arXiv:2610.13128 · Agent 智能体
Locate, Judge, Read: Role-Specialized Tool Agents over a Knowledge-Point Index for Open-Domain Multimodal Document QA
Locate, Judge, Read: Role-Specialized Tool Agents over a Knowledge-Point Index for Open-Domain Multimodal Document QA
Locate, Judge, Read: Role-Specialized Tool Agents over a Knowledge-Point Index for Open-Domain Multimodal Document QA
- 类型：arxiv
- 标识：2610.13128
- 链接：http://arxiv.org/abs/2610.13128v1
- 主分类：agent
- 形态：method
- 被引：0
- 被引来源：Semantic Scholar
- S2被引：0
- 影响力被引：0
- TLDR：KnowForge, a pipeline separated by job: a hybrid-retrieval knowledge-point index; a Locator agent that can only search; a Locator agent that can only search; a code-scheduled bridge judge that forces the missing second hop; a typed, size-bounded evidence pool that merges every search's ranking; a Deep Reader agent that can only read text, images and tables; and a synthesizer with a format checker.
- 副分类：rag
- 待LLM分类：否
- 标题中文：Locate, Judge, Read: Role-Specialized Tool Agents over a Knowledge-Point Index for Open-Domain Multimodal Document QA
- TLDR中文：KnowForge，一个按职责分离的流水线：混合检索的知识点索引；只能搜索的 Locator agent；由代码调度的桥接判定器，用于强制补齐缺失的第二跳；类型化、大小有界的证据池，用于融合各次搜索的排序；只能阅读文本、图像和表格的 Deep Reader agent；以及带格式检查器的合成器。
- 来源文件：
- /inbox/tom/_candidates/2026-10-13-rag-retrieval-reranking-candidates.json
- [S2 enrich]