The joint sentiment-topic (JST) model is a weakly-supervised hierarchical Bayesian model for detecting document-level sentiment and extracting sentiment-bearing topics from text simultaneously. The only supervision used for JST learning is a small set of domain-independent sentiment clues; no labelled documents are used.
The JST source code (written in C++) can be downloaded here.
Lin, C., He, Y., Everson, R. and Rueger, S. (2011) Weakly-Supervised Joint Sentiment-Topic Detection from Text, IEEE Transactions on Knowledge and Data Engineering (TDKE).
Lin, C. and He, Y. (2009) Joint Sentiment/Topic Model for Sentiment Analysis, The 18th ACM Conference on Information and Knowledge Management (CIKM), Hong Kong, China.