Dual Quaternion Embeddings for Link Prediction

The applications of knowledge graph have received much attention in the field artificial intelligence. quality graphs is, however, often influenced by missing facts. To predict facts, various solid transformation based models been proposed mapping into low dimensional spaces. However, most existing approaches ignore that there are multiple relations between two entities, which is common real world. In order to address this challenge, we propose a novel approach called DualQuatE maps entities and dual quaternion space. Specifically, represented pure quaternions modeled on combination rotation translation from head tail entities. After utilize interactions different translations rotations distinguish Experimental results exhibit performance competitive compared state-of-the-art models.