Counterfactual Explanations for Hypergraph Neural Networks accepted at CIKM 2026

We are pleased to share that “Counterfactual Explanations for Hypergraph Neural Networks” by Fabiano Veglianti, Lorenzo Antonelli, and Gabriele Tolomei has been accepted at CIKM 2026.

Hypergraph neural networks can represent interactions among several entities at once, but it can be difficult to understand why they make a particular prediction. The paper introduces CF-HyperGNNExplainer, which asks what small change to a hypergraph would cause the model to change its prediction. It produces explanations by removing node–hyperedge connections or entire hyperedges, revealing which higher-order relationships mattered to the decision.

Experiments on hypergraph benchmarks show that the method generates valid, concise counterfactual explanations. Read the paper on arXiv.