Word Class Prediction of Ambiguous and Unknown Words of Punjabi Language Using Bi-gram Methods

Ambiguous and unknown words are found in every language. Ambiguous words are the words having different meaning in different sentences depending upon the context of the sentence. Assigning the correct word class to these ambiguous words is the fundamental task in almost all the NLP applications. A lot of work has been done on this and a lot of work is still to be done. Many statistical and rule based techniques has been applied to assign the correct word class to the word having ambiguous word class. Most commonly used statistical techniques are HMM (Hidden Markov Model), SVM (Support Vector Machine), ME (Maximum Entropy), CRF (Conditional Random Field) and N-gram based techniques. In this research paper a bigram technique has been discussed to assign the correct word class to the ambiguous and unknown words of Punjabi language. A tag set proposed by TDIL has been used to assign the correct word class to the ambiguous and unknown words. Keywords-Ambiguous words, word class, Unknown words, Bi-gram technique, TDIL proposed Punjabi tag set.