Elena Cabrio: The FBK researcher teaching machines the value of argumentation

Now a full professor at Université Côte d’Azur in France, Elena Cabrio exemplifies how the professional growth she experienced at Fondazione Bruno Kessler helped lay the foundation for an international career.

Her journey, which began with a humanities education in language sciences in Turin and led to the forefront of European computer science, offers valuable inspiration for today’s young researchers navigating rapidly evolving technologies and the renewed importance of the humanities in shaping, guiding, and advancing them.

Her experience also demonstrates how FBK’s research ecosystem has become an international point of reference in the complex field of natural language processing.

We met with Cabrio to retrace the key stages of her career and reflect on the qualities that enabled her to turn challenges into opportunities, advancing not only her own career but also the broader scientific community.

Linguistic curiosity as a catalyst for technology

“Two experiences,” Cabrio recalls, “marked the beginning of my journey: a university course in 2006 that shifted my focus toward computational linguistics, and my experience at Loquendo (T-Lab), under the supervision of Morena Danieli, which acted as a professional accelerator. This was not merely an internship, but the systematic dismantling of the boundary between linguistic abstraction and the electronic generation of the speech signal. My immersion in text-to-speech systems radically redefined the scope of my goals: the synthetic voice emerged not only as a technical output, but as tangible proof that humanistic expertise could guide high-tech industrial production.”

Transformation through doctoral research and computer science

The first defining characteristic of Cabrio’s career is her interdisciplinary curiosity. During her PhD at FBK, under the supervision of Bernardo Magnini, her work evolved from learning annotation techniques and programming to mastering increasingly complex computational paradigms. A pivotal moment came with the transition from Question Answering (the “Qall-Me” project) to  research on textual Entailment (textual inference).

“That transition,” Cabrio explains, “turned what initially seemed like a limitation into an opportunity. Coming from outside computer science, I realized that the semantics of language and inferential reasoning—concepts deeply rooted in a humanities education—were powerful tools for addressing linguistic ambiguity from a computational perspective. It showed that linguistic sensitivity is not an accessory, but a force multiplier for NLP algorithms.”

Leadership, mentorship, and managing academic complexity

A second pillar of Cabrio’s success is her strong commitment to collaborative leadership and mentorship. Today, as head of MARIANNE (“Models and Data for Computational Argumentation in Natural Language“), a team spanning the I3S laboratory at Université Côte d’Azur and Inria, she devotes much of her work to turning competition into collaboration by fostering a culture of knowledge sharing that allows every team member’s strengths to contribute to the success of the group.

“The philosophy we promote within the team rests on three pillars: collaboration over competition (making a deliberate effort to reduce academic toxicity by turning competitive pressure into project synergy); clear authorship protocols (establishing transparent and consistent rules for scientific publications to minimize internal conflicts and ensure recognition for every contributor); and exploratory freedom (a mentoring approach built on trust, delegation, and giving PhD students the freedom to pursue new ideas).”

Building bridges where others saw divides

Another defining characteristic of Cabrio’s career has been her adaptability. After moving to France, she joined Inria at a time when there were no established research groups dedicated to language processing. Rather than seeing this as a limitation, she adapted her expertise and collaborated with specialists in formal semantics to develop innovative approaches, including the translation of natural language into SPARQL —a major challenge in the field of knowledge representation that builds a bridge between human language and databases.

This “bilingual” expertise—combining human language and programming—and her ability to connect different disciplines enabled her to see opportunities where others saw barriers. It also led her to explore, with both passion and rigor, how to enable machines to truly understand human language, rather than the other way around.

That same adaptability later became the driving force behind her pioneering work in argument mining, a cutting-edge field that now has important applications in areas such as medicine and participatory democracy through its ability to identify and structure arguments.

What is argument mining?

Argument mining lies at the intersection of natural language processing (NLP) and formal argumentation. This field of research focuses on analyzing arguments expressed in natural language.

Its core objective is to use data and algorithms to automatically identify and extract arguments from text. Rather than viewing discourse simply as a sequence of words, argument mining treats it as a complex structure in which each argument forms a node within a graph. Researchers analyze the inferential relationships—that is, the logical connections—between different parts of a text to reconstruct its overall reasoning.

The goal of argument mining is to achieve a deeper understanding of language by exploring aspects that current technological models have yet to fully master, including persuasion, pragmatics, and the underlying logic of debate.

Main applications

Key applications include e-democracy (analyzing and structuring public debates and citizens’ contributions in digital participation platforms), medicine (where rigorous analysis of evidence and arguments is essential), and the interpretation of implicit meaning (detecting linguistic nuances such as sarcasm and identifying or mitigating online hate speech).

More broadly, argument mining transforms unstructured text into organized, analyzable data, making it easier to understand how people construct, and defend their ideas.

Value connections: from shared roots to international peer review

The strength of an organization like FBK lies not only in the excellence of its scientific achievements and their societal impact, but also in the culture of reciprocity it fosters and the vitality of its alumni network—a community that creates value through interdisciplinary and intergenerational exchange. Being part of this community means having access to constant inspiration and a real opportunity to engage with mentors who are working at the forefront of international science. In a rapidly evolving technological landscape, where review processes and the quality of research are put to the test by information overload, being able to rely on the expertise of such high-caliber specialists becomes a strategic asset.