GSI:detect at EVALITA 2026: Overview of the Task on Detecting Gender Stereotypes in Italian

Gloria Comandini, Manuela Speranza, Sofia Brenna, Davide Testa, Stefania Cavagnoli, Bernardo Magnini

Keywords: gender stereotypes, perspectivism, linguistic resource, evaluation, LLMs

TL;DR: Overview of the GSI:detect task at EVALITA 2026 on the recognition of gender stereotypes and discussion on participants' results.

Abstract:

GSI:detect is a new shared task for the recognition and classification of gender stereotypes (GSs) presented at EVALITA 2026. The task adopts a perspectivist approach in order to enhance the high subjectivity of GS recognition and analysis on a dataset of challenging short texts in Italian. GSI:detect is organized in: A) a Main Task (GS Detection) in which systems have to assign to a text the GS value, a numerical score that quantifies the extent to which a given text exhibits or refers to a GS; B) an optional Subtask (GS Classification) in which systems, given six pre-defined categories (e.g. role, relational, etc.) must assign one to each text.
Seven teams from academic and non-academic environments took part in the challenge, with a total of 50 submitted runs for the Main Task and a total of 43 submitted runs for the optional Subtask. We present here first an overview of the GSI:detect task, the dataset and the evaluation criteria, then outline and discuss the participants' results.
Content warning: Examples taken from the GSI:detect dataset may contain sensitive, offensive, or otherwise distressing content.

Email Sharing: We authorize the sharing of all author emails with Program Chairs.

Data Release: We authorize the release of our submission and author names to the public in the event of acceptance.

Submission Number: 10