Generating Summaries with Controllable Readability Levels

TL;DR

This paper introduces three novel techniques for generating summaries with precisely controlled readability levels, enabling tailored content for diverse audiences and improving upon existing methods that lack fine-grained control.

Contribution

The work presents three new methods—instruction-based control, reinforcement learning, and lookahead decoding—for fine-grained readability control in summarization tasks.

Findings

Significant improvement in readability control metrics.

Enhanced human judgment scores for summary readability.

Established strong baselines for controllable readability in news summarization.

Abstract

Readability refers to how easily a reader can understand a written text. Several factors affect the readability level, such as the complexity of the text, its subject matter, and the reader's background knowledge. Generating summaries based on different readability levels is critical for enabling knowledge consumption by diverse audiences. However, current text generation approaches lack refined control, resulting in texts that are not customized to readers' proficiency levels. In this work, we bridge this gap and study techniques to generate summaries at specified readability levels. Unlike previous methods that focus on a specific readability level (e.g., lay summarization), we generate summaries with fine-grained control over their readability. We develop three text generation techniques for controlling readability: (1) instruction-based readability control, (2) reinforcement learning to minimize the gap between requested and observed readability and (3) a decoding approach that uses lookahead to estimate the readability of upcoming decoding steps. We show that our generation methods significantly improve readability control on news summarization (CNN/DM dataset), as measured by various readability metrics and human judgement, establishing strong baselines for controllable readability in summarization.