Lexical simplification for the systematic support of cognitive accessibility guidelines

• Autores: Rodrigo Alarcon Garcia
• Directores de la Tesis: Lourdes Moreno (dir. tes.)
• Lectura: En la Universidad Carlos III de Madrid ( España ) en 2022
• Idioma: inglés
• Tribunal Calificador de la Tesis: José Antonio Macías Iglesias (presid.) , Israel González Carrasco (secret.) , Raquel Hervás Ballesteros (voc.)
• Enlaces
  • Tesis en acceso abierto en: e-Archivo
• Resumen

  • The Internet has come a long way in recent years, contributing to the proliferation of large volumes of digitally available information. Through user interfaces we can access these contents, however, they are not accessible to everyone. The main users affected are people with disabilities, who are already a considerable number, but accessibility barriers affect a wide range of user groups and contexts of use in accessing digital information. Some of these barriers are caused by language inaccessibility when texts contain long sentences, unusual words and complex linguistic structures. These accessibility barriers directly affect people with cognitive disabilities.

    For the purpose of making textual content more accessible, there are initiatives such as the Easy Reading guidelines, the Plain Language guidelines and some of the language-specific Web Content Accessibility Guidelines (WCAG). These guidelines provide documentation, but do not specify methods for meeting the requirements implicit in these guidelines in a systematic way. To obtain a solution, methods from the Natural Language Processing (NLP) discipline can provide support for achieving compliance with the cognitive accessibility guidelines for the language.

    The task of text simplification aims at reducing the linguistic complexity of a text from a syntactic and lexical perspective, the latter being the main focus of this Thesis. In this sense, one solution space is to identify in a text which words are complex or uncommon, and in the case that there were, to provide a more usual and simpler synonym, together with a simple definition, all oriented to people with cognitive disabilities.

    With this goal in mind, this Thesis presents the study, analysis, design and development of an architecture, NLP methods, resources and tools for the lexical simplification of texts for the Spanish language in a generic domain in the field of cognitive accessibility. To achieve this, each of the steps present in the lexical simplification processes is studied, together with methods for word sense disambiguation. As a contribution, different types of word embedding are explored and created, supported by traditional and dynamic embedding methods, such as transfer learning methods. In addition, since most of the NLP methods require data for their operation, a resource in the framework of cognitive accessibility is presented as a contribution.