Conditional Generation with a Question-Answering Blueprint

• Published in: Transactions of the Association for Computational Linguistics Vol. 11; pp. 974 - 996
• Main Authors: Narayan, Shashi, Maynez, Joshua, Amplayo, Reinald Kim, Ganchev, Kuzman, Louis, Annie, Huot, Fantine, Sandholm, Anders, Das, Dipanjan, Lapata, Mirella
• Format: Journal Article
• Language: English
• Published: One Broadway, 12th Floor, Cambridge, Massachusetts 02142, USA MIT Press 15.08.2023; MIT Press Journals, The; The MIT Press
• Subjects: Automatic summarization; Automatic text generation; Datasets; Planning; Question answer sequences; Questions
• ISSN: 2307-387X; 2307-387X
• DOI: 10.1162/tacl_a_00583

The ability to convey relevant and faithful information is critical for many tasks in conditional generation and yet remains elusive for neural seq-to-seq models whose outputs often reveal hallucinations and fail to correctly cover important details. In this work, we advocate planning as a useful intermediate representation for rendering conditional generation less opaque and more grounded. We propose a new conceptualization of text plans as a sequence of question-answer (QA) pairs and enhance existing datasets (e.g., for summarization) with a QA operating as a proxy for content selection (i.e., what to say) planning (i.e., in what order). We obtain blueprints automatically by exploiting state-of-the-art question generation technology and convert input-output pairs into input-blueprint-output tuples. We develop Transformer-based models, each varying in how they incorporate the blueprint in the generated output (e.g., as a global plan or iteratively). Evaluation across metrics and datasets demonstrates that blueprint models are more factual than alternatives which do not resort to planning and allow tighter control of the generation output.