Research Of Graph Query On Large Graph Database

Graph,as a general data structure,is able to show the entity’s label and relation. An increasing number of data applications have started to use graph to express,such as social networking,knowledge graphs and genome database etc.Traditional graph query method most relies on subgraph isomorphism and it restrict the scale of graph query.G raph query starts off by transferring graph match based on subgraph isomorphism to graph match based on neighborhood information similarity,so that subgraph isomorphism and graph edit distance could be avoided.This method would adapt to incomplete graph structure and noise and it could increase the level of user-friendliness.Combination of the above problems, futher information on the matching algorithm based on neighbors were studied.Firstly, In order to solve the problem that repetitively calculate the label vector model in Ness.We propose a pruning theory that can quickly remove the node in candidate set basing on neighborhood’s label vector modle. Base on the pruning theory we propose a pollute information propagative model to improved the Ness.We improve the neighborhood’s label vector to add the node mark.It accelerate pruning candidate node in each iterator and improve the Ness’ s efficiencySecondly, we put the pollute information model proposed into the Ne Ma algorithm on neighborhood vector model and transform the model to adapt the Ne Ma. In order to utilize the pollute information on Ne Ma, we use a memorization technique to preserve the target graph node that need to calculate the matching cost. It reduce the iterative computations times and accelerate the propagative matching cost and improve the efficiency of Ne Ma.Finally, it verified effectiveness and efficient of improved NESS and Ne Ma algorithm by experiments on real-word network.