ProBel：结合技术、 spans 与解释的宣传检测
ProBel: Propaganda Detection with Techniques, Spans, and Explanations
浏览论文内容
中文总结 AI 辅助
ProBel 是含阿拉伯语和英语的宣传检测资源，支持多任务，双语多任务模型性能最优，联合训练效果更稳定，将发布相关数据与代码。
中文摘要 AI 辅助
宣传检测包含多个相关预测层级，从句子级决策到技术分类再到 span 识别。然而，目前尚不清楚在阿拉伯语和英语中联合学习时，这些层级的监督信号如何相互作用。我们提出 ProBel，这是一个阿拉伯语和英语资源，它对齐了二分类标签、23 种宣传技术的多标签标注（分为 6 个粗类别）、技术标注的 spans 以及相同新闻句子的参考解释。它包含规模大得多的英语语料库，支持两种语言中匹配的二分类、粗粒度、多标签和 span 级任务。我们在共享设置下评估零样本提示、任务特定微调以及联合训练。单个双语多任务模型取得了最佳整体性能，并且在任务和语言间保持竞争力。跨任务分析表明，迁移效果取决于监督层级：联合分类训练保留二分类性能，而仅 span 训练会削弱句子级预测；联合双语训练产生最稳定的结果，而单语言微调会降低向另一种语言的迁移效果。我们将发布该数据、代码和评估脚本。
英文摘要
Propaganda detection includes several related prediction levels, ranging from sentence-level decisions to technique classification and span identification. However, it remains unclear how supervision at these levels interacts when learned jointly across Arabic and English. We present ProBel, an Arabic and English resource that aligns binary labels, multi-label annotations over 23 propaganda techniques grouped into six coarse categories, technique-labeled spans, and reference explanations for the same news sentences. It includes a substantially larger English collection and supports matched binary, coarse-grained, multi-label, and span-level tasks in both languages. We evaluate zero-shot prompting, task-specific fine-tuning, and joint training under a shared setup. A single bilingual multi-task model achieves the best overall performance and remains competitive across tasks and languages. Cross-task analysis shows that transfer depends on the supervision level. Joint classification training preserves binary performance, whereas span-only training can weaken sentence-level prediction. Joint bilingual training yields the most stable results, while monolingual fine-tuning can reduce transfer to the other language. We will release the data, code, and evaluation scripts.
发表机构
- Qatar Computing Research Institute（卡塔尔计算研究所）
- University of Padova（帕多瓦大学）
机构由 AI 辅助整理，请以论文原文为准。
