A REVIEW OF ENHANCED IGBO LANGUAGE NAMED ENTITY RECOGNITION USING MULTILINGUAL MODELS

Abstract

The problem of Named Entity Recognition (NER) for Igbo language, one of Nigeria’s major languages still exists because of limitations in Natural Language Processing (NLP) tools. This paper reviews progress in Named Entity Recognition (NER) for low-resource African languages, with focus on Igbo. NER is a key task in NLP that involves identifying entities such as persons, locations, organizations, and dates. While high-resource languages have benefited from large datasets and advanced models, African languages face difficulties due to limited corpora, complex grammar, and inconsistent writing standards. Recent efforts, including MasakhaNER and IgboNER 2.0, have begun to address these gaps by releasing annotated datasets and baseline systems. Transformer models such as XLM-R and AfriBERTa have further shown that multilingual pretraining can improve recognition in Igbo, though challenges remain in annotation quality and coverage of entity types. This review highlights advances in dataset development, annotation practices, and the use of multilingual models, and shows how extending Igbo NER to five entity types with XLM-R can strengthen both research and application.