On Gradient Weighting Based Fuzzy Clustering for Graph Data

This paper introduce a gradient weighting based fuzzy clustering algorithm for graph data, in which the clustering process can be regarded as an optimization for objective function. During the process of iteration, the partition matrix is updated by a convex combination of partition information with respect to attribute information and the closeness information between partition information and relational information. On these bases, the iteration process is constructed with the help of fuzzy c-means clustering algorithm. Moreover, its validity is illustrated by a real graph data—Books about US politics, not only in cluster validity indices aspect but also in runtime aspect.