Communication Dans Un Congrès Année : 2024

KGConv, a Conversational Corpus grounded in Wikidata

Quentin Brabant
  • Fonction : Auteur
Lina M Rojas-Barahona
Gwénolé Lecorvé

Résumé

We present KGConv, a large corpus of 71k English conversations where each question-answer pair is grounded in a Wikidata fact. The conversations were generated automatically: in particular, questions were created using a collection of 10,355 templates; subsequently, the naturalness of conversations was improved by inserting ellipses and coreference into questions, via both handcrafted rules and a generative rewriting model. The dataset thus provides several variants of each question (12 on average), organized into 3 levels of conversationality. We provide baselines for the task of Knowledge-Based Conversational Question Generation. KGConv can further be used for other generation and analysis tasks such as single-turn question generation from Wikidata triples, question rewriting, question answering from conversation or from knowledge graphs and quiz generation.
Fichier principal
Vignette du fichier
2024.lrec-main.850.pdf (595.47 Ko) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)
Licence

Dates et versions

hal-04854981 , version 1 (24-12-2024)

Licence

Identifiants

  • HAL Id : hal-04854981 , version 1

Citer

Quentin Brabant, Lina M Rojas-Barahona, Gwénolé Lecorvé, Claire Gardent. KGConv, a Conversational Corpus grounded in Wikidata. The 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024), LREC, May 2024, Turin, Italy. ⟨hal-04854981⟩
7 Consultations
2 Téléchargements

Partager

More