Context Deployed Sentiment Analysis Using Hybrid Lexicon

Sentiment analysis refers to the study of attitudes or opinions. Sentiment mining is the drawing out of polarity of text, its features and the time at which the attitude was conveyed. Lexicon dependent techniques involve drawing out of polarities of term from the lexicon and aggravation of the obtained scores to determine the comprehensive sentiment of textual data. Lexicon based approaches plays vital role with respect to the large coverage of terms. The unsupervised machine learning methods rarely takes into account the appearance of emoticons, modifiers, negation terms, general purpose lexicon and domain specific lexicon while analyzing the polarity of text. In this paper, the lexicon based approaches plays an active role regarding the aforementioned aspects. Here we focus on handling of contextual polarity of text wherein which the prior polarity of the term expressed in the lexicon may be different from the polarity expressed in the text. Experimental results give evidence in the performance improvement of the proposed system in terms of accuracy, recall and precision when compared with the existing systems.