Unverified50% confidenceFactExact time
RAPTOR builds a tree: it clusters related text chunks, generates a summary for each cluster, then clusters and summarizes the summaries, all the way up; at query time it retrieves across levels, capturing both fine-grained details and high-level synthesis.
1
Sources
50%
Confidence
Long-term
Relevance
10/18/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedRAPTOR is a hierarchical retrieval framework proposed by the Stanford team in 2024, whose full name is Recursive Abstractive Processing for Tree-Organized Retrieval.73% similarUnverifiedThe tree index works by splitting long documents into leaf nodes, summarizing adjacent leaf node content to generate parent nodes that aggregate layer by layer to form a summary tree, and during retrieval traversing top-down from the root node to locate text segments.72% similarUnverified父子块策略将文档同时索引为大块和小块，检索时用小块精准定位，生成时返回大块提供完整上下文61% similarUnverifiedParagraph-level retrieval splits documents into chunks of around a few hundred characters, each chunk being vectorized separately and indexed, in order to improve recall precision and directly locate specific content.59% similarUnverifiedDeckhand adopts a RAG (retrieval-augmented generation) architecture: it first vectorizes the user's question, retrieves the most semantically relevant passages from the knowledge base, and then hands them to the large model to generate an answer.58% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1063911API
curl https://kongchang.com/api/v1/knowledge/claims/1063911MCP
get_claim(id=1063911)