Research On Mining Method Of Maximal Motif Clique For Dynamic Heterogeneous Information Network

Heterogeneous information network is a graph that associates vertices with type labels,which is used to describe complex restriction semantics between different types of objects,such as geographic social networks and biological networks.Given the restriction relationship between different types of vertices,the maximal motif clique is a "holistic subgraph" that conforms to this restriction relationship.By discovering maximal motif cliques,it is possible to find closely related groups that meet specific restricted relationships on heterogeneous information networks.Considering that heterogeneous information networks are frequently updated in practical applications,and the existing maximal motif cliques mining algorithms do not support the efficient mining of maximal motif cliques on dynamic graphs,this paper studies the efficient mining of maximal motif cliques on dynamic graphs.The specific research content is as follows:First,a maximal motif clique algorithm UMMD that supports immediate update is proposed.The algorithm is mainly composed of edge addition strategy and edge reduction strategy.When adding and subtracting edges,first determine whether it can be terminated early.The basic idea is to check whether the motif induced subgraph has changed.If not,it means that the maximal motif clique does not change,then terminate early;otherwise,according to the newly designed update strategy,Only handle the maximal motif cliques related to the two endpoints,thereby narrowing the scope of the update and avoiding the re-enumeration operation.Secondly,based on the UMMD algorithm,an optimized and efficient algorithm UMMD+ is proposed.Compared with UMMD,the improvement of UMMD+ is reflected in: the use of optimized inverted index reduces the redundant calculation of UMMD and avoids the problem of unsynchronized update of ordinary inverted index.By accessing the inverted table,we can quickly find the object to be processed in the process of updating a maximal motif clique,which speeds up query processing and improves update efficiency.Finally,experiments are conducted based on 20 real heterogeneous information network data sets.The experimental results show that the algorithm proposed in this paper can efficiently support the mining of maximal motif cliques on dynamic heterogeneous information networks.