트렉

  • 설명 :

TREC(Text REtrieval Conference) 질문 분류 데이터 세트에는 교육 세트에 5500개의 레이블이 지정된 질문과 테스트 세트에 대한 또 다른 500개의 질문이 포함되어 있습니다. 데이터 세트에는 6개의 레이블, 47개의 레벨 2 레이블이 있습니다. 각 문장의 평균 길이는 10개, 어휘 크기는 8700개입니다. 데이터는 USC에서 발행한 4,500개의 영어 질문(Hovy et al., 2001), 소수의 희귀 클래스에 대해 수동으로 구성된 약 500개의 질문, 894개의 TREC 8 및 894개의 출처에서 수집되었습니다. TREC 9 질문 및 테스트 세트 역할을 하는 TREC 10의 500개 질문.

나뉘다
'test' 500
'train' 5,452
  • 기능 구조 :
FeaturesDict({
    'label-coarse': ClassLabel(shape=(), dtype=int64, num_classes=6),
    'label-fine': ClassLabel(shape=(), dtype=int64, num_classes=47),
    'text': Text(shape=(), dtype=string),
})
  • 기능 문서 :
특징 수업 모양 D타입 설명
풍모Dict
거친 레이블 클래스 레이블 int64
라벨 벌금 클래스 레이블 int64
텍스트 텍스트
  • 인용 :
@inproceedings{li-roth-2002-learning,
    title = "Learning Question Classifiers",
    author = "Li, Xin  and
      Roth, Dan",
    booktitle = "{COLING} 2002: The 19th International Conference on Computational Linguistics",
    year = "2002",
    url = "https://www.aclweb.org/anthology/C02-1150",
}
@inproceedings{hovy-etal-2001-toward,
    title = "Toward Semantics-Based Answer Pinpointing",
    author = "Hovy, Eduard  and
      Gerber, Laurie  and
      Hermjakob, Ulf  and
      Lin, Chin-Yew  and
      Ravichandran, Deepak",
    booktitle = "Proceedings of the First International Conference on Human Language Technology Research",
    year = "2001",
    url = "https://www.aclweb.org/anthology/H01-1069",
}