Named Entity Recognizer for Telugu language using Hybrid approach

Abstract

The main goal of Named Entity Recognition (NER) is to classify all Named Entities (NE) in a document into predefined classes like Person name, Location name, Organization name and Miscellaneous. This paper outlines Named Entity Recognizer using hybrid approach i.e., combination of Rule based approach and one of the Machine learning technique i.e, Conditional Random Field (CRF). In Rule based approach we have prepared Gazetteer lists for names of persons, locations and organizations; some suffix and prefix features and dictionary consisting 350266 words to recognize the category of named entities. If ambiguity is rised while we are using Rule based approach, we use Machine learning technique i.e., CRF in order to improve the accuracy.