Man-machine interaction method and system based on large language model, computer device and storage medium
By using a human-computer interaction method based on a large language model, combined with a structured knowledge base and an intelligent question-answering model, the problems of low retrieval efficiency, cumbersome meeting management, and delayed compliance response in existing technologies have been solved, achieving an efficient and accurate office experience.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- QINZHOU POWER SUPPLY BUREAU OF GUANGXI POWER GRID CO LTD
- Filing Date
- 2026-04-14
- Publication Date
- 2026-07-17
AI Technical Summary
In existing technologies, keyword matching-based retrieval methods are difficult to accurately locate complex policy clauses, requiring users to manually filter through a large number of documents, resulting in low retrieval efficiency. Manual transcription and minutes organization in meeting management are costly and prone to missing information, while compliance consultation responses are delayed, failing to meet the demand for an efficient and accurate office experience.
A human-computer interaction method based on a large language model is adopted, which uses a structured knowledge base for semantic understanding and contextual retrieval to generate accurate search results; a meeting management model is used for speech recognition and semantic parsing to automatically generate structured meeting minutes; and an intelligent question-answering model is used to match regulatory clauses in the structured knowledge base to generate compliant answers that include the source of the evidence.
It enables precise location within complex institutional clauses, improves retrieval efficiency and accuracy, automates the generation of meeting minutes, reduces the risk of information omissions, shortens the waiting time for compliance decisions, and achieves intelligent enhancement of the office interaction process.
Smart Images
Figure CN122416993A_ABST