File size: 86,420 Bytes
6fa4bc9 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 | {
"paper_id": "P01-1004",
"header": {
"generated_with": "S2ORC 1.0.0",
"date_generated": "2023-01-19T09:29:24.877011Z"
},
"title": "Low-cost, High-performance Translation Retrieval: Dumber is Better",
"authors": [
{
"first": "Timothy",
"middle": [],
"last": "Baldwin",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "Tokyo Institute of Technology",
"location": {
"addrLine": "2-12-1 O-okayama, Meguro-ku",
"postCode": "152-8552",
"settlement": "Tokyo",
"country": "JAPAN"
}
},
"email": "tim@cl.cs.titech.ac.jp"
}
],
"year": "",
"venue": null,
"identifiers": {},
"abstract": "In this paper, we compare the relative effects of segment order, segmentation and segment contiguity on the retrieval performance of a translation memory system. We take a selection of both bag-of-words and segment order-sensitive string comparison methods, and run each over both characterand word-segmented data, in combination with a range of local segment contiguity models (in the form of N-grams). Over two distinct datasets, we find that indexing according to simple character bigrams produces a retrieval accuracy superior to any of the tested word Ngram models. Further, in their optimum configuration, bag-of-words methods are shown to be equivalent to segment ordersensitive methods in terms of retrieval accuracy, but much faster. We also provide evidence that our findings are scalable.",
"pdf_parse": {
"paper_id": "P01-1004",
"_pdf_hash": "",
"abstract": [
{
"text": "In this paper, we compare the relative effects of segment order, segmentation and segment contiguity on the retrieval performance of a translation memory system. We take a selection of both bag-of-words and segment order-sensitive string comparison methods, and run each over both characterand word-segmented data, in combination with a range of local segment contiguity models (in the form of N-grams). Over two distinct datasets, we find that indexing according to simple character bigrams produces a retrieval accuracy superior to any of the tested word Ngram models. Further, in their optimum configuration, bag-of-words methods are shown to be equivalent to segment ordersensitive methods in terms of retrieval accuracy, but much faster. We also provide evidence that our findings are scalable.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Abstract",
"sec_num": null
}
],
"body_text": [
{
"text": "Translation memories (TMs) are a list of translation records (source language strings paired with a unique target language translation), which the TM system accesses in suggesting a list of target language (L2) translation candidates for a given source language (L1) input (Trujillo, 1999; Planas, 1998) . Translation retrieval (TR) is a description of this process of selecting from the TM a set of translation records (TRecs) of maximum L1 similarity to a given input. Typically in example-based machine translation, either a single TRec is retrieved from the TM based on a match with the overall L1 input, or the input is partitioned into coherent segments, and individual translations retrieved for each (Sato and Nagao, 1990; Nirenburg et al., 1993) ; this is the first step toward generating a customised translation for the input. With stand-alone TM systems, on the other hand, the system selects an arbitrary number of translation candidates falling within a certain empirical corridor of similarity with the overall input string, and simply outputs these for manual manipulation by the user in fashioning the final translation.",
"cite_spans": [
{
"start": 273,
"end": 289,
"text": "(Trujillo, 1999;",
"ref_id": "BIBREF16"
},
{
"start": 290,
"end": 303,
"text": "Planas, 1998)",
"ref_id": "BIBREF9"
},
{
"start": 708,
"end": 730,
"text": "(Sato and Nagao, 1990;",
"ref_id": "BIBREF12"
},
{
"start": 731,
"end": 754,
"text": "Nirenburg et al., 1993)",
"ref_id": "BIBREF8"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "A key assumption surrounding the bulk of past TR research has been that the greater the match stringency/linguistic awareness of the retrieval mechanism, the greater the final retrieval accuracy will become. Naturally, any appreciation in retrieval complexity comes at a price in terms of computational overhead. We thus follow the lead of Baldwin and Tanaka (2000) in asking the question: what is the empirical effect on retrieval performance of different match approaches? Here, retrieval performance is defined as the combination of retrieval speed and accuracy, with the ideal method offering fast response times at high accuracy.",
"cite_spans": [
{
"start": 340,
"end": 365,
"text": "Baldwin and Tanaka (2000)",
"ref_id": "BIBREF0"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "In this paper, we choose to focus on retrieval performance within a Japanese-English TR context. One key area of interest with Japanese is the effect that segmentation has on retrieval performance. As Japanese is a non-segmenting language (does not explicitly delimit words orthographically), we can take the brute-force approach in treating each string as a sequence of characters (character-based indexing), or alternatively call upon segmentation technology in partitioning each string into words (word-based indexing). Orthogonal to this is the question of sensitivity to segment order. That is, should our match mechanism treat each string as an unorganised multiset of terms (the bag-of-words approach), or attempt to find the match that best preserves the original segment order in the input (the segment order-sensitive approach)?",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "We tackle this issue by implementing a sample of representative bag-of-words and segment ordersensitive methods and testing the retrieval performance of each. As a third orthogonal parameter, we consider the effects of segment contiguity. That is, do matches over contiguous segments provide closer overall translation correspondence than matches over displaced segments? Segment contiguity is either explicitly modelled within the string match mechanism, or provided as an add-in in the form of segment N-grams.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "To preempt the major findings of this paper, over a series of experiments we find that character-based indexing is consistently superior to word-based indexing. Furthermore, the bagof-words methods we test are equivalent in retrieval accuracy to the more expensive segment order-sensitive methods, but superior in retrieval speed. Finally, segment contiguity models provide benefits in terms of both retrieval accuracy and retrieval speed, particularly when coupled with character-based indexing. We thus provide clear evidence that high-performance TR is achievable with naive methods, and moreso that such methods outperform more intricate, expensive methods. That is, the dumber the retrieval mechanism, the better.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "Below, we review the orthogonal parameters of segmentation, segment order and segment contiguity ( \u00a7 2). We then present a range of both bag-of-words and segment order-sensitive string comparison methods ( \u00a7 3) and detail the evaluation methodology ( \u00a7 4). Finally, we evaluate the different methods in a Japanese-English TR context ( \u00a7 5), before concluding the paper ( \u00a7 6).",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "In this section, we review three parameter types that we suggest impinge on TR performance, namely segmentation, segment order, and segment contiguity.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Basic Parameters",
"sec_num": "2"
},
{
"text": "Despite non-segmenting languages such as Japanese not making use of segment delimiters, it is possible to artificially partition off a given string into constituent morphemes through the process of segmentation. We will collectively term the resultant segments as words for the remainder of this paper.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Segmentation",
"sec_num": "2.1"
},
{
"text": "Looking to past research on string comparison methods for TM systems, almost all systems involving Japanese as the source language rely on segmentation (Nakamura, 1989; Sumita and Tsutsumi, 1991; Kitamura and Yamamoto, 1996; Tanaka, 1997) , with Sato (1992) and Sato and Kawase (1994) providing rare instances of character-based systems. This is despite Fujii and Croft (1993) providing evidence from Japanese information retrieval that character-based indexing performs comparably to word-based indexing.",
"cite_spans": [
{
"start": 152,
"end": 168,
"text": "(Nakamura, 1989;",
"ref_id": "BIBREF7"
},
{
"start": 169,
"end": 195,
"text": "Sumita and Tsutsumi, 1991;",
"ref_id": "BIBREF14"
},
{
"start": 196,
"end": 224,
"text": "Kitamura and Yamamoto, 1996;",
"ref_id": "BIBREF3"
},
{
"start": 225,
"end": 238,
"text": "Tanaka, 1997)",
"ref_id": "BIBREF15"
},
{
"start": 246,
"end": 257,
"text": "Sato (1992)",
"ref_id": "BIBREF13"
},
{
"start": 262,
"end": 284,
"text": "Sato and Kawase (1994)",
"ref_id": "BIBREF11"
},
{
"start": 354,
"end": 376,
"text": "Fujii and Croft (1993)",
"ref_id": "BIBREF1"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Segmentation",
"sec_num": "2.1"
},
{
"text": "In analogous research, Baldwin and Tanaka (2000) compared characterand word-based indexing within a Japanese-English TR context and found character-based indexing to hold a slight empirical advantage.",
"cite_spans": [
{
"start": 23,
"end": 48,
"text": "Baldwin and Tanaka (2000)",
"ref_id": "BIBREF0"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Segmentation",
"sec_num": "2.1"
},
{
"text": "The most obvious advantage of character-based indexing over word-based indexing is that there is no pre-processing overhead. Other arguments for character-based indexing over word-based indexing are that we: (a) avoid the need to commit ourselves to a particular analysis type in the case of ambiguity or unknown words; (b) avoid the need for stemming/lemmatisation; and (c) to a large extent get around problems related to the normalisation of lexical alternation. Note that all methods described below are applicable to both word-and character-based indexing. To avoid confusion between the two lexeme types, we will collectively refer to the elements of indexing as segments.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Segmentation",
"sec_num": "2.1"
},
{
"text": "Our expectation is that TRecs that preserve the segment order observed in the input string will provide closer-matching translations than TRecs containing those same segments in a different order.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Segment Order",
"sec_num": "2.2"
},
{
"text": "As far as we are aware, there is no TM system operating from Japanese that does not rely on word/segment/character order to some degree. Tanaka (1997) uses pivotal content words identified by the user to search through the TM and locate TRecs which contain those same content words in the same order and preferably the same segment distance apart. Nakamura (1989) similarly gives preference to TRecs in which the content words contained in the original input occur in the same linear order, although there is the scope to back off to TRecs which do not preserve the original word order. Sumita and Tsutsumi (1991) take the opposite tack in iteratively filtering out NPs and adverbs to leave only functional words and matrix-level predicates, and find TRecs which contain those same key words in the same ordering, preferably with the same segment types between them in the same numbers. Sato and Kawase (1994) employ a more local model of character order in modelling similarity according to N-grams fashioned from the original string.",
"cite_spans": [
{
"start": 137,
"end": 150,
"text": "Tanaka (1997)",
"ref_id": "BIBREF15"
},
{
"start": 348,
"end": 363,
"text": "Nakamura (1989)",
"ref_id": "BIBREF7"
},
{
"start": 587,
"end": 613,
"text": "Sumita and Tsutsumi (1991)",
"ref_id": "BIBREF14"
},
{
"start": 887,
"end": 909,
"text": "Sato and Kawase (1994)",
"ref_id": "BIBREF11"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Segment Order",
"sec_num": "2.2"
},
{
"text": "Given the input \u03b1 1 \u03b1 2 \u03b1 3 \u03b1 4 , we would expect that of \u03b1 1 \u03b2 1 \u03b1 2 \u03b2 2 \u03b1 3 \u03b2 3 \u03b1 4 and \u03b1 1 \u03b1 2 \u03b1 3 \u03b1 4 \u03b2 1 \u03b2 2 \u03b2 3 , the latter would provide a translation more reflective of the translation for the input. This intuition is captured either by embedding some contiguity weighting facility within the string match mechanism (in the case of weighted sequential correspondence -see below), or providing an independent model of segment contiguity in the form of segment N-grams.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Segment contiguity",
"sec_num": "2.3"
},
{
"text": "The particular N-gram orders we test are simple unigrams (1-grams), pure bigrams (2-grams), and mixed unigrams/bigrams. These N-gram models are implemented as a pre-processing stage, following segmentation (where applicable). All this involves is mutating the original strings into Ngrams of the desired order, while preserving the original segment order and segmentation schema. From the Japanese string \u2022 \u2022 [natu\u2022no\u2022ame] \"summer rain\", 1 for example, we would generate the following variants (common to both characterand word-based indexing):",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Segment contiguity",
"sec_num": "2.3"
},
{
"text": "1-gram: \u2022 \u2022 2-gram: \u2022 Mixed 1/2-gram: \u2022 \u2022 \u2022 \u2022",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Segment contiguity",
"sec_num": "2.3"
},
{
"text": "As the starting point for evaluation of the three parameter types targeted in this research, we take two bag-of-words (segment orderoblivious) and three segment order-sensitive methods, thereby modelling the effects of segment order (un)awareness. We then run each method over both segmented and unsegmented data in combination with the various N-gram models proposed above, to capture the full range of parameter settings.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "String Comparison Methods",
"sec_num": "3"
},
{
"text": "The particular bag-of-word approaches we target are the vector space model (Manning and Sch\u00fctze, 1999, p300) and \"token intersection\". For segment order-sensitive approaches, we test 3-operation edit distance and similarity, and also \"weighted sequential correspondence\".",
"cite_spans": [
{
"start": 75,
"end": 108,
"text": "(Manning and Sch\u00fctze, 1999, p300)",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "String Comparison Methods",
"sec_num": "3"
},
{
"text": "All methods are formulated to operate over an arbitrary wt schemata, although in L1 string comparison throughout this paper, we assume that any segment made up entirely of punctuation is given a wt of 0, and any other segment a wt of 1.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "String Comparison Methods",
"sec_num": "3"
},
{
"text": "All methods are subject to a threshold on translation utility, and in the case that the threshold is not achieved, the null string is returned. The various thresholds are as follows:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "String Comparison Methods",
"sec_num": "3"
},
{
"text": "Comparison method Threshold Vector space model 0.5 Token intersection 0.4 3-operation edit distance len(IN ) 3-operation edit similarity 0.4 Weighted seq. correspondence 0.2 where IN is the input string, and len is the conventional segment length operator. Various optimisations were made to each string comparison method to reduce retrieval time, of the type described by Baldwin and Tanaka (2000) . While the details are beyond the scope of this paper, suffice to say that the segment order-sensitive methods benefited from the greatest optimisation, and that little was done to accelerate the already quick bag-of-words methods.",
"cite_spans": [
{
"start": 373,
"end": 398,
"text": "Baldwin and Tanaka (2000)",
"ref_id": "BIBREF0"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "String Comparison Methods",
"sec_num": "3"
},
{
"text": "Within our implementation of the vector space model (VSM), the segment content of each string is described as a vector, made up of a single dimension for each segment type occurring within S or T . The value of each vector component is given as the weighted frequency of that type according to its wt value. The string similarity of S and T is then defined as the cosine of the angle between vectors S and T , respectively, calculated as:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Vector Space Model",
"sec_num": null
},
{
"text": "cos( S, T ) = S \u2022 T | S|| T | = j sj tj j s 2 j j t 2 j",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Vector Space Model",
"sec_num": null
},
{
"text": "The token intersection of S and T is defined as the cumulative intersecting frequency of segment types appearing in each of the strings, normalised according to the combined segment lengths of S and T using Dice's coefficient. Formally, this equates to:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Token Intersection",
"sec_num": null
},
{
"text": "tint (S, T ) = 2 \u00d7 e\u2208S,T min freq S (e), freq T (e) len(S) + len(T )",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Token Intersection",
"sec_num": null
},
{
"text": "where each e is a segment occurring in either S or T , freq S (e) is defined as the wt-based frequency of segment type e occurring in string S, and len(S) is the segment length of string S, that is the wtbased count of segments contained in S (similarly for T ).",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Token Intersection",
"sec_num": null
},
{
"text": "Essentially, the segment-based 3-operation edit distance between strings S and T is the minimum number of primitive edit operations on single segments required to transform S into T (and vice versa). The three edit operations are segment equality (segments s i and t j are identical), segment deletion (delete segment s i ) and segment insertion (insert segment a into a given position in string S). The cost associated with each operation is determined by the wt values of the operand segments, with the exception of segment equality which is defined to have a fixed cost of 0.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "3-op Edit Distance and Similarity",
"sec_num": null
},
{
"text": "Dynamic programming (DP) techniques are used to determine the minimum edit distance between a given string pair, following the classic 4-operation edit distance formulation of Wagner and Fisher (1974) . 2 For 3-operation edit distance, the edit distance between strings S = s 1 s 2 ...s m and",
"cite_spans": [
{
"start": 176,
"end": 200,
"text": "Wagner and Fisher (1974)",
"ref_id": "BIBREF17"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "3-op Edit Distance and Similarity",
"sec_num": null
},
{
"text": "T = t 1 t 2 ...t n is defined as D 3op (S, T ): D3op(S, T ) = d3(m, n) d3(i, j) = \uf8f1 \uf8f4 \uf8f4 \uf8f2 \uf8f4 \uf8f4 \uf8f3 0 if i = 0 \u2227 j = 0 d3(0, j \u2212 1) + wt (tj ) if i = 0 \u2227 j = 0 d3(i \u2212 1, 0) + wt (si) if i = 0 \u2227 j = 0 min d3(i \u2212 1, j) + wt (si), d3(i, j \u2212 1) + wt (tj ), m3(i, j) otherwise m3(i, j) = d3(i \u2212 1, j \u2212 1) if si = sj \u221e otherwise",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "3-op Edit Distance and Similarity",
"sec_num": null
},
{
"text": "It is possible to normalise operation edit distance D 3op into 3-operation edit similarity S 3op by way of:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "3-op Edit Distance and Similarity",
"sec_num": null
},
{
"text": "S3op (S, T ) = 1\u2212 D3op (S, T ) len(S) + len(T )",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "3-op Edit Distance and Similarity",
"sec_num": null
},
{
"text": "Weighted sequential correspondence (originally proposed in Baldwin and Tanaka (2000) ) goes one step further than edit distance in analysing not only segment sequentiality, but also the contiguity of matching segments.",
"cite_spans": [
{
"start": 59,
"end": 84,
"text": "Baldwin and Tanaka (2000)",
"ref_id": "BIBREF0"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Weighted Sequential Correspondence",
"sec_num": null
},
{
"text": "Weighted sequential correspondence associates an incremental weight (orthogonal to our wt weights) with each matching segment assessing the contiguity of left-neighbouring segments, in the manner described by Sato (1992) for characterbased matching. Namely, the kth segment of a matched substring is given the multiplicative weight min(k, Max ), where Max is a positive integer. This weighting up of contiguous matches is facilitated through the DP algorithm given below:",
"cite_spans": [
{
"start": 209,
"end": 220,
"text": "Sato (1992)",
"ref_id": "BIBREF13"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Weighted Sequential Correspondence",
"sec_num": null
},
{
"text": "Sw(S, T ) = s(m, n) s(i, j) = 0 if i = 0 \u2228 j = 0 max s(i \u2212 1, j), s(i, j \u2212 1), s(i \u2212 1, j \u2212 1) + mw(i, j) otherwise mw(i, j) = cm(i, j) \u00d7 wt (i) if si = sj 0 otherwise cm(i, j) = 0 if i = 0 \u2228 j = 0 \u2228 si = tj min(Max , cm(i \u2212 1, j \u2212 1) + 1) otherwise",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Weighted Sequential Correspondence",
"sec_num": null
},
{
"text": "The final similarity is determined as:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Weighted Sequential Correspondence",
"sec_num": null
},
{
"text": "WSC (S, T ) = 2 \u00d7 Sw(S, T ) lenWSC (S) + lenWSC (T )",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Weighted Sequential Correspondence",
"sec_num": null
},
{
"text": "where len WSC (S) is the weighted length of S, defined as:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Weighted Sequential Correspondence",
"sec_num": null
},
{
"text": "lenWSC (S) = m i=1 wt (si) \u00d7 min(Max , i) 4 Evaluation Specifications",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Weighted Sequential Correspondence",
"sec_num": null
},
{
"text": "As our main dataset, we used 3033 unique Japanese-English TRecs extracted from construction machinery field reports for the purposes of this research. Most TRecs comprise a single sentence, with an average Japanese character length of 27.7 and English word length of 13.3. Importantly, our dataset constitutes a controlled language, that is, a given word will tend to be translated identically across all usages, and only a limited range of syntactic constructions are employed.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Details of the Dataset",
"sec_num": "4.1"
},
{
"text": "In secondary evaluation of retrieval performance over differing data sizes, we extracted 61,236 Japanese-English TRecs from the JEIDA parallel corpus (Isahara, 1998) , which is made up of government white papers. The alignment granularity of this second corpus is much coarser than for the first corpus, with a single TRec often extending over multiple sentences. The average Japanese character length of each TRec is 76.3, and the average English word length is 35.7. The language used in the JEIDA corpus is highly constrained, although not as controlled as that in the first corpus.",
"cite_spans": [
{
"start": 150,
"end": 165,
"text": "(Isahara, 1998)",
"ref_id": "BIBREF2"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Details of the Dataset",
"sec_num": "4.1"
},
{
"text": "The construction of TRecs from both corpora was based on existing alignment data, and no further effort was made to subdivide partitions.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Details of the Dataset",
"sec_num": "4.1"
},
{
"text": "For Japanese word-based indexing, segmentation was carried out primarily with ChaSen v2.0 (Matsumoto et al., 1999) , and where specifically mentioned, JUMAN v3.5 (Kurohashi and Nagao, 1998) and ALTJAWS 3 were also used.",
"cite_spans": [
{
"start": 90,
"end": 114,
"text": "(Matsumoto et al., 1999)",
"ref_id": "BIBREF6"
},
{
"start": 162,
"end": 189,
"text": "(Kurohashi and Nagao, 1998)",
"ref_id": "BIBREF4"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Details of the Dataset",
"sec_num": "4.1"
},
{
"text": "Retrieval accuracy was determined by way of 10-fold semi-stratified cross validation over the dataset. As part of this, all Japanese strings of length 5 characters or less were extracted from the dataset, and cross validation was performed over the residue, including the shorter strings in the training data (i.e. TM) on each iteration.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Semi-stratified Cross Validation",
"sec_num": "4.2"
},
{
"text": "In N-fold stratified cross validation, the dataset is divided into N equally-sized partitions of uniform class distribution. Evaluation is then carried out N times, taking each partition as the heldout test data, and the remaining partitions as the training data on each iteration; the overall accuracy is averaged over the N data configurations. As our dataset is not pre-classified according to a discrete class description, we are not able to perform true data stratification over the class distribution. Instead, we carry out \"semi-stratification\" over the L1 segment lengths of the TRecs.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Semi-stratified Cross Validation",
"sec_num": "4.2"
},
{
"text": "Evaluation of retrieval accuracy is carried out according to a modified version of the method proposed by Baldwin and Tanaka (2000) . The first step in this process is to determine the set of \"optimal\" translations by way of the same basic TR procedure as described above, except that we use the held-out translation for each input to search through the L2 component of the TM. As for L1 TR, a threshold on translation utility is then applied to ascertain whether the optimal translations are similar enough to the model translation to be of use, and in the case that this threshold is not achieved, the empty string is returned as the sole optimal translation.",
"cite_spans": [
{
"start": 106,
"end": 131,
"text": "Baldwin and Tanaka (2000)",
"ref_id": "BIBREF0"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Evaluation of the Output",
"sec_num": "4.3"
},
{
"text": "Next, we proceed to ascertain whether the actual system output coincides with one of the optimal translations, and rate the accuracy of each method according to the proportion of optimal outputs. If multiple outputs are produced, we select from among them randomly. This guarantees a unique translation output and differs from the methodology of Baldwin and Tanaka (2000) , who judged the system output to be \"correct\" if the potentially multiple set of top-ranking outputs contained an optimal translation, placing methods with greater fan-out of outputs at an advantage.",
"cite_spans": [
{
"start": 346,
"end": 371,
"text": "Baldwin and Tanaka (2000)",
"ref_id": "BIBREF0"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Evaluation of the Output",
"sec_num": "4.3"
},
{
"text": "So as to filter out any bias towards a given string comparison method in TR, we determine translation optimality based on both 3-operation edit distance (operating over English word bigrams) and also weighted sequential correspondence (operating over English word unigrams). We then derive the final translation accuracy as the average of the accuracies from the respective evaluation sets. Here again, our approach differs from that of Baldwin and Tanaka (2000) , who based determination of translation optimality exclusively on 3-operation edit distance (operating over word unigrams), a method which we found to produce a strong bias toward 3-operation edit distance in L1 TR.",
"cite_spans": [
{
"start": 437,
"end": 462,
"text": "Baldwin and Tanaka (2000)",
"ref_id": "BIBREF0"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Evaluation of the Output",
"sec_num": "4.3"
},
{
"text": "In determining translation optimality, all punctuation and stop words were first filtered out of each L2 (English) string, and all remaining segments scored at a wt of 1. Stop words are defined as those contained within the SMART (Salton, 1971 ) stop word list. 4 Perhaps the main drawback of our approach to evaluation is that we assume a unique model translation for each input, where in fact, multiple translations of equivalent quality could reasonably be expected to exist. In our case, however, both corpora represent relatively controlled languages and language use is hence highly predictable. The proposed evaluation methodology is thus justified.",
"cite_spans": [
{
"start": 230,
"end": 243,
"text": "(Salton, 1971",
"ref_id": "BIBREF10"
},
{
"start": 262,
"end": 263,
"text": "4",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Evaluation of the Output",
"sec_num": "4.3"
},
{
"text": "In this section, we test our five string comparison methods over the construction machinery corpus, under both character-and word-based indexing, and with each of unigrams, bigrams and mixed unigrams/bigrams. The retrieval accuracies and times for the different string comparison methods are presented in Figs. 1 and 2 , respectively. Here and in subsequent graphs, \"VSM\" refers to the vector space model, \"TINT\" to token intersection, \"3opD\" to 3-op edit distance, \"3opS\" to 3-op edit similarity, and \"WSC\" to weighted sequential correspondence; the bag-of-words methods are labelled in italics and the segment ordersensitive methods in bold. In Figs. 1 and 2 , results for the three N-gram models are presented separately, within each of which, the data is sectioned off into the different string comparison methods.",
"cite_spans": [],
"ref_spans": [
{
"start": 305,
"end": 318,
"text": "Figs. 1 and 2",
"ref_id": "FIGREF0"
},
{
"start": 647,
"end": 660,
"text": "Figs. 1 and 2",
"ref_id": "FIGREF0"
}
],
"eq_spans": [],
"section": "Basic evaluation",
"sec_num": "5.1"
},
{
"text": "Weighted sequential correspondence was tested with a unigram model only, due to its inbuilt modelling of segment contiguity. Bars marked with an asterisk indicate a statistically significant 5 gain over the corresponding indexing paradigm (i.e. character-based indexing vs. word-based indexing for a given string comparison method and N-gram order). Times in Fig. 2 are calibrated relative to 3-operation edit distance with word unigrams, and plotted against a logarithmic time axis. Results to come from these figures can be summarised as follows:",
"cite_spans": [],
"ref_spans": [
{
"start": 359,
"end": 365,
"text": "Fig. 2",
"ref_id": "FIGREF1"
}
],
"eq_spans": [],
"section": "Basic evaluation",
"sec_num": "5.1"
},
{
"text": "\u2022 Character-based indexing is consistently superior to word-based indexing, particularly when combined with bigrams or mixed unigrams/bigrams.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Basic evaluation",
"sec_num": "5.1"
},
{
"text": "\u2022 In terms of raw translation accuracy, there is very little to separate the best of the bag-ofwords methods from the best of the segment order-sensitive methods.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Basic evaluation",
"sec_num": "5.1"
},
{
"text": "\u2022 With character-based indexing, bigrams offer tangible gains in translation accuracy at the same time as greatly accelerating the retrieval process. With word-based indexing, mixed unigrams/bigrams offer the best balance of translation accuracy and computational cost.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Basic evaluation",
"sec_num": "5.1"
},
{
"text": "\u2022 Weighted sequential correspondence is moderately successful in terms of accuracy, but grossly expensive.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Basic evaluation",
"sec_num": "5.1"
},
{
"text": "Based on the above results, we judge bigrams to be the best segment contiguity model for character-based indexing, and mixed unigrams/bigrams to be the best segment contiguity 5 As determined by the paired t test (p < 0.05). While we have been able to confirm the finding of Baldwin and Tanaka (2000) that characterbased indexing is superior to word-based indexing, we are no closer to determining why this should be the case. In the following sections we look to shed some light on this issue by considering each of: (i) the retrieval accuracy for other segmentation systems, (ii) the effects of lexical normalisation, and (iii) the scalability and reproducibility of the given results over different datasets. Finally, we present a brief qualitative explanation for the overall results.",
"cite_spans": [
{
"start": 275,
"end": 300,
"text": "Baldwin and Tanaka (2000)",
"ref_id": "BIBREF0"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Basic evaluation",
"sec_num": "5.1"
},
{
"text": "Above, we observed that segmentation consistently brought about a degradation in translation retrieval for the given dataset. Automated segmentation inevitably leads to errors, which could possibly impinge on the accuracy of word-based indexing. Alternatively, the performance drop could simply be caused somehow by our particular choice of segmentation module, that is ChaSen.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "The effects of segmentation and lexical normalisation",
"sec_num": "5.2"
},
{
"text": "First, we used JUMAN to segment the construction machinery corpus, and evaluated the resultant dataset in the exact same manner as for the ChaSen output. Similarly, we ran a development version of ALTJAWS over the same corpus to produce two datasets, the first simply segmented and the second both segmented and lexically normalised. By lexical normalisation, we mean that each word is converted to its canonical form. The main segment types that normalisation has an effect on are verbs and adjectives (conjugating words), and also loan-word nouns with an optional long final vowel (e.g. monit\u0101 \"monitor\" \u21d2 monita) and words with multiple inter-replaceable kanji realisations (e.g.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "The effects of segmentation and lexical normalisation",
"sec_num": "5.2"
},
{
"text": "[zy\u016bbuN] \"sufficient\" \u21d2 ). The retrieval accuracies for JUMAN, and ALT-JAWS with and without lexical normalisation are presented in Fig. 3 , juxtaposed against the retrieval accuracies for character-based indexing (bigrams) and also ChaSen (mixed unigrams/bigrams) from Section 5.1. Asterisked bars Looking first to the results for JUMAN, there is a gain in accuracy over ChaSen for all string comparison methods. With ALTJAWS, also, a consistent gain in performance is evident with simple segmentation, the degree of which is significantly higher than for JUMAN. The addition of lexical normalisation enhances this effect marginally. Notice that character-based indexing (based on character bigrams) holds a clear advantage over the best of the word-based indexing results for all string comparison methods.",
"cite_spans": [],
"ref_spans": [
{
"start": 132,
"end": 138,
"text": "Fig. 3",
"ref_id": "FIGREF2"
}
],
"eq_spans": [],
"section": "The effects of segmentation and lexical normalisation",
"sec_num": "5.2"
},
{
"text": "Based on the above, we can state that the choice of segmentation system does have a modest impact on retrieval accuracy, but that the effects of lexical normalisation are highly localised. In the following, we look to quantify the relationship between retrieval and segmentation accuracy.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "The effects of segmentation and lexical normalisation",
"sec_num": "5.2"
},
{
"text": "In the next step of evaluation, we took a random sample of 200 TRecs from the original dataset, and ran each of ChaSen, JUMAN and ALTJAWS over the Japanese component of each. We then manually evaluated the output in terms of segment precision and recall, defined respectively as: Segment precision = # correct segs in output Total # segs in output Segment recall = # correct segs in output Total # segs in model data One slight complication in evaluating the output of the three systems is that they adopt incongruent models of conjugation. We thus made allowance for variation in the analysis of verb and adjective complexes, and focused on the segmentation of noun complexes.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "The effects of segmentation and lexical normalisation",
"sec_num": "5.2"
},
{
"text": "A performance breakdown for ChaSen (CS), JUMAN (JM) and ALTJAWS (AJ) is presented in Tab. 1. ALTJAWS was found to outperform the remaining two systems in terms of segment precision, while ChaSen and JUMAN performed at the exact same level of segment precision. Looking next to segment recall, ChaSen significantly outperformed both ALTJAWS and JUMAN. The source of almost all errors in recall, and roughly half of errors in precision for both ChaSen and Table 1 : Segmentation performance JUMAN was katakana sequences such as g\u0113torokku-barubu \"gate-lock valve\", transcribed from English. ALTJAWS, on the other hand, was remarkably successful at segmenting katakana word sequences, achieving a segment precision of 100% and segment recall approaching 99%. This is thought to have been the main cause for the disparity in retrieval accuracy for the three systems, aggravated by the fact that most katakana sequences were key technical terms.",
"cite_spans": [],
"ref_spans": [
{
"start": 454,
"end": 461,
"text": "Table 1",
"ref_id": null
}
],
"eq_spans": [],
"section": "The effects of segmentation and lexical normalisation",
"sec_num": "5.2"
},
{
"text": "To gain an insight into consistency in the case of error, we further calculated the total number of segment types in the output, expecting to find a core set of correctly-analysed segments, of relatively constant size across the different systems, plus an unpredictable component of segment errors, of variable size. The system generating the fewest segment types can thus be said to be the most consistent.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "The effects of segmentation and lexical normalisation",
"sec_num": "5.2"
},
{
"text": "Based on the segment type counts in Tab. 1, ALTJAWS errs more consistently than the remaining two systems, and there is very little to separate ChaSen and JUMAN. This is thought to have had some impact on the inflated retrieval accuracy for ALTJAWS.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "The effects of segmentation and lexical normalisation",
"sec_num": "5.2"
},
{
"text": "To summarise, there would seem to be a direct correlation between segmentation accuracy and retrieval performance, with segmentation accuracy on key terms (katakana sequences) having a particularly keen effect on translation retrieval. In this respect, ALTJAWS is superior to both ChaSen and JUMAN for the target domain. Additionally, complementing segmentation with lexical normalisation would seem to produce meager performance gains. Lastly, despite the slight gains to word-based indexing with the different segmentation systems, it is still significantly inferior to character-based indexing.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "The effects of segmentation and lexical normalisation",
"sec_num": "5.2"
},
{
"text": "All results to date have arisen from evaluation over a single dataset of fixed size. In order to validate the basic findings from above and observe how increases in the data size affect retrieval performance, we next ran the string comparison methods over differing-sized subsets of the JEIDA corpus.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Scalability of performance",
"sec_num": "5.3"
},
{
"text": "We simulate TMs of differing size by randomly splitting the JEIDA corpus into ten partitions, and running the various methods first over partition 1, then over the combined partitions 1 and 2, and so on until all ten partitions are combined together into the full corpus. We tested all string comparison methods other than weighted sequential correspondence over the ten subsets of the JEIDA corpus. Weighted sequential correspondence was excluded from evaluation due to its overall sub-standard retrieval performance. The translation accuracies for the different methods over the ten datasets of varying size, are indicated in Fig. 4 , with each string comparison method tested under character bigrams (\"2-gram \u2212seg\") and mixed word unigrams/bigrams (\"1/2-gram +seg\") as above. The results for token intersection have been omitted from the graph due to their being almost identical to those for VSM. A striking feature of the graph is that it is rightdecreasing, which is essentially an artifact of the inflated length of each TRec (see Section 4.1) and resultant data sparseness. That is, for smaller datasets, in the bulk of cases, no TRec in the TM is similar enough to the input to warrant consideration as a translation candidate (i.e. the translation utility threshold is generally not achieved). For larger datasets, on the other hand, we are having to make more subtle choices as to the final translation candidate.",
"cite_spans": [],
"ref_spans": [
{
"start": 628,
"end": 634,
"text": "Fig. 4",
"ref_id": "FIGREF4"
}
],
"eq_spans": [],
"section": "Scalability of performance",
"sec_num": "5.3"
},
{
"text": "One key trend in Fig. 4 is the superiority of character-over word-based indexing for each of the three string comparison methods, at a relatively constant level as the TM size grows. Also of interest is the finding that there is very little to distinguish bag-of-words from segment ordersensitive methods in terms of retrieval accuracy in their respective best configurations.",
"cite_spans": [],
"ref_spans": [
{
"start": 17,
"end": 23,
"text": "Fig. 4",
"ref_id": "FIGREF4"
}
],
"eq_spans": [],
"section": "Scalability of performance",
"sec_num": "5.3"
},
{
"text": "As with the original dataset from above, 3operation edit similarity was the strongest performer just nosing out (character bigram-based) VSM for line honours, with 3-operation edit distance lagging well behind.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Scalability of performance",
"sec_num": "5.3"
},
{
"text": "Next, we turn to consider the mean unit retrieval times for each method, under the two indexing paradigms. Times are presented in Fig. 5 , plotted once again on a logarithmic scale in order to fit the full fan-out of retrieval times onto a single graph. VSM and 3-operation edit distance were the most consistent performers, both maintaining retrieval speeds in line with those for the original dataset at around or under 1.0 (i.e. the same retrieval time per input as 3-operation edit distance run over word unigrams for the construction machinery dataset). Most importantly, only minor increases in retrieval speed were evident as the TM size increased, which were then reversed for the larger datasets. All three string comparison methods displayed this convex shape, although the final running time for 3-operation edit similarity under character-and word-based indexing was, respectively, around 10 and 100 times slower than that for VSM or 3-operation edit distance over the same dataset.",
"cite_spans": [],
"ref_spans": [
{
"start": 130,
"end": 136,
"text": "Fig. 5",
"ref_id": "FIGREF6"
}
],
"eq_spans": [],
"section": "Scalability of performance",
"sec_num": "5.3"
},
{
"text": "To combine the findings for accuracy and speed, VSM under character-based indexing suggests itself as the pick of the different system configurations, combining both speed and consistent accuracy. That is, it offers the best overall retrieval performance.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Scalability of performance",
"sec_num": "5.3"
},
{
"text": "Above, we established that character-based indexing is superior to word-based indexing for distinct datasets and a range of segmentation modules, even when segmentation is coupled with lexical normalisation. Additionally, we provided evidence to the effect that bag-of-words methods offer superior translation retrieval performance to segment order-sensitive methods. We are still no closer, however, to determining why this should be the case. Here, we seek to provide an explanation for these intriguing results.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Qualitative evaluation",
"sec_num": "5.4"
},
{
"text": "First comparing character-and word-based indexing, we found that the disparity in retrieval accuracy was largely related to the scoring of katakana words, which are significantly longer in character length than native Japanese words. For the construction machinery dataset as analysed with ChaSen, for example, the average character length of katakana words is 3.62, as compared to 2.05 overall. Under word-based indexing, all words are treated equally and character length does not enter into calculations. Thus a katakana word is treated identically to any other word type. Under character-based indexing, on the other hand, the longer the word, the more segments it generates, and a single matching katakana sequence thus tends to contribute more heavily to the final score than other words. Effectively, therefore, katakana sequences receive a higher score than kanji and other sequences, producing a preference for TRecs which incorporate the same katakana sequences as the input. As noted above, katakana sequences generally represent key technical terms, and such weighting thus tends to be beneficial to retrieval accuracy.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Qualitative evaluation",
"sec_num": "5.4"
},
{
"text": "We next examine the reason for the high correlation in retrieval accuracy between bag-of-words and segment order-sensitive methods in their op-timum configurations (i.e. when coupled with character bigrams). Essentially, the probability of a given segment set permuting in different string contexts diminishes as the number of co-occurring segments decreases. That is, for a given string pair, the greater the segment overlap between them (relative to the overall string lengths), the lower the probability that those segments are going to occur in different orderings. This is particularly the case when local segment contiguity is modelled within the segment description, as occurs for the character bigram and mixed word uni/bigram models. For high-scoring matches, therefore, segment order sensitivity becomes largely superfluous, and the slight edge in retrieval accuracy for segment order-sensitive methods tends to come for mid-scoring matches, in the vicinity of the translation utility threshold.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Qualitative evaluation",
"sec_num": "5.4"
},
{
"text": "This research has been concerned with the relative import of segmentation, segment order and segment contiguity on translation retrieval performance. We simulated the effects of word order sensitivity vs. bag-of-words word order insensitivity by implementing a total of five comparison methods: two bag-of-words approaches and three word order-sensitive approaches. Each of these methods was then tested under characterbased and word-based indexing and in combination with a range of N-gram models, and the relative performance of each such system configuration evaluated. Character-based indexing was found to be superior to word-based indexing, particularly when supplemented with a character bigram model.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Conclusion",
"sec_num": "6"
},
{
"text": "We went on to discover a strong correlation between retrieval accuracy and segmentation accuracy/consistency, and that lexical normalisation produces marginal gains in retrieval performance. We further tested the effects of incremental increases in data on retrieval performance, and confirmed our earlier finding that character-based indexing is superior to word-based indexing. At the same time, we discovered that in their best configurations, the retrieval accuracies of our bag-ofwords and segment order sensitive string comparison methods are roughly equivalent, but that the computational overhead for bag-of-words methods to achieve that accuracy is considerably lower than that for segment order sensitive methods.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Conclusion",
"sec_num": "6"
},
{
"text": "Character boundaries (which double as word boundaries in this case) indicated by \"\u2022\".",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
},
{
"text": "The fourth operator in 4-operation edit distance is segment substitution.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
},
{
"text": "http://www.kecl.ntt.co.jp/icl/mtg/resources/altjaws.html",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
},
{
"text": "ftp://ftp.cornell.cs.edu/pub/smart/english.stop",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
}
],
"back_matter": [],
"bib_entries": {
"BIBREF0": {
"ref_id": "b0",
"title": "The effects of word order and segmentation on translation retrieval performance",
"authors": [
{
"first": "T",
"middle": [],
"last": "Baldwin",
"suffix": ""
},
{
"first": "H",
"middle": [],
"last": "Tanaka",
"suffix": ""
}
],
"year": 2000,
"venue": "Proc. of the 18th International Conference on Computational Linguistics (COLING 2000)",
"volume": "",
"issue": "",
"pages": "35--41",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "T. Baldwin and H. Tanaka. 2000. The effects of word order and segmentation on translation re- trieval performance. In Proc. of the 18th Inter- national Conference on Computational Linguistics (COLING 2000), pages 35-41.",
"links": null
},
"BIBREF1": {
"ref_id": "b1",
"title": "A comparison of indexing techniques for Japanese text retrieval",
"authors": [
{
"first": "H",
"middle": [],
"last": "Fujii",
"suffix": ""
},
{
"first": "W",
"middle": [
"B"
],
"last": "Croft",
"suffix": ""
}
],
"year": 1993,
"venue": "Proc. of 16th International ACM-SIGIR Conference on Research and Development in Information Retrieval (SIGIR'93)",
"volume": "",
"issue": "",
"pages": "237--283",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "H. Fujii and W.B. Croft. 1993. A comparison of index- ing techniques for Japanese text retrieval. In Proc. of 16th International ACM-SIGIR Conference on Research and Development in Information Retrieval (SIGIR'93), pages 237-46.",
"links": null
},
"BIBREF2": {
"ref_id": "b2",
"title": "JEIDA's English-Japanese bilingual corpus project",
"authors": [
{
"first": "H",
"middle": [],
"last": "Isahara",
"suffix": ""
}
],
"year": 1998,
"venue": "Proc. of the 1st International Conference on Language Resources and Evaluation (LREC'98)",
"volume": "",
"issue": "",
"pages": "471--81",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "H. Isahara. 1998. JEIDA's English-Japanese bilin- gual corpus project. In Proc. of the 1st Interna- tional Conference on Language Resources and Eval- uation (LREC'98), pages 471-81.",
"links": null
},
"BIBREF3": {
"ref_id": "b3",
"title": "Translation retrieval system using alignment data from parallel texts",
"authors": [
{
"first": "E",
"middle": [],
"last": "Kitamura",
"suffix": ""
},
{
"first": "H",
"middle": [],
"last": "Yamamoto",
"suffix": ""
}
],
"year": 1996,
"venue": "Proc. of the 53rd Annual Meeting of the IPSJ",
"volume": "2",
"issue": "",
"pages": "385--391",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "E. Kitamura and H. Yamamoto. 1996. Translation retrieval system using alignment data from parallel texts. In Proc. of the 53rd Annual Meeting of the IPSJ, volume 2, pages 385-6. (In Japanese).",
"links": null
},
"BIBREF4": {
"ref_id": "b4",
"title": "Nihongo keitaikaiseki sisutemu JUMAN",
"authors": [
{
"first": "S",
"middle": [],
"last": "Kurohashi",
"suffix": ""
},
{
"first": "M",
"middle": [],
"last": "Nagao",
"suffix": ""
}
],
"year": 1998,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "S. Kurohashi and M. Nagao. 1998. Nihongo keitai- kaiseki sisutemu JUMAN [Japanese morphological analysis system JUMAN] version 3.5. Technical re- port, Kyoto University. (In Japanese).",
"links": null
},
"BIBREF5": {
"ref_id": "b5",
"title": "Foundations of Statistical Natural Language Processing",
"authors": [
{
"first": "C",
"middle": [],
"last": "Manning",
"suffix": ""
},
{
"first": "H",
"middle": [],
"last": "Sch\u00fctze",
"suffix": ""
}
],
"year": 1999,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "C. Manning and H. Sch\u00fctze. 1999. Foundations of Statistical Natural Language Processing. MIT Press.",
"links": null
},
"BIBREF6": {
"ref_id": "b6",
"title": "Japanese Morphological Analysis System ChaSen Version 2.0 Manual",
"authors": [
{
"first": "Y",
"middle": [],
"last": "Matsumoto",
"suffix": ""
},
{
"first": "A",
"middle": [],
"last": "Kitauchi",
"suffix": ""
},
{
"first": "T",
"middle": [],
"last": "Yamashita",
"suffix": ""
},
{
"first": "Y",
"middle": [],
"last": "Hirano",
"suffix": ""
}
],
"year": 1999,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Y. Matsumoto, A. Kitauchi, T. Yamashita, and Y. Hi- rano. 1999. Japanese Morphological Analysis Sys- tem ChaSen Version 2.0 Manual. Technical Report NAIST-IS-TR99009, NAIST.",
"links": null
},
"BIBREF7": {
"ref_id": "b7",
"title": "Translation support by retrieving bilingual texts",
"authors": [
{
"first": "N",
"middle": [],
"last": "Nakamura",
"suffix": ""
}
],
"year": 1989,
"venue": "Proc. of the 38th Annual Meeting of the IPSJ",
"volume": "1",
"issue": "",
"pages": "357--365",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "N. Nakamura. 1989. Translation support by retrieving bilingual texts. In Proc. of the 38th Annual Meeting of the IPSJ, volume 1, pages 357-8. (In Japanese).",
"links": null
},
"BIBREF8": {
"ref_id": "b8",
"title": "Two approaches to matching in example-based machine translation",
"authors": [
{
"first": "S",
"middle": [],
"last": "Nirenburg",
"suffix": ""
},
{
"first": "C",
"middle": [],
"last": "Domashnev",
"suffix": ""
},
{
"first": "D",
"middle": [
"J"
],
"last": "Grannes",
"suffix": ""
}
],
"year": 1993,
"venue": "Proc. of the 5th International Conference on Theoretical and Methodological Issues in Machine Translation (TMI-93)",
"volume": "",
"issue": "",
"pages": "47--57",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "S. Nirenburg, C. Domashnev, and D.J. Grannes. 1993. Two approaches to matching in example-based ma- chine translation. In Proc. of the 5th International Conference on Theoretical and Methodological Is- sues in Machine Translation (TMI-93), pages 47- 57.",
"links": null
},
"BIBREF9": {
"ref_id": "b9",
"title": "A Case Study on Memory Based Machine Translation Tools",
"authors": [
{
"first": "E",
"middle": [],
"last": "Planas",
"suffix": ""
}
],
"year": 1998,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "E. Planas. 1998. A Case Study on Memory Based Machine Translation Tools. PhD Fellow Working Paper, United Nations University.",
"links": null
},
"BIBREF10": {
"ref_id": "b10",
"title": "The SMART Retrieval System: Experiments in Automatic Document Processing",
"authors": [
{
"first": "G",
"middle": [],
"last": "Salton",
"suffix": ""
}
],
"year": 1971,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "G. Salton. 1971. The SMART Retrieval System: Experiments in Automatic Document Processing. Prentice-Hall.",
"links": null
},
"BIBREF11": {
"ref_id": "b11",
"title": "A High-Speed Best Match Retrieval Method for Japanese Text",
"authors": [
{
"first": "S",
"middle": [],
"last": "Sato",
"suffix": ""
},
{
"first": "T",
"middle": [],
"last": "Kawase",
"suffix": ""
}
],
"year": 1994,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "S. Sato and T. Kawase. 1994. A High-Speed Best Match Retrieval Method for Japanese Text. Techni- cal Report IS-RR-94-9I, JAIST.",
"links": null
},
"BIBREF12": {
"ref_id": "b12",
"title": "Toward memorybased translation",
"authors": [
{
"first": "S",
"middle": [],
"last": "Sato",
"suffix": ""
},
{
"first": "M",
"middle": [],
"last": "Nagao",
"suffix": ""
}
],
"year": 1990,
"venue": "Proc. of the 13th International Conference on Computational Linguistics (COL-ING '90)",
"volume": "",
"issue": "",
"pages": "247--52",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "S. Sato and M. Nagao. 1990. Toward memory- based translation. In Proc. of the 13th International Conference on Computational Linguistics (COL- ING '90), pages 247-52.",
"links": null
},
"BIBREF13": {
"ref_id": "b13",
"title": "CTM: An example-based translation aid system",
"authors": [
{
"first": "S",
"middle": [],
"last": "Sato",
"suffix": ""
}
],
"year": 1992,
"venue": "Proc. of the 14th International Conference on Computational Linguistics (COL-ING '92)",
"volume": "",
"issue": "",
"pages": "1259--63",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "S. Sato. 1992. CTM: An example-based transla- tion aid system. In Proc. of the 14th International Conference on Computational Linguistics (COL- ING '92), pages 1259-63.",
"links": null
},
"BIBREF14": {
"ref_id": "b14",
"title": "A practical method of retrieving similar examples for translation aid",
"authors": [
{
"first": "E",
"middle": [],
"last": "Sumita",
"suffix": ""
},
{
"first": "Y",
"middle": [],
"last": "Tsutsumi",
"suffix": ""
}
],
"year": 1991,
"venue": "Transactions of the IEICE",
"volume": "",
"issue": "10",
"pages": "1437--1484",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "E. Sumita and Y. Tsutsumi. 1991. A practical method of retrieving similar examples for translation aid. Transactions of the IEICE, J74-D-II(10):1437-47. (In Japanese).",
"links": null
},
"BIBREF15": {
"ref_id": "b15",
"title": "An efficient way of gauging similarity between long Japanese expressions",
"authors": [
{
"first": "H",
"middle": [],
"last": "Tanaka",
"suffix": ""
}
],
"year": 1997,
"venue": "Information Processing Society of Japan SIG Notes",
"volume": "97",
"issue": "",
"pages": "69--74",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "H. Tanaka. 1997. An efficient way of gauging similar- ity between long Japanese expressions. In Informa- tion Processing Society of Japan SIG Notes, volume 97, no. 85, pages 69-74. (In Japanese).",
"links": null
},
"BIBREF16": {
"ref_id": "b16",
"title": "Translation Engines: Techniques for Machine Translation",
"authors": [
{
"first": "A",
"middle": [],
"last": "Trujillo",
"suffix": ""
}
],
"year": 1999,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "A. Trujillo. 1999. Translation Engines: Techniques for Machine Translation. Springer Verlag.",
"links": null
},
"BIBREF17": {
"ref_id": "b17",
"title": "The string-tostring correction problem",
"authors": [
{
"first": "A",
"middle": [],
"last": "Wagner",
"suffix": ""
},
{
"first": "M",
"middle": [],
"last": "Fisher",
"suffix": ""
}
],
"year": 1974,
"venue": "Journal of the ACM",
"volume": "21",
"issue": "1",
"pages": "168--73",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "A. Wagner and M. Fisher. 1974. The string-to- string correction problem. Journal of the ACM, 21(1):168-73.",
"links": null
}
},
"ref_entries": {
"FIGREF0": {
"uris": null,
"text": "Basic retrieval accuracies",
"num": null,
"type_str": "figure"
},
"FIGREF1": {
"uris": null,
"text": "Basic unit retrieval times model for word-based indexing, and for the remainder of this paper, present only these two sets of results.",
"num": null,
"type_str": "figure"
},
"FIGREF2": {
"uris": null,
"text": "Results using different segmentation modules indicate a statistically significant gain in accuracy over ChaSen.",
"num": null,
"type_str": "figure"
},
"FIGREF4": {
"uris": null,
"text": "Retrieval accuracies over datasets of increasing size",
"num": null,
"type_str": "figure"
},
"FIGREF6": {
"uris": null,
"text": "Relative unit retrieval times over datasets of increasing size",
"num": null,
"type_str": "figure"
}
}
}
} |