The paper describes the EXIST 2026 lab on Sexism identification in social networks, that is expected to take place at the CLEF 2026 conference and represents the sixth edition of the EXIST challenge. The lab comprises six tasks in two languages, English and Spanish, which are the same three tasks (textitsexism identification, textitsource intention detection, and textitsexism categorization) applied to two different types of data: memes (image and text) and TikToks (video and text). This multimedia approach will reveal trends and patterns of sexism across different media and user interactions. It will also provide deeper insight into the underlying social dynamics shaping these behaviors. As in previous EXIST editions, the datasets will include annotations from multiple annotators, showing different or even conflicting opinions. This helps models learn from diverse perspectives. The novelty of the 2026 edition lies in the enrichment of the dataset with physiological and neurophysiological signals – specifically, heart rate, eye tracking, and electroencephalogram (EEG) data collected from different subjects while viewing the materials. The aim is twofold: to explore how implicit emotional and cognitive responses correlate with the perception of sexist content, and to evaluate whether these signals can improve the automatic detection and classification of sexism. By incorporating features derived from users’ unconscious reactions, machine learning models may capture subtle cues of bias or discomfort that are not evident in textual or visual content alone, leading to more accurate and human-aligned systems for identifying sexism across media.
Add the full text or supplementary notes for the publication here using Markdown formatting.