Unsupervised Artificial Neural Nets for Modeling Movie Sentiment

Sentiment mining aims at extracting features on which users express their opinions in order to determine the user's sentiment towards the query object. Movie sentiment in Twitter provides an excellent base upon which to evaluate sentiment mining methodologies both because of the pervasiveness of discussions devoted to movie topics and because of the brevity of expression induced by twitter's 140 word limitation. In this paper we explore movie sentiment expressed in Twitter microblogs. A multi-knowledge based approach is proposed using, Self-Organizing Maps and movie knowledge in order to model opinion across a multi-dimensional sentiment space. We develop a visual model to express this taxonomy of sentiment vocabulary and then apply this model in test data. The results show the effectiveness of the proposed visualization in mining sentiment in the domain of Twitter tweets.