{"paper":{"title":"RAPTOR: Recursive Abstractive Processing for Tree-Organized Retrieval","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Recursive clustering and summarization builds a tree that improves retrieval-augmented reasoning over long documents.","cross_cats":["cs.LG"],"primary_cat":"cs.CL","authors_text":"Aditi Tuli, Anna Goldie, Christopher D. Manning, Parth Sarthi, Salman Abdullah, Shubh Khanna","submitted_at":"2024-01-31T18:30:21Z","abstract_excerpt":"Retrieval-augmented language models can better adapt to changes in world state and incorporate long-tail knowledge. However, most existing methods retrieve only short contiguous chunks from a retrieval corpus, limiting holistic understanding of the overall document context. We introduce the novel approach of recursively embedding, clustering, and summarizing chunks of text, constructing a tree with differing levels of summarization from the bottom up. At inference time, our RAPTOR model retrieves from this tree, integrating information across lengthy documents at different levels of abstractio"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"by coupling RAPTOR retrieval with the use of GPT-4, we can improve the best performance on the QuALITY benchmark by 20% in absolute accuracy.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The recursive clustering and summarization process effectively captures and preserves all relevant information from the original document without significant loss or distortion.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"RAPTOR introduces a tree-organized retrieval method using recursive abstractive summaries, achieving a 20% absolute accuracy improvement on the QuALITY benchmark when paired with GPT-4.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Recursive clustering and summarization builds a tree that improves retrieval-augmented reasoning over long documents.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"363e72ddfd188b5c4b3d91e533f2c9b06efd2f14e3eb35f861aa145e81ededa2"},"source":{"id":"2401.18059","kind":"arxiv","version":1},"verdict":{"id":"3e39f47c-b446-44f2-8a05-1632aa62bd84","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-15T13:02:10.149769Z","strongest_claim":"by coupling RAPTOR retrieval with the use of GPT-4, we can improve the best performance on the QuALITY benchmark by 20% in absolute accuracy.","one_line_summary":"RAPTOR introduces a tree-organized retrieval method using recursive abstractive summaries, achieving a 20% absolute accuracy improvement on the QuALITY benchmark when paired with GPT-4.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The recursive clustering and summarization process effectively captures and preserves all relevant information from the original document without significant loss or distortion.","pith_extraction_headline":"Recursive clustering and summarization builds a tree that improves retrieval-augmented reasoning over long documents."},"references":{"count":124,"sample":[{"doi":"10.1007/3-540-44503-x_27","year":2001,"title":"On the S urprising B ehavior of D istance M etrics in H igh D imensional S pace","work_id":"a97fe484-1064-458f-b442-b0cca45b2093","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2022,"title":"Improving language models by retrieving from trillions of tokens","work_id":"c8e7ce21-2f18-4c10-b0cf-16cd3574fe33","ref_index":7,"cited_arxiv_id":"2112.04426","is_internal_anchor":true},{"doi":"","year":1901,"title":"L anguage M odels are F ew- S hot L earners","work_id":"e21d9ea5-09df-4873-a892-f6f44ddd45ef","ref_index":8,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2022,"title":"PaLM: Scaling Language Modeling with Pathways","work_id":"a94f3ef7-2c49-4445-93fe-6ec16aafd966","ref_index":12,"cited_arxiv_id":"2204.02311","is_internal_anchor":true},{"doi":"10.1145/3077136.3080740","year":2017,"title":"Contextualizing citations for scientific summarization using word embeddings and domain knowledge","work_id":"771354f2-4a97-4e9a-ab55-7fbe34c0a2aa","ref_index":13,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":124,"snapshot_sha256":"533ec12270f67a4525f66fbe7f1d804cd3a8a0cd092f4250f70015eed7dfa3c6","internal_anchors":25},"formal_canon":{"evidence_count":2,"snapshot_sha256":"46400d8dae9eb0ac558d85941795d229d9b4566ce33b36048c2ee2df75745546"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}