Mattia Fumagalli Senior lecturer
Contact
Name and title: Mattia FumagalliSenior lecturer
ORCIDhttps://orcid.org/0000-0003-3385-4769 Länk till annan webbplats.
Workplace: Department of Computer and Systems Sciences Länk till annan webbplats.
Visiting address Nodhuset, Borgarfjordsgatan 12
Postal address Institutionen för data- och systemvetenskap164 25 Kista
About me
I hold a Master's Degree in Philosophy and a Ph.D. in Computer Science, and I have the Italian National Scientific Habilitation (Abilitazione Scientifica Nazionale - ASN) for Logic and Philosopy of Science (Associate Professor). My research spans the fields of Conceptual Modeling, Knowledge Discovery, and Knowledge Representation, with a particular focus on advancing the theoretical foundations of information systems design and developing semi-automated solutions for the creation of conceptual models. Other key research interests include Logic and Philosophy of Language.
I was a member of the KRDB group at the Free University of Bozen-Bolzano, where I lectured in Information Systems Design and Research Methods. I also collaborate with the Services and Cybersecurity Group at the University of Twente, and with the Institute of Cognitive Sciences and Technologies (ISTC-CNR). Additionally, I collaborate with the Knowdive Group at the Department of Information Engineering and Computer Science of the University of Trento, where I obtained my Ph.D. and, from 2015 to 2020, I served as a Teaching Assistant for Computational Logics and Knowledge Graphs and Data Integration.
In recent years a participated in the organisation of several research conferences, such as CAiSE, IJCAI, ER, FOIS, and EDOC. I am also collaborating as an Editorial Board Member of Applied Ontology and Information Systems Journals.
- Object-Oriented Analysis and Design, University of Stockholm
- Enterprise Computing and ERP Systems, University of Stockholm
- Information Systems Design - Lecturer - September 2022 to December 2022, Free University of Bozen-Bolzano
- Information Systems Design - Teaching Assistant - September 2021 to June 2022, Free University of Bozen-Bolzano
- IT Management and Enterprise Modeling - Teaching Assistant - Teaching Assistant - September 2021 to June 2022, Free University of Bozen-Bolzano
- Empirical Research Methods (PhD Seminar) - Lecturer - February 2023, Free University of Bozen-Bolzano
- Computational Logics - Teaching Assistant - September 2016 to May 2020, University of Trento
- Knowledge Graphs and Data Integration - Teaching Assistant - September 2017 to May 2020, University of Trento
All of my research has taken place within the field of conceptual modeling, with particular attention to the key role of ontological analysis for knowledge representation and for the design of information systems, and at the same time with a focus on methods and tools that support conceptual modeling activities and the elicitation of requirements and information relevant for system development. Specifically, in recent years my work has concentrated on four research lines that have often intersected.
First research line. The first line concerns the development of tools for discovering patterns in knowledge graphs and conceptual models. The goal here is to support those who work on knowledge representation and conceptual modeling in extracting frequent subgraphs from corpora of models (such as schemas, knowledge graphs or ontologies) in order to observe regularities that would otherwise be difficult to identify using manual approaches. The contribution of this line involves defining a process composed of several activities, such as collecting and normalizing the input models, defining an encoding capable of preserving all the relevant information, applying Frequent Subgraph Mining algorithms, and analyzing the results with appropriate metrics. In this context, we developed a software prototype and produced work such as, CMiner: An Algorithm to Discover Frequent Structures in Conceptual Models (EDBT, 2026), Pattern Discovery in Conceptual Models Using Frequent Itemset Mining (ER, 2022) and Mining Frequent Structures in Conceptual Models (SoSyM, 2025). The results obtained so far have been recognized as advancing the state of the art in techniques for knowledge discovery in complex and information-rich graphs. Application scenarios for this approach range from biomedicine, for example, graphs of molecules that specialists analyze to identify useful patterns for treatment discovery or correlations among risk factors, to risk analysis, for example graphs representing possible cyberattacks or vulnerabilities in production processes.
Second research line. The second research line concerns the semi-automated creation of knowledge graphs through the integration of Model Finding and inductive logic programming techniques. The main idea is to convert ontologies, or more generally knowledge graphs, into Alloy specifications to simulate the instances allowed by the expressed theory, generate positive and negative examples, and then learn constraints by exploiting the feedback produced by Alloy. The prototype developed in this context supports multiple phases within a single process: processing the input theory, generating plausible instances, marking the generated examples as positive or negative, and inferring missing restrictions, such as cardinalities, disjointness or property dependencies. The resulting output provides the modeler, that is, the person responsible for finalizing the representation, with a set of proposals to make the representation more precise. The approach is described in Conceptual Model Visual Simulation and the Inductive Learning of Missing Domain Constraints (DKE, 2022) and can be seen as an example of a solution that supports cognitively complex activities, such as defining logical constraints that prevent a formal representation of knowledge from allowing unintended instances.
Third research line. The third line has focused on risk modeling. The starting point was the use of domain ontologies to characterize assets, threats, vulnerabilities, controls and security objectives. The focus then shifted to the possibility of integrating, in a synergistic way, classical knowledge representation models, such as ontologies and knowledge graphs, with probabilistic models, such as Bayesian networks. This research line also fits within the broader effort to create solutions that support cognitively complex activities, in this case those associated with risk assessment. To cite some examples, in On the Semantics of Risk Propagation (RCIS, 2023), we outlined the conceptual elements needed to formalize how a risk event can propagate within a graph of events; in An Ontology-Driven Approach for Process-Aware Risk Propagation (SIGAPP, 2023), the attention shifted to the interaction between process models and probabilistic models, with examples from the cybersecurity domain. This line of work has since been expanded through follow-up studies with practical implications and has initiated collaborations with industrial partners.
Fourth research line. The fourth line builds on the experience accumulated in the previous lines and places it within the broader context of information systems design. Specifically, the aim here is to contribute to ongoing efforts to regulate emerging AI technologies. Drawing on our expertise in knowledge representation and on philosophical literature concerning types of explanation, we proposed an approach that reemphasises the notions of purpose and function. Even in the context of so-called general-purpose AI, where it appears difficult to identify a standard rule for appropriate use and operation, it is necessary to establish validation criteria that allow the identification of malfunctions or improper use. In A Teleological Approach to Information Systems Design (Minds and Machines, 2024) and Leveraging Teleological Explanation to Support General-purpose AI Assessment (AI & Society, 2025) we laid the foundations for a teleological approach to the design of information systems, including AI-based systems. As another example of work in this direction, News-Informed Probabilistic Models for AI Risk Analysis (CAiSE, 2026) focuses on the generation of probabilistic models that can be used to support AI risk assessment based on news about AI incidents.
Please refer to my Google Scholar profile.

