A CNNs-based Model for Knowledge Graph Completion

Knowledge Graph Completion (KGC) aims to find the missing relationships between entities. In this paper, we introduce a novel embedding method based on Convolutional Neural Networks (CNNs) for KGC task. Different from the existing model ConvKB, we design three different shapes of filters to produce more useful features, rather than extracting the features by the filters with the same shape of 1 × 3. Experiments show that our model obtains better link prediction results than the previous models on two benchmark datasets WN18RR and FB15k-237.