# Context This is a historical dataset containing 13,087 press releases from the Department of Justice's (DOJ) website https://www.justice.gov/news. The DOJ typically publishes several releases per day and this dataset spans from 2009 to July 2018. The releases contain information such as outcomes of criminal cases, notable actions taken against felons, or other updates about the current administration. This dataset only includes releases categorized as "Press release" and does not contain those which have been labeled as "Speeches". Some releases are tagged with topics or related agencies. The original Python code to scrape the data can be found on GitHub at https://github.com/jbencina/dojreleases # Content The contents are stored as newline delimited JSON records with the following fields: - **id**: Press release number (can be missing if included in contents) - **title**: Title of release - **contents**: Text of release - **date**: Posted date - **topics**: Array of topic tags (if any provided) - **components**: Array of agencies & departments (if any provided) # Acknowledgements All data was sourced from https://www.justice.gov/news # Inspiration The data provides an opportunity for analysis including: - How have the reported topics changed over the years / administrations? - What words tend occur frequently together? - How can documents be clustered using the content of the releases? - Can a predictive text model be trained off the supplied topics? - Use a tool like [Spacy][1] to handle named entities in the releases (names, locations, etc.) [1]: https://spacy.io/usage/linguistic-features#section-named-entities