Text Adaptation to Plain Language and Easy Read via Automatic Post-Editing Cycles
This addresses the need for accessible text adaptation for people with reading difficulties, though it appears incremental as it builds on existing LLM methods.
The researchers tackled the problem of automatically adapting Spanish text to Plain Language and Easy Read formats using iterative post-editing cycles with Large Language Models, achieving first and second place in the CLEARS challenge based on average official metrics.
We describe Vicomtech's participation in the CLEARS challenge on text adaptation to Plain Language and Easy Read in Spanish. Our approach features automatic post-editing of different types of initial Large Language Model adaptations, where successive adaptations are generated iteratively until readability and similarity metrics indicate that no further adaptation refinement can be successfully performed. Taking the average of all official metrics, our submissions achieved first and second place in Plain language and Easy Read adaptation, respectively.