aiqbench analysis
This paper shows that retrieval-augmented in-context demonstrations can lift a frozen LLM's performance on Rhetorical Structure Theory parsing, improving nuclearity and relation F1 without any training. The gains are measured against an untuned strong-LLM baseline, and the authors report that the benefit saturates around fifty exemplars and depends on structural position, so the approach has practical limits. For AI readers, the main takeaway is that a formal linguistic schema can be supplied at inference time rather than learned into weights, though the results are reported on a specialized parsing benchmark and need replication to generalize.
RAG-RST: Retrieved Exemplars as In-Context Discourse Schema for RST Parsing
arXiv:2610.22150v1 Announce Type: new Abstract: In Rhetorical Structure Theory (RST), each edge of a discourse tree carries a nuclearity label, marking which span holds the central content, and a relation drawn from a closed inventory of functional categories. Neither label is a surface property: the same two clauses can be Elaboration or Background depending on rhetorical intent, and the boundary between them is fixed by annotation convention rather than by lexical cues. Predicting these labels therefore requires a formal discourse schema, and exposure to discourse-rich pretraining text does not supply it: given only a target EDU pair, a strong LLM recovers nuclearity at 0.794 F1 and relation at 0.738 F1. We propose RAG-RST, a retrieval-augmented framework that supplies this schema as in-context demonstration rather than through parameter updates, raising nuclearity to 0.846 F1 and relation to 0.788 F1 with no training. Two configurations are compared: a precomputed strategy that labels all EDU pairs before parsing, and a dynamic strategy that retrieves at each reduce step, so that the exemplars reaching the generator depend on the reductions the parser has already committed to. Both outperform existing RST parsers across span, nuclearity, relation, and full tree metrics, with the dynamic variant consistently strongest. Two properties of the retrieved schema emerge: its utility is bounded in number, saturating and then declining beyond fifty exemplars, and it depends on structural position. Both indicate that a formal linguistic schema can be supplied to a foundation model as a demonstration rather than trained into its parameters.
Abstract:In Rhetorical Structure Theory (RST), each edge of a discourse tree carries a nuclearity label, marking which span holds the central content, and a relation drawn from a closed inventory of functional categories. Neither label is a surface property: the same two clauses can be Elaboration or Background depending on rhetorical intent, and the boundary between them is fixed by annotation convention rather than by lexical cues. Predicting these labels therefore requires a formal discourse schema, and exposure to discourse-rich pretraining text does not supply it: given only a target EDU pair, a strong LLM recovers nuclearity at 0.794 F1 and relation at 0.738 F1. We propose RAG-RST, a retrieval-augmented framework that supplies this schema as in-context demonstration rather than through parameter updates, raising nuclearity to 0.846 F1 and relation to 0.788 F1 with no training. Two configurations are compared: a precomputed strategy that labels all EDU pairs before parsing, and a dynamic strategy that retrieves at each reduce step, so that the exemplars reaching the generator depend on the reductions the parser has already committed to. Both outperform existing RST parsers across span, nuclearity, relation, and full tree metrics, with the dynamic variant consistently strongest. Two properties of the retrieved schema emerge: its utility is bounded in number, saturating and then declining beyond fifty exemplars, and it depends on structural position. Both indicate that a formal linguistic schema can be supplied to a foundation model as a demonstration rather than trained into its parameters.
Artificial Intelligence (cs.AI)
arXiv:2610.22150 [cs.AI]
(or arXiv:2610.22150v1 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2610.22150 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 339 words
Related coverage
AI-driven development lifecycle using Amazon Bedrock AgentCore
AWS · Sep 3, 2026
Introducing Mistral OCR 4
Mistral AI · Jun 23, 2026
CacheRAG: A Semantic Caching System for Retrieval-Augmented Generation in Knowledge Graph Question Answering
arXiv · Jun 9, 2026
CacheRAG: A Semantic Caching System for Retrieval-Augmented Generation in Knowledge Graph Question Answering
arXiv · Jun 3, 2026