Líneas de investigación

The proposed research line focuses on the development of technology based on Artificial Intelligence (AI) and Large Language Models (LLM) from an inclusive and sustainable perspective, seeking to advance how these technologies can be improved along with human-computer interaction (HCI). This line addresses several key aspects, such as:

• Accessibility and Inclusion: Research dedicated to creating AI and LLM systems that are universally accessible, with a particular focus on adaptive technologies that can be used by people with diverse abilities, including those with disabilities. Methods to make interaction with AI more accessible and intuitive for less experienced users are explored.
• Personalization and Contextualization: Development of AI and LLM models that not only generate natural language but are also capable of adapting their responses and solutions to the specific needs, preferences, and contexts of users. This includes content personalization, interface adaptability, and the ability to interpret human intentions and emotions more accurately.
• Sustainability: Research on how to make AI and LLMs more energy and computationally efficient. This approach seeks responsible use and minimizing the carbon footprint of AI technologies, developing algorithms that require less computing power and promoting sustainable development practices within the technological community.
• Human-AI Collaboration: Exploration of new models of interaction between humans and AI systems that go beyond assistance to joint problem-solving. We investigate how machines can act not just as tools or assistants but as collaborators that enhance human capabilities.

Research and development of Natural Language Processing (NLP) and Artificial Intelligence (AI) technologies for text simplification, with the goal of making textual content more comprehensible systematically and automatically. Our expertise is particularly highlighted in working with people with cognitive limitations, such as individuals with intellectual disabilities and the elderly, among others. We began our work using more traditional NLP methods, evolving towards the use of transformers and, finally, to Large Language Models (LLMs) that we apply today. This technological progress has allowed us to explore and enhance text simplification through methods, languages, and specialized linguistic resources.

We address different simplification approaches depending on the target audience, having notable expertise in simplification following two similar but different initiatives, such as Plain Language and Easy to Read guidelines. Our work also includes the creation of corpora within the framework of Plain Language and Easy Read, adapting solutions to various domains such as public administrations, education, and health.

Furthermore, we research technologies to support professionals in adapting texts to Easy Read and Plain Language, offering tools for complex word detection, readability evaluation, and summary generation, among other functionalities.

An essential aspect of our research is the constant validation with individuals and the target audience of the work done using Human-Computer Interaction (HCI) methods, ensuring that our solutions are not only technologically advanced but also effectively helpful and accessible to those for whom they are intended.

Highlighted Projects:

• EASIER
• Cognitive and Sensory Accessibility in Video Conferencing Systems (Access2Meet)

Research and development of technologies supporting the design and development of accessible interfaces (websites, apps, among others), optimizing the user experience (UX). We have extensive experience in accessible technologies, ensuring they can be accessed and used by anyone, including those with disabilities. Part of our accessibility and disability experience has been enriched by our long collaboration with the Spanish Center for Subtitling and Audio Description, a centre under the Royal Board on Disability (Ministry of Social Rights and 2030 Agenda) led by Carlos III University of Madrid. Our experience covers various types of disabilities (visual, auditory, physical, intellectual, and seniors) and diverse application domains.

We follow approaches that combine, on the one hand, development in accordance with essential standards and regulatory frameworks and, on the other hand, people-centred designs that capture those interaction requirements not considered in the standards, using accessibility methods and HCI (Human-Computer Interaction).

We have extensive knowledge of accessibility standards (UNE, ETSI, ISO), such as the WCAG, UAAG, ATAG, COGA, and EN 301 549, among others. We work with more traditional HCI methods, such as user-centred designs, as well as co-design approaches, like Design Thinking and the Double Diamond model.

Highlighted Projects:

• Sensory and Cognitive Accessibility in the Communication and Management of Telematic and Telephone Services of the AGE (Access2Citizen)
• Cognitive and Sensory Accessibility to Videoconferencing Systems (Access2Meet )
• A Model-Based Framework for the Development of Accessible Services in e-Government (e-Gobernability)

This line focuses on the development and implementation of advanced technological solutions to overcome the communication barriers and access to information faced by people with disabilities, especially in educational environments.

The goal is not only to facilitate access to digital learning materials in an inclusive manner but also to enhance the overall educational experience through the integration of subtitling services, audio descriptions, and sign language adapted to the needs of students with hearing and visual disabilities. Part of our experience in accessibility and disability has been enriched by our long collaboration with the Spanish Center for Subtitling and Audio Description, a centre under the Royal Board on Disability (Ministry of Social Rights and 2030 Agenda) led by Carlos III University of Madrid.

This research line addresses the challenge of creating accessible and usable learning environments, applying automatic voice recognition techniques, natural language processing, and other artificial intelligence technologies, with the aim of producing accessible and high-quality educational content.

Proyectos más destacados: PICOGAC, Access2Class

Esta línea de investigación se centra en el desarrollo y aplicación de técnicas avanzadas de PLN y algoritmos de aprendizaje profundo para la extracción, análisis y aprovechamiento de información clínica contenida en historias clínicas electrónicas (HCE) no estructuradas.

Busca transformar grandes volúmenes de datos textuales en información estructurada y procesable, con el objetivo de mejorar la práctica clínica, apoyar la toma de decisiones médicas personalizadas, y facilitar la investigación, especialmente en estudios epidemiológicos que requieren la revisión y análisis de un extenso número de registros clínicos.

Este enfoque no solo promete una optimización significativa de los recursos y tiempo en la investigación médica sino también una adaptabilidad a diversos contextos clínicos y patologías, subrayando la versatilidad y potencial transformador de las tecnologías de PLN y deep learning en el ámbito de la salud.

Proyectos más destacados: 

• Extracción de información clínica usando deep learning y técnicas de Big Data (DeepEMR)

Our research lines reflect the journey from our beginnings, showcasing how the foundations laid in areas such as natural language processing, biomedicine, web development, and accessibility have evolved into the current cutting-edge directions.

This journey demonstrates not only our adaptability and growth but also underpins our present investigations. These research lines, which continue to be relevant, are as follows:

Extraction and Retrieval of Information in the Biomedical Domain

Research and development of automatic tools to manage the exponential growth of data and biomedical publications, facilitating the extraction of valuable knowledge for professionals in the field.

Methodological Frameworks for the Development of Accessible Web Applications

Creation and guidelines for the application of accessible applications and websites through the methodological support of AWA, aimed at integrating accessibility as an essential requirement in web development.

Accessibility in the Educational Environment

Research and development of Information and Communication Technologies (ICT) that can overcome accessibility barriers in e-learning and educational content management systems, promoting inclusive education for all.

Question Answering Systems

Research and development of Question Answering Systems (QAS) that allow obtaining precise information through natural language questions, addressing challenges.

Recognition of Named Entities and Temporal Expressions

Research in the recognition and classification of named entities and the automatic processing of temporal information in texts is needed to improve the performance of Natural Language Processing applications.

Conceptual Modeling, Integrity Constraints, and Business Rules

Research on data models to improve the management, adaptability, and integration of information systems, ensuring the accuracy and completeness of databases.