Published February 28, 2018 | Version v1
Dataset Open

Corpus for hybrid question answering

  • 1. LIMSI, CNRS

Description

Question answering corpus, from a compilation of several existing datasets (CLEF and TREC). The questions and answers were annotated by Spotlight. Then a manuel annotation was performed to validate the answers and the paths: 3 = correct, 2 = partial, 1 = related

## Format

Json array of objects, each object representing a question with its answers and other data :

* answerType
* question
* answers : array of answers
* corpusName
* answersUris
* questionUris
* paths : array of path of length 1
* paths2 : array of path of length 2

 

Files

corpusqa.zip

Files (992.9 kB)

Name Size Download all
md5:2be27d633b31193e9c0b9a09a8c3f8d1
992.9 kB Preview Download

Additional details

Related works

Is documented by
10.1145/3184558.3191540 (DOI)