TACENR: Task-Agnostic Contrastive Explanations for Node Representations

TL;DR

TACENR introduces a task-agnostic contrastive explanation method for node representations in graphs, revealing attribute, proximity, and structural features that influence the learned embeddings.

Contribution

The paper presents TACENR, a novel contrastive learning-based approach for explaining node representations without relying on task-specific supervision.

Findings

Proximity and structural features significantly influence node representations.

TACENR effectively identifies impactful features in an unsupervised, task-agnostic manner.

Supervised variant of TACENR performs comparably to existing task-specific explainability methods.

Abstract

Graph representation learning has achieved notable success in encoding graph-structured data into latent vector spaces, enabling a wide range of downstream tasks. However, these node representations remain opaque and difficult to interpret. Existing explainability methods primarily focus on supervised settings or on explaining individual representation dimensions, leaving a critical gap in explaining the overall structure of node representations. In this paper, we propose TACENR (Task-Agnostic Contrastive Explanations for Node Representations), a local explanation method that identifies not only attribute features but also proximity and structural ones that contribute the most in the representation space. TACENR builds on contrastive learning, through which we learn a similarity function in the representation space, revealing which are the features that play an important role in the representation of a node. While our focus is on task-agnostic explanations, TACENR can be applied to supervised scenarios as well. Experimental results demonstrate that proximity and structural features play a significant role in shaping node representations and that our supervised variant performs comparably to existing task-specific approaches in identifying the most impactful features.