File size: 66,043 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 | {
"paper_id": "P01-1041",
"header": {
"generated_with": "S2ORC 1.0.0",
"date_generated": "2023-01-19T09:30:03.226642Z"
},
"title": "Japanese Named Entity Recognition based on a Simple Rule Generator and Decision Tree Learning",
"authors": [
{
"first": "Hideki",
"middle": [],
"last": "Isozaki",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "NTT Communication Science Laboratories",
"location": {
"addrLine": "2-4 Hikaridai, Seika-cho, Souraku-gun",
"postCode": "619-0237",
"settlement": "Kyoto",
"country": "Japan"
}
},
"email": "isozaki@cslab.kecl.ntt.co.jp"
}
],
"year": "",
"venue": null,
"identifiers": {},
"abstract": "Named entity (NE) recognition is a task in which proper nouns and numerical information in a document are detected and classified into categories such as person, organization, location, and date. NE recognition plays an essential role in information extraction systems and question answering systems. It is well known that hand-crafted systems with a large set of heuristic rules are difficult to maintain, and corpus-based statistical approaches are expected to be more robust and require less human intervention. Several statistical approaches have been reported in the literature. In a recent Japanese NE workshop, a maximum entropy (ME) system outperformed decision tree systems and most hand-crafted systems. Here, we propose an alternative method based on a simple rule generator and decision tree learning. Our experiments show that its performance is comparable to the ME approach. We also found that it can be trained more efficiently with a large set of training data and that it improves readability.",
"pdf_parse": {
"paper_id": "P01-1041",
"_pdf_hash": "",
"abstract": [
{
"text": "Named entity (NE) recognition is a task in which proper nouns and numerical information in a document are detected and classified into categories such as person, organization, location, and date. NE recognition plays an essential role in information extraction systems and question answering systems. It is well known that hand-crafted systems with a large set of heuristic rules are difficult to maintain, and corpus-based statistical approaches are expected to be more robust and require less human intervention. Several statistical approaches have been reported in the literature. In a recent Japanese NE workshop, a maximum entropy (ME) system outperformed decision tree systems and most hand-crafted systems. Here, we propose an alternative method based on a simple rule generator and decision tree learning. Our experiments show that its performance is comparable to the ME approach. We also found that it can be trained more efficiently with a large set of training data and that it improves readability.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Abstract",
"sec_num": null
}
],
"body_text": [
{
"text": "Named entity (NE) recognition is a task in which proper nouns and numerical information in a document are detected and classi-fied into categories such as person, organization, location, and date. NE recognition plays an essential role in information extraction systems (see MUC documents (1996) ) and question answering systems (see TREC-QA documents, http://trec.nist.gov/). When you want to know the location of the Taj Mahal, traditional IR techniques direct you to relevant documents but do not directly answer your question. NE recognition is essential for finding possible answers from documents. Although it is easy to build an NE recognition system with mediocre performance, it is difficult to make it reliable because of the large number of ambiguous cases. For instance, we cannot determine whether \"Washington\" is a person's name or a location's name without the necessary context.",
"cite_spans": [
{
"start": 289,
"end": 295,
"text": "(1996)",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "There are two major approaches to building NE recognition systems. The first approach employs hand-crafted rules. It is well known that handcrafted systems are difficult to maintain because it is not easy to predict the effect of a small change in a rule. The second approach employs a statistical method, which is expected to be more robust and to require less human intervention. Several statistical methods have been reported in the literature (Bikel et al., 1999; Borthwick, 1999; Sekine et al., 1998; .",
"cite_spans": [
{
"start": 447,
"end": 467,
"text": "(Bikel et al., 1999;",
"ref_id": "BIBREF2"
},
{
"start": 468,
"end": 484,
"text": "Borthwick, 1999;",
"ref_id": "BIBREF3"
},
{
"start": 485,
"end": 505,
"text": "Sekine et al., 1998;",
"ref_id": "BIBREF15"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "IREX (Information Retrieval and Extraction Exercise, (Sekine and Eriguchi, 2000; IRE, 1999) ) was held in 1999, and fifteen systems participated in the formal run of the Japanese NE excercise. In the formal run, participants were requested to tag two data sets (GENERAL and AR-REST), and their scores were compared in terms of F-measure, i.e., the harmonic mean of 'recall' and 'precision' defined as follows. recall = x/(the number of correct NEs) precision = x/(the number of NEs extracted by the system) where x is the number of NEs correctly extracted and classified by the system. GENERAL was the larger test set, and its best system was a hand-crafted one that attained F=83.86%. The second best system (F=80.05%) was also hand-crafted but enhanced with transformation-based error-driven learning. The third best system (F=77.37%) was Borthwick's ME system enhanced with hand-crafted rules and dictionaries (1999) . Thus, the best three systems used quite different approaches.",
"cite_spans": [
{
"start": 53,
"end": 80,
"text": "(Sekine and Eriguchi, 2000;",
"ref_id": "BIBREF14"
},
{
"start": 81,
"end": 91,
"text": "IRE, 1999)",
"ref_id": null
},
{
"start": 913,
"end": 919,
"text": "(1999)",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "In this paper, we propose an alternative approach based on a simple rule generator and decision tree learning (RG+DT). Our experiments show that its performance is comparable to the ME method, and we found that it can be trained more efficiently with a large set of training data. By adding in-house data, the proposed system's performance was improved by several points, while a standard ME toolkit crashed.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "When we try to extract NEs in Japanese, we encounter several problems that are not serious in English. It is relatively easy to detect English NEs because of capitalization. In Japanese, there is no such useful hint. Proper nouns and common nouns look very similar. In English, it is also easy to tokenize a sentence because of inter-word spacing. In Japanese, inter-word spacing is rarely used. We can use an off-the-shelf morphological analyzer for tokenization, but its word boundaries may differ from the corresponding NE boundaries in the training data. For instance, a morphological analyzer may divide a four-character expression OO-SAKA-SHI-NAI into two words OO-SAKA (= Osaka) and SHI-NAI (= in the city), but the training data would be tagged as <LOCATION>OO-SAKA-SHI</LO-CATION>NAI (= in <LOCATION>Osaka City </LOCATION>). Moreover, unknown words are often divided excessively or incorrectly because an analyzer tries to interpret a sentence as a sequence of known words. Throughout this paper, the typewriter-style font is used for Japanese, and hyphens indicate character boundaries. Different types of characters are used in Japanese: hiragana, katakana, kanji, symbols, numbers, and letters of the Roman alphabet. We use 17 character types for words, e.g., single-kanji, all-kanji, all-katakana, all-uppercase, float (for floating point numbers), small-integer (up to 4 digits).",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1"
},
{
"text": "Our RG+DT system (Fig. 1) generates a recognition rule from each NE in the training data. Then, the rule is refined by decision tree learning. By applying the refined recognition rules to a new document, we get NE candidates. Then, nonoverlapping candidates are selected by a kind of longest match method.",
"cite_spans": [],
"ref_spans": [
{
"start": 17,
"end": 25,
"text": "(Fig. 1)",
"ref_id": "FIGREF0"
}
],
"eq_spans": [],
"section": "Methodology",
"sec_num": "2"
},
{
"text": "In our method, each tokenized NE is converted to a recognition rule that is essentially a sequence of part-of-speech (POS) tags in the NE. For instance, OO-SAKA-GIN-KOU (= Osaka Bank) is tokenized into two words: OO-SAKA:allkanji:location-name (= Osaka) and GIN-KOU:all-kanji:common-noun (= Bank), where location-name and common-noun are POS tags. In this case, we get the following recognition rule. Here, '*' matches anything. *:*:location-name, *:*:common-noun -> ORGANIZATION However, this rule is not very good. For instance, OO-SAKA-WAN (= Osaka Bay) follows this pattern, but it is a location's name. GIN-KOU and WAN strongly imply ORGANIZATION and LOCATION, respectively. Thus, the last word of an NE is often a head that is more useful than other words for the classification. Therefore, we register the last word into a suffix dictionary for each non-numerical NE class (i.e., ORGANIZA-TION, PERSON, LOCATION, and ARTIFACT) in order to accept only reliable candidates. If the last word appears in two or more different NE, we call it a reliable NE suffix. We register only reliable ones. In the above examples, the last words were common nouns. However, the last word can also be a proper noun. For instance, we will get the following rule from <ORGANIZATION>OO-SAKA-TO-YO-TA</ORGANIZATION> (= Osaka Toyota) because Japanese POS taggers know that TO-YO-TA is an organization name (a kind of proper noun). *:*:location-name, *:*:org-name -> ORGANIZATION,0,0",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Generation of recognition rules",
"sec_num": "2.1"
},
{
"text": "Since Yokohama Honda and Kyoto Sony also follow this pattern, the second element *:*:org-name should not be restricted to the words in the training data. Therefore, we do not restrict proper nouns by a suffix dictionary, and we do not restrict numbers either.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Generation of recognition rules",
"sec_num": "2.1"
},
{
"text": "In addition, the first or last word of an NE may contain an NE boundary as we described before (SHI</LOCATION>NAI). In this case, we can get OO-SAKA-SHI by removing no character of the first word OO-SAKA and one character of the last word SHI-NAI. Accordingly, this modification can be represented by two integers: 0,1.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Generation of recognition rules",
"sec_num": "2.1"
},
{
"text": "Furthermore, one-word NEs are different from other NEs in the following respects.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Generation of recognition rules",
"sec_num": "2.1"
},
{
"text": "The word is usually a proper noun, an unknown word, or a number; otherwise, it is an exceptional case.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Generation of recognition rules",
"sec_num": "2.1"
},
{
"text": "The character type of a one-word NE gives a useful hint for its classification. For instance, all-uppercase words (e.g., IOC) are often classified as ORGANIZATION.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Generation of recognition rules",
"sec_num": "2.1"
},
{
"text": "Since unknown words are often proper nouns, we assume they are tagged as misc-proper-noun.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Generation of recognition rules",
"sec_num": "2.1"
},
{
"text": "If the training data contains <ORGANIZATION>I-O-C</ORGANIZATION> and I-O-C (= IOC) is an unknown word, we will get I-O-C:alluppercase:misc-proper-noun.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Generation of recognition rules",
"sec_num": "2.1"
},
{
"text": "By considering these facts, we modify the above rule generation. That is, we replace every word in an NE and its character type by '*' to get the left-hand side of the corresponding recognition rule except the following cases.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Generation of recognition rules",
"sec_num": "2.1"
},
{
"text": "A word that contains an NE boundary If the first or last word of the NE contains an NE boundary (e.g, SHI</LOCATION>NAI), the word is not replaced by '*'. The number of characters to be deleted is also recorded in the right-hand side of the recognition rule.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Generation of recognition rules",
"sec_num": "2.1"
},
{
"text": "One-word NE The following exceptions are applied to one-word NEs. If the word is a proper noun or a number, its character type is not replaced by '*'. Otherwise, the word is not replaced by '*'.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Generation of recognition rules",
"sec_num": "2.1"
},
{
"text": "The last word of a longer NE The following exceptions are applied to the last word of a non-numerical NE that is composed of two or more words when the word is neither a proper noun nor a number. If the last word is a reliable NE suffix (i.e., it appears in two or more different NEs in the class), its information (i.e., the last word, its character type, and its POS tag) is registered into a suffix dictionary for the NE class. The last word of the recognition rule must be an element of the suffix dictionary. Unreliable NE suffixes are not replaced by '*'. Suffixes of numerical NEs (i.e., DATE, TIME, MONEY, PERCENT) are not replaced, either. Now, we obtain the following recognition rules from the above examples.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Generation of recognition rules",
"sec_num": "2.1"
},
{
"text": "*:all-uppercase:misc-proper-noun -> ORGANIZATION,0,0. *:*:location-name, SHI-NAI:*:common-noun -> LOCATION,0,1. *:*:location-name, *:*:common-noun -> ORGANIZATION,0,0.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Generation of recognition rules",
"sec_num": "2.1"
},
{
"text": "The first rule extracts CNN as an organization. The second rule extracts YOKO-HAMA-SHI (= Yokohama City) from YOKO-HAMA-SHI-NAI (= in Yokohama City). The third rule extracts YOKO-HAMA-GIN-KOU (= Yokohama Bank) as an organization. Note that, in this rule, the second element (*:*:common-noun) is constrained by the suffix dictionary for ORGANIZATION because it is neither a proper noun nor a number. Hence, the rule does not match YOKO-HAMA-WAN (= Yokohama Bay). If the suffix dictionary also happens to have KOU-KOU:all-kanji: commmon-noun (= senior high school), the rule also matches YOKO-HAMA-KOU-KOU (= Yokohama Senior High School).",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Generation of recognition rules",
"sec_num": "2.1"
},
{
"text": "IREX introduced <ARTIFACT> for product names, prizes, pacts, books, and fine arts, among other nouns. Titles of books and fine arts are often long and have atypical word patterns. However, they are often delimited by a pair of symbols that correspond to quotation marks in English. Some atypical organization names are also delimited by these symbols. In order to extract such a long NE, we concatenate all words within a pair of such symbols into one word. We employ the first and last word of the quoted words as extra features. In addition, we do not regard the quotation symbols as adjacent words because they are constant and lack semantic meaning.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Generation of recognition rules",
"sec_num": "2.1"
},
{
"text": "When a large amount of training data is given, thousands of recognition rules are generated. For efficiency, we compile these recognition rules by using a hash table that converts a hash key into a list of relevant rules that have to be examined. We make this hash table as follows. If the lefthand side of a rule contains only one element, the element is used as a hash key and its rule identifier is appended to the corresponding rule list. If the left-hand side contains two or more elements, the first two elements are concatenated and used as a hash key and its rule identifier is appended to the corresponding rule list. After this compilation, we can efficiently apply all of the rules to a new document. By taking the first two elements into consideration, we can reduce the number of rules that need to be examined.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Generation of recognition rules",
"sec_num": "2.1"
},
{
"text": "Some recognition rules are not reliable. For instance, we get the following rule when a person's name is incorrectly tagged as a location's name by a POS tagger.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Refinement of recognition rules",
"sec_num": "2.2"
},
{
"text": "*:all-kanji:location-name -> PERSON,0,0 Therefore, we have to consider a way to refine the recognition rules. By applying each recognition rule to the untagged training data, we can obtain NE candidates for the rule. By comparing the candidates with the given answer for the training data, we can classify them into positive examples and negative examples for the recognition rule. Consequently, we can apply decision tree learning to classify these examples correctly. We represent each example by a list of features: words in the NEs, \u00a1 preceding words, \u00a2 succeeding words, their character types, and their POS tags. If we consider one preceding word and two succeeding words, the feature list for a two-word named entity (\u00a3 is a boolean value that indicates whether it is a positive example. If a feature value appears less than three times in the examples, it is replaced by a dummy constant. We also replace numbers by dummy constants because most numerical NEs follow typical patterns, and their specific values are often useless for NE recognition.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Refinement of recognition rules",
"sec_num": "2.2"
},
{
"text": "EQUATION",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [
{
"start": 0,
"end": 8,
"text": "EQUATION",
"ref_id": "EQREF",
"raw_str": "\u00a5 \u00a4 \u00a7 \u00a6 \u00a3 \u00a9 ) will be \u00a3 \u00a9 , \u00a9 , \u00a9 , \u00a3 \u00a4 , \u00a4 , \u00a4 , \u00a3 \u00a9 , \u00a9 , \u00a9 , \u00a3 , ,",
"eq_num": ", \u00a3 , , , \" ! ,"
}
],
"section": "Refinement of recognition rules",
"sec_num": "2.2"
},
{
"text": "Here, we discuss handling short NEs. For example, NO-O-BE-RU-SHOU-SEN-KOU-I-IN-KAI (= the Nobel Prize Selection Committee) is an organization's name that contains a person's name NO-O-BE-RU (= Nobel) and an artifact name NO-O-BE-RU-SHOU (= Nobel Prize), but <PERSON>NO-O-BE-RU</PER-SON> and <ARTIFACT>NO-O-BE-RU-SHOU </ARTIFACT> are incorrect in this case. If the training data contain NO-O-BE-RU as both positive and negative examples of a person's name, the decision tree learner will be confused. They are rejected because there is a longer named entity and overlapping tags are not allowed. We do not have to change our knowledge that Nobel is a person's name. Therefore, we remove such negative examples caused by longer NEs. Consequently, the decision tree may fail to reject <PERSON> NO-O-BE-RU</PERSON>, but it will disappear in the final output because we use a longest match method for arbitration.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Refinement of recognition rules",
"sec_num": "2.2"
},
{
"text": "For readability, we translate each decision tree into a set of production rules by c4.5rules (Quinlan, 1993) . Throughout this paper, we call them dt-rules (Fig. 1 ) in order to distinguish them from recognition rules. Thus, each recognition rule is enhanced by a set of dt-rules. The dt-rules removes unlikely candidates.",
"cite_spans": [
{
"start": 93,
"end": 108,
"text": "(Quinlan, 1993)",
"ref_id": "BIBREF11"
}
],
"ref_spans": [
{
"start": 156,
"end": 163,
"text": "(Fig. 1",
"ref_id": "FIGREF0"
}
],
"eq_spans": [],
"section": "Refinement of recognition rules",
"sec_num": "2.2"
},
{
"text": "Once the refined rules are generated, we can apply them to a new document. This obtains a large number of NE candidates (Fig. 1) . Since overlapping tags are not allowed, we use a kind of leftto-right longest match method. First, we compare their starting points and select the earliest ones. If two or more candidates start at the same point, their ending points are compared and the longest candidate is selected. Therefore, the candidates overlapping the selected candidate are removed from the candidate set. This procedure is repeated until the candidate set becomes empty.",
"cite_spans": [],
"ref_spans": [
{
"start": 120,
"end": 128,
"text": "(Fig. 1)",
"ref_id": "FIGREF0"
}
],
"eq_spans": [],
"section": "Arbitration of candidates",
"sec_num": "2.3"
},
{
"text": "The rank of a candidate starting at the ( th word boundary and ending at the ) -th word boundary can be represented by a pair 0 1 ( 2 \u00a6 3 4 ) 6 5",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Arbitration of candidates",
"sec_num": "2.3"
},
{
"text": ". The beginning of a sentence is the zeroth word boundary, and the first word ends at the first word boundary, etc. Then, the selected candidate should have the minimum rank according to the lexicographical ordering of 0 1 ( 7 \u00a6 8 3 4 ) 6 5",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Arbitration of candidates",
"sec_num": "2.3"
},
{
"text": ". When a candidate starts or ends within a word (e.g., SHI-NAI), we assume that the entire word is a member of the candidate for the definition of 0 1 ( 2 \u00a6 3 4 ) 6 5 . According to this ordering, two candidates can have the same rank. One of them might assert that a certain word is an organization's name and another candidate might assert that it is a person's name. In order to apply the most frequently used rule, we extend this ordering by 0 1 ( 2 \u00a6 3 4 ) 6 \u00a6 8 3 @ 9 B A 5",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Arbitration of candidates",
"sec_num": "2.3"
},
{
"text": ", where 9 C A is the number of positive examples for the rule D .",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Arbitration of candidates",
"sec_num": "2.3"
},
{
"text": "In order to compare our method with the ME approach, we also implement an ME system based on Ristad's toolkit (1997). Borthwick's (1999) and Uchimoto's (2000) We use the following features for each word in the training data: the word itself, \u00a1 preceding words, \u00a2 succeeding words, their character types, and their POS tags. By following Uchimoto, we disregard words that appear fewer than five times and other features that appear fewer than three times.",
"cite_spans": [
{
"start": 118,
"end": 136,
"text": "Borthwick's (1999)",
"ref_id": "BIBREF3"
},
{
"start": 141,
"end": 158,
"text": "Uchimoto's (2000)",
"ref_id": "BIBREF16"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Maximum entropy system",
"sec_num": "2.4"
},
{
"text": "Then, the ME-based classifier gives a probability for each class to each word in a new sentence. Finally, the Viterbi algorithm (see textbooks, e.g., (Allen, 1995) ) enhanced with consistency checking (e.g., PERS ON-EN D should follow PER SON-BEGI N or PERS ON-M IDDLE) determines the best combination for the entire sentence.",
"cite_spans": [
{
"start": 150,
"end": 163,
"text": "(Allen, 1995)",
"ref_id": "BIBREF0"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Maximum entropy system",
"sec_num": "2.4"
},
{
"text": "We generate the word boundary rewriting rules as follows. First, the NE boundaries inside a word are assumed to be at the nearest word boundary outside the named entity. Hence, SHI</LOCATION>NAI is rewritten as SHI-NAI</LOCATION>. Accordingly, SHI-NAI is classified as LOC ATION -END. The original NE boundary is recorded for the pair SHI-NAI/ LOCATION -END, If SHI-NAI/LOCATION-END is found in the output of the Viterbi algorithm, it is rewritten as SHI</LOCATION>NAI. Since rewriting rules from rare cases can be harmful, we employ a rewriting rule only when the rule correctly works for more than 50% of the word/class pairs in the training data.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Maximum entropy system",
"sec_num": "2.4"
},
{
"text": "Now, we compare our method with the ME system. We used the standard IREX training data (CRL NE 1.4 MB and NERT 30 KB) and the formal run test data (GENERAL and AR-REST). When human annotators were not sure, they used <OPTIONAL POSSIBILITY=...> where POSSIBILITY is a list of possible NE classes. We also used 7.4 MB of in-house NE data that did not contain optional tags. All of the training data (all = CRL NE+NERT+in-house) were based on the Mainichi Newspaper's 1994 and 1995 CD-ROMs. Table 1 shows the details. We removed an optional tag when its possibility list contains NONE, which means this part is accepted without a tag. Otherwise, we selected the majority class in the list. As a result, 56 NEs were added to CRL NE.",
"cite_spans": [],
"ref_spans": [
{
"start": 488,
"end": 495,
"text": "Table 1",
"ref_id": "TABREF3"
}
],
"eq_spans": [],
"section": "Results",
"sec_num": "3"
},
{
"text": "For tokenization, we used chasen 2.2.1 (http:// chasen. aist-nara. ac. jp/). It has about 90 POS tags and large proper noun dictionaries (persons = 32,167, organizations = 16,610, locations = 67,296, miscellaneous proper nouns = 26,106). (Large dictionaries sometimes make the extraction of NEs difficult. If OO-SAKA-GIN-KOU is registered as a single word, GIN-KOU is not extracted as an organization suffix from this example.) We tuned chasen's parameters for NE recognition. In order to avoid the excessive division of unknown words (see Introduction), we reduced the cost for unknown words (30000 R 7000). We also changed its setting so that an unknown word are classified as a misc-proper-noun.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Results",
"sec_num": "3"
},
{
"text": "Then, we compared the above methods in terms of the averaged F-measures by 5-fold crossvalidation of CRL NE data. by removing bad templates with fewer positive examples than negative ones.) Thus, the two methods returned similar results. However, we cannot expect good performance for other documents because CRL NE is limited to January, 1995. Figure 2 compares these systems by using the formal run data. We cannot show the ME results for the large training data because Ristad's toolkit crashes even on a 2 GB memory machine. According to this graph, the RG+DT system's scores are comparable to those of the ME system. When all the training data was used, RG+DT's F-measure for GENERAL was 87.43%. We also examined RG+DT's variants. When we replaced character types of one-word NEs by '*', the score dropped to 86.79%. When we did not replace any character type by '*' at all, the score was 86.63%. RG+DT/n in the figure is a variant that also applies suffix dictionary to numerical NE classes.",
"cite_spans": [],
"ref_spans": [
{
"start": 345,
"end": 353,
"text": "Figure 2",
"ref_id": "FIGREF2"
}
],
"eq_spans": [],
"section": "Results",
"sec_num": "3"
},
{
"text": "When we used tokenized CRL NE for training, the RG+DT system's training time was about 3 minutes on a Pentium III 866 MHz 256 MB memory Linux machine. This performance is much faster than that of the ME system, which takes a few hours; this difference cannot be explained by the fact that the ME system is implemented on a slower machine. When we used all of the training data, the training time was less than one hour and the processing time of tokenized GENERAL (79 KB before tokenization) was about 14 seconds.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Results",
"sec_num": "3"
},
{
"text": "Before the experiments, we did not expect that the RG+DT system would perform very well because the number of possible combinations of POS tags increases exponentially with respect to the num- When we compare the RG+DT method with other statistical methods, its advantage is its readability and independence of generated rules. When using cascaded rules, a small change in a rule can damage another rule's functionality. On the other hand, the recognition rules of our system are not cascaded (Fig. 1) . Therefore, rewriting a recognition rule does not influence the performance of other rules at all. Moreover, dt-rules are usually very simple. When all of the training data were used, most of the RG+DT's recognition rules had a simple additional constraint that always accepts (65%) or rejects (16%) candidates. This result also implies the usefulness of our rule generator. Only 2% of the recognition rules have 10 or more dt-rules. For instance, the following recognition rule has dozens of dt-rules. *:all-katakana:misc-proper-noun -> PERSON,0,0.",
"cite_spans": [],
"ref_spans": [
{
"start": 493,
"end": 501,
"text": "(Fig. 1)",
"ref_id": "FIGREF0"
}
],
"eq_spans": [],
"section": "Discussion",
"sec_num": "4"
},
{
"text": "However, they are easy to understand as follows. We can explain this tendency as follows. Short NEs like 'Washington' are often ambiguous, but longer NEs like 'Washington State University' are less ambiguous. Thus, short recognition rules often have dozens of dt-rules, whereas long rules have simple constraints.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Discussion",
"sec_num": "4"
},
{
"text": "Some NE systems use decision tree learning to classify a word. Sekine's system (1998) is similar to the above ME systems, but C4.5 (Quinlan, 1993 ) is used instead. A similar system participated in IREX, but failed to show good performance. Borthwick (1999) explained the reason for this tendency. When he added lexical questions (e.g., whether the current word is ( or not) to Sekine's system, C4.5 crashed with CRL NE. Accordingly, the decision tree systems did not directly use words as features. Instead, they used a word's memberships in their word lists. Cowie (1995) interprets a decision tree deterministically and uses heuristic rewriting rules to get consistent results. Baluja's system (2000) simply determines whether a word is in an NE or not and does not classify it. On the other hand, Paliouras (2000) uses decision tree learning for classification of a noun phrase by assuming that named entities are noun phrases. Gallippi (1996) employs hundreds of hand-crafted templates as features for decision tree learning. Brill's rule generation method (Brill, 2000) is not used for NE tasks, but it might be useful.",
"cite_spans": [
{
"start": 131,
"end": 145,
"text": "(Quinlan, 1993",
"ref_id": "BIBREF11"
},
{
"start": 241,
"end": 257,
"text": "Borthwick (1999)",
"ref_id": "BIBREF3"
},
{
"start": 561,
"end": 573,
"text": "Cowie (1995)",
"ref_id": "BIBREF6"
},
{
"start": 681,
"end": 703,
"text": "Baluja's system (2000)",
"ref_id": null
},
{
"start": 801,
"end": 817,
"text": "Paliouras (2000)",
"ref_id": "BIBREF10"
},
{
"start": 932,
"end": 947,
"text": "Gallippi (1996)",
"ref_id": "BIBREF7"
},
{
"start": 1062,
"end": 1075,
"text": "(Brill, 2000)",
"ref_id": "BIBREF4"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Discussion",
"sec_num": "4"
},
{
"text": "Recently, unsupervised or minimally supervised models have been proposed (Collins and Singer, 2000; ).",
"cite_spans": [
{
"start": 73,
"end": 99,
"text": "(Collins and Singer, 2000;",
"ref_id": "BIBREF5"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Discussion",
"sec_num": "4"
},
{
"text": "Collins' system is not a full NE system and Utsuro's score is not very good yet, but they represent interesting directions.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Discussion",
"sec_num": "4"
},
{
"text": "As far as we can tell, Japanese NE recognition technology has not yet matured. Conventional decision tree systems have not shown good performance. The maximum entropy method is competitive, but adding more training data causes problems. In this paper, we presented an alternative method based on decision tree learning and longest match. According to our experiments, this method's performance is comparable to that of the maximum entropy system, and it can be trained more efficiently. We hope our method can be applicable to other languages.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Conclusions",
"sec_num": "5"
}
],
"back_matter": [
{
"text": "I would like to thank Yutaka Sasaki, Kiyotaka Uchimoto, Tsuneaki Kato, Eisaku Maeda, Shigeru Katagiri, Kenichiro Ishii, and anonymous reviewers.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Acknowledgement",
"sec_num": null
}
],
"bib_entries": {
"BIBREF0": {
"ref_id": "b0",
"title": "Natural Language Understanding 2nd",
"authors": [
{
"first": "James",
"middle": [],
"last": "Allen",
"suffix": ""
}
],
"year": 1995,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "James Allen. 1995. Natural Language Understanding 2nd. Ed. Benjamin Cummings.",
"links": null
},
"BIBREF1": {
"ref_id": "b1",
"title": "Applying Machine Learning for High Performance Named-Entity Extraction",
"authors": [
{
"first": "Shumeet",
"middle": [],
"last": "Baluja",
"suffix": ""
},
{
"first": "Vibhu",
"middle": [],
"last": "Mittal",
"suffix": ""
},
{
"first": "Rahul",
"middle": [],
"last": "Sukthankar",
"suffix": ""
}
],
"year": 2000,
"venue": "Computational Intelligence",
"volume": "16",
"issue": "4",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Shumeet Baluja, Vibhu Mittal, and Rahul Sukthankar. 2000. Applying Machine Learning for High Perfor- mance Named-Entity Extraction. Computational Intelligence, 16(4).",
"links": null
},
"BIBREF2": {
"ref_id": "b2",
"title": "An algorithm that learns what's in a name",
"authors": [
{
"first": "M",
"middle": [],
"last": "Daniel",
"suffix": ""
},
{
"first": "Richard",
"middle": [],
"last": "Bikel",
"suffix": ""
},
{
"first": "Ralph",
"middle": [
"M"
],
"last": "Schwartz",
"suffix": ""
},
{
"first": "",
"middle": [],
"last": "Weischedel",
"suffix": ""
}
],
"year": 1999,
"venue": "Machine Learning",
"volume": "34",
"issue": "",
"pages": "211--231",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Daniel M. Bikel, Richard Schwartz, and Ralph M. Weischedel. 1999. An algorithm that learns what's in a name. Machine Learning, 34(1-3):211-231.",
"links": null
},
"BIBREF3": {
"ref_id": "b3",
"title": "A Maximum Entropy Approach to Named Entity Recognition",
"authors": [
{
"first": "Andrew",
"middle": [],
"last": "Borthwick",
"suffix": ""
}
],
"year": 1999,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Andrew Borthwick. 1999. A Maximum Entropy Ap- proach to Named Entity Recognition. Ph.D. thesis, New York University.",
"links": null
},
"BIBREF4": {
"ref_id": "b4",
"title": "Pattern-based disambiguation for natural language processing",
"authors": [
{
"first": "Eric",
"middle": [],
"last": "Brill",
"suffix": ""
}
],
"year": 2000,
"venue": "Proceedings of EMNLP/VLC-2000",
"volume": "",
"issue": "",
"pages": "1--8",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Eric Brill. 2000. Pattern-based disambiguation for natural language processing. In Proceedings of EMNLP/VLC-2000, pages 1-8.",
"links": null
},
"BIBREF5": {
"ref_id": "b5",
"title": "Unsupervised models for named entity classification",
"authors": [
{
"first": "Michael",
"middle": [],
"last": "Collins",
"suffix": ""
},
{
"first": "Yoram",
"middle": [],
"last": "Singer",
"suffix": ""
}
],
"year": 2000,
"venue": "Proceedings of EMNLP/VLC",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Michael Collins and Yoram Singer. 2000. Unsuper- vised models for named entity classification. In Proceedings of EMNLP/VLC.",
"links": null
},
"BIBREF6": {
"ref_id": "b6",
"title": "CRL/NMSU description of the CRL/NMSU system used for MUC-6",
"authors": [
{
"first": "Jim",
"middle": [],
"last": "Cowie",
"suffix": ""
}
],
"year": 1995,
"venue": "Proceedings of the Sixth Message Understanding Conference",
"volume": "",
"issue": "",
"pages": "157--166",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Jim Cowie. 1995. CRL/NMSU description of the CRL/NMSU system used for MUC-6. In Proceed- ings of the Sixth Message Understanding Confer- ence, pages 157-166. Morgan Kaufmann.",
"links": null
},
"BIBREF7": {
"ref_id": "b7",
"title": "Learning to recognize names accross lanugages",
"authors": [
{
"first": "Anthony",
"middle": [
"F"
],
"last": "Gallippi",
"suffix": ""
}
],
"year": 1996,
"venue": "Proceedings of the International Conference on Computational Linguistics",
"volume": "",
"issue": "",
"pages": "424--429",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Anthony F. Gallippi. 1996. Learning to recognize names accross lanugages. In Proceedings of the In- ternational Conference on Computational Linguis- tics, pages 424-429.",
"links": null
},
"BIBREF8": {
"ref_id": "b8",
"title": "Proceedings of the IREX Workshop",
"authors": [
{
"first": "",
"middle": [],
"last": "Irex Comittee",
"suffix": ""
}
],
"year": 1999,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "IREX Comittee. 1999. Proceedings of the IREX Workshop (in Japanese).",
"links": null
},
"BIBREF9": {
"ref_id": "b9",
"title": "Proceedings of the Sixth Message Understanding Conference",
"authors": [
{
"first": "",
"middle": [],
"last": "Muc-6",
"suffix": ""
}
],
"year": 1996,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "MUC-6. 1996. Proceedings of the Sixth Message Un- derstanding Conference. Morgan Kaufmann.",
"links": null
},
"BIBREF10": {
"ref_id": "b10",
"title": "Learning decision trees for named-entity recognition and classification",
"authors": [
{
"first": "Georgios",
"middle": [],
"last": "Paliouras",
"suffix": ""
},
{
"first": "Vangelis",
"middle": [],
"last": "Karkaletsis",
"suffix": ""
},
{
"first": "Georgios",
"middle": [],
"last": "Petasis",
"suffix": ""
},
{
"first": "Constantine",
"middle": [
"D"
],
"last": "Spyropoulos",
"suffix": ""
}
],
"year": 2000,
"venue": "ECAI Workshop on Machine Learning for Information Extraction",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Georgios Paliouras, Vangelis Karkaletsis, Georgios Petasis, and Constantine D. Spyropoulos. 2000. Learning decision trees for named-entity recogni- tion and classification. In ECAI Workshop on Ma- chine Learning for Information Extraction.",
"links": null
},
"BIBREF11": {
"ref_id": "b11",
"title": "C4.5: Programs for Machine Learning",
"authors": [
{
"first": "J",
"middle": [],
"last": "",
"suffix": ""
},
{
"first": "Ross",
"middle": [],
"last": "Quinlan",
"suffix": ""
}
],
"year": 1993,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "J. Ross Quinlan. 1993. C4.5: Programs for Machine Learning. Morgan Kaufmann Publishers.",
"links": null
},
"BIBREF12": {
"ref_id": "b12",
"title": "Maximum entropy modeling toolkit, release 1.5 Beta",
"authors": [
{
"first": "Eric",
"middle": [],
"last": "Sven Ristad",
"suffix": ""
}
],
"year": 1997,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Eric Sven Ristad, 1997. Maximum entropy modeling toolkit, release 1.5 Beta. ftp:// ftp. cs. princeton. edu/ pub/ packages/ memt, January.",
"links": null
},
"BIBREF13": {
"ref_id": "b13",
"title": "Named entity chunking techniques in supervised learning for Japanese named entity recognition",
"authors": [
{
"first": "Manabu",
"middle": [],
"last": "Sassano",
"suffix": ""
},
{
"first": "Takehito",
"middle": [],
"last": "Utsuro",
"suffix": ""
}
],
"year": 2000,
"venue": "Proceedings of the International Conference on Computational Linguistics",
"volume": "",
"issue": "",
"pages": "705--711",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Manabu Sassano and Takehito Utsuro. 2000. Named entity chunking techniques in supervised learning for Japanese named entity recognition. In Proceed- ings of the International Conference on Computa- tional Linguistics, pages 705-711.",
"links": null
},
"BIBREF14": {
"ref_id": "b14",
"title": "Japanese named entity extraction evaluation -analysis of results",
"authors": [
{
"first": "Satoshi",
"middle": [],
"last": "Sekine",
"suffix": ""
},
{
"first": "Yoshio",
"middle": [],
"last": "Eriguchi",
"suffix": ""
}
],
"year": 2000,
"venue": "Proceedings of 18th International Conference on Computational Linguistics",
"volume": "",
"issue": "",
"pages": "1106--1110",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Satoshi Sekine and Yoshio Eriguchi. 2000. Japanese named entity extraction evaluation -analysis of results -. In Proceedings of 18th International Conference on Computational Linguistics, pages 1106-1110.",
"links": null
},
"BIBREF15": {
"ref_id": "b15",
"title": "A decision tree method for finding and classifying names in Japanese texts",
"authors": [
{
"first": "Satoshi",
"middle": [],
"last": "Sekine",
"suffix": ""
},
{
"first": "Ralph",
"middle": [],
"last": "Grishman",
"suffix": ""
},
{
"first": "Hiroyuki",
"middle": [],
"last": "Shinnou",
"suffix": ""
}
],
"year": 1998,
"venue": "Proceedings of the Sixth Workshop on Very Large Corpora",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Satoshi Sekine, Ralph Grishman, and Hiroyuki Shin- nou. 1998. A decision tree method for finding and classifying names in Japanese texts. In Proceedings of the Sixth Workshop on Very Large Corpora.",
"links": null
},
"BIBREF16": {
"ref_id": "b16",
"title": "Named entity extraction based on a maximum entropy model and transformation rules",
"authors": [
{
"first": "Kiyotaka",
"middle": [],
"last": "Uchimoto",
"suffix": ""
},
{
"first": "Qing",
"middle": [],
"last": "Ma",
"suffix": ""
},
{
"first": "Masaki",
"middle": [],
"last": "Murata",
"suffix": ""
},
{
"first": "Hiromi",
"middle": [],
"last": "Ozaku",
"suffix": ""
},
{
"first": "Masao",
"middle": [],
"last": "Utiyama",
"suffix": ""
},
{
"first": "Hitoshi",
"middle": [],
"last": "Isahara",
"suffix": ""
}
],
"year": 2000,
"venue": "Journal of Natural Language Processing",
"volume": "7",
"issue": "",
"pages": "63--90",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Kiyotaka Uchimoto, Qing Ma, Masaki Murata, Hi- romi Ozaku, Masao Utiyama, and Hitoshi Isahara. 2000. Named entity extraction based on a maxi- mum entropy model and transformation rules (in Japanese). Journal of Natural Language Process- ing, 7(2):63-90.",
"links": null
},
"BIBREF17": {
"ref_id": "b17",
"title": "Minimally supervised Japanese named entity recognition: Resources and evaluation",
"authors": [
{
"first": "Takehito",
"middle": [],
"last": "Utsuro",
"suffix": ""
},
{
"first": "Manabu",
"middle": [],
"last": "Sassano",
"suffix": ""
}
],
"year": 2000,
"venue": "Proceedings of the Second International Conference on Language Resources and Evaluation",
"volume": "",
"issue": "",
"pages": "1229--1236",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Takehito Utsuro and Manabu Sassano. 2000. Min- imally supervised Japanese named entity recogni- tion: Resources and evaluation. In Proceedings of the Second International Conference on Language Resources and Evaluation, pages 1229-1236.",
"links": null
}
},
"ref_entries": {
"FIGREF0": {
"text": "Rough sketch of RG+DT system",
"type_str": "figure",
"uris": null,
"num": null
},
"FIGREF2": {
"text": "Comparison of RG+DT systems and Max. Ent. system ber of words in an NE. However, the above results are encouraging. Its performance is comparable to the ME system. Why did it work so well? First, the percentage of long NEs is negligible. 91% of the NEs in the training data have at most three words. Second, the POS tags frequently used in NEs are limited.",
"type_str": "figure",
"uris": null,
"num": null
},
"TABREF1": {
"content": "<table><tr><td>O RGAN IZAT ION, E PERS ON, LOC ATIO N, ARTI FACT, DATE, TIM E,</td></tr><tr><td>MON EY, PERC ENTF SING LEF Q P O THERF . For instance, the words H G B EG IN, MID DLE, END, I E I E in \"President <PERSON> George Herbert Walker</td></tr><tr><td>Bush </PERSON>\" are classified as follows:</td></tr><tr><td>President = OTHE R, George = PERS ON-BE GIN,</td></tr><tr><td>Herbert = PERSO N-MI DDLE, Walker = PER SON-</td></tr><tr><td>MIDD LE, Bush = PER SON-END.</td></tr></table>",
"num": null,
"text": "ME systems are quite similar but differ in details. They regarded Japanese NE recognition as a classification problem of a word. The first word of a person name is classified as PERS ON-B EGIN. The last word is classified as PERS ON-E ND. Other words in the person's name (if any) are classified as PERS ON-M IDDL E. If the person's name is composed of only one word, it is classified as PERS ON-S INGLE. Similar labels are given to all other classes such as LOCATION. Non-NE words are classified as OTHE R. Thus, every word is classified into 33 classes, i.e.,",
"type_str": "table",
"html": null
},
"TABREF3": {
"content": "<table><tr><td colspan=\"2\">: Data used for comparison</td></tr><tr><td>attained 81.18% for</td><td>0 \u1e84 \" \u00a6 \" 5</td></tr></table>",
"num": null,
"text": "",
"type_str": "table",
"html": null
}
}
}
} |