EVWSD-ITA (Enhanced Visual Word Sense Disambiguation for Italian) is a shared task proposed at the EVALITA 2026 campaign. While traditional Visual Word Sense Disambiguation (VWSD) focuses on broad semantic distinctions, EVWSD-ITA introduces a more rigorous challenge by requiring systems to perform fine-grained disambiguation. The task involves selecting the most appropriate image for a target word within a given context, specifically designed to include "hard negatives"—co-hyponyms that share a common hypernym but represent distinct concepts. The dataset was meticulously constructed, comprising a manually validated test set. In this report, we showcase the dataset construction procedure and the results of the task.
EVWSD-ITA at EVALITA 2026: Overview of the Enhanced Visual Word Sense Disambiguation for Italian Task
Elio Musacchio
;
2026-01-01
Abstract
EVWSD-ITA (Enhanced Visual Word Sense Disambiguation for Italian) is a shared task proposed at the EVALITA 2026 campaign. While traditional Visual Word Sense Disambiguation (VWSD) focuses on broad semantic distinctions, EVWSD-ITA introduces a more rigorous challenge by requiring systems to perform fine-grained disambiguation. The task involves selecting the most appropriate image for a target word within a given context, specifically designed to include "hard negatives"—co-hyponyms that share a common hypernym but represent distinct concepts. The dataset was meticulously constructed, comprising a manually validated test set. In this report, we showcase the dataset construction procedure and the results of the task.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
