q_re_cc

  • Descripción :

Un conjunto de datos que contiene 14 000 conversaciones con 81 000 pares de preguntas y respuestas. QReCC se basa en preguntas de TREC CAsT, QuAC y Google Natural Questions.

Separar Ejemplos
'test' 16,451
'train' 63,501
  • Estructura de características :
FeaturesDict({
   
'answer': Text(shape=(), dtype=string),
   
'answer_url': Text(shape=(), dtype=string),
   
'context': Sequence(Text(shape=(), dtype=string)),
   
'conversation_id': Scalar(shape=(), dtype=int32),
   
'question': Text(shape=(), dtype=string),
   
'question_rewrite': Text(shape=(), dtype=string),
   
'source': Text(shape=(), dtype=string),
   
'turn_id': Scalar(shape=(), dtype=int32),
})
  • Documentación de características :
Rasgo Clase Forma Tipo D Descripción
CaracterísticasDict
responder Texto cadena
respuesta_url Texto cadena
contexto Secuencia (Texto) (Ninguno,) cadena
conversación_id Escalar int32 El id de la conversación.
pregunta Texto cadena
pregunta_reescribir Texto cadena
fuente Texto cadena La fuente original de los datos, ya sea QuAC, CAsT o Natural Questions
turno_id Escalar int32 El id del turno de conversación, dentro de una conversación.
  • Cita :
@article{qrecc,
  title
={Open-Domain Question Answering Goes Conversational via Question Rewriting},
  author
={Anantha, Raviteja and Vakulenko, Svitlana and Tu, Zhucheng and Longpre, Shayne and Pulman, Stephen and Chappidi, Srinivas},
  journal
={Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies},
  year
={2021}
}