Construction method of universal embedding framework of multi-semantic heterogeneous graph

Abstract

The invention discloses a method for constructing a universal embedding framework of a multi-semantic heterogeneous graph, which comprises the following steps of: 1, constructing a neighborhood exploration strategy alpha-exploration, and smoothly splicing two exploration strategies, namely, DFS and BFS, so as to adapt to different heterogeneous network structures; 2, based on alpha-exploration, constructing an HNSE model, wherein the HNSE model comprises an alpha-exploration neighborhood exploration layer, a multi-semantic learning layer and a node classification layer; and learning low-dimensional embedding of the nodes while heterogeneous information and semantic information of the nodes are reserved; 3, realizing a multi-layer HNSE model in a residual form, and connecting a full-connection output layer behind the multi-layer HNSE model; and 4, constructing three expansion strategies of the HNSE. According to the method, each vertex of the multi-semantic heterogeneous graph is embedded by aggregating adjacent / meta-path neighbor nodes of different types, and a node aggregation sampling strategy combining meta-path neighbors and direct neighbors is designed for the HNSE, so that a multi-head attention mechanism in the HNSE is guided, and capture of node multi-semantic information is improved by utilizing meta-paths.

Description

technical field [0001] The invention relates to the field of graph neural networks, in particular to a method for constructing a general embedding framework of multi-semantic heterogeneous graphs. Background technique [0002] The graph embedding work realizes applications such as node classification and link prediction on the topological graph by extracting the deep feature representation of the nodes in the graph. With the multi-modality of various network structures, the latest graph embedding methods have gradually abandoned homogeneous information. The network modeling method focuses on modeling these interconnected graph data as a heterogeneous information network composed of different types of nodes and edges, and uses the comprehensive structural information and rich semantic information in the network for more accurate knowledge discovery. . Compared with homogeneous networks, multiple types of objects and relationships coexist in heterogeneous networks, which cont...

AI Technical Summary

Problems solved by technology

[0005] (2) The multi-semantic description problem brought about by the rich information in the heterogeneous network

[0006] Multi-semantic heterogeneous graphs (Multi-Semantic Heterogeneous Graphs) are more complex than ordinary heterogeneous graphs. On the basis of having multiple edge attributes / node attributes, each node contains multiple semantic information at the same time. It can be seen that the different link information a node participates in is likely to represent the multiple semantics contained in the node, but this consideration has two shortcomings: First, this idea weakens one of the most basic elements of graph embedding work: we Work on a complex graph, not a set of links

[0008] The graph convolutional network propagates the structural information of the graph layer by layer by performing the neighborhood convolution operation, and frees the graph embedding method from the trouble of link-guided learning; but both focus on improving the basic graph neural network model, such as introducing an autoencoder, or Some progress has been made in improving the sampling work of nodes, but at the same time, it shows that meta-path is a difficult element to choose in heterogeneous graph representation learning. In general, meta-path splits the original structure of the graph. However, it greatly simplifies the description of multiple semantic information, which makes the current multi-semantic heterogeneous network embedding model unable to capture both graph structure and multiple semantics.

Images