Simple-QE: Better Automatic Quality Estimation for Text Simplification

TL;DR

Simple-QE is a BERT-based quality estimation model for text simplification that correlates well with human judgments and does not require reference texts, making it practical for real-world use.

Contribution

The paper introduces Simple-QE, a novel reference-free quality estimation model for text simplification based on BERT, adapted from summarization QE work.

Findings

Simple-QE correlates strongly with human quality judgments.

It accurately predicts the complexity of human-written texts.

The model is practical for real-world applications without needing references.

Abstract

Text simplification systems generate versions of texts that are easier to understand for a broader audience. The quality of simplified texts is generally estimated using metrics that compare to human references, which can be difficult to obtain. We propose Simple-QE, a BERT-based quality estimation (QE) model adapted from prior summarization QE work, and show that it correlates well with human quality judgments. Simple-QE does not require human references, which makes the model useful in a practical setting where users would need to be informed about the quality of generated simplifications. We also show that we can adapt this approach to accurately predict the complexity of human-written texts.