Research map / Vision-language and multimodal / Cross-modal learning
MMSD3.0: A Multi-Image Benchmark for Real-World Multimodal Sarcasm Detection
Haochen Zhao, Yuyao Kong, Yongxiu Xu, Gaopeng Gou, Hongbo Xu, Yubin Wang, Haoliang Zhang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add Dual-Stage Bridge Module (ablation) | CIRM · MMSD3.0 | F1: 81.41 -> 84.42 | Improved |
| add Relevance-Guided Fusion Module (ablation) | CIRM · MMSD3.0 | F1: 81.36 -> 84.42 | Improved |
| add image positional encoding and masking (ablation) | CIRM · MMSD3.0 | F1: 83.25 -> 84.42 | Improved |
| add OCR text features (ablation) | CIRM · MMSD3.0 | F1: 81.59 -> 84.42 | Improved |
| keep emoji features (ablation) | CIRM · MMSD3.0 | F1: 82.31 -> 84.42 | Improved |
| add Pre-Bridge stage (ablation) | CIRM · MMSD3.0 | F1: 81.81 -> 84.42 | Improved |
| add sequential modeling module (ablation) | CIRM · MMSD3.0 | F1: 81.23 -> 84.42 | Improved |
| add Post-Bridge stage (ablation) | CIRM · MMSD3.0 | F1: 81.99 -> 84.42 | Improved |
| randomly shuffle image order at test (ablation) | CIRM · MMSD3.0 | F1: 84.42 -> 83.51 | Hurt |
| truncate text to first 15 tokens (ablation) | CIRM · MMSD3.0 long text | accuracy: 86.40 -> 73.40 | Hurt |
| mixing weight alpha 0.3 between cosine and learned relevance (ablation) | CIRM RGFM · MMSD3.0 | F1: best F1 84.42 at alpha=0.3 | Improved |
| use multi-image-aware CIRM instead of single-image methods with tiling | multimodal sarcasm detectors · MMSD3.0 | accuracy: 82.20 -> 85.16 | Improved |
Closest papers by meaning
- G^2SAM: Graph-Based Global Semantic Awareness Method for Multimodal Sarcasm Detection (AAAI 2024 · similarity 0.89)
- DocMSU: A Comprehensive Benchmark for Document-Level Multimodal Sarcasm Understanding (AAAI 2024 · similarity 0.89)
- Well, Now We Know! Unveiling Sarcasm: Initiating and Exploring Multimodal Conversations with Reasoning (AAAI 2024 · similarity 0.89)
- Debiasing Multimodal Sarcasm Detection with Contrastive Learning (AAAI 2024 · similarity 0.89)
- Is Sarcasm Detection a Step-by-Step Reasoning Process in Large Language Models? (AAAI 2025 · similarity 0.86)
- II-Bench: An Image Implication Understanding Benchmark for Multimodal Large Language Models (NeurIPS 2024 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
