Knowledge Graph Embedding by Translating on Hyperplanes
Встраивание графа знаний посредством трансляций на гиперплоскостях
2014-06-21
SCID: 54.1/xfx7hmub
Discuss with AI
TransETransHknowledge graph embeddinglink predictiontranslating on hyperplanes
Figures from the paper
Abstract (AI)
We deal with embedding a large scale knowledge graph composed of entities and relations into a continuous vector space. TransE is a promising method proposed recently, which is very efficient while achieving state-of-the-art predictive performance. We discuss some mapping properties of relations which should be considered in embedding, such as reflexive, one-to-many, many-to-one, and many-to-many. We note that TransE does not do well in dealing with these properties. Some complex models are capable of preserving these mapping properties but sacrifice efficiency in the process. To make a good trade-off between model capacity and efficiency, in this paper we propose TransH which models a relation as a hyperplane together with a translation operation on it. In this way, we can well preserve the above mapping properties of relations with almost the same model complexity of TransE. Additionally, as a practical knowledge graph is often far from completed, how to construct negative examples to reduce false negative labels in training is very important. Utilizing the one-to-many/many-to-one mapping property of a relation, we propose a simple trick to reduce the possibility of false negative labeling. We conduct extensive experiments on link prediction, triplet classification and fact extraction on benchmark datasets like WordNet and Freebase. Experiments show TransH delivers significant improvements over TransE on predictive accuracy with comparable capability to scale up.
Key Findings
1
The paper proposes a simple negative-sampling trick using one-to-many/many-to-one relation properties to reduce false negative labels during training.
2
TransE struggles to model relation mapping properties such as reflexive, one-to-many, many-to-one, and many-to-many.
3
TransH achieves significant improvements over TransE in predictive accuracy on benchmark datasets (WordNet, Freebase) while retaining comparable scalability and model complexity.
4
TransH maintains nearly the same efficiency as TransE but improves capacity to handle complex relation mappings, evidenced by experiments on link prediction, triplet classification, and fact extraction.
5
TransH models each relation as a hyperplane plus a translation on that hyperplane, enabling better preservation of those mapping properties.
Research Object
Knowledge graph embedding model for entities and relations (TransH)
Research Subject
Modeling relations as hyperplanes with translations to preserve mapping properties (reflexive, one-to-many, many-to-one, many-to-many) and improve link prediction/triplet classification accuracy while maintaining efficiency
Publication Details
Publication Date
2014-06-21
Journal
Publisher
ISSN
Cited by
3982
Open access PDF
Access Type
Author Information
Download PDF
Subscribe to digest
References available in scid.ai2
Cited by19
Learning Entity and Relation Embeddings for Knowledge Graph Completion2015
A Survey on Knowledge Graphs: Representation, Acquisition, and Applications2021
Knowledge Graph Embedding: A Survey of Approaches and Applications2017
Knowledge Graphs2021
Knowledge Graph Embedding via Dynamic Mapping Matrix2015
A Survey on Knowledge Graph-Based Recommender Systems2020
DeepPath: A Reinforcement Learning Method for Knowledge Graph Reasoning2017
Knowledge Graph Embedding for Link Prediction: A Comparative Analysis2021
Knowledge Graphs: Opportunities and Challenges2023
Holographic Embeddings of Knowledge Graphs2016
TransG : A Generative Model for Knowledge Graph Embedding2016
Knowledge Graph Completion: A Review2020
Constructing knowledge graphs and their biomedical applications2020
Enriching contextualized language model from knowledge graph for biomedical information extraction2020
A Re-evaluation of Knowledge Graph Completion Methods2020
Knowledge graph and knowledge reasoning: A systematic review2022
Construction of Knowledge Graphs: Current State and Challenges2024
Triple Classification for Scholarly Knowledge Graph Completion2021
Research hypothesis generation over scientific knowledge graphs2025