Human rights defenders need evidence and research to advocate for justice. However, many common bodies of human rights information, such as legislation, policies and jurisprudence, remain trapped in difficult-to-access PDFs and stacks of paper. By gaining access to this information, human rights defenders can see patterns of violence, preserve collective memory, and leverage the most important legal precedents across the world.

HURIDOCS, a nonprofit that helps human rights groups gather, organize and use information, created an open source tool called Uwazi. This tool helps human rights defenders manage their collections of information, including documents, evidence, cases, and complaints. By applying machine learning, Uwazi extracts, explores and connects human rights information collections, advancing access to information at an unprecedented speed and scale.

Before using the machine learning services within Uwazi, human rights reviewers spent two to three months updating their database every time new information became available. With machine learning, the same task takes only one week. For example, the team at Plan International's Girls' Rights Platform had to meticulously categorize and annotate paragraphs and documents. Now, Uwazi offers highly accurate suggestions which can be accepted, modified or rejected, saving enormous time. HURIDOCS has seen similar results across pilot partners around the world, and they are now working to make the services usable as a feature for all users.