Controllable Summarization with Constrained Markov Decision Process

• Published in: Transactions of the Association for Computational Linguistics Vol. 9; pp. 1213 - 1232
• Main Authors: Chan, Hou Pong, Wang, Lu, King, Irwin
• Format: Journal Article
• Language: English
• Published: One Rogers Street, Cambridge, MA 02142-1209, USA MIT Press 05.11.2021; MIT Press Journals, The; The MIT Press
• Subjects: Automatic summarization; Constraints; Controllability; Fines & penalties; Linguistics; Markov analysis; Markov processes; Performance evaluation; Summaries
• ISSN: 2307-387X; 2307-387X
• DOI: 10.1162/tacl_a_00423

We study , which allows users to gain control on a particular attribute (e.g., length limit) of the generated summaries. In this work, we propose a novel training framework based on (CMDP), which conveniently includes a along with a set of , to facilitate better summarization control. The reward function encourages the generation to resemble the human-written reference, while the constraints are used to explicitly prevent the generated summaries from violating user-imposed requirements. Our framework can be applied to control important attributes of summarization, including , , and , as we devise specific constraints for each of these aspects. Extensive experiments on popular benchmarks show that our CMDP framework helps generate informative summaries while complying with a given attribute’s requirement.