Author migration classification method for scientific cooperation heterogeneous network

Abstract

The invention discloses an author migration classification method for a scientific cooperation heterogeneous network, and the method comprises the following steps of: acquiring scientific cooperation data, and constructing a heterogeneous scientific cooperation network; classifying each conference in the heterogeneous scientific cooperation network according to themes, and grouping each author according to the conference category to which the paper published by the author belongs; carrying out neighbor node sampling on the heterogeneous scientific cooperation network; conducting neighbor information aggregation on the sampled neighbor nodes; carrying out training and learning on the heterogeneous scientific cooperation network to obtain heterogeneous network representation; training an author classifier using the training dataset representing vectors and categories by the author; and classifying the trained author classifier for unknown authors. According to the method, sampling of direct neighbor nodes and indirect neighbor nodes is realized based on meta-paths, and meanwhile, migration classification of authors has a relatively good effect by adopting multiple information modes.

Description

technical field [0001] The invention belongs to the field of social network science and technology, and relates to an author migration classification method for heterogeneous networks of scientific cooperation. Background technique [0002] In the real world, connections between individuals are ubiquitous, and these complex connections can be described by different forms of networks (social networks, citation networks, protein molecular interaction networks, power networks, etc.). As a common data carrier, data in the form of the network exists in all aspects of society. Deep mining of meaningful information in the network has very high academic value and potential application value. Although the structures and forms of various networks are complex and diverse, their basic components are nodes and edges. Therefore, the analysis and research on complex networks has universal significance. Since the network contains a large number of nodes and edges, it takes a lot of time to...

AI Technical Summary

Problems solved by technology

However, in the current network representation research, there is still a key problem: how to aggregate the information of the neighbors of the target node together? Most graph neural networks define the neighbors of a node as its directly connected nodes, and iteratively aggregate the neighbor information with a distance greater than 2 through multi-layer nesting

Images