Heterogeneous Information Network Embedding for Meta Path based Proximity

TL;DR

This paper introduces a method for embedding heterogeneous information networks that captures meta path based proximity, enabling better representation of complex, multi-typed graphs for search and mining tasks.

Contribution

It proposes a novel embedding approach specifically designed for heterogeneous networks, addressing the limitations of existing methods that focus only on homogeneous networks.

Findings

Effective preservation of meta path based proximity in embeddings

Improved performance in network search and mining tasks

Demonstrated applicability to complex heterogeneous networks

Abstract

A network embedding is a representation of a large graph in a low-dimensional space, where vertices are modeled as vectors. The objective of a good embedding is to preserve the proximity between vertices in the original graph. This way, typical search and mining methods can be applied in the embedded space with the help of off-the-shelf multidimensional indexing approaches. Existing network embedding techniques focus on homogeneous networks, where all vertices are considered to belong to a single class.