A Semantic Wiki for Language Learning: The Case of the Baoulé Language

This article presents BAOULE-WIKI, an innovative project aimed at preserving and teaching the Baoulé language through a semantic wiki that combines community collaboration and semantic web technologies. Based on the alarming findings of UNESCO and FORTIN on the imminent disappearance of African languages and the lack of suitable digital resources, the study analyses the limitations of existing initiatives (SweetWiki, Aragonais, Yoruba, Ummto) and proposes an original approach based on the automatic detection of homographs, a major problem in tonal languages. A mathematical model for recognising and eliminating homographic duplicates was designed, implemented and experimentally validated using a Python algorithm, achieving an accuracy of over 90% and a recall of around 80%, confirming the feasibility of reliable homograph differentiation. This initial contribution guarantees lexical integrity and paves the way for integration into a functional semantic wiki prototype. The results demonstrate the effectiveness of the approach while highlighting limitations related to subtle phonetic variations and multimedia integration. The article concludes with prospects for enriching the project, including the addition of audio-visual resources, extension to other Ivorian languages (Bété, Dioula, Senoufo), the use of machine learning techniques to refine tone detection, and interconnection with open knowledge bases such as Wikidata and DBpedia. Ultimately, BAOULE-WIKI aims to become a model for the digital preservation and transmission of endangered African languages. The simulation conducted on an annotated corpus yielded an accuracy of 92% and a recall of 80%, confirming the relevance of our model in the automatic differentiation of homographs. Furthermore, integrating this model into an OWL ontology and a Semantic MediaWiki prototype strengthens lexical coherence and paves the way for educational and collaborative use of BAOULE-WIKI.