Authors: Jinqing Lian, Xinyi Liu, Yingxia Shao, Yang Dong, Ming Wang, Zhang Wei, Tianqi Wan, Ming Dong, Hailin Yan
Published on: May 01, 2024
Impact Score: 7.4
Arxiv code: Arxiv:2405.00527
Summary
- What is new: ChatBI introduces a new approach for the NL2BI task, focusing on dealing with schemas with large numbers of columns, which was a limitation in existing NL2SQL technologies.
- Why this is important: The challenge in converting natural language queries into SQL, especially in BI scenarios with complex schemas and a high number of columns.
- What the research proposes: ChatBI uses a phased process flow and combines view technology with a smaller model for efficient schema linking, enabling complex SQL generation.
- Results: Deployed on Baidu’s data platform, ChatBI showed significant improvement in practicality, versatility, and efficiency over existing NL2SQL technologies.
Technical Details
Technological frameworks used: ChatBI
Models used: Smaller, cheaper machine learning models for view selection and schema linking
Data used: Real BI scenario data tables and queries from Baidu’s platform
Potential Impact
This technology could disrupt markets that rely on business intelligence tools, particularly benefiting companies looking to make SQL database interaction more accessible to non-expert users.
Want to implement this idea in a business?
We have generated a startup concept here: DataQuerist.
Leave a Reply