Recent work has shown that large language models (LLMs) are capable of generating summaries zero-shot—i.e., without explicit supervision—that, under human assessment, are often comparable or even preferred to manually composed reference summaries. However, this prior work has focussed almost exclusively on evaluating news article summarization. How do zero-shot summarizers perform in other (potentially more specialized) domains?In this work we evaluate zero-shot generated summaries across specialized domains including: biomedical articles, and legal bills (in addition to standard news benchmarks for reference). We focus especially on the factuality of outputs. We acquire annotations from domain experts to identify inconsistencies in summaries and systematically categorize these errors. We analyze whether the prevalence of a given domain in the pretraining corpus affects extractiveness and faithfulness of generated summaries of articles in this domain. We release all collected annotations to facilitate additional research toward measuring and realizing factually accurate summarization, beyond news articles (The dataset can be downloaded from https://anonymous.4open.science/r/zero_shot_faceval_domains-9B83)
more »
« less
SQuALITY: Building a Long-Document Summarization Dataset the Hard Way
Summarization datasets are often assembled either by scraping naturally occurring public-domain summaries -- which are nearly always in difficult-to-work-with technical domains -- or by using approximate heuristics to extract them from everyday text -- which frequently yields unfaithful summaries. In this work, we turn to a slower but more straightforward approach to developing summarization benchmark data: We hire highly-qualified contractors to read stories and write original summaries from scratch. To amortize reading time, we collect five summaries per document, with the first giving an overview and the subsequent four addressing specific questions. We use this protocol to collect SQuALITY, a dataset of question-focused summaries built on the same public-domain short stories as the multiple-choice dataset QuALITY (Pang et al., 2021). Experiments with state-of-the-art summarization systems show that our dataset is challenging and that existing automatic evaluation metrics are weak indicators of quality.
more »
« less
- Award ID(s):
- 2046556
- NSF-PAR ID:
- 10441669
- Date Published:
- Journal Name:
- Proceedings of the 2022 Conference on Empirical Methods in Natural Language Processing
- Page Range / eLocation ID:
- 1139-1156
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Progress in summarizing long texts is inhibited by the lack of appropriate evaluation frameworks. A long summary that appropriately covers the facets of that text must also present a coherent narrative, but current automatic and human evaluation methods fail to identify gaps in coherence. In this work, we introduce SNaC, a narrative coherence evaluation framework for fine-grained annotations of long summaries. We develop a taxonomy of coherence errors in generated narrative summaries and collect span-level annotations for 6.6k sentences across 150 book and movie summaries. Our work provides the first characterization of coherence errors generated by state-of-the-art summarization models and a protocol for eliciting coherence judgments from crowdworkers. Furthermore, we show that the collected annotations allow us to benchmark past work in coherence modeling and train a strong classifier for automatically localizing coherence errors in generated summaries. Finally, our SNaC framework can support future work in long document summarization and coherence evaluation, including improved summarization modeling and post-hoc summary correction.more » « less
-
Tabular data provide answers to a significant portion of search queries. However, reciting an entire result table is impractical in conversational search systems. We propose to generate natural language summaries as answers to describe the complex information contained in a table. Through crowdsourcing experiments, we build a new conversation-oriented, open-domain table summarization dataset. It includes annotated table summaries, which not only answer questions but also help people explore other information in the table. We utilize this dataset to develop automatic table summarization systems as SOTA baselines. Based on the experimental results, we identify challenges and point out future research directions that this resource will support.more » « less
-
Abstractive summarization models often generate inconsistent summaries containing factual errors or hallucinated content. Recent works focus on correcting factual errors in generated summaries via post-editing. Such correction models are trained using adversarial non-factual summaries constructed using heuristic rules for injecting errors. However, generating non-factual summaries using heuristics often does not generalize well to actual model errors. In this work, we propose to generate hard, representative synthetic examples of non-factual summaries through infilling language models. With this data, we train a more robust fact-correction model to post-edit the summaries to improve factual consistency. Through quantitative and qualitative experiments on two popular summarization datasets— CNN/DM and XSum—we show that our approach vastly outperforms prior methods in correcting erroneous summaries. Our model—FactEdit—improves factuality scores by over ~11 points on CNN/DM and over ~31 points on XSum on average across multiple summarization models, producing more factual summaries while maintaining competitive summarization quality.more » « less
-
Abstract Eliciting informative user opinions from online reviews is a key success factor for innovative product design and development. The unstructured, noisy, and verbose nature of user reviews, however, often complicate large-scale need finding in a format useful for designers without losing important information. Recent advances in abstractive text summarization has created the opportunity to systematically generate opinion summaries from online reviews to inform the early stages of product design and development. However, two knowledge gaps hinder the applicability of opinion summarization methods in practice. First, there is a lack of formal mechanisms to guide the generative process with respect to different categories of product attributes and user sentiments. Second, the annotated training datasets needed for supervised training of abstractive summarization models are often difficult and costly to create. This article addresses these gaps by (1) devising an efficient computational framework for abstractive opinion summarization guided by specific product attributes and sentiment polarities, and (2) automatically generating a synthetic training dataset that captures various degrees of granularity and polarity. A hierarchical multi-instance attribute-sentiment inference mode is developed for assembling a high-quality synthetic dataset, which is utilized to fine-tune a pretrained language model for abstractive summary generation. Numerical experiments conducted on a large dataset scraped from three major e-Commerce retail store for apparel and footwear products indicate the performance, feasibility, and potentials of the developed framework. Several directions are provided for future exploration in the area of automated opinion summarization for user-centered design.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Conference Paper:
- https://doi.org/10.18653/v1/2022.emnlp-main.75
