A NOVEL TECHNIQUE OF SENTIMENT ANALYSIS BY SANDWICH OF DATA MINING AND IMAGE PROCESSING

Abstract: When a text is accessible in a digital format, it is a task in the field of text mining to accurately predict the sentiment of a text or the intention with which the text was written. In this paper, we are trying to develop a system that will analyse the intention with which a particular text is written. Sentiment analysis becomes a tedious job in the domain of text mining as the problem of polysemy (one word with different meanings) and synonymy (many words with the same meaning) arises frequently. Our intention is to take this process one step ahead by analysing and verifying the sentiment using image processing in order to identify the intention with which a sender is sending a particular text message. By processing the image, it will help to improve the accuracy of sentiment analysis as the facial expression is captured while the sender is sending the message. The facial expression will account for the mood of the sender at that instance which enhances the sentiment analysis. This system will give a report of all the activities of the sender and will eventually detect the mood of the sender just with the help of the messages that he has sent.

Keywords: Data mining, Image processing, Emotion, Sentiment analysis, Natural Language Processing (NLP), ADAboost, Region of Interest (ROI