Feature oriented sentiment analysis in social networking sites to track malicious campaigners

Social networking websites are considered as major sources of opinions and views of the public on the prevalent social issues at a given point in time. Websites like the Twitter¹ reflect the public views through its millions of messages posted by its users world wide, whenever a controversial issue arises in the society. It is during this time that we observe significant amount of malicious, violent contents going viral over the internet. In this paper we propose a technique that applies sentiment analysis on data from Twitter and measures the sentiments of posts in order to identify the origin of malicious contents. This is achieved by taking into account the influence of the posts on the public as well. The prominent feature of our work is the technique that is used for feature-oriented sentiment analysis. This involves an algorithm that parses a given tweet and builds a Dependency tree of each sentence in order to effectively identify the sentiment of the tweet. The working and the scope of our techniques are illustrated with a case study and associated results.