Saved in:
Bibliographic Details
Main Author: Gupta, Kshitij
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2502.07747
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913686286761984
author Gupta, Kshitij
author_facet Gupta, Kshitij
contents We present a novel data set, WhoDunIt, to assess the deductive reasoning capabilities of large language models (LLM) within narrative contexts. Constructed from open domain mystery novels and short stories, the dataset challenges LLMs to identify the perpetrator after reading and comprehending the story. To evaluate model robustness, we apply a range of character-level name augmentations, including original names, name swaps, and substitutions with well-known real and/or fictional entities from popular discourse. We further use various prompting styles to investigate the influence of prompting on deductive reasoning accuracy. We conduct evaluation study with state-of-the-art models, specifically GPT-4o, GPT-4-turbo, and GPT-4o-mini, evaluated through multiple trials with majority response selection to ensure reliability. The results demonstrate that while LLMs perform reliably on unaltered texts, accuracy diminishes with certain name substitutions, particularly those with wide recognition. This dataset is publicly available here.
format Preprint
id arxiv_https___arxiv_org_abs_2502_07747
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle WHODUNIT: Evaluation benchmark for culprit detection in mystery stories
Gupta, Kshitij
Computation and Language
Artificial Intelligence
We present a novel data set, WhoDunIt, to assess the deductive reasoning capabilities of large language models (LLM) within narrative contexts. Constructed from open domain mystery novels and short stories, the dataset challenges LLMs to identify the perpetrator after reading and comprehending the story. To evaluate model robustness, we apply a range of character-level name augmentations, including original names, name swaps, and substitutions with well-known real and/or fictional entities from popular discourse. We further use various prompting styles to investigate the influence of prompting on deductive reasoning accuracy. We conduct evaluation study with state-of-the-art models, specifically GPT-4o, GPT-4-turbo, and GPT-4o-mini, evaluated through multiple trials with majority response selection to ensure reliability. The results demonstrate that while LLMs perform reliably on unaltered texts, accuracy diminishes with certain name substitutions, particularly those with wide recognition. This dataset is publicly available here.
title WHODUNIT: Evaluation benchmark for culprit detection in mystery stories
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2502.07747