Aug 2026· Proceedings of the 32nd ACM SIGKDD Conference on Knowledge Discovery and Data Mining V.2· pp. 3424-3435· 0 citations· 12 references
Abstract
Interpreting predictions of temporal graph neural networks (TGNNs) is challenging: structural patterns are entangled with temporal dynamics and node-level activity. Existing methods often overemphasize frequent or recent interactions, producing explanations that conflate structural influence with temporal or behavioral confounding and are not grounded by motif-level evidence. We propose CATGX, a confounder-aware framework for explaining temporal graph predictions through motif-level reasoning. CATGX models temporal interaction mechanisms as motif occurrences, and explicitly treats temporal context and entity activity as observed confounders. By abstracting causal factors into temporal motif, context, and entity codebooks, CATGX applies an adjustment-inspired scoring scheme that compares motif-level influence to isolate structural contributions that persist across comparable conditions. To support fast explanation generation, CATGX integrates graph approximate nearest neighbor (ANN) sampling strategy as an unbiased motif occurrence statistics estimator, which preserves unbiased estimation through importance weighting. The entire pipeline operates in a training-free, model-agnostic manner and achieves bounded, low polynomial-time cost. Experiments demonstrate that CATGX strikes a good balance to generate explanations that are faithful, grounded and confounder-aware, and outperform existing TGNN explainers in efficiency with 83x speed up.
A novel theoretical model, namely the multiway autoregressive (MARS) model, which characterizes multiple evolutionary paths by capturing dependencies within and across two core factors underlying diverse evolutionary mechanisms is proposed, which develops a general DGNN framework, a multiway autoregressive network (MAN).
Ping He, Xiao-hua Xu· IEEE Transactions on Neural...· 0 citations
Much of the recent work on traffic forecasting relies on two implicit simplifications: representing the road network as an undirected graph and adopting temporally symmetric encoders within the historical observation window. These choices may obscure asymmetric predictive relations among sensors and the chronological ordering of intermediate temporal representations. CausalST addresses these limitations jointly. For spatial modeling, separate source and destination embeddings are assigned to each sensor to construct a learned directed adjacency under road-topology constraints. On the temporal side, dilated causal convolutions serve as a temporal inductive bias, ensuring that the representation at each historical position depends only on current and earlier observations. Delay-aware graph propagation aggregates current and lagged neighbor states over predefined discrete lags. A parallel spatial attention branch captures nonlocal dependencies beyond the physical topology, while a gated fusion unit adaptively integrates the temporal, graph-propagation, and spatial-attention representations. Experiments on PeMS03, PeMS04, and PeMS08 show that CausalST achieves the lowest MAPE among all compared methods while remaining competitive in MAE and RMSE. Ablation results reveal non-additive interactions between directed weighting and delayed aggregation, with the full configuration attaining the lowest MAE on both evaluated datasets.
A novel Dual-Channel Hybrid Graph Neural Network (HyGNN) that jointly models temporal dynamics and high-order structural dependencies and robustly captures the nonlinear interplay between mobility and sociality.
Liang Chen, Xiang Li, Guiyuan Jiang et al.· 0 citations
LiFTER turns future-link forecasting into a verifiable grounded computation and achieves competitive historical-negative forecasting and the highest macro explanation ac- curacy and deletion fidelity across four CTDG benchmarks.
Explainable recommendation has been conceptualized as a joint ranking task encompassing both items and explanations within contemporary recommender system research. The modeling of user–item–explanation triplets can be effectively facilitated by Graph Neural Networks (GNNs) due to their powerful representation learning capabilities. However, various observable and unobservable confounding factors, such as the user’s mood at the time of purchase and product popularity, may cause users to make decisions that diverge from their genuine preferences. These confounding variables significantly affect GNN models, resulting in inconsistent or inaccurate representations of the relationships among users, items, and explanations. To address these challenges, we propose Causality-enhanced Graph Contrastive Learning for Explainable Recommendation (CGCLER). This method enables item–explanation joint ranking by distinguishing causal and confounding features at the graph-node representation level. Guided by the backdoor adjustment principle, CGCLER further introduces a backdoor-inspired graph contrastive learning objective that constructs representation-level perturbation views by combining causal features with randomly sampled confounding features, thereby encouraging representations that are less affected by varying confounding contexts. The effectiveness of CGCLER is evaluated through experiments on three publicly available datasets.
The Structure-Guided Spatiotemporal Attention Graph Neural Network is proposed, offering a mechanistic account of the model's decision-making process while ensuring robust forecasting by aligning attention-based reasoning with identified macroscopic dependencies and preventing over-reliance on ephemeral local noise.
Xuan He, Can Li, Wan-Jing Ma· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
