Headers-as-context chunking improves retrieval MRR@5 by 23.8% over a fixed-threshold baseline, but the paper's core result is that LLM dual-annotator relevance judgment is unstable across text-transform ablations (kappa 0.45 to 0.04).
Raptor: Recursive abstractive processingfortree-organizedretrieval
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
other 1
citation-polarity summary
fields
cs.SE 1years
2026 1verdicts
CONDITIONAL 1roles
other 1polarities
unclear 1representative citing papers
citing papers explorer
-
Structure-Aware Semantic Chunking with Title-Chain Prefixes: A 1600-Query Evaluation and the Measurement Trap in Text-Transform Ablations
Headers-as-context chunking improves retrieval MRR@5 by 23.8% over a fixed-threshold baseline, but the paper's core result is that LLM dual-annotator relevance judgment is unstable across text-transform ablations (kappa 0.45 to 0.04).