Unsupervised Lexical Simplification with Context Augmentation
This work addresses lexical simplification for multiple languages, offering an unsupervised approach that could benefit applications like text accessibility, though it appears incremental as it builds on existing methods and data.
The paper tackles lexical simplification by proposing an unsupervised method that uses monolingual data and pre-trained language models to generate substitutes, achieving substantial performance improvements over other unsupervised systems on the TSAR-2022 shared task across English, Portuguese, and Spanish, and setting new state-of-the-art results on SWORDS.
We propose a new unsupervised lexical simplification method that uses only monolingual data and pre-trained language models. Given a target word and its context, our method generates substitutes based on the target context and also additional contexts sampled from monolingual data. We conduct experiments in English, Portuguese, and Spanish on the TSAR-2022 shared task, and show that our model substantially outperforms other unsupervised systems across all languages. We also establish a new state-of-the-art by ensembling our model with GPT-3.5. Lastly, we evaluate our model on the SWORDS lexical substitution data set, achieving a state-of-the-art result.