umn-cyber/mafindo-2024

--- license: mit --- # Codename Deephoax: Datasets This repository contains the datasets used in the Codename Deephoax project, initialized by UMN Cybersecurity Laboratory researchers. This datasets is a collection of >15k articles, containing both real and fake news. The real news dataset contains news articles from reputable news sources such as TEMPO, Kompas, etc. The fake news dataset contains news articles from CekFakta.com; a community-maintained repository of fake news articles by Mafindo that are verified fake by experts. The datasets are split into three types; `all`, `train`, and `test`. The all dataset contains all the news articles, the `train` dataset contains 80% of the news articles, and the `test` dataset contains 20% of the news articles. The original datasets is still available in the `mafindo` and `news` directories.