ACQAD: A Dataset for Arabic Complex Question Answering - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2022

ACQAD: A Dataset for Arabic Complex Question Answering

Résumé

In this paper, we tackle the problem of Arabic complex Question Answering (QA), where models are required to reason over multiple documents to find the answer. Indeed, no Arabic dataset is available for this type of questions. To fill this lack, we propose a new approach to automatically generate a dataset for Arabic complex question answering task. The proposed approach is based on using an effective workflow with a set of templates. The generated dataset, denoted as ACQAD, contains more than 118k questions, covering both comparison and multi-hop types. Each question-answer pair is decomposed into a set of single-hop questions, allowing QA systems to reduce question complexity and explain the reasoning steps. We then provide a statistical analysis of the produced dataset. Afterwards, we will make the corpus available to the international community.
Fichier principal
Vignette du fichier
ICCSAITCS_2022_paper_9032.pdf (650.46 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-03992129 , version 1 (16-02-2023)

Identifiants

  • HAL Id : hal-03992129 , version 1

Citer

Abdellah Hamouda Sidhoum, M’hamed Mataoui, Faouzi Sebbak, Kamel Smaïli. ACQAD: A Dataset for Arabic Complex Question Answering. International Conference on Cyber Security, Artificial Inteligence and Theoretical Computer Science, Dec 2022, Boumerdès, Algeria. ⟨hal-03992129⟩
99 Consultations
390 Téléchargements

Partager

Gmail Facebook X LinkedIn More