On the Representation and Embedding of Knowledge Bases beyond Binary Relations
By Jianfeng Wen, Jianxin Li, Yongyi Mao, Shini Chen, Richong Zhang. 2016
153 citations recorded by the source corpus.
The models developed to date for knowledge base embedding are all based on the assumption that the relations contained in knowledge bases are binary. For the training and testing of these embedding models, multi-fold (or n-ary) relational data are converted to triples (e.g., in FB15K dataset) and interpreted as instances of binary relations. This paper presents a canonical representation of knowledge bases containing multi-fold relations. We show that the existing embedding models on the popular FB15K datasets correspond to a sub-optimal modelling framework, resulting in a loss of structural information. We advocate a novel modelling framework, which models multi-fold relations directly using this canonical representation. Using this framework, the existing TransH model is generalized to a new model, m-TransH. We demonstrate experimentally that m-TransH outperforms TransH by a large margin, thereby establishing a new state of the art.
Subfields: Artificial Intelligence
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
- Pascal Poupart
- Rishab Goel
- Ivan Kobyzev
- Peter Forsyth
- Kshitij Jain
- Akshay Sethi
- Denilson Barbosa
- Antonio Matinata
Citations in the corpus
- A Comprehensive Survey of Graph Embedding: Problems, Techniques, and Applications (2017, neutral): “TransD [125] kMrhhþ r Mrttk22 TranSparse [126] kM r ður Þhþ r M rðurÞtk 2 l1=2 m-TransH [127] k P r2MðRrÞ arðrÞPnr ðtðrÞÞ þ brk (2); t 2 NMðRrÞ”
- Relational Representation Learning for Dynamic (Knowledge) Graphs: A Survey (2019, background): “Some recent works study ways of handling such hyperedges for static (knowledge) graphs [233, 73, 239, 10].”
- Knowledge Graph Embedding for Link Prediction (2020, methodology): “We also note that models such as m-TransH [68] and the recent HypE [15] have tried to circumvent these issues by developing systems that can explicitly learn hyperedges.”