Link prediction method based on heterogeneous network representation learning

Abstract

The invention discloses a link prediction method based on heterogeneous network representation learning, and belongs to the field of network analysis. The method includes: capturing the structure andsemantics of the heterogeneous network through random walk based on a meta-path, and generating a training sequence; training a skip-gram model by using the training sequence so as to learn network features, and mapping network nodes to a low-dimensional vector; obtaining a link prediction result by processing the similarity of the node vectors; according to the invention, the problem that most ofcurrent link prediction algorithms cannot be applied to the heterogeneous network is solved, and rich information contained in different nodes of the heterogeneous network can be obtained. Most prediction is based on neighborhood local information and structural features, the requirement for the aggregation coefficient of the network is high, the sparse network prediction effect is poor, random walk based on meta-paths can be used for sparse networks, node vectors are established in a unified mode through overall learning of the networks, and calculation is simplified.

Description

technical field [0001] The invention belongs to the field of network analysis and relates to a link prediction method, in particular to a link prediction method based on heterogeneous network representation learning. Background technique [0002] There are a large number of complex systems in the real world that can be abstracted into the form of networks, so the research on networks has become a research hotspot in many disciplines and fields in recent years. In the era of big data, with the emergence of a large number of social networks and information networks, the research on link prediction is closely related to the structure and evolution of the network. At the same time, the study of link prediction can also theoretically help to understand the mechanism of complex network evolution. Therefore, link prediction has become an important direction of data mining research. [0003] But there are some problems in the current research on link prediction. On the one hand, ...

AI Technical Summary

Problems solved by technology

On the other hand, most link prediction methods are based on local information and structural features of the neighborhood, which have high requirements on the aggregation coefficient of the network, poor prediction effect on sparse networks, and high complexity, which is not friendly to large-scale networks.

Images