Attribute graph community search method and system based on representation learning

Abstract

The invention relates to an attribute graph community search method and system based on representation learning. According to the method, the method comprises the steps of adopting a query node as a starting point of random walk, adopting the attribute information and structure information of the node for guiding jump of the random walk, and obtaining a node sequence and an attribute keyword sequence; performing node representation learning on the node sequence and the attribute keyword sequence to obtain topological representation and attribute representation of nodes, and combining the topological representation and the attribute representation as node representation information; reconstructing the original graph data according to the similarity between the nodes to obtain a reconstructed graph; establishing a community model based on the reconstructed graph and the node representation information; and searching a target community meeting requirements by taking the established community model as guidance. According to the method, the calculation scale is reduced, the representation learning process is more accordant with the characteristics of local characteristics concerned by community search, the information of the nodes on the community level is fused, the node feature mining accuracy is improved, and the established community model gives consideration to the interpretability, quality and efficiency of community discovery.

Description

technical field [0001] The invention belongs to the field of information technology, and in particular relates to a method and system for searching communities in attribute graphs based on representation learning. Background technique [0002] Many complex information systems in the real world can be modeled in the form of graphs (or networks), such as social networks. Compared with traditional modeling methods, data organized in a graph structure can intuitively describe the potential connections between entities, and the content is richer. With the rapid development of information technology, the scale of data to be processed is also increasing rapidly. How to efficiently mine the data that users are interested in on the graph has become very important. Community search provides a new idea for efficient and personalized map data mining. It can return the community structure containing the query node according to the query node given by the user, that is, a set of data clo...

AI Technical Summary

Problems solved by technology

However, in practical applications, graph data often changes over time, and the global learning method is not suitable for dynamic graphs; at the same time, for community search tasks, it is usually only necessary to focus on the local area near the query node, and the global The learning method will generate a lot of redundant information and reduce the overall efficiency of community search

In addition, the existing methods treat all nodes equally in the representation learning stage, and do not consider the information of nodes at the community level, which will lead to nodes having different status and weights.

Images