![](https://valgrai.eu/wp-content/uploads/2023/10/lenguaje_3.png)
![](https://valgrai.eu/wp-content/uploads/2023/10/lenguaje_3.png)
Adapting Court Rulings to Easy Reading Using Deep Learning
Judgments modifying the capacity to act of individuals restrict, in some circumstances, the exercise of legal rights without limiting the capacity to be holders of those rights. In many cases, people who have their capacity to act are limited in whole or in part also find it difficult to read and fully understand judgments that limit their capacity.
There are initiatives led by foundations and non-profit associations to manually adapt court rulings that limit people’s ability to act to what is known as Easy Reading. Easy Reading includes a set of guidelines and recommendations related to the writing of texts and the validation of their comprehensibility, aimed at making information accessible to people with reading comprehension difficulties. Despite this effort, the number of adapted judgments is still very small.
The scarcity of adapted sentences makes it unfeasible to use classical natural language processing techniques or to learn sentence simplification models using deep learning. However, in recent years the use of neural language models based on deep learning has shown its power to establish semantic relationships between words or phrases. The use of these language models and the establishment of semantic distances between judicial rulings and rights that may be limited allows the creation of automatic models for the adaptation of court rulings to Easy Reading.
Applications:
• Adaptation of court rulings that limit the ability to act to Easy Reading.
Technical advantages:
•Adaptation is carried out automatically.
Relevant Experience:
• A prototype and controlled evaluation of this technique have been successfully carried out with the help of the ICAV and the Espurna Foundation.”