基于动态语义分块与混合检索的法规问答生成方法及系统
By employing dynamic semantic segmentation and hybrid retrieval methods, the challenges of semantic understanding and timeliness management in the legal Q&A system in the field of discipline inspection and supervision have been solved, enabling accurate and reliable extraction and output of legal information and ensuring the authority and traceability of the answers.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Filing Date
- 2026-05-22
- Publication Date
- 2026-07-17
AI Technical Summary
Existing technologies are insufficient for accurate and reliable semantic understanding and timeliness management of massive, dynamic, and structurally complex legal texts in professional fields such as discipline inspection and supervision. This results in incomplete or inaccurate retrieval and poses a risk of sensitive data leakage.
The method adopts dynamic semantic segmentation and hybrid retrieval. By acquiring legal and regulatory data and historical case data, text parsing and metadata processing are performed. Combined with dense vector representation and timeliness attributes, hybrid retrieval and deep semantic interaction are carried out to generate structured legal question and answer results, and the results are traced and formatted.
It enables the accurate and reliable extraction of semantically complete and timely legal information from massive and dynamic legal texts, generating authoritative and traceable question-and-answer results, and ensuring the semantic coherence and legal validity of the answers.
Smart Images
Figure CN122412554A_ABST