Artificial Intelligence

About the Domain

Despite a decade-long history, the Artificial Intelligence (AI) domain is shaping the future of multiple areas — from natural language processing to predictive modelling and decision support systems. With growing impact in sectors like healthcare, transport, and manufacturing, AI is already part of our daily lives through virtual assistants and recommendation systems. At INESC TEC, we explore the frontiers of AI, from fundamental research to the development of innovative technologies. To achieve this, we bring together scientists from various fields, driving new discoveries and applications.

We feature a wide range of competences that allow us to develop intelligent and adaptive solutions, capable of extracting knowledge from large volumes of data and supporting decision-making in complex, real-time contexts. Some of these areas of expertise include:

Machine Learning

Natural Language Processing

Knowledge extraction from continuous data streams

Decision Support Systems

Research Challenges

Our research work in Artificial Intelligence focuses on creating reusable resources and advanced models for complex tasks, ensuring efficiency in both learning and implementation. We aim to develop AI in a transparent and understandable way, enabling humans to analyse, learn from, and contribute to it. We also explore new ways to improve AI perception in dynamic, noisy, and multimodal environments, making it more adaptable and robust to tackle real-world challenges.

Main Achievements

Our research in Artificial Intelligence led to internationally recognised advances: from automatic knowledge extraction to transparency and explainability in AI models. With proven impact across academia, industry, and society, our developments are widely acknowledged and applied across multiple fields. Some of our main achievements include:

Multilingual keyword extraction

We developed YAKE!, an innovative system for automatic keyword extraction, used by organisations like the National Library of Finland and Volkswagen. Considered state-of-the-art, it has been cited in over 1000 scientific publications and is used in more than 800 projects.

Transparent Artificial Intelligence

We created innovative methods for visually explaining AI models, ensuring privacy, intelligibility, and relevance of evidence. Our work includes pioneering techniques for the anonymisation of medical images, which have gained wide recognition - with one of the most cited reviews in the field (over 1400 citations).

Ordinal data classification

We are a global reference in learning with ordinal data and rankings, being the first to objectively define the concept of ordinal classification and propose new methods in this area. Our work has received over 1000 citations and continues to shape research worldwide.