Named Entity Recognition Using Part-of-Speech Rules for Telugu

Part-of-Speech (POS) plays an important role in identifying and classifying Named Entities (NEs) for any language, especially, for inflected and agglutinating languages like Telugu. The main objective of our work is to generate POS-based rules to identify named entities in Telugu language using Decision Tree classifier. The corpus is generated by crawling through Telugu newspaper websites, which consists of 54457 words and each of these words are manually annotated with NEs tags, namely person, location, organization, and others. We have achieved competent performance by the generated POS rules and they help to get deeper insights into NER for Telugu.