Overview of SSD-2026 at IberLEF 2026: Social Support Detection and Target Identification in Social Media

Resumen

Social support in online communication involves expressions of encouragement, care, solidarity, and empathy toward individuals or communities. This paper overviews the Social Support Detection shared task (SSD-2026) at IberLEF 2026, which evaluates NLP systems for detecting support in English and Spanish YouTube comments, identifying its target, and classifying the supported community. The task comprised three hierarchical subtasks and attracted 72 participants, 387 submissions, and 18 system description papers. Approaches ranged from classical machine learning to transformer-based, multi-task, byte-level, and LLM-assisted methods. Results were strong for binary detection but more limited for fine-grained classification under class imbalance, with Spanish systems generally outperforming English ones.