User Engagement Prediction Using Tweets

Abstract

People are spending huge amount of time on social media platforms these days. Through this they get engage in various real-world activities, either for awareness or just for participation. Twitter has 330 million active users, who generates around 6,000 tweet per second. This forms a huge corpus of data that is widely available for analysis, monitoring and research. Different forms of user behavior can be studied with this data. Analysis in this paper shows how simple machine learning and natural language processing techniques can be used to predict user interests based on his/her past tweets. The paper proposes to use a keyword extraction and semantic clustering based approach to do the analysis. The proposed approach has been tested on a dataset of 1,69,000 tweets and has achieved an accuracy of 80%.