A Comparative Study of Twitter Sentiment Analysis Methods for Live Applications
Angel Cambero · RIT Scholar Works (Rochester Institute of Technology) · 2016
Sentiment Analysis is growing exponentially due to the importance of the automation in mining, extracting and processing information in order to determine the general opinion of a person. The problem that this thesis proposes to address is to determine what methods are more suitable to extract subjective impressions in real time from Twitter. For live applications, since the opinions collected from Twitter are limited to certain amount of characters and it will happen in a real-time environment, this provides an interesting scenario; we will test using both the Machine Learning Approach and the Lexicon-based Approach, and then combine them in an effort to increase the accuracy. In order to test the real-time factor, I will implement a web service with the purpose of collecting real-time feedback from Twitter in real-time, which will be later processed and analyzed for accuracy and realtime performance.