A multi-hop Shapley-based framework for graph convolutional network node classification explanation

doi:10.1016/j.asoc.2025.113615

A multi-hop Shapley-based framework for graph convolutional network node classification explanation

2025-08-27

0

PRE

AI

DOI：10.1016/j.asoc.2025.113615

Original

Original request for help

Save

Abstract

En 中文

• Propose a computational graph construction method to integrate multi-hop edges. • Introduce a sampling strategy to sample coalitions at diverse scales. • Improve GCN confidence in predicting the correct class with key edges.

Keywords:

computational graph

multi-hop edges

sampling strategy

coalition sampling

GCN confidence

A multi-hop Shapley-based framework for graph convolutional network node classification explanation

Abstract

Journal

Organization