Strategies for Arabic Readability Modeling

Juan Piñeros Liberato, Bashar Alhafni, Muhamed Al Khalil, Nizar Habash

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Automatic readability assessment is relevant to building NLP applications for education, content analysis, and accessibility. However, Arabic readability assessment is a challenging task due to Arabic’s morphological richness and limited readability resources. In this paper, we present a set of experimental results on Arabic readability assessment using a diverse range of approaches, from rule-based methods to Arabic pretrained language models. We report our results on a newly created corpus at different textual granularity levels (words and sentence fragments). Our results show that combining different techniques yields the best results, achieving an overall macro F1 score of 86.7 at the word level and 87.9 at the fragment level on a blind test set. We make our code, data, and pretrained models publicly available.

Original languageEnglish (US)
Title of host publicationArabicNLP 2024 - 2nd Arabic Natural Language Processing Conference, Proceedings of the Conference
EditorsNizar Habash, Houda Bouamor, Ramy Eskander, Nadi Tomeh, Ibrahim Abu Farha, Ahmed Abdelali, Samia Touileb, Injy Hamed, Yaser Onaizan, Bashar Alhafni, Wissam Antoun, Salam Khalifa, Hatem Haddad, Imed Zitouni, Badr AlKhamissi, Rawan Almatham, Khalil Mrini
PublisherAssociation for Computational Linguistics (ACL)
Pages55-66
Number of pages12
ISBN (Electronic)9798891761322
StatePublished - 2024
Event2nd Arabic Natural Language Processing Conference, ArabicNLP 2024 - Bangkok, Thailand
Duration: Aug 16 2024 → …

Publication series

NameArabicNLP 2024 - 2nd Arabic Natural Language Processing Conference, Proceedings of the Conference

Conference

Conference2nd Arabic Natural Language Processing Conference, ArabicNLP 2024
Country/TerritoryThailand
CityBangkok
Period8/16/24 → …

ASJC Scopus subject areas

  • Language and Linguistics
  • Computational Theory and Mathematics
  • Software
  • Linguistics and Language

Fingerprint

Dive into the research topics of 'Strategies for Arabic Readability Modeling'. Together they form a unique fingerprint.

Cite this