BanMANI
收藏资源简介:
BanMANI是一个针对孟加拉语社交媒体新闻操纵识别的数据集,由南佛罗里达大学等机构创建。该数据集包含800条社交媒体内容,与500篇参考新闻文章相关联,用于训练和评估模型识别新闻操纵的能力。数据集通过半自动方法生成,利用ChatGPT和孟加拉语NER系统辅助,由人工标注者验证。BanMANI旨在解决孟加拉语社交媒体中新闻操纵的问题,为低资源语言提供数据支持,以提升现有NLP系统的性能和训练新模型。
BanMANI is a dataset dedicated to identifying news manipulation in Bengali social media, developed by institutions including the University of South Florida. It contains 800 social media content entries associated with 500 reference news articles, and is utilized for training and evaluating models' capacity to recognize news manipulation. The dataset was generated via a semi-automatic method, assisted by ChatGPT and Bengali NER systems, and validated by human annotators. BanMANI aims to address the issue of news manipulation in Bengali social media, providing data support for low-resource languages to enhance the performance of existing NLP systems and support the training of new models.

- 1BanMANI: A Dataset to Identify Manipulated Social Media News in Bangla南佛罗里达大学 · 2023年



