Opinion Mining through Enhanced Lexicon Approach

Abstract

Abstract

The World Wide Web’s rapid growth has enhanced online communication. One of the approaches to putting user views is the use of social networking sites. The user’s opinion needs to be analyzed by studying user’s input text using techniques of machine learning with subjectivity, emotion, sentiment analysis or polarity calculation. Most of the existing work in the field of opinion mining has largely focused on identifying sentiment polarity as positive, negative, or neutral. This paper presents an evaluation of sentiments from the 2016 Kashmir Unrest tweets to find people’s opinions as peace or protest during the unrest using enhanced lexicon approach. Lexicon-based approach has three methods of opinion mining: manual based method, dictionary-based method and corpus-based method. Dictionary-based method is used in this paper where two lexicons, the lexicon for peace and the lexicon for protest are created and then expanded using Word2Vec to determine opinion of general public during 2016 Kashmir Unrest.