一种围绕故事主题的多轮追问方法和系统
By employing multi-dimensional scoring and context management mechanisms, the shortcomings of existing multi-turn dialogue systems have been addressed, enhancing the depth of interaction and user experience. This has enabled accurate evaluation of user responses and personalized follow-up questions, thereby strengthening the system's robustness.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- E-SURFING DIGITAL LIFE TECH CO LTD
- Filing Date
- 2025-09-01
- Publication Date
- 2026-07-17
AI Technical Summary
Existing multi-turn dialogue systems suffer from rigid questioning strategies, weak context management, low emotional matching, insufficient personalization, and poor robustness, resulting in poor interaction depth and user experience.
A multi-dimensional scoring system is constructed, including completeness deviation value, emotional relevance deviation value, extensibility score and story relevance score. The dialogue context is managed by combining a tree structure and the RAPTOR recursive summarization algorithm. A sentiment analysis module and a linkage mechanism with story sentiment nodes are developed to realize user preference recording and dynamic adaptation. A fuzzy response mechanism is constructed and optimized through reinforcement learning.
It enables accurate evaluation of user responses, generates appropriate follow-up questions, improves the depth of interaction and user experience, avoids repeated questioning, enhances emotional matching and personalization, and improves the robustness of the system.
Smart Images
Figure CN121119147B_ABST