Nowadays, data mining is widely used in almost every field to discover hidden information in large amounts of data that is being continuously generated. For any recommendation system to work accurately, it is important to build profile of the user which is done which is done by extracting personal profiles from various sources such as watched movies and the rating given, purchased items, items liked by users, etc. Movie choice is something very personal and hence it is difficult to accurately recommend movies to user. Current movie recommendation systems use collaborative and content based filtering. This method aims to investigate a different method to build personal profiles using information obtained from Twitter to provide personalized movie recommendation service. For a Twitter user, our method utilizes tweets of the user from which important keywords are extracted using Sentiment Analysis and TF-IDF (Term Frequency - Inverse Document frequency) to build a personal profile. The usefulness of this method is validated by implementing a prototype movie recommendation service and by performing a user study. Using TF-IDF, mapping algorithm and finding users preference of genre, we finally get a score for each movie in the database. We recommend trending movies if the user hasn’t tweeted anything before or his tweets aren’t relevant for recommendation. This alleviates the cold start problem  . The prediction accuracy of movie recommendation is predicted against a small group of users.