An innovative six-class classification model and web application in Python that is able to predict sentiments of real-time tweets
around you based on keyword searched.
Spring 2013
Skills: Python, Django, Twitter Streaming API, machine learning, data mining
Lexicon: Bingliu's List, MPQA, FrameNet, WordNet, list in Emotion Intelligence
Course: CIS630, Advanced Natural Language Processing, Spring 2013, University of Pennsylvania
Teamwork: Yayang Tian, Tao Feng, Chun Chen
- Tweets Corpus: Presented a method for automatically collecting recent tweets with different emotions. Created a large tweets corpus consisting of five emotions: ”happy, sad, angry, afraid, ashamed”.
- Six-class Classification Model: Adopted various methods for tweets affect classification and outperform baseline approach by 21.197%. Conducted experimental evaluations on real-time Tweets and showed the importance for stemming, affect dictionary, smiley, information gain, and SVM in five-class classification.
- Web Application: Build a web application that can classify and summarize emotions on Twitter in real-time based on user-specified keyword.