Research map / Vision-language and multimodal / Multimodal LLMs
Evidence Triangulation for Multimodal Fact-Checking in the Wild
Stefanos-Iordanis Papadopoulos, Zacharias Chrysidis, Christos Koutlis, Symeon Papadopoulos, Panagiotis Petrantonakis
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 image input (ablation) | TRENT · X-POSE random split (full test) | macro-F1: 57.62 -> 63.10 | Improved |
| add external evidence (ablation) | TRENT · X-POSE random split (full test) | macro-F1: 59.23 -> 63.10 | Improved |
| use VLM-extracted excerpts instead of full-article embeddings (ablation) | TRENT · X-POSE random split (full test) | macro-F1: 58.88 -> 63.10 | Improved |
| use excerpts instead of VLM summaries (ablation) | TRENT · X-POSE random split (full test) | macro-F1: 60.33 -> 63.10 | Improved |
| use CLIP text encoder instead of multi-E5 (ablation) | TRENT · X-POSE random split (full test) | macro-F1: 60.17 -> 63.10 | Improved |
| add evidence credibility filter (ablation) | TRENT · X-POSE random split (full test) | macro-F1: 61.35 -> 63.10 | Improved |
| add evidence reranking (ablation) | TRENT · X-POSE random split (h>=90%) | macro-F1: 69.99 -> 70.83 | Improved |
| add image-retrieved evidence (ablation) | TRENT · X-POSE random split (full test) | macro-F1: 61.75 -> 63.10 | Improved |
| add text-retrieved evidence (ablation) | TRENT · X-POSE random split (h>=80%) | macro-F1: 66.13 -> 67.36 | Mixed |
| add relational fusion instead of concatenation (ablation) | TRENT · X-POSE random split (full test) | macro-F1: 60.41 -> 63.10 | Improved |
| use three triangulated streams instead of intra-modality consistency (ablation) | TRENT · X-POSE random split (full test) | macro-F1: 58.83 -> 63.10 | Improved |
| add each of the three cross-attention streams (ablation) | TRENT · X-POSE random split (full test) | macro-F1: 58.50 -> 63.10 (text-image stream) | Improved |
Closest papers by meaning
- Pioneering Explainable Video Fact-Checking with a New Dataset and Multi-role Multimodal Model Approach (AAAI 2025 · similarity 0.88)
- DEFAME: Dynamic Evidence-based FAct-checking with Multimodal Experts (ICML 2025 · similarity 0.88)
- AVerImaTeC: A Dataset for Automatic Verification of Image-Text Claims with Evidence from the Web (NeurIPS 2025 · similarity 0.87)
- DAMMFND: Domain-Aware Multimodal Multi-view Fake News Detection (AAAI 2025 · similarity 0.86)
- SNIFFER: Multimodal Large Language Model for Explainable Out-of-Context Misinformation Detection (CVPR 2024 · similarity 0.86)
- Reinforced Adaptive Knowledge Learning for Multimodal Fake News Detection (AAAI 2024 · similarity 0.86)
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.
