Data: A Key Protection Tool for Refugees

When crisis hits, those forced to flee become very vulnerable. They don’t have the protection of their countries – it is often their own governments that threatens and persecutes them – and depend on international assistance to guarantee their security and basic rights.

In 2013, as part of UNHCR’s efforts to ensure the safety and protection of refugees, the organization launched Refworld.org – an online platform with more than 175,000 monthly users worldwide that provides a vast collection of legal documents and reports to support those making legal decisions on refugee and statelessness status. But the legal documents on the platform lacked uniformity and required significant resources to organize, manage and update.

To respond to this need, The Hive – USA for UNHCR’s innovation lab  – and the UNHCR Division of International Protection Team embarked on a project that used machine learning to automate the manual annotations of legal documents. These annotations help enhance navigation and searchability between documents to ultimately assist legal aid workers, lawyers, judges, and UNHCR staff to quickly find relevant information helpful in building legal arguments in protecting people of concern.

The Hive prepared and converted more than 4,500 legal PDF files into text and used various machine learning models in order to design an optimal prototype capable of surfacing legal information in the most accurate manner. The prototype allows users to find legal information faster and guarantees consistency across the Refworld.org platform.