Context Convergence Improves Answering Inferential Questions

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Mozafari, Jamshid, Piryani, Bhawna, Jatowt, Adam
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914559774687232
author Mozafari, Jamshid
Piryani, Bhawna
Jatowt, Adam
author_facet Mozafari, Jamshid
Piryani, Bhawna
Jatowt, Adam
contents While Large Language Models (LLMs) are widely used in open-domain Question Answering (QA), their ability to handle inferential questions-where answers must be derived rather than directly retrieved-remains still underexplored. This study investigates how the structure and quality of passages influence LLM performance on such questions. We focus on convergence, a measure of how effectively sentences (hints) eliminate incorrect answers, as a criterion for constructing passages. Using subsets of the TriviaHG dataset, we form passages by combining sentences with varying convergence levels and evaluate six LLMs of different sizes and architectures. Our results show that passages built from higher convergence sentences lead to substantially better answer accuracy than those selected by cosine similarity, indicating that convergence captures meaningful relevance for inferential reasoning. Additionally, ordering sentences by descending convergence slightly improves performance, suggesting that LLMs tend to prioritize earlier, information-rich cues. These findings highlight convergence as a practical signal for guiding passage construction and analyzing inferential reasoning behavior in LLMs.
format Preprint
id arxiv_https___arxiv_org_abs_2605_12370
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Context Convergence Improves Answering Inferential Questions
Mozafari, Jamshid
Piryani, Bhawna
Jatowt, Adam
Computation and Language
Information Retrieval
While Large Language Models (LLMs) are widely used in open-domain Question Answering (QA), their ability to handle inferential questions-where answers must be derived rather than directly retrieved-remains still underexplored. This study investigates how the structure and quality of passages influence LLM performance on such questions. We focus on convergence, a measure of how effectively sentences (hints) eliminate incorrect answers, as a criterion for constructing passages. Using subsets of the TriviaHG dataset, we form passages by combining sentences with varying convergence levels and evaluate six LLMs of different sizes and architectures. Our results show that passages built from higher convergence sentences lead to substantially better answer accuracy than those selected by cosine similarity, indicating that convergence captures meaningful relevance for inferential reasoning. Additionally, ordering sentences by descending convergence slightly improves performance, suggesting that LLMs tend to prioritize earlier, information-rich cues. These findings highlight convergence as a practical signal for guiding passage construction and analyzing inferential reasoning behavior in LLMs.
title Context Convergence Improves Answering Inferential Questions
topic Computation and Language
Information Retrieval
url https://arxiv.org/abs/2605.12370