AI RESEARCH

CRAFTQA: A Code-Driven Adaptive Framework for Complex Structured Data Reasoning

arXiv CS.CL

ArXi:2606.02170v1 Announce Type: new Real-world scenarios involve massive heterogeneous structured data (e.g., tables, knowledge graphs), making effective reasoning over such diverse data increasingly important. Unified structured data question answering has emerged as a prominent research trend, aiming to answer natural language questions across different structured data types within a single framework.