HousE: Knowledge Graph Embedding with Householder Parameterization

  • Rui Li ,
  • Chaozhuo Li ,
  • Jianan Zhao ,
  • Di He ,
  • Yiqi Wang ,
  • Yuming Liu ,
  • Hao Sun ,
  • Senzhang Wang ,
  • Weiwei Deng ,
  • Yanming Shen ,
  • ,
  • Qi Zhang

ICML 2022 |

The effectiveness of knowledge graph embedding (KGE) largely depends on the ability to model intrinsic relation patterns and map- ping properties. However, existing approaches can only capture some of them with insufficient modeling capacity. In this work, we propose a more powerful KGE framework named HousE, which in- volves a novel parameterization based on two kinds of Householder transformations: (1) Householder rotations to achieve superior ca- pacity of modeling relation patterns; (2) Householder projections to handle sophisticated relation mapping properties. Theoretically, HousE is capable of modeling crucial relation patterns and mapping properties simultaneously. Besides, HousE is a generalization of existing rotation-based models while extending the rotations to high-dimensional spaces. Empirically, HousE achieves new state- of-the-art performance on five benchmark datasets. Our code is available at https://github.com/anrep/HousE.