Chain of Summaries: Summarization Through Iterative Questioning

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Brach, William, Košťál, Kristián, Poech, Lukas Galke
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866912908864126976
author Brach, William
Košťál, Kristián
Poech, Lukas Galke
author_facet Brach, William
Košťál, Kristián
Poech, Lukas Galke
contents Large Language Models (LLMs) are increasingly using external web content. However, much of this content is not easily digestible by LLMs due to LLM-unfriendly formats and limitations of context length. To address this issue, we propose a method for generating general-purpose, information-dense summaries that act as plain-text repositories of web content. Inspired by Hegel's dialectical method, our approach, denoted as Chain of Summaries (CoS), iteratively refines an initial summary (thesis) by identifying its limitations through questioning (antithesis), leading to a general-purpose summary (synthesis) that can satisfy current and anticipate future information needs. Experiments on the TriviaQA, TruthfulQA, and SQUAD datasets demonstrate that CoS outperforms zero-shot LLM baselines by up to 66\% and specialized summarization methods such as Chain of Density, BRIO and PEGASUS by up to 27\%. CoS-generated summaries yield higher Q\&A performance compared to the source content, while requiring substantially fewer tokens and being agnostic to the specific downstream LLM. CoS thus resembles an appealing option for website maintainers to make their content more accessible for LLMs, while retaining possibilities for human oversight.
format Preprint
id arxiv_https___arxiv_org_abs_2511_15719
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Chain of Summaries: Summarization Through Iterative Questioning
Brach, William
Košťál, Kristián
Poech, Lukas Galke
Artificial Intelligence
Computation and Language
Large Language Models (LLMs) are increasingly using external web content. However, much of this content is not easily digestible by LLMs due to LLM-unfriendly formats and limitations of context length. To address this issue, we propose a method for generating general-purpose, information-dense summaries that act as plain-text repositories of web content. Inspired by Hegel's dialectical method, our approach, denoted as Chain of Summaries (CoS), iteratively refines an initial summary (thesis) by identifying its limitations through questioning (antithesis), leading to a general-purpose summary (synthesis) that can satisfy current and anticipate future information needs. Experiments on the TriviaQA, TruthfulQA, and SQUAD datasets demonstrate that CoS outperforms zero-shot LLM baselines by up to 66\% and specialized summarization methods such as Chain of Density, BRIO and PEGASUS by up to 27\%. CoS-generated summaries yield higher Q\&A performance compared to the source content, while requiring substantially fewer tokens and being agnostic to the specific downstream LLM. CoS thus resembles an appealing option for website maintainers to make their content more accessible for LLMs, while retaining possibilities for human oversight.
title Chain of Summaries: Summarization Through Iterative Questioning
topic Artificial Intelligence
Computation and Language
url https://arxiv.org/abs/2511.15719