The field of multimodal analysis is rapidly evolving, with a focus on developing innovative methods for understanding and detecting hate speech in online content. Research is shifting towards leveraging large vision-language models to improve multimodal understanding, with applications in meme analysis and hate detection. Notable papers include CAMU, which introduces a novel framework for multimodal hate detection, and MemeBLIP2, which presents a lightweight multimodal system for detecting harmful memes. Additionally, papers such as Detecting and Mitigating Hateful Content in Multimodal Memes with Vision-Language Models propose effective methods for transforming hateful content in memes, promoting safer online environments.
Advances in Multimodal Analysis and Hate Detection
Sources
CAMU: Context Augmentation for Meme Understanding
Clustering of return words in languages of interval exchanges
A Unified MDL-based Binning and Tensor Factorization Framework for PDF Estimation
Factor Analysis with Correlated Topic Model for Multi-Modal Data
Score-Debiased Kernel Density Estimation
Sequence Reconstruction for Sticky Insertion/Deletion Channels
How Cohesive Are Community Search Results on Online Social Networks?: An Experimental Evaluation
How Group Lives Go Well
Sequence Reconstruction under Channels with Multiple Bursts of Insertions or Deletions
MemeBLIP2: A novel lightweight multimodal system to detect harmful memes
Mining and Intervention of Social Networks Information Cocoon Based on Multi-Layer Network Community Detection
Efficiently Finding All Minimal and Shortest Absent Subsequences in a String
Belief System Dynamics as Network of Single Layered Neural Network
Clustering Internet Memes Through Template Matching and Multi-Dimensional Similarity
Detecting and Mitigating Hateful Content in Multimodal Memes with Vision-Language Models
Algorithmic Collective Action with Two Collectives