- Neurosymbolic Systems
- Agentic AI
- Agent-based Intelligent Systems Engineering
- Agent-Based Simulation of Complex Systems
Neurosymbolic Systems. The emergence of new machine learning approaches, together with the rapid expansion of Large Language Models (LLMs) and Generative AI techniques, raises above all the issue of the reliability of intelligent systems and the trust that can be placed in them. Only the integration of symbolic and subsymbolic techniques within neurosymbolic systems can lead to the development of intelligent systems that combine the remarkable efficiency and effectiveness of subsymbolic approaches with the correctness and verifiability of results that characterize classical AI technologies.
Agentic AI and Tools. The rise of the agent paradigm in contemporary intelligent technologies demands the development of new models that are cognitively and operationally aligned with both agent–human social interaction and agent–environment interaction. Meta-models such as A&A (Agents and Artefacts), originally conceived within the classical Multi-Agent Systems (MAS) domain, can provide a novel, coherent, and effective conceptual framework for the design, integration, and use of tools in modern agentic AI systems.
Agent-based Intelligent Systems Engineering. While mainstream software engineering, grounded in object-oriented abstractions, has largely matured and exposed its inherent limitations, emerging intelligent technologies both require and stimulate the development of new methodological approaches. This research seeks to advance novel methods, tools, and Agent-Oriented Software Engineering (AOSE) methodologies for the design and engineering of contemporary intelligent systems.
Agent-based complex system simulation. MAS represent today the main paradigm for simulation of complex systems such as biosystems, social systems, etc, promoting the integration of traditional top-down approaches based on mathematical models with bottom-up computational approaches.
