EXPLORING THE EFFECT OF BAG-OF-WORDS AND BAG-OF-BIGRAM FEATURES ON TURKISH WORD SENSE DISAMBIGUATION

Authors : Bahar İLGEN, Eşref ADALI

Pages : 5-10

View : 10 | Download : 7

Publication Date : 2014-12-21

Article Type : Research Paper

Abstract :Feature selection in Word Sense Disambiguation insert ignore into journalissuearticles values(WSD); is as important as the selection of algorithm to remove sense ambiguity. Bag-of-word insert ignore into journalissuearticles values(BoW); features comprise the information of neighbors around the ambiguous target word without considering any relation between words. In this study, we investigate the effect of BoW features and Bag-of-bigrams insert ignore into journalissuearticles values(BoB); on Turkish WSD and compare the results with the collocational features. The results suggest that BoW features yield better accuracy for all the cases. According to the comparison results, collocational features are more effective than both BoW and the BoB features on disambiguation of word senses.
Keywords : Word Sense Disambiguation, feature selection, feature selection, supervised methods, bag of word features