Knowledge Graph Completion Method Based on Fusing Association Information

Knowledge graph is a carrier of knowledge. The knowledge completion task to predict the links between entities make more complete. Currently, method based on representation learning ignores association information each relation in multiple-step paths and direct relations, head tail entity types relations. In this study, we extract utilize these associated information, propos AiTransE model for completion, which uses frequency calculate degree with matching obtain them, Finally, two degrees are linearly weighted merged then introduced into objective function, so that can give different attention triples, improve performance. experiment link prediction was carried out Tibet Animal Husbandry Dataset WN18 dataset, compared TransE, TransH, TransR, other models. experimental results show has significant improvement over models indicators Hits@10 Mean Rank.