Learning Graph Embeddings from WordNet-based Similarity Measures

TL;DR

path2vec is a novel graph embedding method that learns dense representations approximating user-defined similarity measures, excelling in semantic tasks and outperforming existing baselines in efficiency and accuracy.

Contribution

The paper introduces path2vec, a new method for learning graph embeddings that approximate various similarity measures, including those beyond simple graph structure.

Findings

Outperforms strong graph embedding baselines in semantic similarity tasks

Computationally efficient, significantly faster than direct distance computations

Yields competitive results in word sense disambiguation

Abstract

We present path2vec, a new approach for learning graph embeddings that relies on structural measures of pairwise node similarities. The model learns representations for nodes in a dense space that approximate a given user-defined graph distance measure, such as e.g. the shortest path distance or distance measures that take information beyond the graph structure into account. Evaluation of the proposed model on semantic similarity and word sense disambiguation tasks, using various WordNet-based similarity measures, show that our approach yields competitive results, outperforming strong graph embedding baselines. The model is computationally efficient, being orders of magnitude faster than the direct computation of graph-based distances.