TransRT-RMP: An Rotation Translation and Relationship Mapping Attribute-Based Representation Learning Model

Knowledge reasoning, knowledge question re-sponding, and knowledge retrieval have all benefited from existing knowledge representation learning models. They are, never-theless, insufficient when dealing with complex relationships. As a result, this study introduces TransRT-Rmp,a knowledge representation learning model based on rotation translation and relationship mapping features. To begin, relationships and entities are embedded in low-dimensional vector spaces using hyperplane projection. Second, the translation rules have been improved by including rotation translation rules, which eliminates the typical constraint of head entity+relationship=tail entity. Then, relation-ship mapping features were added to improve the model's capac-ity to handle complex relationships. The quality of producing negative example triplets can be enhanced during model training by replacing semantically related things, which increases sample diversity and improves the model's generalization ability and robustness. Finally, results from experiments on publicly available datasets FB15K,WN18, WN11, and FB13 show that our technique performs well in MeanRank Hits@10 In terms of ACC indicators, the most positive outcomes were obtained.