Heterogeneous information network embedding method and system, terminal and storage medium

Abstract

The invention discloses a heterogeneous information network embedding method and system, a terminal and a storage medium. The method comprises the following steps: obtaining multiple types of objectsand connection information, and establishing a heterogeneous information network; establishing a meta-path through the object or connection type and the relationship between the object or connection type and the object or connection type; a high-order Markov chain is defined through any element path, and random walk is guided through the high-order Markov chain; and guiding the embedding of the heterogeneous information network through the random walk distribution. The problems of non-stationary random walk and meta-path limitation in a heterogeneous information network embedding method can besolved.

Description

technical field [0001] The invention relates to the field of heterogeneous information networks, in particular to a method, system, terminal and storage medium for embedding heterogeneous information networks. Background technique [0002] In practical applications, many networks, such as modern social media, academic networks, patient and drug supervision networks, and knowledge graphs, often require the use of heterogeneous information networks (Heterogeneous information network, HIN) to represent these networks. If we use the network embedding method to extract the semantics of different meta-paths, we will be able to perform many downstream analyzes of specific applications of heterogeneous information networks, such as link prediction or recommendation, object classification, similarity retrieval, etc. in heterogeneous information networks. [0003] Currently, meta-path-driven object embedding models have been developed inspired by the DeepWalk algorithm in homogeneous...

AI Technical Summary

Problems solved by technology

The above method has many problems. For example, the first-order Markov chain used to guide the random walk path may cause non-stationary random walks in heterogeneous information networks; secondly, although it states that random walks can be applied to arbitrary path, but in fact only the meta path of the object type with the same beginning and end can be realized

Images