Sentiment Analysis by Three Side Review for Text Messages using Trust Based Collective View Prediction

Now a day's digital communication plays a vital role in the communication era. In the evolution of mobile based communication such as voice, text, MMS; there is a possibility of mistreat the medium such as exchanging the offensive words during the text based communication. In this paper, sentiment classification focuses on mobile based SMS with its maximum text character of 160 as a Bag of Words (BOW) and analyze the BOW for finding the possibility of offensive language or unpronounced words that is not in the pseudo antonym dictionary. Using natural language processing, the three side review is made for the text message such as original, reverse and break through words (BTW) to extent the possibility of finding the nature of words. A trust based collective (TBC) view prediction algorithm is proposed to classify the text by considering the three sides of one review and also extend the framework from polarity (positive or negative) classification to three-class (positive, negative or neutral) classification. For removing dependency on an external antonym dictionary for review reversion a corpus based method is developed to construct a pseudo-antonym dictionary. In this approach, exchange ideas between individual and groups can rich in good text or vocabulary and share their thoughts as future memories.