleAlias"   
     */
    KmsKeyId?: KmsKeyId;
  }
  export interface DocumentClassifierProperties {
    /**
     * The Amazon Resource Name (ARN) that identifies the document classifier.
     */
    DocumentClassifierArn?: DocumentClassifierArn;
    /**
     * The language code for the language of the documents that the classifier was trained on.
     */
    LanguageCode?: LanguageCode;
    /**
     * The status of the document classifier. If the status is TRAINED the classifier is ready to use. If the status is FAILED you can see additional information about why the classifier wasn't trained in the Message field.
     */
    Status?: ModelStatus;
    /**
     * Additional information about the status of the classifier.
     */
    Message?: AnyLengthString;
    /**
     * The time that the document classifier was submitted for training.
     */
    SubmitTime?: Timestamp;
    /**
     * The time that training the document classifier completed.
     */
    EndTime?: Timestamp;
    /**
     * Indicates the time when the training starts on documentation classifiers. You are billed for the time interval between this time and the value of TrainingEndTime. 
     */
    TrainingStartTime?: Timestamp;
    /**
     * The time that training of the document classifier was completed. Indicates the time when the training completes on documentation classifiers. You are billed for the time interval between this time and the value of TrainingStartTime.
     */
    TrainingEndTime?: Timestamp;
    /**
     * The input data configuration that you supplied when you created the document classifier for training.
     */
    InputDataConfig?: DocumentClassifierInputDataConfig;
    /**
     *  Provides output results configuration parameters for custom classifier jobs.
     */
    OutputDataConfig?: DocumentClassifierOutputDataConfig;
    /**
     * Information about the document classifier, including the number of documents used for training the classifier, the number of documents used for test the classifier, and an accuracy rating.
     */
    ClassifierMetadata?: ClassifierMetadata;
    /**
     * The Amazon Resource Name (ARN) of the AWS Identity and Management (IAM) role that grants Amazon Comprehend read access to your input data.
     */
    DataAccessRoleArn?: IamRoleArn;
    /**
     * ID for the AWS Key Management Service (KMS) key that Amazon Comprehend uses to encrypt data on the storage volume attached to the ML compute instance(s) that process the analysis job. The VolumeKmsKeyId can be either of the following formats:   KMS Key ID: "1234abcd-12ab-34cd-56ef-1234567890ab"    Amazon Resource Name (ARN) of a KMS Key: "arn:aws:kms:us-west-2:111122223333:key/1234abcd-12ab-34cd-56ef-1234567890ab"   
     */
    VolumeKmsKeyId?: KmsKeyId;
    /**
     *  Configuration parameters for a private Virtual Private Cloud (VPC) containing the resources you are using for your custom classifier. For more information, see Amazon VPC. 
     */
    VpcConfig?: VpcConfig;
  }
  export type DocumentClassifierPropertiesList = DocumentClassifierProperties[];
  export interface DominantLanguage {
    /**
     * The RFC 5646 language code for the dominant language. For more information about RFC 5646, see Tags for Identifying Languages on the IETF Tools web site.
     */
    LanguageCode?: String;
    /**
     * The level of confidence that Amazon Comprehend has in the accuracy of the detection.
     */
    Score?: Float;
  }
  export interface DominantLanguageDetectionJobFilter {
    /**
     * Filters on the name of the job.
     */
    JobName?: JobName;
    /**
     * Filters the list of jobs based on job status. Returns only jobs with the specified status.
     */
    JobStatus?: JobStatus;
    /**
     * Filters the list of jobs based on the time that the job was submitted for processing. Returns only jobs submitted before the specified time. Jobs are returned in ascending order, oldest to newest.
     */
    SubmitTimeBefore?: Timestamp;
    /**
     * Filters the list of jobs based on the time that the job was submitted for processing. Returns only jobs submitted after the specified time. Jobs are returned in descending order, newest to oldest.
     */
    SubmitTimeAfter?: Timestamp;
  }
  export interface DominantLanguageDetectionJobProperties {
    /**
     * The identifier assigned to the dominant language detection job.
     */
    JobId?: JobId;
    /**
     * The name that you assigned to the dominant language detection job.
     */
    JobName?: JobName;
    /**
     * The current status of the dominant language detection job. If the status is FAILED, the Message field shows the reason for the failure.
     */
    JobStatus?: JobStatus;
    /**
     * A description for the status of a job.
     */
    Message?: AnyLengthString;
    /**
     * The time that the dominant language detection job was submitted for processing.
     */
    SubmitTime?: Timestamp;
    /**
     * The time that the dominant language detection job completed.
     */
    EndTime?: Timestamp;
    /**
     * The input data configuration that you supplied when you created the dominant language detection job.
     */
    InputDataConfig?: InputDataConfig;
    /**
     * The output data configuration that you supplied when you created the dominant language detection job.
     */
    OutputDataConfig?: OutputDataConfig;
    /**
     * The Amazon Resource Name (ARN) that gives Amazon Comprehend read access to your input data.
     */
    DataAccessRoleArn?: IamRoleArn;
    /**
     * ID for the AWS Key Management Service (KMS) key that Amazon Comprehend uses to encrypt data on the storage volume attached to the ML compute instance(s) that process the analysis job. The VolumeKmsKeyId can be either of the following formats:   KMS Key ID: "1234abcd-12ab-34cd-56ef-1234567890ab"    Amazon Resource Name (ARN) of a KMS Key: "arn:aws:kms:us-west-2:111122223333:key/1234abcd-12ab-34cd-56ef-1234567890ab"   
     */
    VolumeKmsKeyId?: KmsKeyId;
    /**
     *  Configuration parameters for a private Virtual Private Cloud (VPC) containing the resources you are using for your dominant language detection job. For more information, see Amazon VPC. 
     */
    VpcConfig?: VpcConfig;
  }
  export type DominantLanguageDetectionJobPropertiesList = DominantLanguageDetectionJobProperties[];
  export type Double = number;
  export interface EntitiesDetectionJobFilter {
    /**
     * Filters on the name of the job.
     */
    JobName?: JobName;
    /**
     * Filters the list of jobs based on job status. Returns only jobs with the specified status.
     */
    JobStatus?: JobStatus;
    /**
     * Filters the list of jobs based on the time that the job was submitted for processing. Returns only jobs submitted before the specified time. Jobs are returned in ascending order, oldest to newest.
     */
    SubmitTimeBefore?: Timestamp;
    /**
     * Filters the list of jobs based on the time that the job was submitted for processing. Returns only jobs submitted after the specified time. Jobs are returned in descending order, newest to oldest.
     */
    SubmitTimeAfter?: Timestamp;
  }
  export interface EntitiesDetectionJobProperties {
    /**
     * The identifier assigned to the entities detection job.
     */
    JobId?: JobId;
    /**
     * The name that you assigned the entities detection job.
     */
    JobName?: JobName;
    /**
     * The current status of the entities detection job. If the status is FAILED, the Message field shows the reason for the failure.
     */
    JobStatus?: JobStatus;
    /**
     * A description of the status of a job.
     */
    Message?: AnyLengthString;
    /**
     * The time that the entities detection job was submitted for processing.
     */
    SubmitTime?: Timestamp;
    /**
     * The time that the entities detection job completed
     */
    EndTime?: Timestamp;
    /**
     * The Amazon Resource Name (ARN) that identifies the entity recognizer.
     */
    EntityRecognizerArn?: EntityRecognizerArn;
    /**
     * The input data configuration that you supplied when you created the entities detection job.
     */
    InputDataConfig?: InputDataConfig;
    /**
     * The output data configuration that you supplied when you created the entities detection job. 
     */
    OutputDataConfig?: OutputDataConfig;
    /**
     * The language code of the input documents.
     */
    LanguageCode?: LanguageCode;
    /**
     * The Amazon Resource Name (ARN) that gives Amazon Comprehend read access to your input data.
     */
    DataAccessRoleArn?: IamRoleArn;
    /**
     * ID for the AWS Key Management Service (KMS) key that Amazon Comprehend uses to encrypt data on the storage volume attached to the ML compute instance(s) that process the analysis job. The VolumeKmsKeyId can be either of the following formats:   KMS Key ID: "1234abcd-12ab-34cd-56ef-1234567890ab"    Amazon Resource Name (ARN) of a KMS Key: "arn:aws:kms:us-west-2:111122223333:key/1234abcd-12ab-34cd-56ef-1234567890ab"   
     */
    VolumeKmsKeyId?: KmsKeyId;
    /**
     *  Configuration parameters for a private Virtual Private Cloud (VPC) containing the resources you are using for your entity detection job. For more information, see Amazon VPC. 
     */
    VpcConfig?: VpcConfig;
  }
  export type EntitiesDetectionJobPropertiesList = EntitiesDetectionJobProperties[];
  export interface Entity {
    /**
     * The level of confidence that Amazon Comprehend has in the accuracy of the detection.
     */
    Score?: Float;
    /**
     * The entity's type.
     */
    Type?: EntityType;
    /**
     * The text of the entity.
     */
    Text?: String;
    /**
     * A character offset in the input text that shows where the entity begins (the first character is at position 0). The offset returns the position of each UTF-8 code point in the string. A code point is the abstract character from a particular graphical representation. For example, a multi-byte UTF-8 character maps to a single code point.
     */
    BeginOffset?: Integer;
    /**
     * A character offset in the input text that shows where the entity ends. The offset returns the position of each UTF-8 code point in the string. A code point is the abstract character from a particular graphical representation. For example, a multi-byte UTF-8 character maps to a single code point. 
     */
    EndOffset?: Integer;
  }
  export interface EntityRecognizerAnnotations {
    /**
     *  Specifies the Amazon S3 location where the annotations for an entity recognizer are located. The URI must be in the same region as the API endpoint that you are calling.
     */
    S3Uri: S3Uri;
  }
  export type EntityRecognizerArn = string;
  export interface EntityRecognizerDocuments {
    /**
     *  Specifies the Amazon S3 location where the training documents for an entity recognizer are located. The URI must be in the same region as the API endpoint that you are calling.
     */
    S3Uri: S3Uri;
  }
  export interface EntityRecognizerEntityList {
    /**
     * Specifies the Amazon S3 location where the entity list is located. The URI must be in the same region as the API endpoint that you are calling.
     */
    S3Uri: S3Uri;
  }
  export interface EntityRecognizerEvaluationMetrics {
    /**
     * A measure of the usefulness of the recognizer results in the test data. High precision means that the recognizer returned substantially more relevant results than irrelevant ones. 
     */
    Precision?: Double;
    /**
     * A measure of how complete the recognizer results are for the test data. High recall means that the recognizer returned most of the relevant results.
     */
    Recall?: Double;
    /**
     * A measure of how accurate the recognizer results are for the test data. It is derived from the Precision and Recall values. The F1Score is the harmonic average of the two scores. The highest score is 1, and the worst score is 0. 
     */
    F1Score?: Double;
  }
  export interface EntityRecognizerFilter {
    /**
     * The status of an entity recognizer.
     */
    Status?: ModelStatus;
    /**
     * Filters the list of entities based on the time that the list was submitted for processing. Returns only jobs submitted before the specified time. Jobs are returned in descending order, newest to oldest.
     */
    SubmitTimeBefore?: Timestamp;
    /**
     * Filters the list of entities based on the time that the list was submitted for processing. Returns only jobs submitted after the specified time. Jobs are returned in ascending order, oldest to newest.
     */
    SubmitTimeAfter?: Timestamp;
  }
  export interface EntityRecognizerInputDataConfig {
    /**
     * The entity types in the input data for an entity recognizer. A maximum of 12 entity types can be used at one time to train an entity recognizer.
     */
    EntityTypes: EntityTypesList;
    /**
     * S3 location of the documents folder for an entity recognizer
     */
    Documents: EntityRecognizerDocuments;
    /**
     * S3 location of the annotations file for an entity recognizer.
     */
    Annotations?: EntityRecognizerAnnotations;
    /**
     * S3 location of the entity list for an entity recognizer.
     */
    EntityList?: EntityRecognizerEntityList;
  }
  export interface EntityRecognizerMetadata {
    /**
     *  The number of documents in the input data that were used to train the entity recognizer. Typically this is 80 to 90 percent of the input documents.
     */
    NumberOfTrainedDocuments?: Integer;
    /**
     *  The number of documents in the input data that were used to test the entity recognizer. Typically this is 10 to 20 percent of the input documents.
     */
    NumberOfTestDocuments?: Integer;
    /**
     * Detailed information about the accuracy of an entity recognizer.
     */
    EvaluationMetrics?: EntityRecognizerEvaluationMetrics;
    /**
     * Entity types from the metadata of an entity recognizer.
     */
    EntityTypes?: EntityRecognizerMetadataEntityTypesList;
  }
  export type EntityRecognizerMetadataEntityTypesList = EntityRecognizerMetadataEntityTypesListItem[];
  export interface EntityRecognizerMetadataEntityTypesListItem {
    /**
     * Type of entity from the list of entity types in the metadata of an entity recognizer. 
     */
    Type?: AnyLengthString;
    /**
     * Detailed information about the accuracy of the entity recognizer for a specific item on the list of entity types. 
     */
    EvaluationMetrics?: EntityTypesEvaluationMetrics;
    /**
     * Indicates the number of times the given entity type was seen in the training data. 
     */
    NumberOfTrainMentions?: Integer;
  }
  export interface EntityRecognizerProperties {
    /**
     * The Amazon Resource Name (ARN) that identifies the entity recognizer.
     */
    EntityRecognizerArn?: EntityRecognizerArn;
    /**
     *  The language of the input documents. All documents must be in the same language. Only English ("en") is currently supported.
     */
    LanguageCode?: LanguageCode;
    /**
     * Provides the status of the entity recognizer.
     */
    Status?: ModelStatus;
    /**
     *  A description of the status of the recognizer.
     */
    Message?: AnyLengthString;
    /**
     * The time that the recognizer was submitted for processing.
     */
    SubmitTime?: Timestamp;
    /**
     * The time that the recognizer creation completed.
     */
    EndTime?: Timestamp;
    /**
     * The time that training of the entity recognizer started.
     */
    TrainingStartTime?: Timestamp;
    /**
     * The time that training of the entity recognizer was completed.
     */
    TrainingEndTime?: Timestamp;
    /**
     * The input data properties of an entity recognizer.
     */
    InputDataConfig?: EntityRecognizerInputDataConfig;
    /**
     *  Provides information about an entity recognizer.
     */
    RecognizerMetadata?: EntityRecognizerMetadata;
    /**
     *  The Amazon Resource Name (ARN) of the AWS Identity and Management (IAM) role that grants Amazon Comprehend read access to your input data.
     */
    DataAccessRoleArn?: IamRoleArn;
    /**
     * ID for the AWS Key Management Service (KMS) key that Amazon Comprehend uses to encrypt data on the storage volume attached to the ML compute instance(s) that process the analysis job. The VolumeKmsKeyId can be either of the following formats:   KMS Key ID: "1234abcd-12ab-34cd-56ef-1234567890ab"    Amazon Resource Name (ARN) of a KMS Key: "arn:aws:kms:us-west-2:111122223333:key/1234abcd-12ab-34cd-56ef-1234567890ab"   
     */
    VolumeKmsKeyId?: KmsKeyId;
    /**
     *  Configuration parameters for a private Virtual Private Cloud (VPC) containing the resources you are using for your custom entity recognizer. For more information, see Amazon VPC. 
     */
    VpcConfig?: VpcConfig;
  }
  export type EntityRecognizerPropertiesList = EntityRecognizerProperties[];
  export type EntityType = "PERSON"|"LOCATION"|"ORGANIZATION"|"COMMERCIAL_ITEM"|"EVENT"|"DATE"|"QUANTITY"|"TITLE"|"OTHER"|string;
  export type EntityTypeName = string;
  export interface EntityTypesEvaluationMetrics {
    /**
     * A measure of the usefulness of the recognizer results for a specific entity type in the test data. High precision means that the recognizer returned substantially more relevant results than irrelevant ones. 
     */
    Precision?: Double;
    /**
     * A measure of how complete the recognizer results are for a specific entity type in the test data. High recall means that the recognizer returned most of the relevant results.
     */
    Recall?: Double;
    /**
     * A measure of how accurate the recognizer results are for for a specific entity type in the test data. It is derived from the Precision and Recall values. The F1Score is the harmonic average of the two scores. The highest score is 1, and the worst score is 0. 
     */
    F1Score?: Double;
  }
  export type EntityTypesList = EntityTypesListItem[];
  export interface EntityTypesListItem {
    /**
     * Entity type of an item on an entity type list.
     */
    Type: EntityTypeName;
  }
  export type Float = number;
  export type IamRoleArn = string;
  export interface InputDataConfig {
    /**
     * The Amazon S3 URI for the input data. The URI must be in same region as the API endpoint that you are calling. The URI can point to a single input file or it can provide the prefix for a collection of data files.  For example, if you use the URI S3://bucketName/prefix, if the prefix is a single file, Amazon Comprehend uses that file as input. If more than one file begins with the prefix, Amazon Comprehend uses all of them as input.
     */
    S3Uri: S3Uri;
    /**
     * Specifies how the text in an input file should be processed:    ONE_DOC_PER_FILE - Each file is considered a separate document. Use this option when you are processing large documents, such as newspaper articles or scientific papers.    ONE_DOC_PER_LINE - Each line in a file is considered a separate document. Use this option when you are processing many short documents, such as text messages.  
     */
    InputFormat?: InputFormat;
  }
  export type InputFormat = "ONE_DOC_PER_FILE"|"ONE_DOC_PER_LINE"|string;
  export type Integer = number;
  export type JobId = string;
  export type JobName = string;
  export type JobStatus = "SUBMITTED"|"IN_PROGRESS"|"COMPLETED"|"FAILED"|"STOP_REQUESTED"|"STOPPED"|string;
  export interface KeyPhrase {
    /**
     * The level of confidence that Amazon Comprehend has in the accuracy of the detection.
     */
    Score?: Float;
    /**
     * The text of a key noun phrase.
     */
    Text?: String;
    /**
     * A character offset in the input text that shows where the key phrase begins (the first character is at position 0). The offset returns the position of each UTF-8 code point in the string. A code point is the abstract character from a particular graphical representation. For example, a multi-byte UTF-8 character maps to a single code point.
     */
    BeginOffset?: Integer;
    /**
     * A character offset in the input text where the key phrase ends. The offset returns the position of each UTF-8 code point in the string. A code point is the abstract character from a particular graphical representation. For example, a multi-byte UTF-8 character maps to a single code point.
     */
    EndOffset?: Integer;
  }
  export interface KeyPhrasesDetectionJobFilter {
    /**
     * Filters on the name of the job.
     */
    JobName?: JobName;
    /**
     * Filters the list of jobs based on job status. Returns only jobs with the specified status.
     */
    JobStatus?: JobStatus;
    /**
     * Filters the list of jobs based on the time that the job was submitted for processing. Returns only jobs submitted before the specified time. Jobs are returned in ascending order, oldest to newest.
     */
    SubmitTimeBefore?: Timestamp;
    /**
     * Filters the list of jobs based on the time that the job was submitted for processing. Returns only jobs submitted after the specified time. Jobs are returned in descending order, newest to oldest.
     */
    SubmitTimeAfter?: Timestamp;
  }
  export interface KeyPhrasesDetectionJobProperties {
    /**
     * The identifier assigned to the key phrases detection job.
     */
    JobId?: JobId;
    /**
     * The name that you assigned the key phrases detection job.
     */
    JobName?: JobName;
    /**
     * The current status of the key phrases detection job. If the status is FAILED, the Message field shows the reason for the failure.
     */
    JobStatus?: JobStatus;
    /**
     * A description of the status of a job.
     */
    Message?: AnyLengthString;
    /**
     * The time that the key phrases detection job was submitted for processing.
     */
    SubmitTime?: Timestamp;
    /**
     * The time that the key phrases detection job completed.
     */
    EndTime?: Timestamp;
    /**
     * The input data configuration that you supplied when you created the key phrases detection job.
     */
    InputDataConfig?: InputDataConfig;
    /**
     * The output data configuration that you supplied when you created the key phrases detection job.
     */
    OutputDataConfig?: OutputDataConfig;
    /**
     * The language code of the input documents.
     */
    LanguageCode?: LanguageCode;
    /**
     * The Amazon Resource Name (ARN) that gives Amazon Comprehend read access to your input data.
     */
    DataAccessRoleArn?: IamRoleArn;
    /**
     * ID for the AWS Key Management Service (KMS) key that Amazon Comprehend uses to encrypt data on the storage volume attached to the ML compute instance(s) that process the analysis job. The VolumeKmsKeyId can be either of the following formats:   KMS Key ID: "1234abcd-12ab-34cd-56ef-1234567890ab"    Amazon Resource Name (ARN) of a KMS Key: "arn:aws:kms:us-west-2:111122223333:key/1234abcd-12ab-34cd-56ef-1234567890ab"   
     */
    VolumeKmsKeyId?: KmsKeyId;
    /**
     *  Configuration parameters for a private Virtual Private Cloud (VPC) containing the resources you are using for your key phrases detection job. For more information, see Amazon VPC. 
     */
    VpcConfig?: VpcConfig;
  }
  export type KeyPhrasesDetectionJobPropertiesList = KeyPhrasesDetectionJobProperties[];
  export type KmsKeyId = string;
  export type LanguageCode = "en"|"es"|"fr"|"de"|"it"|"pt"|"ar"|"hi"|"ja"|"ko"|"zh"|"zh-TW"|string;
  export interface ListDocumentClassificationJobsRequest {
    /**
     * Filters the jobs that are returned. You can filter jobs on their names, status, or the date and time that they were submitted. You can only set one filter at a time.
     */
    Filter?: DocumentClassificationJobFilter;
    /**
     * Identifies the next page of results to return.
     */
    NextToken?: String;
    /**
     * The maximum number of results to return in each page. The default is 100.
     */
    MaxResults?: MaxResultsInteger;
  }
  export interface ListDocumentClassificationJobsResponse {
    /**
     * A list containing the properties of each job returned.
     */
    DocumentClassificationJobPropertiesList?: DocumentClassificationJobPropertiesList;
    /**
     * Identifies the next page of results to return.
     */
    NextToken?: String;
  }
  export interface ListDocumentClassifiersRequest {
    /**
     * Filters the jobs that are returned. You can filter jobs on their name, status, or the date and time that they were submitted. You can only set one filter at a time.
     */
    Filter?: DocumentClassifierFilter;
    /**
     * Identifies the next page of results to return.
     */
    NextToken?: String;
    /**
     * The maximum number of results to return in each page. The default is 100.
     */
    MaxResults?: MaxResultsInteger;
  }
  export interface ListDocumentClassifiersResponse {
    /**
     * A list containing the properties of each job returned.
     */
    DocumentClassifierPropertiesList?: DocumentClassifierPropertiesList;
    /**
     * Identifies the next page of results to return.
     */
    NextToken?: String;
  }
  export interface ListDominantLanguageDetectionJobsRequest {
    /**
     * Filters that jobs that are returned. You can filter jobs on their name, status, or the date and time that they were submitted. You can only set one filter at a time.
     */
    Filter?: DominantLanguageDetectionJobFilter;
    /**
     * Identifies the next page of results to return.
     */
    NextToken?: String;
    /**
     * The maximum number of results to return in each page. The default is 100.
     */
    MaxResults?: MaxResultsInteger;
  }
  export interface ListDominantLanguageDetectionJobsResponse {
    /**
     * A list containing the properties of each job that is returned.
     */
    DominantLanguageDetectionJobPropertiesList?: DominantLanguageDetectionJobPropertiesList;
    /**
     * Identifies the next page of results to return.
     */
    NextToken?: String;
  }
  export interface ListEntitiesDetectionJobsRequest {
    /**
     * Filters the jobs that are returned. You can filter jobs on their name, status, or the date and time that they were submitted. You can only set one filter at a time.
     */
    Filter?: EntitiesDetectionJobFilter;
    /**
     * Identifies the next page of results to return.
     */
    NextToken?: String;
    /**
     * The maximum number of results to return in each page. The default is 100.
     */
    MaxResults?: MaxResultsInteger;
  }
  export interface ListEntitiesDetectionJobsResponse {
    /**
     * A list containing the properties of each job that is returned.
     */
    EntitiesDetectionJobPropertiesList?: EntitiesDetectionJobPropertiesList;
    /**
     * Identifies the next page of results to return.
     */
    NextToken?: String;
  }
  export interface ListEntityRecognizersRequest {
    /**
     * Filters the list of entities returned. You can filter on Status, SubmitTimeBefore, or SubmitTimeAfter. You can only set one filter at a time.
     */
    Filter?: EntityRecognizerFilter;
    /**
     * Identifies the next page of results to return.
     */
    NextToken?: String;
    /**
     *  The maximum number of results to return on each page. The default is 100.
     */
    MaxResults?: MaxResultsInteger;
  }
  export interface ListEntityRecognizersResponse {
    /**
     * The list of properties of an entity recognizer.
     */
    EntityRecognizerPropertiesList?: EntityRecognizerPropertiesList;
    /**
     * Identifies the next page of results to return.
     */
    NextToken?: String;
  }
  export interface ListKeyPhrasesDetectionJobsRequest {
    /**
     * Filters the jobs that are returned. You can filter jobs on their name, status, or the date and time that they were submitted. You can only set one filter at a time.
     */
    Filter?: KeyPhrasesDetectionJobFilter;
    /**
     * Identifies the next page of results to return.
     */
    NextToken?: String;
    /**
     * The maximum number of results to return in each page. The default is 100.
     */
    MaxResults?: MaxResultsInteger;
  }
  export interface ListKeyPhrasesDetectionJobsResponse {
    /**
     * A list containing the properties of each job that is returned.
     */
    KeyPhrasesDetectionJobPropertiesList?: KeyPhrasesDetectionJobPropertiesList;
    /**
     * Identifies the next page of results to return.
     */
    NextToken?: String;
  }
  export type ListOfDetectDominantLanguageResult = BatchDetectDominantLanguageItemResult[];
  export type ListOfDetectEntitiesResult = BatchDetectEntitiesItemResult[];
  export type ListOfDetectKeyPhrasesResult = BatchDetectKeyPhrasesItemResult[];
  export type ListOfDetectSentimentResult = BatchDetectSentimentItemResult[];
  export type ListOfDetectSyntaxResult = BatchDetectSyntaxItemResult[];
  export type ListOfDominantLanguages = DominantLanguage[];
  export type ListOfEntities = Entity[];
  export type ListOfKeyPhrases = KeyPhrase[];
  export type ListOfSyntaxTokens = SyntaxToken[];
  export interface ListSentimentDetectionJobsRequest {
    /**
     * Filters the jobs that are returned. You can filter jobs on their name, status, or the date and time that they were submitted. You can only set one filter at a time.
     */
    Filter?: SentimentDetectionJobFilter;
    /**
     * Identifies the next page of results to return.
     */
    NextToken?: String;
    /**
     * The maximum number of results to return in each page. The default is 100.
     */
    MaxResults?: MaxResultsInteger;
  }
  export interface ListSentimentDetectionJobsResponse {
    /**
     * A list containing the properties of each job that is returned.
     */
    SentimentDetectionJobPropertiesList?: SentimentDetectionJobPropertiesList;
    /**
     * Identifies the next page of results to return.
     */
    NextToken?: String;
  }
  export interface ListTagsForResourceRequest {
    /**
     * The Amazon Resource Name (ARN) of the given Amazon Comprehend resource you are querying. 
     */
    ResourceArn: ComprehendArn;
  }
  export interface ListTagsForResourceResponse {
    /**
     * The Amazon Resource Name (ARN) of the given Amazon Comprehend resource you are querying.
     */
    ResourceArn?: ComprehendArn;
    /**
     * Tags associated with the Amazon Comprehend resource being queried. A tag is a key-value pair that adds as a metadata to a resource used by Amazon Comprehend. For example, a tag with "Sales" as the key might be added to a resource to indicate its use by the sales department. 
     */
    Tags?: TagList;
  }
  export interface ListTopicsDetectionJobsRequest {
    /**
     * Filters the jobs that are returned. Jobs can be filtered on their name, status, or the date and time that they were submitted. You can set only one filter at a time.
     */
    Filter?: TopicsDetectionJobFilter;
    /**
     * Identifies the next page of results to return.
     */
    NextToken?: String;
    /**
     * The maximum number of results to return in each page. The default is 100.
     */
    MaxResults?: MaxResultsInteger;
  }
  export interface ListTopicsDetectionJobsResponse {
    /**
     * A list containing the properties of each job that is returned.
     */
    TopicsDetectionJobPropertiesList?: TopicsDetectionJobPropertiesList;
    /**
     * Identifies the next page of results to return.
     */
    NextToken?: String;
  }
  export type MaxResultsInteger = number;
  export type ModelStatus = "SUBMITTED"|"TRAINING"|"DELETING"|"STOP_REQUESTED"|"STOPPED"|"IN_ERROR"|"TRAINED"|string;
  export type NumberOfTopicsInteger = number;
  export interface OutputDataConfig {
    /**
     * When you use the OutputDataConfig object with asynchronous operations, you specify the Amazon S3 location where you want to write the output data. The URI must be in the same region as the API endpoint that you are calling. The location is used as the prefix for the actual location of the output file. When the topic detection job is finished, the service creates an output file in a directory specific to the job. The S3Uri field contains the location of the output file, called output.tar.gz. It is a compressed archive that contains the ouput of the operation.
     */
    S3Uri: S3Uri;
    /**
     * ID for the AWS Key Management Service (KMS) key that Amazon Comprehend uses to encrypt the output results from an analysis job. The KmsKeyId can be one of the following formats:   KMS Key ID: "1234abcd-12ab-34cd-56ef-1234567890ab"    Amazon Resource Name (ARN) of a KMS Key: "arn:aws:kms:us-west-2:111122223333:key/1234abcd-12ab-34cd-56ef-1234567890ab"    KMS Key Alias: "alias/ExampleAlias"    ARN of a KMS Key Alias: "arn:aws:kms:us-west-2:111122223333:alias/ExampleAlias"   
     */
    KmsKeyId?: KmsKeyId;
  }
  export interface PartOfSpeechTag {
    /**
     * Identifies the part of speech that the token represents.
     */
    Tag?: PartOfSpeechTagType;
    /**
     * The confidence that Amazon Comprehend has that the part of speech was correctly identified.
     */
    Score?: Float;
  }
  export type PartOfSpeechTagType = "ADJ"|"ADP"|"ADV"|"AUX"|"CONJ"|"CCONJ"|"DET"|"INTJ"|"NOUN"|"NUM"|"O"|"PART"|"PRON"|"PROPN"|"PUNCT"|"SCONJ"|"SYM"|"VERB"|string;
  export type S3Uri = string;
  export type SecurityGroupId = string;
  export type SecurityGroupIds = SecurityGroupId[];
  export interface SentimentDetectionJobFilter {
    /**
     * Filters on the name of the job.
     */
    JobName?: JobName;
    /**
     * Filters the list of jobs based on job status. Returns only jobs with the specified status.
     */
    JobStatus?: JobStatus;
    /**
     * Filters the list of jobs based on the time that the job was submitted for processing. Returns only jobs submitted before the specified time. Jobs are returned in ascending order, oldest to newest.
     */
    SubmitTimeBefore?: Timestamp;
    /**
     * Filters the list of jobs based on the time that the job was submitted for processing. Returns only jobs submitted after the specified time. Jobs are returned in descending order, newest to oldest.
     */
    SubmitTimeAfter?: Timestamp;
  }
  export interface SentimentDetectionJobProperties {
    /**
     * The identifier assigned to the sentiment detection job.
     */
    JobId?: JobId;
    /**
     * The name that you assigned to the sentiment detection job
     */
    JobName?: JobName;
    /**
     * The current status of the sentiment detection job. If the status is FAILED, the Messages field shows the reason for the failure.
     */
    JobStatus?: JobStatus;
    /**
     * A description of the status of a job.
     */
    Message?: AnyLengthString;
    /**
     * The time that the sentiment detection job was submitted for processing.
     */
    SubmitTime?: Timestamp;
    /**
     * The time that the sentiment detection job ended.
     */
    EndTime?: Timestamp;
    /**
     * The input data configuration that you supplied when you created the sentiment detection job.
     */
    InputDataConfig?: InputDataConfig;
    /**
     * The output data configuration that you supplied when you created the sentiment detection job.
     */
    OutputDataConfig?: OutputDataConfig;
    /**
     * The language code of the input documents.
     */
    LanguageCode?: LanguageCode;
    /**
     * The Amazon Resource Name (ARN) that gives Amazon Comprehend read access to your input data.
     */
    DataAccessRoleArn?: IamRoleArn;
    /**
     * ID for the AWS Key Management Service (KMS) key that Amazon Comprehend uses to encrypt data on the storage volume attached to the ML compute instance(s) that process the analysis job. The VolumeKmsKeyId can be either of the following formats:   KMS Key ID: "1234abcd-12ab-34cd-56ef-1234567890ab"    Amazon Resource Name (ARN) of a KMS Key: "arn:aws:kms:us-west-2:111122223333:key/1234abcd-12ab-34cd-56ef-1234567890ab"   
     */
    VolumeKmsKeyId?: KmsKeyId;
    /**
     *  Configuration parameters for a private Virtual Private Cloud (VPC) containing the resources you are using for your sentiment detection job. For more information, see Amazon VPC. 
     */
    VpcConfig?: VpcConfig;
  }
  export type SentimentDetectionJobPropertiesList = SentimentDetectionJobProperties[];
  export interface SentimentScore {
    /**
     * The level of confidence that Amazon Comprehend has in the accuracy of its detection of the POSITIVE sentiment.
     */
    Positive?: Float;
    /**
     * The level of confidence that Amazon Comprehend has in the accuracy of its detection of the NEGATIVE sentiment.
     */
    Negative?: Float;
    /**
     * The level of confidence that Amazon Comprehend has in the accuracy of its detection of the NEUTRAL sentiment.
     */
    Neutral?: Float;
    /**
     * The level of confidence that Amazon Comprehend has in the accuracy of its detection of the MIXED sentiment.
     */
    Mixed?: Float;
  }
  export type SentimentType = "POSITIVE"|"NEGATIVE"|"NEUTRAL"|"MIXED"|string;
  export interface StartDocumentClassificationJobRequest {
    /**
     * The identifier of the job.
     */
    JobName?: JobName;
    /**
     * The Amazon Resource Name (ARN) of the document classifier to use to process the job.
     */
    DocumentClassifierArn: DocumentClassifierArn;
    /**
     * Specifies the format and location of the input data for the job.
     */
    InputDataConfig: InputDataConfig;
    /**
     * Specifies where to send the output files.
     */
    OutputDataConfig: OutputDataConfig;
    /**
     * The Amazon Resource Name (ARN) of the AWS Identity and Access Management (IAM) role that grants Amazon Comprehend read access to your input data.
     */
    DataAccessRoleArn: IamRoleArn;
    /**
     * A unique identifier for the request. If you do not set the client request token, Amazon Comprehend generates one.
     */
    ClientRequestToken?: ClientRequestTokenString;
    /**
     * ID for the AWS Key Management Service (KMS) key that Amazon Comprehend uses to encrypt data on the storage volume attached to the ML compute instance(s) that process the analysis job. The VolumeKmsKeyId can be either of the following formats:   KMS Key ID: "1234abcd-12ab-34cd-56ef-1234567890ab"    Amazon Resource Name (ARN) of a KMS Key: "arn:aws:kms:us-west-2:111122223333:key/1234abcd-12ab-34cd-56ef-1234567890ab"   
     */
    VolumeKmsKeyId?: KmsKeyId;
    /**
     * Configuration parameters for an optional private Virtual Private Cloud (VPC) containing the resources you are using for your document classification job. For more information, see Amazon VPC. 
     */
    VpcConfig?: VpcConfig;
  }
  export interface StartDocumentClassificationJobResponse {
    /**
     * The identifier generated for the job. To get the status of the job, use this identifier with the operation.
     */
    JobId?: JobId;
    /**
     * The status of the job:   SUBMITTED - The job has been received and queued for processing.   IN_PROGRESS - Amazon Comprehend is processing the job.   COMPLETED - The job was successfully completed and the output is available.   FAILED - The job did not complete. For details, use the operation.   STOP_REQUESTED - Amazon Comprehend has received a stop request for the job and is processing the request.   STOPPED - The job was successfully stopped without completing.  
     */
    JobStatus?: JobStatus;
  }
  export interface StartDominantLanguageDetectionJobRequest {
    /**
     * Specifies the format and location of the input data for the job.
     */
    InputDataConfig: InputDataConfig;
    /**
     * Specifies where to send the output files.
     */
    OutputDataConfig: OutputDataConfig;
    /**
     * The Amazon Resource Name (ARN) of the AWS Identity and Access Management (IAM) role that grants Amazon Comprehend read access to your input data. For more information, see https://docs.aws.amazon.com/comprehend/latest/dg/access-control-managing-permissions.html#auth-role-permissions.
     */
    DataAccessRoleArn: IamRoleArn;
    /**
     * An identifier for the job.
     */
    JobName?: JobName;
    /**
     * A unique identifier for the request. If you do not set the client request token, Amazon Comprehend generates one.
     */
    ClientRequestToken?: ClientRequestTokenString;
    /**
     * ID for the AWS Key Management Service (KMS) key that Amazon Comprehend uses to encrypt data on the storage volume attached to the ML compute instance(s) that process the analysis job. The VolumeKmsKeyId can be either of the following formats:   KMS Key ID: "1234abcd-12ab-34cd-56ef-1234567890ab"    Amazon Resource Name (ARN) of a KMS Key: "arn:aws:kms:us-west-2:111122223333:key/1234abcd-12ab-34cd-56ef-1234567890ab"   
     */
    VolumeKmsKeyId?: KmsKeyId;
    /**
     * Configuration parameters for an optional private Virtual Private Cloud (VPC) containing the resources you are using for your dominant language detection job. For more information, see Amazon VPC. 
     */
    VpcConfig?: VpcConfig;
  }
  export interface StartDominantLanguageDetectionJobResponse {
    /**
     * The identifier generated for the job. To get the status of a job, use this identifier with the operation.
     */
    JobId?: JobId;
    /**
     * The status of the job.    SUBMITTED - The job has been received and is queued for processing.   IN_PROGRESS - Amazon Comprehend is processing the job.   COMPLETED - The job was successfully completed and the output is available.   FAILED - The job did not complete. To get details, use the operation.  
     */
    JobStatus?: JobStatus;
  }
  export interface StartEntitiesDetectionJobRequest {
    /**
     * Specifies the format and location of the input data for the job.
     */
    InputDataConfig: InputDataConfig;
    /**
     * Specifies where to send the output files.
     */
    OutputDataConfig: OutputDataConfig;
    /**
     * The Amazon Resource Name (ARN) of the AWS Identity and Access Management (IAM) role that grants Amazon Comprehend read access to your input data. For more information, see https://docs.aws.amazon.com/comprehend/latest/dg/access-control-managing-permissions.html#auth-role-permissions.
     */
    DataAccessRoleArn: IamRoleArn;
    /**
     * The identifier of the job.
     */
    JobName?: JobName;
    /**
     * The Amazon Resource Name (ARN) that identifies the specific entity recognizer to be used by the StartEntitiesDetectionJob. This ARN is optional and is only used for a custom entity recognition job.
     */
    EntityRecognizerArn?: EntityRecognizerArn;
    /**
     * The language of the input documents. All documents must be in the same language. You can specify any of the languages supported by Amazon Comprehend: English ("en"), Spanish ("es"), French ("fr"), German ("de"), Italian ("it"), or Portuguese ("pt"). If custom entities recognition is used, this parameter is ignored and the language used for training the model is used instead.
     */
    LanguageCode: LanguageCode;
    /**
     * A unique identifier for the request. If you don't set the client request token, Amazon Comprehend generates one.
     */
    ClientRequestToken?: ClientRequestTokenString;
    /**
     * ID for the AWS Key Management Service (KMS) key that Amazon Comprehend uses to encrypt data on the storage volume attached to the ML compute instance(s) that process the analysis job. The VolumeKmsKeyId can be either of the following formats:   KMS Key ID: "1234abcd-12ab-34cd-56ef-1234567890ab"    Amazon Resource Name (ARN) of a KMS Key: "arn:aws:kms:us-west-2:111122223333:key/1234abcd-12ab-34cd-56ef-1234567890ab"   
     */
    VolumeKmsKeyId?: KmsKeyId;
    /**
     * Configuration parameters for an optional private Virtual Private Cloud (VPC) containing the resources you are using for your entity detection job. For more information, see Amazon VPC. 
     */
    VpcConfig?: VpcConfig;
  }
  export interface StartEntitiesDetectionJobResponse {
    /**
     * The identifier generated for the job. To get the status of job, use this identifier with the operation.
     */
    JobId?: JobId;
    /**
     * The status of the job.    SUBMITTED - The job has been received and is queued for processing.   IN_PROGRESS - Amazon Comprehend is processing the job.   COMPLETED - The job was successfully completed and the output is available.   FAILED - The job did not complete. To get details, use the operation.   STOP_REQUESTED - Amazon Comprehend has received a stop request for the job and is processing the request.   STOPPED - The job was successfully stopped without completing.  
     */
    JobStatus?: JobStatus;
  }
  export interface StartKeyPhrasesDetectionJobRequest {
    /**
     * Specifies the format and location of the input data for the job.
     */
    InputDataConfig: InputDataConfig;
    /**
     * Specifies where to send the output files.
     */
    OutputDataConfig: OutputDataConfig;
    /**
     * The Amazon Resource Name (ARN) of the AWS Identity and Access Management (IAM) role that grants Amazon Comprehend read access to your input data. For more information, see https://docs.aws.amazon.com/comprehend/latest/dg/access-control-managing-permissions.html#auth-role-permissions.
     */
    DataAccessRoleArn: IamRoleArn;
    /**
     * The identifier of the job.
     */
    JobName?: JobName;
    /**
     * The language of the input documents. You can specify any of the primary languages supported by Amazon Comprehend: German ("de"), English ("en"), Spanish ("es"), French ("fr"), Italian ("it"), or Portuguese ("pt"). All documents must be in the same language.
     */
    LanguageCode: LanguageCode;
    /**
     * A unique identifier for the request. If you don't set the client request token, Amazon Comprehend generates one.
     */
    ClientRequestToken?: ClientRequestTokenString;
    /**
     * ID for the AWS Key Management Service (KMS) key that Amazon Comprehend uses to encrypt data on the storage volume attached to the ML compute instance(s) that process the analysis job. The VolumeKmsKeyId can be either of the following formats:   KMS Key ID: "1234abcd-12ab-34cd-56ef-1234567890ab"    Amazon Resource Name (ARN) of a KMS Key: "arn:aws:kms:us-west-2:111122223333:key/1234abcd-12ab-34cd-56ef-1234567890ab"   
     */
    VolumeKmsKeyId?: KmsKeyId;
    /**
     *  Configuration parameters for an optional private Virtual Private Cloud (VPC) containing the resources you are using for your key phrases detection job. For more information, see Amazon VPC. 
     */
    VpcConfig?: VpcConfig;
  }
  export interface StartKeyPhrasesDetectionJobResponse {
    /**
     * The identifier generated for the job. To get the status of a job, use this identifier with the operation.
     */
    JobId?: JobId;
    /**
     * The status of the job.    SUBMITTED - The job has been received and is queued for processing.   IN_PROGRESS - Amazon Comprehend is processing the job.   COMPLETED - The job was successfully completed and the output is available.   FAILED - The job did not complete. To get details, use the operation.  
     */
    JobStatus?: JobStatus;
  }
  export interface StartSentimentDetectionJobRequest {
    /**
     * Specifies the format and location of the input data for the job.
     */
    InputDataConfig: InputDataConfig;
    /**
     * Specifies where to send the output files. 
     */
    OutputDataConfig: OutputDataConfig;
    /**
     * The Amazon Resource Name (ARN) of the AWS Identity and Access Management (IAM) role that grants Amazon Comprehend read access to your input data. For more information, see https://docs.aws.amazon.com/comprehend/latest/dg/access-control-managing-permissions.html#auth-role-permissions.
     */
    DataAccessRoleArn: IamRoleArn;
    /**
     * The identifier of the job.
     */
    JobName?: JobName;
    /**
     * The language of the input documents. You can specify any of the primary languages supported by Amazon Comprehend: German ("de"), English ("en"), Spanish ("es"), French ("fr"), Italian ("it"), or Portuguese ("pt"). All documents must be in the same language.
     */
    LanguageCode: LanguageCode;
    /**
     * A unique identifier for the request. If you don't set the client request token, Amazon Comprehend generates one.
     */
    ClientRequestToken?: ClientRequestTokenString;
    /**
     * ID for the AWS Key Management Service (KMS) key that Amazon Comprehend uses to encrypt data on the storage volume attached to the ML compute instance(s) that process the analysis job. The VolumeKmsKeyId can be either of the following formats:   KMS Key ID: "1234abcd-12ab-34cd-56ef-1234567890ab"    Amazon Resource Name (ARN) of a KMS Key: "arn:aws:kms:us-west-2:111122223333:key/1234abcd-12ab-34cd-56ef-1234567890ab"   
     */
    VolumeKmsKeyId?: KmsKeyId;
    /**
     * Configuration parameters for an optional private Virtual Private Cloud (VPC) containing the resources you are using for your sentiment detection job. For more information, see Amazon VPC. 
     */
    VpcConfig?: VpcConfig;
  }
  export interface StartSentimentDetectionJobResponse {
    /**
     * The identifier generated for the job. To get the status of a job, use this identifier with the operation.
     */
    JobId?: JobId;
    /**
     * The status of the job.    SUBMITTED - The job has been received and is queued for processing.   IN_PROGRESS - Amazon Comprehend is processing the job.   COMPLETED - The job was successfully completed and the output is available.   FAILED - The job did not complete. To get details, use the operation.  
     */
    JobStatus?: JobStatus;
  }
  export interface StartTopicsDetectionJobRequest {
    /**
     * Specifies the format and location of the input data for the job.
     */
    InputDataConfig: InputDataConfig;
    /**
     * Specifies where to send the output files. The output is a compressed archive with two files, topic-terms.csv that lists the terms associated with each topic, and doc-topics.csv that lists the documents associated with each topic
     */
    OutputDataConfig: OutputDataConfig;
    /**
     * The Amazon Resource Name (ARN) of the AWS Identity and Access Management (IAM) role that grants Amazon Comprehend read access to your input data. For more information, see https://docs.aws.amazon.com/comprehend/latest/dg/access-control-managing-permissions.html#auth-role-permissions.
     */
    DataAccessRoleArn: IamRoleArn;
    /**
     * The identifier of the job.
     */
    JobName?: JobName;
    /**
     * The number of topics to detect.
     */
    NumberOfTopics?: NumberOfTopicsInteger;
    /**
     * A unique identifier for the request. If you do not set the client request token, Amazon Comprehend generates one.
     */
    ClientRequestToken?: ClientRequestTokenString;
    /**
     * ID for the AWS Key Management Service (KMS) key that Amazon Comprehend uses to encrypt data on the storage volume attached to the ML compute instance(s) that process the analysis job. The VolumeKmsKeyId can be either of the following formats:   KMS Key ID: "1234abcd-12ab-34cd-56ef-1234567890ab"    Amazon Resource Name (ARN) of a KMS Key: "arn:aws:kms:us-west-2:111122223333:key/1234abcd-12ab-34cd-56ef-1234567890ab"   
     */
    VolumeKmsKeyId?: KmsKeyId;
    /**
     * Configuration parameters for an optional private Virtual Private Cloud (VPC) containing the resources you are using for your topic detection job. For more information, see Amazon VPC. 
     */
    VpcConfig?: VpcConfig;
  }
  export interface StartTopicsDetectionJobResponse {
    /**
     * The identifier generated for the job. To get the status of the job, use this identifier with the DescribeTopicDetectionJob operation.
     */
    JobId?: JobId;
    /**
     * The status of the job:    SUBMITTED - The job has been received and is queued for processing.   IN_PROGRESS - Amazon Comprehend is processing the job.   COMPLETED - The job was successfully completed and the output is available.   FAILED - The job did not complete. To get details, use the DescribeTopicDetectionJob operation.  
     */
    JobStatus?: JobStatus;
  }
  export interface StopDominantLanguageDetectionJobRequest {
    /**
     * The identifier of the dominant language detection job to stop.
     */
    JobId: JobId;
  }
  export interface StopDominantLanguageDetectionJobResponse {
    /**
     * The identifier of the dominant language detection job to stop.
     */
    JobId?: JobId;
    /**
     * Either STOP_REQUESTED if the job is currently running, or STOPPED if the job was previously stopped with the StopDominantLanguageDetectionJob operation.
     */
    JobStatus?: JobStatus;
  }
  export interface StopEntitiesDetectionJobRequest {
    /**
     * The identifier of the entities detection job to stop.
     */
    JobId: JobId;
  }
  export interface StopEntitiesDetectionJobResponse {
    /**
     * The identifier of the entities detection job to stop.
     */
    JobId?: JobId;
    /**
     * Either STOP_REQUESTED if the job is currently running, or STOPPED if the job was previously stopped with the StopEntitiesDetectionJob operation.
     */
    JobStatus?: JobStatus;
  }
  export interface StopKeyPhrasesDetectionJobRequest {
    /**
     * The identifier of the key phrases detection job to stop.
     */
    JobId: JobId;
  }
  export interface StopKeyPhrasesDetectionJobResponse {
    /**
     * The identifier of the key phrases detection job to stop.
     */
    JobId?: JobId;
    /**
     * Either STOP_REQUESTED if the job is currently running, or STOPPED if the job was previously stopped with the StopKeyPhrasesDetectionJob operation.
     */
    JobStatus?: JobStatus;
  }
  export interface StopSentimentDetectionJobRequest {
    /**
     * The identifier of the sentiment detection job to stop.
     */
    JobId: JobId;
  }
  export interface StopSentimentDetectionJobResponse {
    /**
     * The identifier of the sentiment detection job to stop.
     */
    JobId?: JobId;
    /**
     * Either STOP_REQUESTED if the job is currently running, or STOPPED if the job was previously stopped with the StopSentimentDetectionJob operation.
     */
    JobStatus?: JobStatus;
  }
  export interface StopTrainingDocumentClassifierRequest {
    /**
     * The Amazon Resource Name (ARN) that identifies the document classifier currently being trained.
     */
    DocumentClassifierArn: DocumentClassifierArn;
  }
  export interface StopTrainingDocumentClassifierResponse {
  }
  export interface StopTrainingEntityRecognizerRequest {
    /**
     * The Amazon Resource Name (ARN) that identifies the entity recognizer currently being trained.
     */
    EntityRecognizerArn: EntityRecognizerArn;
  }
  export interface StopTrainingEntityRecognizerResponse {
  }
  export type String = string;
  export type StringList = String[];
  export type SubnetId = string;
  export type Subnets = SubnetId[];
  export type SyntaxLanguageCode = "en"|"es"|"fr"|"de"|"it"|"pt"|string;
  export interface SyntaxToken {
    /**
     * A unique identifier for a token.
     */
    TokenId?: Integer;
    /**
     * The word that was recognized in the source text.
     */
    Text?: String;
    /**
     * The zero-based offset from the beginning of the source text to the first character in the word.
     */
    BeginOffset?: Integer;
    /**
     * The zero-based offset from the beginning of the source text to the last character in the word.
     */
    EndOffset?: Integer;
    /**
     * Provides the part of speech label and the confidence level that Amazon Comprehend has that the part of speech was correctly identified. For more information, see how-syntax.
     */
    PartOfSpeech?: PartOfSpeechTag;
  }
  export interface Tag {
    /**
     * The initial part of a key-value pair that forms a tag associated with a given resource. For instance, if you want to show which resources are used by which departments, you might use “Department” as the key portion of the pair, with multiple possible values such as “sales,” “legal,” and “administration.” 
     */
    Key: TagKey;
    /**
     *  The second part of a key-value pair that forms a tag associated with a given resource. For instance, if you want to show which resources are used by which departments, you might use “Department” as the initial (key) portion of the pair, with a value of “sales” to indicate the sales department. 
     */
    Value?: TagValue;
  }
  export type TagKey = string;
  export type TagKeyList = TagKey[];
  export type TagList = Tag[];
  export interface TagResourceRequest {
    /**
     * The Amazon Resource Name (ARN) of the given Amazon Comprehend resource to which you want to associate the tags. 
     */
    ResourceArn: ComprehendArn;
    /**
     * Tags being associated with a specific Amazon Comprehend resource. There can be a maximum of 50 tags (both existing and pending) associated with a specific resource. 
     */
    Tags: TagList;
  }
  export interface TagResourceResponse {
  }
  export type TagValue = string;
  export type Timestamp = Date;
  export interface TopicsDetectionJobFilter {
    /**
     * 
     */
    JobName?: JobName;
    /**
     * Filters the list of topic detection jobs based on job status. Returns only jobs with the specified status.
     */
    JobStatus?: JobStatus;
    /**
     * Filters the list of jobs based on the time that the job was submitted for processing. Only returns jobs submitted before the specified time. Jobs are returned in descending order, newest to oldest.
     */
    SubmitTimeBefore?: Timestamp;
    /**
     * Filters the list of jobs based on the time that the job was submitted for processing. Only returns jobs submitted after the specified time. Jobs are returned in ascending order, oldest to newest.
     */
    SubmitTimeAfter?: Timestamp;
  }
  export interface TopicsDetectionJobProperties {
    /**
     * The identifier assigned to the topic detection job.
     */
    JobId?: JobId;
    /**
     * The name of the topic detection job.
     */
    JobName?: JobName;
    /**
     * The current status of the topic detection job. If the status is Failed, the reason for the failure is shown in the Message field.
     */
    JobStatus?: JobStatus;
    /**
     * A description for the status of a job.
     */
    Message?: AnyLengthString;
    /**
     * The time that the topic detection job was submitted for processing.
     */
    SubmitTime?: Timestamp;
    /**
     * The time that the topic detection job was completed.
     */
    EndTime?: Timestamp;
    /**
     * The input data configuration supplied when you created the topic detection job.
     */
    InputDataConfig?: InputDataConfig;
    /**
     * The output data configuration supplied when you created the topic detection job.
     */
    OutputDataConfig?: OutputDataConfig;
    /**
     * The number of topics to detect supplied when you created the topic detection job. The default is 10. 
     */
    NumberOfTopics?: Integer;
    /**
     * The Amazon Resource Name (ARN) of the AWS Identity and Management (IAM) role that grants Amazon Comprehend read access to your job data. 
     */
    DataAccessRoleArn?: IamRoleArn;
    /**
     * ID for the AWS Key Management Service (KMS) key that Amazon Comprehend uses to encrypt data on the storage volume attached to the ML compute instance(s) that process the analysis job. The VolumeKmsKeyId can be either of the following formats:   KMS Key ID: "1234abcd-12ab-34cd-56ef-1234567890ab"    Amazon Resource Name (ARN) of a KMS Key: "arn:aws:kms:us-west-2:111122223333:key/1234abcd-12ab-34cd-56ef-1234567890ab"   
     */
    VolumeKmsKeyId?: KmsKeyId;
    /**
     * Configuration parameters for a private Virtual Private Cloud (VPC) containing the resources you are using for your topic detection job. For more information, see Amazon VPC. 
     */
    VpcConfig?: VpcConfig;
  }
  export type TopicsDetectionJobPropertiesList = TopicsDetectionJobProperties[];
  export interface UntagResourceRequest {
    /**
     *  The Amazon Resource Name (ARN) of the given Amazon Comprehend resource from which you want to remove the tags. 
     */
    ResourceArn: ComprehendArn;
    /**
     * The initial part of a key-value pair that forms a tag being removed from a given resource. For example, a tag with "Sales" as the key might be added to a resource to indicate its use by the sales department. Keys must be unique and cannot be duplicated for a particular resource. 
     */
    TagKeys: TagKeyList;
  }
  export interface UntagResourceResponse {
  }
  export interface VpcConfig {
    /**
     * The ID number for a security group on an instance of your private VPC. Security groups on your VPC function serve as a virtual firewall to control inbound and outbound traffic and provides security for the resources that you’ll be accessing on the VPC. This ID number is preceded by "sg-", for instance: "sg-03b388029b0a285ea". For more information, see Security Groups for your VPC. 
     */
    SecurityGroupIds: SecurityGroupIds;
    /**
     * The ID for each subnet being used in your private VPC. This subnet is a subset of the a range of IPv4 addresses used by the VPC and is specific to a given availability zone in the VPC’s region. This ID number is preceded by "subnet-", for instance: "subnet-04ccf456919e69055". For more information, see VPCs and Subnets. 
     */
    Subnets: Subnets;
  }
  /**
   * A string in YYYY-MM-DD format that represents the latest possible API version that can be used in this service. Specify 'latest' to use the latest possible version.
   */
  export type apiVersion = "2017-11-27"|"latest"|string;
  export interface ClientApiVersions {
    /**
     * A string in YYYY-MM-DD format that represents the latest possible API version that can be used in this service. Specify 'latest' to use the latest possible version.
     */
    apiVersion?: apiVersion;
  }
  export type ClientConfiguration = ServiceConfigurationOptions & ClientApiVersions;
  /**
   * Contains interfaces for use with the Comprehend client.
   */
  export import Types = Comprehend;
}
export = Comprehend;