aiqbench analysis
This research proposes a novel cross-attention fusion framework to enhance the detection of political intent in Bengali memes by aligning visual regions with semantic tokens. By integrating domain-specific lexicons and leveraging Vision-Language Models, the study achieves state-of-the-art performance on low-resource language benchmarks while providing interpretability insights into multimodal grounding. The work advances machine learning techniques for handling noisy, stylized text-image pairs critical to understanding social media sentiment in underrepresented linguistic contexts.
Token-Region Guided Cross-Attention Fusion for Multimodal Affect Interpretation
arXiv:2607.23493v1 Announce Type: cross Abstract: Automated analysis of multimodal content on social networks has become a critical task for understanding public sentiment and information diffusion in the digital age. However, classifying internet memes remains computationally challenging due to the intricate interplay between visual cues and embedded, often stylized, text, particularly in low-resource languages like Bengali Language. This paper addresses the detection of political intent in Bengali memes by introducing Multimodal Cross-Attention Fusion framework. We first leverage a Vision-Language Model to extract high-fidelity OCR text from noisy meme images. Subsequently, we encode visual and textual features and synthesize them through a cross-modal multi-head attention mechanism that aligns semantic tokens with visual regions. We also investigate the integration of a domain-specific political lexicon as a knowledge prior. Experimental evaluation on the PoliMemeDecode1 dataset shows that our attention-based fusion significantly outperforms unimodal baselines and standard concatenation methods, achieving a state-of-the-art Macro-F1 of approximately 0.94. Interpretability analyzes further confirm that the model effectively learns to ground textual semantics in visual evidence.
Abstract:Automated analysis of multimodal content on social networks has become a critical task for understanding public sentiment and information diffusion in the digital age. However, classifying internet memes remains computationally challenging due to the intricate interplay between visual cues and embedded, often stylized, text, particularly in low-resource languages like Bengali Language. This paper addresses the detection of political intent in Bengali memes by introducing Multimodal Cross-Attention Fusion framework. We first leverage a Vision-Language Model to extract high-fidelity OCR text from noisy meme images. Subsequently, we encode visual and textual features and synthesize them through a cross-modal multi-head attention mechanism that aligns semantic tokens with visual regions. We also investigate the integration of a domain-specific political lexicon as a knowledge prior. Experimental evaluation on the PoliMemeDecode1 dataset shows that our attention-based fusion significantly outperforms unimodal baselines and standard concatenation methods, achieving a state-of-the-art Macro-F1 of approximately 0.94. Interpretability analyzes further confirm that the model effectively learns to ground textual semantics in visual evidence.
Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
arXiv:2607.23493 [cs.CV]
(or arXiv:2607.23493v1 [cs.CV] for this version)
https://doi.org/10.48550/arXiv.2607.23493 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 262 words
Related coverage
MVFA: A Multi-View Text-Guided Multimodal Fusion LLM Adapter for Sentiment Analysis and Emotion Recognition
arXiv · Sep 9, 2026
Modality Discrepancy Transformer for Ambivalence and Hesitancy Recognition
arXiv · Sep 18, 2026
Salient Knowledge Pathways: Sparse Cross-Modal Routing for Efficient Knowledge-Intensive Multimodal Question Answering
arXiv · Jul 29, 2026
Multi-Context Fusion Transformer for Pedestrian Crossing Intention Prediction in Urban Environments
arXiv · Aug 25, 2026