Research Areas
Our areas of research include AI and Data Engineering, especially their meeting points, where we design, develop and study AI and traditional systems to work with large and complex data.
Check out our main fields and their founded projects below, or visit the Publications page to see our published papers for a detailed, academical view of our work.
Research Area
Data-Centric AI
We develop machine learning methods designed for real-world data, where noise, scarcity, heterogeneity, and limited annotation represent the main bottlenecks. Our research focuses on improving the quality and usability of data throughout the learning pipeline.
Funded Projects
Research Area
Natural Language Understanding & Knowledge Graph Enrichment
We study methods for extracting, structuring, and enriching knowledge from unstructured and semi-structured data by combining machine learning, large language models, and symbolic representations. Our work includes natural language understanding, entity and relation extraction, knowledge graph enrichment, and hybrid approaches that integrate statistical and symbolic reasoning.
Funded Projects
HybridAI: a hybrid approach to Natural Language Understanding
Research Area
Interpretable AI
We investigate models and methodologies that make machine learning systems more transparent, understandable, and trustworthy. Our work focuses on interpretable representations, explainability, and decision-support settings in which model behavior must be communicated clearly to human users.
A key component of this research is Symbolic Regression, a machine learning approach that automatically discovers mathematical expressions to model data.
Research Area
Information Sharing, Interoperability, and Retrieval
We study methods for integrating, organizing, and retrieving heterogeneous information across distributed and complex environments. This includes semantic interoperability, information access, and techniques for supporting collaboration and knowledge exchange across systems and domains.
Research Area
Non-Conventional Data Management
We also work on data management approaches for complex, heterogeneous, and non-traditional data settings, including scenarios in which standard relational assumptions are insufficient. This line of research provides methodological foundations for several of the group’s applied and interdisciplinary activities.