On contraction method to cleansing duplicates in graph

Abstract

With the quick development of the linked data, graph data explosion has become a challenging problem. Duplicates also exit in graph data. Duplicates detection is a hotspot in the study of heterogeneous data integration and information retrieval. However, attentions are seldom paid to duplicates cleansing after detecting. Due to the complexity and relevance, duplicates in graph could not remove immediately. It is more important need some special methods for duplicates cleansing. In this paper, we study the problem and give a solution, named on contraction method to cleansing graph duplicates. The proposed method introduces graph contraction for duplicates cleansing. According to the situations, we give different solutions. Experiments on publication datasets show that the proposed method efficient ensures the relationship and stability in the graph.