ABSTRACT
Homogeneous relational data consist of only one type of data objects. In the literature, a special case of homogeneous relational data clustering has been studied as the graph partitioning problem. However, the research on the general case is still limited. In this chapter, we propose a general model based on the graph approximation to learn relation-pattern-based cluster structures from a graph. The model generalizes the traditional graph partitioning approaches and is applicable to learning the various cluster structures.
