Abstract:
Online Social Media now become a part of human life. Twitter is the most famous micro blogging
site that is used by major portion of crowd in the world. This site provides us the opportunity to
interact with more number of people in less time. Using the twitter data we want to get some
useful knowledge that helps humanity. In this thesis, we try to get the interest of users of
certain location based on tweets on certain topics (like entertainment, politics, sports, technology,
business, etc.). The motivation for this work is to help users to recommending things more
accurately. We first analyze the sentiments of tweets and then classify the tweets according to
topics. Using any one of the algorithm does not provide us good interest result and hence we
combine both the methods. By combining these, we are able to get the topic in which users are
positively and negatively interested. We have done this experiment on million of tweets which we
have collected for thousands of users and show their interest graph. The results are satisfactory
and it validates the proposed approach.
