VERSE: Versatile Graph Embeddings from Similarity Measures

TL;DR

VERSE is a versatile graph embedding method that explicitly preserves vertex similarity distributions, outperforming existing techniques in accuracy, efficiency, and scalability across various data mining tasks.

Contribution

The paper introduces VERSE, a novel graph embedding approach that explicitly incorporates similarity measures, offering a simple, memory-efficient, and scalable solution with superior performance.

Findings

Outperforms state-of-the-art methods in link prediction and classification.

Achieves better precision and recall on benchmark datasets.

Demonstrates superior time and space efficiency, especially with the scalable variant.

Abstract

Embedding a web-scale information network into a low-dimensional vector space facilitates tasks such as link prediction, classification, and visualization. Past research has addressed the problem of extracting such embeddings by adopting methods from words to graphs, without defining a clearly comprehensible graph-related objective. Yet, as we show, the objectives used in past works implicitly utilize similarity measures among graph nodes. In this paper, we carry the similarity orientation of previous works to its logical conclusion; we propose VERtex Similarity Embeddings (VERSE), a simple, versatile, and memory-efficient method that derives graph embeddings explicitly calibrated to preserve the distributions of a selected vertex-to-vertex similarity measure. VERSE learns such embeddings by training a single-layer neural network. While its default, scalable version does so via sampling similarity information, we also develop a variant using the full information per vertex. Our experimental study on standard benchmarks and real-world datasets demonstrates that VERSE, instantiated with diverse similarity measures, outperforms state-of-the-art methods in terms of precision and recall in major data mining tasks and supersedes them in time and space efficiency, while the scalable sampling-based variant achieves equally good results as the non-scalable full variant.