Published June 14, 2021 | Version 1.0
Conference paper Open

ExQuestions: An Expanded Factual Corpus for Question Answering over Knowledge Graphs

  • 1. Federal University of Ceará
  • 2. University of Fortaleza

Description

ExQuestions dataset contains 128,578 pairs in SPARQL building from entities extracted by Wikipedia corpus.  Since 104,791 generated from the expansion promoted with the help of the common sense base. The dataset has 7,350 entities generated, being 4,579 of the subject type and 2,771 of the object type. The predicates contained in the base are 83 and the number of generated templates is 25,831. The generated natural language questions reached a total of 233,369.

Files

testJSON.json

Files (207.6 MB)

Name Size Download all
md5:35cdc9de0776130033a3fbda750704ac
49.1 MB Preview Download
md5:a13e3ef8d05dc7ae91151a67ed367354
158.6 MB Preview Download