TransRESCAL: A Dense Feature Model for Knowledge Graph Completion

Knowledge Graph Completion (KGC) is a hot-spot in artificial intelligence research. Dozens of hundreds of models are devised to this end. The translation-based and factorization- based approaches are very popular ones for their efficiency and simplicity. TransE and RESCAL are the canonical ones in both research lines. However, the simplicity also limits the expressivity of the model to achieve better performance. To improve the results, we couple the two approaches to form a more expressive one named TransRESCAL. It is capable of distinguishing head from tail entity embedding due to TransE has the ability to be capable of doing this. Therefore, the proposed model is capable of clarifying the fact that the position of head and tail entity embeddings in score function of factorization-based models. It is beneficial for further understanding the model for KGC task. Hence, the proposed model is not only capable of capturing structural feature by TransE but also semantic ones by RESCAL. It potentially captures denser features than every single model, which is a competitive alternative to the deep neural network (DNN)-based models for KGC tasks.