EVENT BASED SENTENCE LEVEL INTERPRETATION OF SENTIMENT VARIATION IN TWITTER DATA

Twitter is one of the most popular micro blogging sites used by people to express their opinions. Text mining is the area where automatically data is mined for extracting features etc for different purposes. Interpretation of public opinion in micro blogging site is a challenging problem since it has noise data and other unnecessary tweets. The current systems focus on removing these challenges along with the sentiment extraction and modeling. Also the existing system focus on topic related extraction. We move ahead to the sentence level extraction with the help of existing methods. In this paper we propose a combination of enhanced RCB-LDA method, NLP, event based analysis and text summarization. RCB-LDA is used to automatically extract the sentiments within a variation period. NLP is used for finding the meaning of sentiments in the tweets. Event based analysis analyzes the sentiment related to each other by using text summarization. Event based analysis group the sentiment together to relate each other by summarizing tweets. Finally a candidate is assigned to which related ones are combined together so that it will be the most important reason behind the sentiment variation.