## Data description ### Files - `antiracist_6topic_public_corpus.jsonl`: Antiracist text from Reddit and Medium, in JSON lines format. This file is sampled to match properties of the corresponding filtered white supremacist corpus used in experiments in the paper. Email `yoder@cs.cmu.edu` if you are interested in unfiltered data and/or tweet IDs from antiracist organizations. - `neutral_6topic_public_corpus.jsonl`: "Neutral" data with similar topics as the white supremacist data but without the white supremacist ideology, in JSON lines format. This file is sampled to match properties of the corresponding filtered white supremacist corpus used in experiments in the paper. Email `yoder@cs.cmu.edu` if you are interested in unfiltered data and/or tweet IDs to complete this dataset. ### Fields - `id`: A unique ID for that post/text, in a `<dataset>_<number>` format. - `text`: tokenized, lowercased, with some source artifacts removed - `dataset`: name of the source dataset or data dump - `source`: the platform source, such as Medium - `domain`: domain of the platform source. Values include `forum`, `chat`, `long-form` (article or book section), `offline_propaganda`, and `tweet` - `timestamp`: If available, the timestamp of the text from the source