arXiv：cs.CL· Priyank Jayraj, Poonam Goyal, Navneet Goyal·· 6 小时前AI 评分42
NavTree：无需 LLM 摘要的树导航，长文档 QA 的分层检索匹配成本研究
Tree Navigation Without LLM Summaries: A Matched-Cost Study of Hierarchical Retrieval for Long-Document QA
AI 导读
NavTree 是一种仅用叶节点的检索器，在 chunk 上构建确定性平衡线段树，索引时零 LLM 调用，靠混合词法与稠密向量的前沿游走从根下降到叶，只向 reader 输出叶 chunk。
正文
Abstract:Retrieval-augmented generation grounds language models in external context, but for long documents flat top-$k$ retrieval can cluster on a single region and miss complementary evidence. RAPTOR-style summary trees address this by recursively clustering chunks and using a language model to summarize each cluster at indexing time, then ranking summary nodes alongside raw chunks at query time. We show the main benefit of summary trees in long-document QA can come from navigation rather than the generated summary content. We introduce NavTree, a leaves-only retriever that builds a deterministic balanced segment tree over chunks (zero language-model calls at indexing) and uses the tree purely as a navigation scaffold: a hybrid lexical-and-dense frontier walk, anchored on top retrieved leaves, descends from the root and emits only leaf chunks to the reader. On a matched-cost evaluation against flat retrievers and an extractive re-implementation of RAPTOR, NavTree is the strongest matched-cost hierarchical retriever in our evaluated grid and ties the strongest flat baseline. On long-document multi-hop QA, it is the only hierarchical method that significantly beats BM25 on a class-vs-class basis, corroborated by a reader-free retrieval-recall check. A matched-reader replication of the published abstractive RAPTOR variant, given strong cluster summaries, still loses to NavTree at every multi-chunk budget, at zero indexing cost. The ranking carries across stronger and open-weight readers, a stronger encoder, and a full factorial that isolates leaves-only emission as the structural lever.
| Subjects: | Computation and Language (cs.CL); Information Retrieval (cs.IR) |
| Cite as: | arXiv:2610.06902 [cs.CL] |
| (or arXiv:2610.06902v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2610.06902 arXiv-issued DOI via DataCite |
Submission history
From: Priyank Jayraj [view email]
[v1]
Wed, 30 Sep 2026 08:42:45 UTC (9,192 KB)
来源：arXiv：cs.CL · arxiv.org