AttitudeMiner: Mining Attitude from Online Discussions

This demonstration presents AttitudeMiner, a system for mining attitude from online discussions. AttitudeMiner uses linguistic techniques to analyze the text exchanged between participants of online discussion threads at different levels of granularity: the word level, the sentence level, the post level, and the thread level. The goal of this analysis is to identify the polarity of the attitude the discussants carry towards one another. Attitude predictions are used to construct a signed network representation of the discussion thread. In this network, each discussant is represented by a node. An edge connects two discussants if they exchanged posts. The sign (positive or negative) of the edge is set based on the polarity of the attitude identified in the text associated with the edge. The system can be used in different applications such as: word polarity identification, identifying attitudinal sentences and their signs, signed social network extraction from text, subgroup detect in discussion. The system is publicly available for download and has an online demonstration at http://clair.eecs.umich.edu/AttitudeMiner/.