MODAN: An Ontology-Motivated Model for Annotation of Noun Class Systems to Enhance Meaningful NLP

Keet, C. Maria and Mahlaza, Zola and Khan, Zubeida C. (2026) MODAN: An Ontology-Motivated Model for Annotation of Noun Class Systems to Enhance Meaningful NLP, Proceedings of 16th International Conference on Formal Ontology in Information Systems, 21–25 September 2026, Vitória, ES, Brazil.

Text Paper-228.pdf - Published Version Download (1MB)

Abstract

Natural Language Processing (NLP) for African languages, specifically for tasks such as Machine Translation and Human-Robot Interaction, may require knowledge-infused approaches. In such contexts, especially for Niger-Congo B (NCB) languages, representation of noun class agreement in the sentences is crucial to ensure grammatical correct output. However, there is no linguistic ontology to support such work. A foundational ontology could assist NLP tasks, but existing foundational ontologies do not include linguistic entities and the key desiderata only very partially; hence, it was not possible to use them either. To solve this problem, we developed a new logic-based model (MODAN) that may be used directly or as a module of a linguistic ontology, to accurately represent and annotate NCB noun class systems for better computational use, and therewith the nouns. We evaluated MODAN against noun class systems in different language families and found that it covers most of those noun class semantics as well, therewith increasing the prospect of improving NLP systems well beyond the African languages.

Item Type:	Conference paper
Subjects:	Computing methodologies > Artificial intelligence > Natural language processing > Language resources Information systems > Information retrieval > Document representation > Ontologies
Date Deposited:	20 Oct 2026 07:48
Last Modified:	20 Oct 2026 07:48
URI:	https://pubs.cs.uct.ac.za/id/eprint/1824

Actions (login required)

View Item