IMPOLS Shared Task at EVALITA 2026: Implicit Contents in Political Speech

IMPOLS shared task at EVALITA 2026 focuses on detecting implicit contents in Italian political speech using NLP and multimodal approaches.

The IMPOLS (IMplicit contents in POLitical Speech) shared task is announced within the EVALITA 2026 campaign. IMPOLS focuses on detecting and classifying implicit contents in political speech that are non-bona fide true.

Key aspects of the task include:

• Binary detection of questionable implicit contents
• Binary classification between implicatures and presuppositions
• Multiclass classification of implicatures

The task is monolingual (Italian) and multimodal, providing both speech and manually revised transcription. Participation is encouraged in one or more subtasks using textual or multimodal data.

For more information, visit the IMPOLS website or watch the IMPOLS demo video. The EVALITA 2026 campaign details can be found on the EVALITA website.

Important dates include the evaluation window from 1st to 8th December 2025 and the final workshop on 26-27th February 2026 in Bari, Italy.

Tags: IMPOLS, EVALITA 2026, NLP, political speech analysis, implicit content detection, multimodal task, Italian language processing