构建可导航的结构化知识维基，让智能体像查百科一样精准找信息。
SearchWiki: Learning to Build and Navigate Knowledge Wikis for Active Information Seeking
- 将文档组织成三层次维基：概览、主题页、来源记录，支持逐步精炼检索
- 训练90亿参数智能体在8个领域任务中超越同规模基线模型
- 适合需要深度检索与跨文档推理的研究者或知识系统开发者
传统检索增强生成将语料库视为无序文本块，忽略文档层级与跨文档结构。我们提出SearchWiki框架，将语料库合成层次化、类型化、可导航的维基，并训练一个名为WikiResearcher-9B的智能体，通过多轮工具使用实现信息检索。该维基包含三个层级：文档概览、跨文档主题页和页面级源记录，支持在初始检索失败时逐步精炼结果。我们采用基于策略的强化学习优化智能体的导航策略，奖励函数综合考虑答案正确性、检索质量与轨迹效率。在ViDoRe-V3（8个领域）、FinanceBench及记忆基准（LoCoMo、LongMemEval、PersonaMem-v2）上的评估表明，经强化学习调优的Qwen 9B模型——WikiResearcher-9B，显著优于同规模未训练基线，并达到或超过更大外部模型性能。结果表明，对结构化语料库进行学习型导航，是优于扁平检索的更优方案。
原文摘要 · Abstract (English)
Flat retrieval-augmented generation treats a corpus as a bag of chunks, discarding document hierarchy and cross document structure. We introduce SearchWiki, a harness framework that synthesizes a corpus into a hierarchical, typed, navigable wiki and trains an agent, WikiResearcher-9B, to retrieve information through multi-turn tool use. The wiki organizes knowledge into three layers - document overviews, cross- document topic pages, and page-level source records; enabling progressive refinement of retrieval when initial lookup misses. We optimize the agent's navigation policy with on-policy reinforcement learning with a multi-component reward function balancing answer correctness, retrieval quality and trajectory efficiency. Evaluation on ViDoRe-V3 (8 domains), FinanceBench, and memory benchmarks (LoCoMo, LongMemEval, PersonaMem-v2) shows that WikiResearcher- 9B which is our RL-tuned Qwen 9B model, significantly outperforms same-size untrained baselines and exceeds or matches larger external models. SearchWiki paired with WikiResearcher-9B demonstrates that learned navigation over structured corpora is a superior alternative to flat retrieval.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。