Core Concepts
TransERR is a translation-based knowledge graph embedding method that utilizes efficient relation rotation in the hypercomplex-valued space to enhance translation freedom and model various relation patterns effectively.
Abstract
TransERR introduces a translation-based approach in the hypercomplex-valued space for knowledge graph embedding.
It adapts relation rotation with unit quaternions to minimize translation distance and enhance translation freedom.
Mathematical proofs are provided to demonstrate TransERR's ability to model symmetry, antisymmetry, inversion, composition, and subrelation patterns.
Experimental results on benchmark datasets validate the effectiveness and generalization of TransERR.
TransERR outperforms existing models in encoding large-scale datasets with fewer parameters.
The model is available on GitHub for further exploration.
Stats
TransERR는 지식 그래프 임베딩을 위해 효율적인 관계 회전을 사용합니다.
TransERR는 하이퍼복소값 공간에서 번역 기반 방법론을 도입합니다.
Quotes
"TransERR encodes knowledge graphs in the hypercomplex-valued space, enabling higher translation freedom."
"Mathematical proofs demonstrate TransERR's ability to model various relation patterns effectively."