Case Based Construal using Minimal Features to Decipher Ambiguityin Punjabi Language

A minimal feature set is taken in order to find the closest context of the given ambiguous word by using case based construal model. The model proposed in this paper uses case-based reasoning to sort out the similar cases with vectors of size two (bigram), three (trigram) and four (n-gram) using Euclidean similarity function. These cases are then subjected to three different classifiers, namely - Bayes, k-Nearest Neighbor, and Decision Tree, to decipher the ambiguity.Vectorization eases the process of disambiguation as compared to when full sentences are processed. The similarity function helps to find similar cases of the given ambiguous word whereas the three classifiers helps in finding the right context of the given ambiguous word. Upon experimentation, Decision Tree classifier has achieved an accuracy of 84.88% using pre-bigram vectors.