Easy Reading of Automatically Generated Texts

Abstract

Technology has managed to incorporate itself into our lives on a daily basis, reaching a turning point where its absence seems inconceivable to us. However, although these technologies are used by numerous people, technological accessibility aims to ensure that any resource, through any medium, is available to all people, whether or not they have any type of disability. Although there are several techniques to achieve cognitive accessibility, in this project we will make use of the Easy Reading Methodology.
The Easy Reading Methodology provides guidelines to facilitate the comprehension of materials for different groups, such as people with cognitive disabilities, foreigners with limited language skills, among others. Currently, the Easy Reading Methodology is applied manually to achieve the cognitive accessibility of materials.
This project proposes the development of several methods and their corresponding software application to automatically apply a subset of the Easy Reading guidelines to texts in Spanish containing written cardinal numbers, percentages, fractions and parentheses. Specifically, all numbers written in letters will be replaced by digits, decimal numbers will be replaced by whole numbers, and large numbers will be rounded and words such as ≪almost≫ or ≪more than≫ will be added to reduce the complexity in reading them. As for percentages and fractions, they will be replaced by words that preserve the meaning of the sentence and, in addition, we will reduce the complexity of the sentences by eliminating parentheses and the content within them.
For this we have made use of Natural Language Processing (NLP) techniques for computer understanding of human language and rule-based programming for substitutions. Likewise, we have developed a web application with Django following some of the requirements exposed in the Easy Reading Methodology for the design, which will add the required accessibility to our application.