Dynamic link prediction depth model based on multiple view and application

Abstract

The invention discloses a dynamic link prediction depth model based on multiple views and an application, and the dynamic link prediction depth model comprises a pooling module which is used for classifying and aggregating nodes with similar features and attributes in an initial network into new nodes to form a new network, and achieving the dimension reduction of the initial network; the input of the GCN-attention module is connected with the output of the pooling module, and a GCN-attention module, used for carrying out feature extraction on the input network by adopting an attention mechanism to obtain a node representation vector with spatial information and time information at the same time; a depooling module, and the input of the depooling module being connected with the output of the GCN-attention module, used for depooling the input node representation vector and outputting a dynamic link probability prediction matrix. The model can realize link prediction of the dynamic network. The model is mainly applied to prediction of dynamic links in a social network, a communication network, a traffic network, a scientific cooperation network or a social security network.

Description

technical field [0001] The invention belongs to the field of network science, and in particular relates to a multi-view-based dynamic link prediction depth model and its application. Background technique [0002] Dynamic networks are ubiquitous in practical applications, including social networks, transportation networks, industrial systems, and biology. The structure of these networks changes over time, resulting in the addition or deletion of nodes or links, and the prediction of the link status at a future moment is called dynamic link prediction. [0003] Compared with static networks, dynamic networks not only need to consider the spatial structure characteristics of the network at each moment, but also consider the temporal characteristics of the network evolution process at previous moments, in order to more accurately predict the state of the network structure in the future. In recent years, methods for extracting spatial representations of nodes via graph neural ne...

AI Technical Summary

Problems solved by technology

Since the network status at each moment has a certain correlation in time, it is more scientific and reliable to predict the link in the future according to the network status at multiple previous moments, Xiaoyi Li et al. , etal.A deep learning approach to link prediction in dynamic networks[C]//Proceedings of the 2014

Images