Digital Humanist
Post‑doctoral researcher at Vrije Universiteit Amsterdam, working on cognitive computing, and the extraction of commonsense knowledge from multimodal sources, adopting techniques from applied ontology, knowledge representation, and neuro-symbolic AI. I build ontologies that formalize commonsense knowledge: moral and cultural values, emotions, social norms, and embodied cognition.
Amsterdam, The Netherlands
My latest personal achievement: managing to cite Anna Karenina, by Leo Tolstoy, in my latest paper, accepted at EMNLP 2026.
I maintained an average of six pastéis de nata a day while attending NeSy 2026 in Lisbon, where I presented “The Bigger Picture: Neuro‑Symbolic Knowledge Graph Enrichment and Anticipatory Inference from Multimodal Content”.
Hi, I’m Stefano. I’m a Digital Humanist and currently a post-doctoral researcher at Vrije Universiteit Amsterdam. My main research areas are Knowledge Representation, the Semantic Web, and AI — with a particular focus on emotions, moral and cultural values, commonsense knowledge, social norms, and embodied cognition.
I was born in Turin, Italy, in 1993. An early interest in language and semantic technologies led me to Alma Mater Studiorum in Bologna, where I graduated in Linguistics Sciences with a thesis on Post-truth, populism and knowledge extraction — A frame semantics approach to Indirect Hate Speech. My PhD thesis in Digital Humanities, Ethics in the Flesh — Formalizing Moral Values with Frame Semantics and Embodied Cognition, focuses on cognitive semantics and the formalization of moral values through frame semantics and embodied cognition.
I firmly believe in empathy: the cognitive, imaginative ability to put oneself in someone else’s shoes. The unknown is just an entity not yet framed in a network of meaningful semantic relationships — and every conversation, personal or professional, is a chance to practice framing it.
Some ontologies about higher layers of meaning.
Moral & cultural values — modular ontology
A modular ontology of moral and cultural values, integrating Schwartz's Basic Human Values, Graham and Haidt's Moral Foundations Theory, and Curry's Moral Molecules. Each value is modeled as a conceptual frame, aligned to DOLCE, and built with reusable Ontology Design Patterns as part of the Framester knowledge graph.
View on GitHub →
Emotions as semantic frames
An OWL, frame-based ontology that treats emotions as semantic frames with roles capturing the different facets of emotional experience. EFO integrates multiple emotion theories as cross-linked modules and supports multimodal datasets, including emotional speech and facial expression.
View on GitHub →
Image schemas & embodied cognition
The Image Schema Abstraction And Cognition modular ontology formalizes Image Schemas — dynamic conceptual building blocks arising from recurring sensorimotor interaction with the physical world — and integrates them with linguistic and factual knowledge.
View on GitHub →
PhD thesis — values, emotions & embodied cognition
An explainable detector, built on the FRED tool, that exploits knowledge graph patterns to extract emotions, values, and image schemas from text and infer how they combine. The example above is generated automatically from the sentence “The police finally stopped the awful killer.”
Larger image →
Side project — knowledge graph for tabletop rules
A knowledge graph formalizing the Dungeons & Dragons 5th edition rule system, built out of genuine nerd necessity: character bonuses, monster stats, and rules, all queryable in one place.
View on GitHub →What I reach for day to day.