Feature based clustering considering context dependent words
Sonal Garg, Dilip Kumar Sharma · 2015
With the increase in the popularity of e-commerce, there is large amount of opinions available on the web. So there is a need to generate clustered summary of products based on features. Most of the opinions contain opinion words which has same polarity in all contexts. But there are some opinion words called context dependent words which have different polarity in different context. So there is a need to determine the polarity of ambiguous words (context dependent words) efficiently and effectively and then generate the aspect based summary. In this paper we used k nearest neighbor classifier to determine the polarity of context dependent words. Then short summary is produced for that particular product based on each feature. Experimental results show that our method gives better result in comparison to existing method.