Information processing apparatus
By segmenting and associating Japanese documents into paragraphs, generating numerical vectors and registering them in the knowledge base, the problem of insufficient search accuracy in the knowledge base is solved, thus improving the accuracy of the chatbot's responses.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- TOYOTA JIDOSHA KK
- Filing Date
- 2025-11-27
- Publication Date
- 2026-06-02
AI Technical Summary
In chatbots that use large-scale language models and Japanese document knowledge bases, the search accuracy of the knowledge base may decrease, leading to insufficient accuracy in the responses.
By dividing Japanese documents into paragraph units and establishing connections based on the relationships between paragraphs, multiple blocks are generated, which are then converted into numerical vectors and registered in a knowledge base. The search department extracts blocks related to the question text and their associated blocks from the knowledge base as search results.
This improves the search accuracy of the knowledge base, thereby enhancing the response accuracy of large-scale language models and ensuring accurate answers to user questions.
Smart Images
Figure CN122132519A_ABST