A Morphology Method for Determining the Number of Clusters Present in Spectral Co-clustering Documents and Words

A new algorithm for clustering documents and words simultaneously has recently been presented. As most spectral clustering algorithms, the prior knowledge of the number of clusters present is required. In this paper, we explore a method based on morphology for determining the number of clusters present in the given dataset for co-clustering documents and words. The proposed method employs some refined feature extraction techniques, which mainly include a VAT (Visual Assessment of Cluster Tendency) image representation of input matrix generated by spectral co-clustering documents and words, and the texture information obtained by filtering the VAT image. The number of clusters present in co-clustering documents and words is finally reported by computing the eigengap of gray-scale matrix of filtered image. Our experimental results show that the proposed method works well in practice.