Text processing, device and medium for civil aviation document review
By utilizing semantic relevance thresholds and explanatory text splicing methods in civil aviation document review, the problem of low semantic matching between rule texts and related regulations texts was solved, achieving efficient text retrieval and review.
CN122389882APending Publication Date: 2026-07-14CHINA ACAD OF CIVIL AVIATION SCI & TECH
View PDF 0 Cites 0 Cited by
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-05-11
- Publication Date
- 2026-07-14
AI Technical Summary
Technical Problem
In the current technology for reviewing civil aviation documents, the semantic matching degree between rule texts and relevant regulations is low, which makes it impossible to effectively recall relevant regulations and affects review efficiency.
Method used
By obtaining the semantic relevance between the target rule text to be reviewed and the superior rule text, if it is lower than the threshold, the explanatory text is concatenated for secondary matching, and the domain term set is used for sorting to improve the recall rate and review efficiency.
Benefits of technology
It improves the recall rate of highly abstract rule texts, reduces the workload of reviewers, and ensures the relevance and accuracy of the recall.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN122389882A_ABST
Abstract
The application provides a text processing method for civil aviation file auditing, and relates to the field of text processing. The method comprises the following steps: obtaining a first semantic correlation between a target to-be-audited rule text and each upper rule text; if P1i >= Y1, the i-th upper rule text is determined as a target upper rule text; otherwise, a second semantic correlation P2i between the target to-be-audited rule text and a first spliced text corresponding to the i-th upper rule text is obtained; if P2i >= Y2, the i-th upper rule text is determined as the target upper rule text; a third semantic correlation between the target to-be-audited rule text and a second spliced text corresponding to each target upper rule text is used to sort the target upper rule text, and the target to-be-audited rule text and the sorted target upper rule text are output to an auditing module; and the application can effectively match highly abstract or generalized upper rule texts.
Need to check novelty before this filing date? Find Prior Art