On Improving Summarization Factual Consistency from Natural Language Feedback
This work addresses the issue of factual errors in AI-generated summaries, which is critical for users relying on accurate information, though it is incremental as it builds on existing feedback methods.
The paper tackled the problem of improving factual consistency in summarization by collecting a dataset (DeFacto) with human feedback, and showed that fine-tuned language models can leverage this dataset to enhance summary factual consistency, while large language models struggled in zero-shot tasks requiring controllable generation.
Despite the recent progress in language generation models, their outputs may not always meet user expectations. In this work, we study whether informational feedback in natural language can be leveraged to improve generation quality and user preference alignment. To this end, we consider factual consistency in summarization, the quality that the summary should only contain information supported by the input documents, as the user-expected preference. We collect a high-quality dataset, DeFacto, containing human demonstrations and informational natural language feedback consisting of corrective instructions, edited summaries, and explanations with respect to the factual consistency of the summary. Using our dataset, we study three natural language generation tasks: (1) editing a summary by following the human feedback, (2) generating human feedback for editing the original summary, and (3) revising the initial summary to correct factual errors by generating both the human feedback and edited summary. We show that DeFacto can provide factually consistent human-edited summaries and further insights into summarization factual consistency thanks to its informational natural language feedback. We further demonstrate that fine-tuned language models can leverage our dataset to improve the summary factual consistency, while large language models lack the zero-shot learning ability in our proposed tasks that require controllable text generation.