Controlling Pre-trained Language Models for Grade-Specific Text Simplification

TL;DR

This paper investigates how to better control pre-trained language models for grade-specific text simplification by predicting instance-specific edit operations, leading to improved simplification quality.

Contribution

It introduces a novel method that predicts edit operations per instance for more effective grade-specific text simplification, surpassing previous corpus-level approaches.

Findings

Instance-specific control improves simplification quality.

Predicting edit operations enhances adequacy and simplicity.

The proposed method outperforms corpus-level heuristics.

Abstract

Text simplification (TS) systems rewrite text to make it more readable while preserving its content. However, what makes a text easy to read depends on the intended readers. Recent work has shown that pre-trained language models can simplify text using a wealth of techniques to control output simplicity, ranging from specifying only the desired reading grade level, to directly specifying low-level edit operations. Yet it remains unclear how to set these control parameters in practice. Existing approaches set them at the corpus level, disregarding the complexity of individual inputs and considering only one level of output complexity. In this work, we conduct an empirical study to understand how different control mechanisms impact the adequacy and simplicity of text simplification systems. Based on these insights, we introduce a simple method that predicts the edit operations required for simplifying a text for a specific grade level on an instance-per-instance basis. This approach improves the quality of the simplified outputs over corpus-level search-based heuristics.