Mood Swing Analyser: A Dynamic Sentiment Detection Approach

Abstract

This paper presents the mood swing analyzer— a novel dynamic sentiment analysis approach that determines the swings in the mood of its user by following a purely unsupervised machine learning technique. This approach uses an internal model to detect the polarity of the sentiments automatically and classify them into clusters based on K-means algorithm hence eradicating the need for normalization. In reaction to a high deviation in the users mood obtained the concept of appropriate message dropping has been proposed. Detailed algorithmic explanation along with the experimental results is well illustrated in this paper. This paper also discusses an extension of this approach in the real world to stop suicidal attempts due to cyber depression.