On the Interpretability and Evaluation of Graph Representation Learning

TL;DR

This paper investigates interpretability in graph representation learning, proposing methods and a framework to evaluate how well embeddings capture graph structures, with experiments across various graph types and parameters.

Contribution

It introduces new interpretability methods and a comprehensive evaluation framework for comparing graph embedding algorithms and hyperparameters.

Findings

Embedding training parameters influence structural recovery

Proposed methods improve understanding of encoded information

Framework enables systematic comparison of algorithms

Abstract

With the rising interest in graph representation learning, a variety of approaches have been proposed to effectively capture a graph's properties. While these approaches have improved performance in graph machine learning tasks compared to traditional graph techniques, they are still perceived as techniques with limited insight into the information encoded in these representations. In this work, we explore methods to interpret node embeddings and propose the creation of a robust evaluation framework for comparing graph representation learning algorithms and hyperparameters. We test our methods on graphs with different properties and investigate the relationship between embedding training parameters and the ability of the produced embedding to recover the structure of the original graph in a downstream task.