Hypergraph parallel multilateral kernel maintenance method

Abstract

The invention discloses a hypergraph parallel multilateral graph core maintenance method, which comprises the following steps of S1, calculating core values of all nodes in a hypergraph, and taking the core value of the node with the minimum core value in a hyperedge as the core value of the hyperedge; s2, calculating NI (u) of the node u, wherein the NI (u) is a hyperedge set in which hyperedge core values in hyperedges where the node u is located are greater than or equal to the core value of the node u; s3, calculating the EI (e) of the hyperedge e, wherein the EI (e) is the minimum value of the NI of the node of which the core value of the node in the hyperedge is equal to the core value of the hyperedge; and S4, parallel processing of kernel maintenance increment or kernel maintenance decrement is carried out. The method has the advantage that when the edges are inserted and deleted in the graph, correct updating of the graph kernel is ensured.

Description

technical field [0001] The present application relates to the technical field of data mining, in particular to a graph kernel maintenance method for hypergraphs. Background technique [0002] A graph is a structured data that allows us to model complex relationships between objects. In recent years, more and more data can be modeled as graphs, and graph data is ubiquitous in social networks, biology, chemistry, and other fields. Graph kernel data mining is an important direction in graph analysis. Its purpose is to find the most connected set of points in the graph, and this can be used for things like targeted advertising, link prediction, and community detection. [0003] Existing methods for the maintenance of graph kernels for dynamic hypergraphs are mainly accomplished through kernel decomposition calculation, which is inefficient, wastes time, and performs useless calculations for most points. Or only approximate kernel maintenance of dynamic hypergraph can be maint...

AI Technical Summary

Problems solved by technology

[0003] The existing method for the maintenance of the graph kernel of the dynamic hypergraph is mainly completed through the kernel decomposition calculation. This method is inefficient and wastes time. For most points, it performs useless calculations.

Or can only maintain the approximate kernel maintenance of the dynamic hypergraph, and cannot get an accurate point

Or the existing algorithm can only perform unilateral dynamic kernel maintenance of the hypergraph, which has low efficiency and low practicability

Images