PID-GNN: Probabilistically disentangled Laplacian renormalization for uncertainty-aware social recommendation

Liangli Yang

doi:10.1016/j.physa.2026.131798

PID-GNN: Probabilistically disentangled Laplacian renormalization for uncertainty-aware social recommendation

2026-06-30

0

PRE

AI

L

Liangli Yang

Y

Yutai Zhang

S

Siqing Pang

X

Xinyue Sun

Y

Yihua Zhou

Y

Yi-Xiu Kong *

DOI：10.1016/j.physa.2026.131798

Original

Original request for help

Save

Abstract

En 中文

• PID-GNN infers node receptivity and expressivity via variational Bayesian modeling. • Role-conditioned Laplacian reweighting normalizes receiver and sender degrees. • Ciao and Epinions tests with ablation and four-state analyses support PID-GNN.

PID-GNN: Probabilistically disentangled Laplacian renormalization for uncertainty-aware social recommendation

Abstract

Journal

Organization

Cited Papers

Citing Papers