Text Adaptation & Readability
January 1, 2022
·
1 min read
I develop models and resources to adapt text for different audiences and contexts, from sentence-level simplification to document-level plain language, with a focus on language learners, low-literacy readers, and general audiences. My work combines system development with rigorous human-centred evaluation, ensuring outputs are assessed by and with the people they are designed for.
- 💬 Sentence and document simplification with controllable outputs (e.g. ASSET, CEFR-controlled simplification)
- 📊 Evaluation of simplification metrics, with insights into their human-alignment and limitations (e.g. meta-evaluation of metrics)
- 🧪 Benchmark creation (e.g. TSAR Shared Task, BLESS, SimpleTICO-19)
- 🛠️ Tools: EASSE, readability predictors for multiple languages
Funded project: TINE: Plain Language Adaptation of Multilingual Science
