MemeSense: An Adaptive In-Context Framework for Social Commonsense Driven Meme Moderation

TL;DR

MemeSense is an adaptive framework that combines visual and textual analysis with commonsense cues to improve moderation of subtle, culturally nuanced harmful memes, outperforming existing methods on benchmark datasets.

Contribution

This paper introduces MemeSense, a novel in-context framework that enhances meme moderation by integrating multimodal understanding and commonsense reasoning, addressing limitations of traditional text-based systems.

Findings

Achieves up to 35% higher semantic similarity on benchmarks.

Improves BERTScore by 9% for non-textual memes.

Demonstrates effectiveness across multiple datasets.

Abstract

Online memes are a powerful yet challenging medium for content moderation, often masking harmful intent behind humor, irony, or cultural symbolism. Conventional moderation systems "especially those relying on explicit text" frequently fail to recognize such subtle or implicit harm. We introduce MemeSense, an adaptive framework designed to generate socially grounded interventions for harmful memes by combining visual and textual understanding with curated, semantically aligned examples enriched with commonsense cues. This enables the model to detect nuanced complexed threats like misogyny, stereotyping, or vulgarity "even in memes lacking overt language". Across multiple benchmark datasets, MemeSense outperforms state-of-the-art methods, achieving up to 35% higher semantic similarity and 9% improvement in BERTScore for non-textual memes, and notable gains for text-rich memes as well. These results highlight MemeSense as a promising step toward safer, more context-aware AI systems for real-world content moderation. Code and data available at: https://github.com/sayantan11995/MemeSense