Complex Embeddings for Simple Link Prediction
This work addresses the problem of automatically understanding large knowledge bases for researchers and practitioners in statistical relational learning, though it is incremental as it builds on existing latent factorization methods.
The paper tackles link prediction in knowledge bases by using complex-valued embeddings for latent factorization, which can model symmetric and antisymmetric relations and outperforms state-of-the-art models like Neural Tensor Network and Holographic Embeddings on standard benchmarks.
In statistical relational learning, the link prediction problem is key to automatically understand the structure of large knowledge bases. As in previous studies, we propose to solve this problem through latent factorization. However, here we make use of complex valued embeddings. The composition of complex embeddings can handle a large variety of binary relations, among them symmetric and antisymmetric relations. Compared to state-of-the-art models such as Neural Tensor Network and Holographic Embeddings, our approach based on complex embeddings is arguably simpler, as it only uses the Hermitian dot product, the complex counterpart of the standard dot product between real vectors. Our approach is scalable to large datasets as it remains linear in both space and time, while consistently outperforming alternative approaches on standard link prediction benchmarks.