Multimodal Sexism Identification and Characterization using Large Language Models and Gradient Boosting
This work addresses the challenge of automated sexism detection in multimodal social media content, but the findings are incremental and domain-specific.
The authors developed a multimodal system for sexism detection in memes and short-form videos, combining gradient boosting with LLM-derived features. Their approach achieved strong results on memes but revealed that compact feature selection for videos did not generalize to unseen test data, where unfiltered features performed better.
We present the AILS-NTUA submission to the EXIST 2026 Lab at CLEF, addressing multimodal sexism identification and characterization in memes (Task 2) and short-form videos (Task 3). Our system follows a feature-engineered late-fusion pipeline built around gradient-boosted regression models and hierarchical post-processing. For memes, we combine visual, textual, demographic, biometric, and LLM-derived semantic indicators designed to capture high-level cues such as stereotyping, objectification, irony, and misogyny. For videos, we investigate the effect of feature selection, frame-based visual representations, OCR-based textual features, acoustic descriptors, and sensor-derived metadata. Development results show that focused LLM-derived semantic cues improve meme sexism identification, while video performance is highly sensitive to feature dimensionality and cross-modal noise. For videos, development results favor compact feature selection, but official test results show that this conclusion does not fully transfer to unseen data, where the unfiltered representation generalizes better. Overall, our findings highlight the usefulness of targeted semantic feature engineering for static memes and the need for more robust temporal modeling in noisy short-form video settings.