Self-Explainable Graph Transformer for Link Sign Prediction
Lu Li, Jiale Liu, Xingyu Ji, Maojun Wang, Zeyu Zhang
TL;DR
This work introduces SE-SGformer, a self-explainable Transformer-based model for signed graphs aimed at link sign prediction. It fuses three encodings—centrality, adjacency, and signed random walk—with a Transformer encoder to learn node representations, and couples this with an explainable prediction module that identifies $K$-nearest positive and $K$-farthest negative neighbors using an SRWR-derived diffusion matrix to generate human-interpretable explanations. The authors prove that the signed random walk encoding is more expressive than shortest-path encoding and that the overall Transformer with signed random walk encoding is more expressive than SGNN baselines, and they validate these claims with experiments on eight real-world datasets, achieving notable gains in both prediction and explainability. The approach advances practical trust in SGNNs by coupling accurate edge-sign predictions with explanation signals that highlight influential neighbor relationships, and it lays groundwork for further exploration of self-explainable graph architectures in signed domains.
Abstract
Signed Graph Neural Networks (SGNNs) have been shown to be effective in analyzing complex patterns in real-world situations where positive and negative links coexist. However, SGNN models suffer from poor explainability, which limit their adoptions in critical scenarios that require understanding the rationale behind predictions. To the best of our knowledge, there is currently no research work on the explainability of the SGNN models. Our goal is to address the explainability of decision-making for the downstream task of link sign prediction specific to signed graph neural networks. Since post-hoc explanations are not derived directly from the models, they may be biased and misrepresent the true explanations. Therefore, in this paper we introduce a Self-Explainable Signed Graph transformer (SE-SGformer) framework, which can not only outputs explainable information while ensuring high prediction accuracy. Specifically, We propose a new Transformer architecture for signed graphs and theoretically demonstrate that using positional encoding based on signed random walks has greater expressive power than current SGNN methods and other positional encoding graph Transformer-based approaches. We constructs a novel explainable decision process by discovering the $K$-nearest (farthest) positive (negative) neighbors of a node to replace the neural network-based decoder for predicting edge signs. These $K$ positive (negative) neighbors represent crucial information about the formation of positive (negative) edges between nodes and thus can serve as important explanatory information in the decision-making process. We conducted experiments on several real-world datasets to validate the effectiveness of SE-SGformer, which outperforms the state-of-the-art methods by improving 2.2\% prediction accuracy and 73.1\% explainablity accuracy in the best-case scenario.