Multi-scale Attributed Node Embedding

TL;DR

This paper introduces multi-scale attributed node embedding algorithms that effectively capture local attribute distributions using random walks, improving performance on social and web graph tasks.

Contribution

It proposes a novel multi-scale embedding method (MUSAE) that encodes neighborhood information at different scales and provides theoretical insights into the matrix factorization underlying the embeddings.

Findings

Algorithms are robust and computationally efficient.

Outperform comparable models on social networks and web graphs.

Theoretical proof links embeddings to matrix factorization of pointwise mutual information.

Abstract

We present network embedding algorithms that capture information about a node from the local distribution over node attributes around it, as observed over random walks following an approach similar to Skip-gram. Observations from neighborhoods of different sizes are either pooled (AE) or encoded distinctly in a multi-scale approach (MUSAE). Capturing attribute-neighborhood relationships over multiple scales is useful for a diverse range of applications, including latent feature identification across disconnected networks with similar attributes. We prove theoretically that matrices of node-feature pointwise mutual information are implicitly factorized by the embeddings. Experiments show that our algorithms are robust, computationally efficient and outperform comparable models on social networks and web graphs.