关键词: 基于Apriori的图挖掘算法, 子图同构, 图结构数据, 三维坐标, 生理活性

Abstract: Apriori-based Graph Mining （AGM） algorithm is simple and it is based on recursion statistics. When graph data set is very large, due to sub-graph isomorphism problem, so many redundant sub-graphs would be generated while generating candidate sub-graphs, which increases computation time. In this paper, an improved method was proposed to reduce redundant sub-graph candidates by adding extra constraints and use three-dimensional coordinate to calculate the distance between each vertex of a graph, which was added to the edge label for handling three-dimensional graph structured data. In this paper, chemical compounds were analyzed by the improved algorithm to describe their three-dimensional chemical structure and correlation with physiological activity and the computation time on different conditions were examined. The experimental results prove that the improved algorithm cuts down the computation time with more edge labels and improves the efficiency.

Key words: Apriori-based Graph Mining （AGM） algorithm, sub-graph isomorphism, graph structured data, three-dimensional coordinate, physiological activity