An ontology-based business semantic modeling method and system
By using ontology-based semantic segmentation and graph structure organization, combined with the improved OneKE model and ReAct algorithm, the accuracy and consistency issues of semantic modeling in existing technologies are solved, achieving efficient semantic modeling and governance.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHUZHIYUN (BEIJING) TECHNOLOGY CO LTD
- Filing Date
- 2026-04-22
- Publication Date
- 2026-07-21
AI Technical Summary
Existing data modeling methods struggle to accurately identify the true relationships between business concepts and attributes in complex business scenarios. Semantic candidate results are often confused or missing, mapping relationships are incomplete or mismatched, semantic model structures are incomplete and inconsistent, and semantic governance is inefficient.
An ontology-based approach is adopted to generate a set of semantic fragments and a set of data structures through semantic fragment segmentation and structural parsing. The improved OneKE model is used to extract semantics through a term hierarchy stacking mechanism. The GraphRAG algorithm is combined to construct a semantic association graph, and the ReAct algorithm is used to construct candidate mapping paths. Finally, a business semantic ontology structure and a set of mapping relationships are generated, and semantic verification and abnormal path backtracking are performed.
It improves the accuracy of semantic extraction and the completeness of associations, enhances the consistency of mapping relationships and semantic governance capabilities, and ensures the accuracy and usability of the semantic model.
Smart Images

Figure CN122433739A_ABST