一种财务数据智能问数的自然语言交互处理系统和方法
By constructing a structured financial knowledge base and intent recognition module, and combining keyword and vector retrieval, accurate SQL query statements are generated. This solves the problems of operational complexity and inaccurate results in existing BI tools for financial data queries, achieving a balance between the flexibility of natural language interaction and the accuracy of data queries, and improving the efficiency and reliability of financial data queries.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- COSCO SHIPPING GROUP FINANCE CO LTD
- Filing Date
- 2026-04-01
- Publication Date
- 2026-07-17
AI Technical Summary
Existing business intelligence (BI) tools have high operational barriers, low query efficiency, and cannot flexibly adapt to users' diverse natural language query needs in financial data queries. Furthermore, the results are inaccurate and fail to effectively integrate corporate financial business experience and knowledge resources.
A structured financial knowledge base is constructed, and an intent recognition module and an SQL generation module are combined. Through a two-way retrieval method that combines keyword retrieval and vector retrieval, the user's intent is identified and accurate SQL query statements are generated. The system includes a structured financial knowledge base construction module, an intent recognition module, an SQL generation module, and a result output module, and realizes natural language interactive processing.
It lowers the operational threshold for financial data queries, improves query efficiency and accuracy, supports users to ask questions directly using natural language, fully integrates enterprise knowledge resources, and ensures the reliability and security of query results.
Smart Images
Figure CN122412445A_ABST