正在补充深度解读，当前内容可以先阅读
论文解决了什么问题
Data agents enable natural-language analytics over organizational workspaces, where relevant evidence may be scattered across databases, structured files, long documents, and multimedia. Existing benchmarks largely isolate structured querying, retrieval, or open-ended analysis, leaving heterogeneous evidence discovery,...
适合谁阅读
Agent 系统多模态视频生成Benchmark / 评测
可核验的原论文来源和作者
- 作者
- 作者信息暂未从原始元数据中确认
- 来源
- arXiv
- 论文 ID
- 2608.03451