Simplicity Level Estimate (SLE): A Learned Reference-Less Metric for Sentence Simplification
This addresses the problem of evaluating sentence simplification for researchers and practitioners, offering a reference-less metric that improves upon current methods.
The paper tackles the challenge of automatic evaluation for sentence simplification by proposing a new learned metric called Simplicity Level Estimate (SLE), which focuses on simplicity and outperforms almost all existing metrics in correlation with human judgments.
Automatic evaluation for sentence simplification remains a challenging problem. Most popular evaluation metrics require multiple high-quality references -- something not readily available for simplification -- which makes it difficult to test performance on unseen domains. Furthermore, most existing metrics conflate simplicity with correlated attributes such as fluency or meaning preservation. We propose a new learned evaluation metric (SLE) which focuses on simplicity, outperforming almost all existing metrics in terms of correlation with human judgements.