Domain Based Named Entity Recognition Using Naive Bayes Classification

Abstract

In this paper we propose a methodology to perform named entity recognition on Tamil text which speaks about various temples in Tamil Nadu. The approach is to preprocess the text by tokenizing, parse the text to find the parts of speech and to perform named entity recognition as a classification problem with the help of Traditional Naïve Bayes algorithm. The statistical processing framework makes use of the dictionary created from the training data which belongs to predefined labels of named entities. Our research primarily focused on the domain specific named entities (Temple name, location swami name), temporal entities (date and time) and numbers. Our experiments on the presented system provided us with desirable results.

Keywords: Named Entity Recognition Syntactic parsing Temple domain Naïve Bayes