Recognizing Named Entities Based on Ontologies in Kazakh Language Dataset

Abstract

This study looks into a novel approach for enhancing named entity recognition (NER) in the Kazakh language. Using an IOB2-annotated dataset, the study proposes a specialized ontology for capturing Kazakh language and culture traits. Methodologically, the article integrates this ontology with the dataset by mapping tokens to IOB2 annotations. The main findings demonstrate the utility of ontology-driven NER, as judged by accuracy, recall, and F1 score metrics. The research addresses annotation difficulties by showing how ontological augmentation enhances awareness to regional disparities. Finally, the work contributes to the field of NER by proposing a contextually aware ontology for extracting semantic insights from IOB2-annotated tokens in Kazakh language literature. The approach improves information extraction while taking into account the Kazakh language's linguistic complexity.