A Frequent Subgraph Mining Method Based on Graphics Processor Parallel Computing

Abstract

The invention discloses a frequent subgraph mining method based on the parallel computing of a graphics processor, divides each thread block block through the GPU, evenly distributes the frequent edges to different threads for parallel processing, and obtains different extensions through the rightmost extension The subgraph returns the graph mining data set obtained by each thread to each thread block; finally, data communication is carried out between the GPU and the memory, and the result is returned to the CPU for processing. The invention provides a feasible and effective graph mining method, which optimizes the performance of graph mining in an intensive big data environment, improves the efficiency of graph mining, provides fast and reliable data information for scientific research analysis, market research, etc., and realizes A Parallel Mining Method on the CUDA Unified Computing Device Architecture.

Description

technical field [0001] The invention relates to a frequent subgraph mining method based on parallel computing of a graphics processor, so as to improve the efficiency of graph data mining. Background technique [0002] With the continuous deepening and development of the field of data mining, graph data mining has attracted more and more attention from researchers, so graph mining has become a new research direction of data mining and machine learning. This research has great potential value in many fields in real life, such as protein structure analysis in bioinformatics, genome identification, connection between entities in social networks, Web content mining in Web analysis, Web link structure analysis, text information retrieval, etc. [0003] At present, the research work on graph data mining at home and abroad can be mainly divided into the following four categories: 1. Research on graph matching; 2. Research on keyword query in graph data; 3. Research on frequent sub...

AI Technical Summary

Problems solved by technology

There are several drawbacks to this traditional mining method: first, the mining process requires complex isomorphism verification, and the subgraph isomorphism problem is actually an NP-complete problem, and its calculation is complex and computationally intensive; second, in the mining process , a large number of repeated calculations will be performed, wasting resources; third, due to the huge amount of data, the load of the single CPU processor platform is too large, and the internal storage unit of the CPU is not enough to use

Images