NLP is concerned with the analysis, interpretation and generation of human language and its automation in technology. Research in this area focuses on multilingual LLMs, speech recognition and processing, machine translation, and retrieval-augmented language generation.
Using computational techniques and algorithms to automatically translate text or speech from one language to another.
Key researchers: André Martins, Chryssa Zerva
Using advanced AI models that can understand, generate, and translate text in multiple languages. Multilingual LLMs are trained on data from many different languages.
Key researchers: André Martins, Bruno Martins
Advancing computational techniques for processing, understanding, and generating human speech and language.
Key researchers: Alberto Abad, Isabel Trancoso
Developing methods for searching and retrieving relevant information from a large collection of structure (such as a database, documents, or the web), and unstructured data (like text, images, or videos).
Key researchers: Bruno Martins, Arlindo Oliveira
We develop advanced algorithms for the analysis and understanding of mixed media documents, combining text, images, and other modalities. We are particularly interested in exploring the internal representations of multimodal language models, aiming to uncover how these systems integrate and process diverse types of information and how these internal representations can be mapped to the internal representations in human brains. In parallel, we also examine the intellectual property implications that arise in the training and deployment of such models, contributing to the broader discussion on ethical and legal challenges in AI. Finally, we are exploring the integration of reasoning abilities in multimodal language models and their applications in agentic artificial intelligence.
Key researchers: Arlindo Oliveira, Bruno Martins