STORYSUMM: Evaluating Faithfulness in Story Summarization

TL;DR

This paper introduces STORYSUMM, a new dataset for evaluating faithfulness in story summarization, revealing current automatic metrics' limitations and emphasizing the need for diverse evaluation methods.

Contribution

The paper presents STORYSUMM, a novel dataset with localized faithfulness labels for stories, and demonstrates the inadequacy of existing automatic metrics in detecting inconsistencies.

Findings

Human annotations often miss inconsistencies in faithfulness.

Existing automatic metrics achieve less than 70% accuracy on the dataset.

Diverse evaluation approaches are necessary for reliable faithfulness assessment.

Abstract

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.