ReMMD：用於多模態虛假信息檢測的真實多語言多圖像智能體驗證

摘要

多模態虛假信息檢測日益重要，因為病毒式貼文現在結合了長篇多語言敘述、多張圖片、混合來源，以及細微的文字-圖片框架錯誤。現有的基準評測和方法仍難以完全應對這一場景：它們通常僅限於短標題、單張圖片、二元標籤或單一操縱來源，而在真實情況下的證據搜索中，代理式驗證仍耗費高昂。我們提出ReMMD，一個針對多模態虛假信息檢測的真實多語言多圖片代理式驗證框架。ReMMD包含ReMMDBench，一個真實世界的多模態虛假信息檢測基準評測，涵蓋500個樣本、2,756張圖片、五種單語語言、兩種跨語言設定、三種文本長度級別、多圖片貼文、五類真實性標籤、八類扭曲標籤、證據來源及推理說明。它還包含ReMMD-Agent，一個具備持久記憶的驗證器，能將貼文分解為原子觀點，建立可重複使用的證據集，並預測結構化的L1/L2/L3輸出。在封閉源系統、開放式大型視覺語言模型、MMD-Agent與T2-Agent的比較中，ReMMD-Agent獲得了最佳的五類真實性分類表現，使用GPT-5.2時準確率達41.80%，宏觀F1值達39.12%，同時與MMD-Agent相比成本降低17.5%，與T2-Agent相比降低79.9%。該專案網址為 https://dang-ai.github.io/ReMMD。

English

Multimodal misinformation detection is increasingly important because viral posts now combine long multilingual narratives, several images, mixed provenance, and subtle text--image framing errors. Existing benchmarks and methods remain poorly matched to this setting: they usually isolate short captions, single images, binary labels, or one manipulation source, while agentic verification remains costly under realistic evidence search. We present ReMMD, a realistic multilingual multi-image agentic verification framework for multimodal misinformation detection. ReMMD includes ReMMDBench, a real-world multimodal misinformation detection benchmark with 500 samples, 2,756 images, five monolingual languages, two cross-lingual settings, three text-length tiers, multi-image posts, five-way veracity labels, eight distortion labels, evidence provenance, and rationales. It also includes ReMMD-Agent, a persistent-memory verifier that decomposes posts into atomic points, builds a reusable evidence set, and predicts structured L1/L2/L3 outputs. Across proprietary systems, open LVLMs, MMD-Agent, and T2-Agent, ReMMD-Agent obtains the best five-way veracity performance, with 41.80% accuracy and 39.12% macro-F1 using GPT-5.2, while reducing cost by 17.5% relative to MMD-Agent and 79.9% relative to T2-Agent. The project is available at https://dang-ai.github.io/ReMMD.