Part-Of-Speech Tagging for Urdu in Scarce Resource Mix Maximum Entropy Modelling System

Abstract: The area of automated Part-of-speech tagging has been developed over the last few decades by involvement from several researchers. Many new models have been introduced to improve the effectiveness of the tagger and to build the POS taggers for several languages. In this paper we develop an approach for Urdu POS tagging in scarce resource. We use Maximum Entropy (ME) modelling system [1][2], Morphological analyser(MA) [3]and stemmer[4] for automatic POS Tagging. Maximum Entropy model is a very flexible method of statistical modelling which handles the data sparse problem. Under this model, a natural combination of several features can be easily incorporated. Maximum Entropy based methods can deal with various sets has common characteristics features. We mix MA with ME model, we proposed different models ME, ME+Suf, ME+MA, ME+Suf+MA. These models are tested and results were analysed.

Keywords: Maximum Entropy Model, Morphological analyser, Stemmer, Urdu Language, NLP.