A Technique to Handle Negation in Sentiment Analysis on Movie Reviews

Day by day there is seen large growth in sentiment rich social and electronic media. Researchers have increasing interest in vast amount of user data generated such as comments, customer reviews and opinions which can be used to extract valuable information with the help of sentiment analysis and opinion mining. The negation modifiers make the Sentiment Classification approaches suffer and they can completely distort the meaning of the discourse. Hence it becomes mandatory to handle them effectively. Our work provides an approach for the identification and handling negation in unstructured data. The paper proposes and evaluates better modified approach for negation identification in sentiment analysis than the existing identification methods. The provided data which is stored in document is fed into a vector and if the data has negation, it is treated exceptionally. Both syntactic and morphologic negation is handled using dependency parse tree and prefix algorithm respectively. We have attainted an accuracy of 92% from our work.