sem_eval_2020_task_11

SemEval-2020 Task 11 旨在研究如何利用自动算法检测新闻文章中使用的宣传技巧，例如“诉诸人身”、“转移视线”和“贴标签”等。它提供了英文新闻文章样本，规模较小，包含训练集（371条）、验证集（75条）和测试集（90条）。数据集中，文本被标注了宣传片段的起始和结束位置，以及具体的宣传技巧类别，例如“诉诸权威”、“诉诸恐惧-偏见”等。该数据集支持文本分类和token分类任务，并提供标准化数据操作。虽然数据集的授权许可、来源和创建过程等信息尚不明确，但它为研究宣传技巧检测提供了宝贵资源。

SemEval-2020 Task 11 aims to explore the application of automated algorithms for detecting propaganda techniques employed in news articles, including ad hominem attacks, red herrings, and labeling, among others. It provides a small-scale corpus of English news articles, which is split into a training set (371 instances), a validation set (75 instances), and a test set (90 instances). In this dataset, texts are annotated with the start and end positions of propaganda spans as well as their corresponding specific propaganda technique categories, such as appeal to authority, appeal to fear-prejudice, etc. This dataset supports both text classification and token classification tasks, and offers standardized data processing operations. While information regarding the dataset’s license, original sources, and creation process remains unclear, it serves as a valuable resource for research on propaganda technique detection.