Regularized bi-directional co-clustering

The simultaneous clustering of documents and words, known as co-clustering, has proved to be more effective than one-sided in dealing with sparse high-dimensional datasets. By their nature, text data are also generally unbalanced directional. Recently, the von Mises–Fisher (vMF) mixture model was proposed handle while harnessing directional nature text. In this paper, we propose a general co-clustering framework based on matrix formulation vMF model-based co-clustering. This leads flexible for that can easily incorporate both word–word semantic relationships document–document similarities. contrast existing methods, which use an additive incorporation similarities, bi-directional multiplicative regularization better encapsulates underlying structure. Extensive evaluations various real-world datasets demonstrate superior performance our approach over baseline competitive terms results co-cluster topic coherence.