Develop a transparent AI system that (i) processes legal documents (e.g., tax audit acts) to extract judicial principles of law; (ii) integrates domain-specific contextual knowledge, such as EU legal principles traceable back to the processed documents; (iii) produce a human-comprehensible explanation on the extraction process that is grounded on the integrated knowledge.

Developing LLM-based solutions for industrial use cases that speed up existing user assistance systems while guaranteeing privacy.

Worked on two research papers on rumour verification and LLMs reasoning.

Developed horizontal matchmaking solutions for ranking textual sources. These solutions should be compliant with AI-on-demand online services. Critical requirements concern textual sources maintenance.

Developed NLP solutions for large-scale heterogeneous document matching. In particular, I’ve researched strategies for combining unstructured knowledge (e.g., textual documents, researcher profiles, user requests) and ontology knowledge graphs.

Developed a ranking system for scientific literature based on Argument Mining features.

Developed of AI dialogue argumentative chatbots on scientific literature. We investigated the impact of arguments in scientific papers to boost the capabilities of a dialogue agent.

Developed anomaly detection solutions based on forecasting neural networks and encoder-decoder architectures.