Discovering Functional Communities in Social Media

We propose an approach for discovering functional communities in social media by identifying groups of users who interact with similar content, represented as dense biclusters in a user-content matrix. We present a heuristic algorithm to efficiently search the space of possible co-clusterings for one which maximizes the value of a given metric, along with a new class of co-clustering metrics that are more suitable for this task than existing metrics. We evaluate our approach using synthetic and real-world datasets.