STORYSUMM: Evaluating Faithfulness in Story Summarization

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Subbiah, Melanie, Ladhak, Faisal, Mishra, Akankshya, Adams, Griffin, Chilton, Lydia B., McKeown, Kathleen
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916668815441920
author Subbiah, Melanie
Ladhak, Faisal
Mishra, Akankshya
Adams, Griffin
Chilton, Lydia B.
McKeown, Kathleen
author_facet Subbiah, Melanie
Ladhak, Faisal
Mishra, Akankshya
Adams, Griffin
Chilton, Lydia B.
McKeown, Kathleen
contents Human evaluation has been the gold standard for checking faithfulness in abstractive summarization. However, with a challenging source domain like narrative, multiple annotators can agree a summary is faithful, while missing details that are obvious errors only once pointed out. We therefore introduce a new dataset, STORYSUMM, comprising LLM summaries of short stories with localized faithfulness labels and error explanations. This benchmark is for evaluation methods, testing whether a given method can detect challenging inconsistencies. Using this dataset, we first show that any one human annotation protocol is likely to miss inconsistencies, and we advocate for pursuing a range of methods when establishing ground truth for a summarization dataset. We finally test recent automatic metrics and find that none of them achieve more than 70% balanced accuracy on this task, demonstrating that it is a challenging benchmark for future work in faithfulness evaluation.
format Preprint
id arxiv_https___arxiv_org_abs_2407_06501
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle STORYSUMM: Evaluating Faithfulness in Story Summarization
Subbiah, Melanie
Ladhak, Faisal
Mishra, Akankshya
Adams, Griffin
Chilton, Lydia B.
McKeown, Kathleen
Artificial Intelligence
Computation and Language
Human evaluation has been the gold standard for checking faithfulness in abstractive summarization. However, with a challenging source domain like narrative, multiple annotators can agree a summary is faithful, while missing details that are obvious errors only once pointed out. We therefore introduce a new dataset, STORYSUMM, comprising LLM summaries of short stories with localized faithfulness labels and error explanations. This benchmark is for evaluation methods, testing whether a given method can detect challenging inconsistencies. Using this dataset, we first show that any one human annotation protocol is likely to miss inconsistencies, and we advocate for pursuing a range of methods when establishing ground truth for a summarization dataset. We finally test recent automatic metrics and find that none of them achieve more than 70% balanced accuracy on this task, demonstrating that it is a challenging benchmark for future work in faithfulness evaluation.
title STORYSUMM: Evaluating Faithfulness in Story Summarization
topic Artificial Intelligence
Computation and Language
url https://arxiv.org/abs/2407.06501