Full Professor of Data Science and Artificial Intelligence. Director of the Master in AI & Data Analytics for Business at Milano-Bicocca.
Conferences I'm involved in, as PC or senior PC.
I am Full Professor in Computer Science at University of Milan-Bicocca. I’m director of the Master in AI and Data Analytics for Business at Milano-Bicocca, Italy.
My work sits at the intersection of explainable AI, large language models, and labour-market intelligence — with a long-standing interest in how AI tools can be made transparent, evaluable, and useful in real-world settings.
"If you want to go fast go alone. If you want to go far go together." — Proud to work with the brilliant PhD students and researchers of AI@CRISP-UNIMIB.
Head of CRISP · Big Data Analytics
Researcher · AI & Computational Linguistics
Researcher · XAI
PostDoc · NLP
PostDoc · Economics
PostDoc · NLP
Alessia De Santo
PostDoc · NLP
Antonio Serino
PhD · Big Data Analytics for Business
Daniele Potertì
PhD · Big Data Analytics for Business
Simone Adobati
PhD · Big Data Analytics for Business
PhD CS → Mario Negri Institute
Simone D'Amico
PhD BDABF → AI Engineer, Capgemini
Anna Velyka
PostDoc · NLP
PhD CS → Data Sci. Mgr, Intesa SanPaolo
Open-source tools and libraries built with the team.
Two-stage AI evaluation pipeline assessing the capability of LLMs, image-processing systems, and robotics to perform specific occupational tasks.
Unsupervised method for embedding alignment via semantic anchor selection — generates bilingual or monolingual word-pair dictionaries without prior similarity assumption.
Pipeline for Semantic-Functional Alignment Scores. Analyses and links sparse-autoencoder features by combining co-occurrence and semantic similarity of explanations.
Unified Python framework for intrinsic evaluation of word-embedding algorithms — similarity, analogy, categorisation, outlier detection, plus the holistic PCE metric.
Global, model-agnostic, contrastive explainer for any tabular or text classifier — provides contrastive explanations of how two ML models differ.
Model-agnostic, global, time-contrastive explainer for any text classifier. Available as a Python tool on GitHub.
Methodology for taxonomy refinement via word embeddings — evaluates the best embedding by its ability to represent taxonomic similarity relations.
Models and enriches DBLP as a graph database for graph-based queries and social-network analysis on academic publications.
Planner for linear and nonlinear continuous PDDL+ models with processes and events. Handles huge state spaces through a disk-based algorithm.
Planner built on UPMurphi using graph-based heuristics to speed up plan synthesis (King's College London Planning Group).
Large-scale benchmark of 10K+ multiple-choice questions for evaluating LLM understanding of Italian culture, commonsense reasoning, and linguistic proficiency in a morphologically rich language. 12 domains.
Structured benchmark for evaluating LLM proficiency on Italian student competencies via the national INVALSI tests — 405 questions across 6 educational grades.
AI Communications · Associate Editor
Reviewer for: Artificial Intelligence · Cognitive Computation · Applied Soft Computing · Applied Intelligence · Future Generation Computer Systems · Knowledge-Based Systems · Computers in Industry · Expert Systems with Applications.
Email (preferred): fabio.mercorio@unimib.it
Office hours: by appointment. Students book here.
Room 2043, Building U7 · Department of Statistics & Quantitative Methods · CRISP Research Centre
University of Milan-Bicocca · Viale dell'Innovazione 10 · 20126 Milano · Italy · (+39) 02 644 82170
