Semantic Interaction Matching Network for Few-shot Knowledge Graph Completion

The prosperity of knowledge graphs (KG), as well as related downstream applications, have raised the urgent request of knowledge graph completion techniques for fully supporting the KG reasoning tasks, especially under the circumstance of training data scarcity. Though large efforts have been made on solving this challenge via few-shot learning tools, they mainly focus on simply aggregating entity neighbors to represent few-shot references, while the enhancement from latent semantic correlation within neighbors has been largely ignored. To that end, in this paper, we propose a novel few-shot learning solution, named as Semantic Interaction Matching network (SIM), which applies Transformer framework to enhance the entity representation with capturing semantic interaction between entity neighbors. Specifically, we first design entity-relation fusion module to adaptively encode neighbors with incorporating relation representation. Along this line, Transformer layers are integrated to capture latent correlation within neighbors, as well as the semantic diversification of the support set. Finally, a similarity score is attentively estimated with attention mechanism. Extensive experiments on two public benchmark datasets demonstrate that our model outperforms a variety of state-of-the-art methods with a significant margin.