CoSD: Collaborative stance detection with contrastive heterogeneous topic graph learning

Yinghan Cheng, Qi Zhang, Chongyang Shi, Liang Xiao, Shufeng Hao, Liang Hu

Abstract: Highlights•We mark the first attempt to propose a collaborative stance detection framework CoSD.•We establish a heterogeneous topic graph to correlate texts, targets, and stances.•We elaborately design CPA to capture informative multi-hop collaborative signals.