面向科研业务的税务政策语义解析与自动适配的方法
By employing a terminology alignment method based on embedded boxes and hyperbolic embedded coordinates, along with a Sheaf consistency diagnostic mechanism, the problems of low efficiency and high error rate in traditional tax policy parsing methods are solved. This enables intelligent parsing and automatic adaptation of tax policies in scientific research operations, ensuring semantic accuracy and execution traceability.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- CHINA ELECTRIC POWER RESEARCH INSTITUTE CO LTD
- Filing Date
- 2025-11-07
- Publication Date
- 2026-07-17
AI Technical Summary
Traditional methods of interpreting tax policies rely on manual interpretation and static rule coding, which makes it difficult to handle complex policy provisions and semantic diversity, resulting in low policy application efficiency, high error rate, and difficulty in keeping up with the frequency of policy updates.
By constructing a terminology alignment method that combines embedded boxes with hyperbolic embedded coordinates, we achieve precise alignment between policy clauses and scientific research business terms. We adopt a Sheaf consistency diagnostic mechanism to ensure semantic consistency across documents and contexts, refine clause obligations into obligation slot sets, and use a stable mating strategy to generate action lists and process checkpoint rules.
It enables intelligent parsing and automatic adaptation of tax policies in scientific research operations, ensuring semantic accuracy, traceability of execution, and system adaptability, reducing manual comparison costs, and improving the transparency and consistency of policy implementation.
Smart Images
Figure CN121745104B_ABST