Word-Sense Disambiguation using maximum entropy model
Niladri Chatterjee, R. Misra · 2009 Proceeding of International Conference on Methods and Models in Computer Science (ICM2CS) · 2009
Natural languages are typically replete with homographs, words which have more than one meaning. Consequently, machine understanding of natural language sentences sometimes suffers from certain ambiguities in getting the correct sense of a word in a given sentence. In this work we present a trainable model for word sense disambiguation (WSD) for resolving this ambiguity. The proposed model applies concepts of information theory to find the appropriate sense of a word when the context is known. Given a training text tagged with the correct senses of a particular word, our model learns to classify each occurrence of the target word with its correct sense in the unseen text.