Integrating Entity and Attribute for Object Similarity

In this paper, we propose an object similarity algorithm (OSA) to address the problem of similarity calculation in the complex graph network by integrating the entity and attribute similarity of the graph node. The proposed method can solve the similarity judgment of the objects that are lack of links or missing attributes by accident or intention. Experimental results have showed that our proposed object similarity method has the advantage of assessing the similarity of objects comprehensively, which can correct the one-sided judgment errors and improves the ability to distinguish similarity of graph objects; the method can also be used for homogeneous or heterogeneous networks.