Text Simplification in RRI.

Responsible Research and Innovation (RRI) entails doing science, information and communication with society and for society carrying out an involvement of society during the process of research and innovation.

Our interests converge towards the development of interfaces based on language for modern day information technologies applications for modifying or finding information, issuing commands, or present output results in a way easy to understand.

In this proposal, we focus on natural language processing, in particular in the text simplification process to transform a text into a similar text that is easier to read. To do this, one has to identify what causes difficulties to readers and define different transformations, mainly aimed at syntactic and lexical constructions that can be applied to the original text to generate a simplified version.

The way of writing or presenting information can exclude many people, especially those who have problems to read and write or to understand. There are different factors as for example limited cultural education, people have cognitive problems or another disability, people with social problems or people whose mother tongue is not the official language of their adoption country which can cause problems. In particular, there are specific groups like deaf people, autistic people, elderly or people with language disorders such as aphasia or dyslexia, who have problems when they access information. In order to make information accessible to all people, we must keep in mind the diversity of the people who will access it.

Our proposed work on automatic simplification of numerical expressions, is a computational implementation based on a generic model of the process. We present the theoretical bases for text simplification along with a generic model, and show the experimental identification of simplification strategies on numerical expressions to decide what kind of changes need to be implemented for our automatic approaches. We have developed and implemented two systems to simplify numerical expressions in English and Spanish that follow the generic model and used the simplification strategies identified in the experimental studies. For both systems an evaluation with experts has been carried out.