Annotating and Modeling Fine-grained Factuality in Summarization

By Tanya Goyal, Greg Durrett. 2021

180 citations recorded by the source corpus.

Recent pre-trained abstractive summarization systems have started to achieve credible performance, but a major barrier to their use in practice is their propensity to output summaries that are not faithful to the input and that contain factual errors. While a number of annotated datasets and statistical models for assessing factuality have been explored, there is no clear picture of what errors are most important to target or where current techniques are succeeding and failing. We explore both synthetic and human-labeled data sources for training models to identify factual errors in summarization, and study factuality at the word-, dependency-, and sentence-level. Our observations are threefold. First, exhibited factual errors differ significantly across datasets, and commonly-used training sets of simple synthetic errors do not reflect errors made on abstractive datasets like XSum. Second, human-labeled data with fine-grained annotations provides a more effective training signal than sentence-level annotations or synthetic data. Finally, we show that our best factuality detection model enables training of more factual XSum summarization models by allowing us to identify non-factual tokens in the training data.

Subfields: Artificial Intelligence, Natural Language Processing, Evaluation & Benchmarks

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Natural Language Processing

In Evaluation & Benchmarks

Citations in the corpus

Repairing the Cracked Foundation: A Survey of Obstacles in Evaluation Practices for Generated Text (2022, background): “Additionally complicating this problem is that artificially introduced errors rarely match errors of real summarization models, which means that metrics trained on synthetic errors may not generalize to real systems (Goyal and Durrett, 2021).”
Improving Faithfulness in Abstractive Summarization with Contrast Candidate Generation and Selection (2021, background): “Our study instead focuses on the more challenging setting (Goyal and Durrett, 2021) where a large part of training data suffers from extrinsic and intrinsic hallucinations, and provides cross-system analysis on the both hallucinations categories.”
FactPEGASUS: Factuality-Aware Pre-training and Fine-tuning for Abstractive Summarization (2022, methodology): “We show that the full model improves three factuality metrics, the token and sentence error of DEP Entail (Goyal and Durrett, 2021) and FactCC, on the downstream datasets of XSum (Narayan et al., 2018), WikiHow (Koupaee and Wang, 2018), and Gigaword (Rush et al., 2015).”
Are Factuality Checkers Reliable? Adversarial Meta-evaluation of Factuality in Summarization (2021, background): “t factuality as different problems: text entailment over sequential (Kryscinski et al., 2020; Goyal and Durrett, 2021a) or tree (Goyal and Durrett, 2020, 2021a) structures; question answering (Wang et al.”
FactGraph: Evaluating Factuality in Summarization with Semantic Graph Representations (2022, unknown): “Third, previous work (DAE, Goyal and Durrett, 2021) treats semantic relations as isolated units, not simultaneously considering the entire semantic structure of both document and summary texts.”
Trusting Your Evidence: Hallucinate Less with Context-aware Decoding (2023, background): “, 2020) and removing noisy training samples (Kang and Hashimoto, 2020; Goyal and Durrett, 2021).”
CLIFF: Contrastive Learning for Improving Faithfulness and Factuality in Abstractive Summarization (2021, background): “Three types of remedies have been proposed: running a separately learned error correction component (Dong et al., 2020), removing noisy training samples (Nan et al., 2021; Goyal and Durrett, 2021), and modifying the Transformer architecture (Huang et al., 2020; Zhu et al., 2021).”
News Summarization and Evaluation in the Era of GPT-3 (2022, background): “Annotator rationales often mentioned misleading or incorrect information as the primarily reason for selecting BRIO as the worst summary, confirming the issues that have been observed with XSum-trained models (Maynez et al., 2020; Pagnoni et al., 2021; Goyal and Durrett, 2021).”
Spurious Correlations in Reference-Free Evaluation of Text Generation (2022, methodology): “We chose Electra in order to match the model architecture in DAE.”
Faithful or Extractive? On Mitigating the Faithfulness-Abstractiveness Trade-off in Abstractive Summarization (2021, methodology): “methods to improve the faithfulness of abstractive summarization systems (Matsumaru et al., 2020; Zhao et al., 2020; Dong et al., 2020; Goyal and Durrett, 2021; Xu et al., 2020; Chen et al., 2021; Zhu et al., 2021).”
A Survey on Hallucination in Large Language Models: Principles, Taxonomy, Challenges, and Open Questions (2023, unknown): “…hallucinations, such as abstractive summarization (Kryscinski et al., 2020; Wang et al., 2020a; Maynez et al., 2020; Fabbri et al., 2021; Goyal and Durrett, 2021; Pagnoni et al., 2021; Tang et al., 2022), data-to-text(Tian et al., 2019; Parikh et al., 2020), and machine translation (Zhou et al.,…”
Factual Error Correction for Abstractive Summaries Using Entity Retrieval (2022, unknown): “Since factual errors frequently appear in entity-level (Goyal and Durrett, 2021), we develop a system that is specialized in correcting entitylevel errors.”
Evaluating Factuality in Text Simplification (2022, background): “This category is similar to extrinsic hallucination in the summarization literature (Maynez et al., 2020; Goyal and Durrett, 2021).”
Correcting Diverse Factual Errors in Abstractive Summarization via Post-Editing and Language Model Infilling (2022, methodology): “We therefore evaluate our model using two metrics - FactCC (Kryscin-ski et al., 2020) and Ent-DAE (Goyal and Durrett, 2021); each captures different error types.”