Conduct a study involving sentiment analysis/opinion mining of social networking or microblogging posts. For example, you could determine how to track a product or public figure’s popularity over time, and possibly also correlate this to real-world events reported in the news media or blogosphere. This project has great potential for nifty visualizations.
For this project you will need access to a corpus of social networking or microblogging posts. You could construct this yourself (using, for example, the Twitter API, or producing your own Facebook app) or you might be able to use one of the publicly available Twitter or social media corpora.
- Florian: Fine-tuning sentiment model.
- Raphael: Crawling reddit.
- Soner: Crawling alternatives.
- Leo: Fine-tuning non-sentiment model.