An unsupervised multi-view contrastive learning framework with attention-based reranking strategy for entity alignment

Yan Liang, Weishan Cai, Minghao Yang, Yuncheng Jiang

Abstract: Highlights•Two graph augmentation methods are proposed for complementary information.•A multi-view contrastive learning method is introduced to minimize the semantic gap.•An attention-based reranking strategy is developed to mine hard entities.•Our method outperforms most supervised and unsupervised methods on three benchmarks.