| Currently, there are network graphs formulated by individual links in many fields such as social network, whose graph data is of great value. With the change of contact between individuals, dynamic graph is formulated. Dynamic graph window query is an important method to analyze the dynamic graph. Previous studies mainly focus on constructing the index of dynamic graph window, and the existing index methods still face the problem of query efficiency and update efficiency which to be improved. At the same time, with the size of the dynamic graph increasing, constructing the index presents a great challenge. In order to further improve query performance and index updating efficiency, it is necessary and significant to study the index problems of large-scale dynamic graph window query.Considering that query efficiency of the existing graph window indexes is not high, by constructing two indexes to calculate the share ratio between graph window queries to improve query efficiency, layered index is designed for large scale dynamic graph window, and the graph window query system is designed to support large-scale dynamic graph. Considering the problem of the efficiency of constructing the index, as the independent data is calculated serially in different stages, parallel calculating method is designed to construct layered index. Considering the index updating efficiency, through designing different types of rules to update the index, the methods of incremental updating index is designed to ensure the correctness and to improve the efficiency of index updates.Finally, the experiments are conducted about the function and the efficiency of large-scale dynamic graph window query system. Experiment results indicate that the layer index designed can support the large-scale dynamic graph window query. With respect to the existing index, the query efficiency is better. With the parallel solution of constructing index, the speed of constructing index is improved. With the incrementally updating index solution, the efficiency of updating index is improved. |
