RAGA: Relation-Aware Graph Attention Networks for Global Entity Alignment

Entity alignment (EA) is the task to discover entities referring same real-world object from different knowledge graphs (KGs), which most crucial step in integrating multi-source KGs. The majority of existing embedding-based entity methods embed and relations into a vector space based on relation triples KGs for local alignment. As these insufficiently consider multiple between entities, structure information has not been fully leveraged. In this paper, we propose novel framework Relation-aware Graph Attention Networks capture interactions relations. Our adopts self-attention mechanism spread then aggregate back entities. Furthermore, global algorithm make one-to-one alignments with fine-grained similarity matrix. Experiments three cross-lingual datasets show that our outperforms state-of-the-art methods.