Tamil POS Tagging using Linear Programming

Abstract

Part of speech (POS) tagging is the process of annotating syntactic categories for each word in a corpus. This paper presents an SVM methodology based on Linear Programming for implementing automatic Tamil POS tagger. We have designed our own tagset consisting of 32 tags for preparing the annotated corpus for Tamil. The features are extracted from a corpus of twenty five thousand sentences and trained with linear programming based SVM. This method, when tested with 10,000 sentences, gave an overall accuracy of 95.63%. Index Terms— Annotated corpus, tokenization, tagging, machine learning.