SKGE: Spherical Knowledge Graph Embedding with Geometric Regularization

The recently introduced Spherical Knowledge Graph Embedding (SKGE) presents a novel approach to knowledge graph embedding by incorporating geometric regularization to constrain entity representations. This method aims to address limitations found in traditional models such as TransE, which often struggle with efficiently modeling complex relationships in multi-relational data. By embedding entities on a spherical manifold, SKGE improves training efficiency and enhances the capacity to capture intricate relational patterns. The approach focuses on better representing multi-relational data structures, which is a key challenge in knowledge graph embedding. Compared to conventional methods, SKGE’s geometric constraints offer a promising direction for more effective and scalable knowledge graph modeling. This development aligns with ongoing research efforts to refine embedding techniques for improved performance in machine learning tasks involving relational data. The work was detailed in a paper published on arXiv in November 2025, contributing to the evolving landscape of artificial intelligence research.