Understand relationships between stock movement and social media sentiment in the company's Twitter posts is appealing and catching attentions of both academic and industrial researchers  nowaday. One of the factors preventing their analysis from great accuracy is the quality of the text data and how to mine it. Our team want to examine some of current models (Logistics, Naïve Bayes and CNN) to mine text data, specifically for enterprises’ tweets. Our research question is: To mine data from companies’ tweets in Twitter, which one is the best method to build a tweet sentiment analysis model with limited supervised data?