A graph neural network representation learning-based structure graph alignment method and a multi-graph joint data mining method

Abstract

The invention relates to a graph neural network representation learning-based structure graph alignment method and a multi-graph joint data mining method. The method comprises the following steps: performing sub-graph sampling on a graph in training data; learning low-dimensional vector representation of nodes in the sub-graphs by using a graph neural network through marked aligned node pairs; calculating the similarity among the nodes according to the low-dimensional vector representation of the nodes in the sub-graphs, aligning the graphs based on the similarity, and finally obtaining a graph neural network with trained parameters; in the speculation stage, obtaining low-dimensional vector representation of each node of two to-be-aligned graphs through the trained graph neural network, then calculating the similarity between the nodes, aligning the two graphs on the basis of the similarity, and then performing joint data mining through aligned multi-graph data. According to the invention, under the supervised setting, the expression performance of the model, the loss function setting, the spatial constraint of the representation vector and the expandability are considered, and the improvement of the existing method is realized.

Description

Technical field [0001] The present invention belongs to the field of information technology, and more particularly to a structural diagram alignment method and a multi-graph combination data mining method based on the study of the map neural network. Background technique [0002] The diagram data describes the relationship between things in the form of points and edges, which can well describe things in the objective world, which can be seen everywhere in production and life, such as social networks, knowledge maps, e-commerce, etc. As the data scale has grown rapidly in recent years, many objective entities cannot be expressed in the same figure, and we often need to tide the links in multiple pictures. Therefore, the graph pair with a necessary preprocessing step for multi-graph combined data mining, has a very wide application and demand. For example, many people today may have an account in multiple social networks. If you want to integrate information about multiple social n...

AI Technical Summary

Problems solved by technology

The problem with this method is that the modeling is too simple, the expression ability is insufficient, and it is not conducive to extending to very large-scale graph data.

[0005] From the perspective of algorithm scalability, since the graph alignment algorithm usually needs to calculate the similarity between two graph nodes, it needs the storage and time complexity of the graph scale square level

Therefore, in the case of limited video memory, if a model is trained on the entire image, it is difficult to ensure the scalability of the algorithm in terms of time and space.

From the perspective of model effect, the existing methods have room for optimization in the model architecture and the loss function that guides the training, but are lacking in expressive ability, resulting in insufficient performance of the model, and the existing methods do not take into account the two graphs. Consistency in Mapped Spaces

Images