Should Graph Convolution Trust Neighbors? A Simple Causal Inference Method

Abstract

Graph Convolutional Network (GCN) is an emerging technique for information retrieval (IR) applications. While GCN assumes the homophily property of a graph, real-world graphs are never perfect: the local structure of a node may contain discrepancy, e.g., the labels of a node's neighbors could vary. This pushes us to consider the discrepancy of local structure in GCN modeling. Existing work approaches this issue by introducing an additional module such as graph attention, which is expected to learn the contribution of each neighbor. However, such module may not work reliably as expected, especially when there lacks supervision signal, e.g., when the labeled data is small. Moreover, existing methods focus on modeling the nodes in the training data, and never consider the local structure discrepancy of testing nodes. This work focuses on the local structure discrepancy issue for testing nodes, which has received little scrutiny. From a novel perspective of causality, we investigate whether a GCN should trust the local structure of a testing node when predicting its label. To this end, we analyze the working mechanism of GCN with causal graph, estimating the causal effect of a node's local structure for the prediction. The idea is simple yet effective: given a trained GCN model, we first intervene the prediction by blocking the graph structure; we then compare the original prediction with the intervened prediction to assess the causal effect of the local structure on the prediction. Through this way, we can eliminate the impact of local structure discrepancy and make more accurate prediction. Extensive experiments on seven node classification datasets show that our method effectively enhances the inference stage of GCN.

Accepted by SIGIR'21

References in corpus (17)

• Distilling the Knowledge in a Neural Network
• Towards Deeper Graph Neural Networks
• Graph Transformer Networks
• Simple and Deep Graph Convolutional Networks
• Deeper Insights into Graph Convolutional Networks for Semi-Supervised Learning
• DeeperGCN: All You Need to Train Deeper GCNs
• Clicks can be Cheating: Counterfactual Recommendation for Mitigating Clickbait Issue
• Bias and Debias in Recommender System: A Survey and Future Directions
• Interventional Few-Shot Learning
• CauseRec: Counterfactual User Sequence Synthesis for Sequential Recommendation
• Investigating and Mitigating Degree-Related Biases in Graph Convolutional Networks
• Graph Wavelet Neural Network
• DeVLBert: Learning Deconfounded Visio-Linguistic Representations
• Multi-hop Attention Graph Neural Network
• Improving Graph Attention Networks with Large Margin-based Constraints
• Graph Inference Learning for Semi-supervised Classification
• Interventional Video Grounding with Dual Contrastive Learning

Cited by in corpus (13)

• Model-Agnostic Counterfactual Reasoning for Eliminating Popularity Bias in Recommender System
• Causal Attention for Interpretable and Generalizable Graph Classification
• Deconfounded Recommendation for Alleviating Bias Amplification
• CauseRec: Counterfactual User Sequence Synthesis for Sequential Recommendation
• User-controllable Recommendation Against Filter Bubbles
• Neighborhood-Enhanced Supervised Contrastive Learning for Collaborative Filtering
• Unbiased Knowledge Distillation for Recommendation
• Debiasing Recommendation with Personal Popularity
• Cross-Domain Contract Element Extraction with a Bi-directional Feedback Clause-Element Relation Network
• Causality-Aware Spatiotemporal Graph Neural Networks for Spatiotemporal Time Series Imputation
• Learning Robust Recommender from Noisy Implicit Feedback
• Uncovering Main Causalities for Long-tailed Information Extraction
• Deconfounded Video Moment Retrieval with Causal Intervention