Sentiment Analysis of Microblogs

In this project we attempt to perform sentiment based classification of Micro-blogs using Machine Learning techniques. Sentiment Analysis of short messages posted on Micro-blogging tools can be helpful in determining the current usability and acceptance of any target product or service. It can help in raising alarms in the wake of sudden shifts in user sentiments or attitude towards the service. We present a system that can read messages relevant to a particular topic from a micro-blogging service such as Twitter, analyze the messages for the sentiments they carry and classify them as neutral,positive or negative. We try out different feature selection and classification algorithms in our search for the best combination. We also evaluate different strategies in order to get the best performance over the imbalanced dataset caused by relatively infrequent changes in user sentiments and to understand the underlying nature of this problem.