Irreflexive and Hierarchical Relations as Translations

Abstract

We consider the problem of embedding entities and relations of knowledge bases in low-dimensional vector spaces. Unlike most existing approaches, which are primarily efficient for modeling equivalence relations, our approach is designed to explicitly model irreflexive relations, such as hierarchies, by interpreting them as translations operating on the low-dimensional embeddings of the entities. Preliminary experiments show that, despite its simplicity and a smaller number of parameters than previous approaches, our approach achieves state-of-the-art performance according to standard evaluation protocols on data from WordNet and Freebase.

Submitted at the ICML 2013 workshop "Structured Learning: Inferring Graphs from Structured and Unstructured Inputs"

References in corpus (3)

• Efficient Estimation of Word Representations in Vector Space
• Learning New Facts From Knowledge Bases With Neural Tensor Networks and Semantic Word Vectors
• A Semantic Matching Energy Function for Learning with Multi-relational Data

Cited by in corpus (6)

• Complex Embeddings for Simple Link Prediction
• Ultrahyperbolic Knowledge Graph Embeddings
• Large Margin Nearest Neighbor Embedding for Knowledge Representation
• Searching to Sparsify Tensor Decomposition for N-ary Relational Data
• Generalizing Tensor Decomposition for N-ary Relational Knowledge Bases
• Empirically Verifying Hypotheses Using Reinforcement Learning