The Arolsen Archives is creating a massive digital memorial for victims of Nazi persecution.
A human-machine collaboration powered by Solutions.AI is helping accelerate the speed and improve the accuracy of document digitization.
People, working alongside Solutions.AI technology on prisoner lists, tracing inquiries and more, have gone from extracting 4 documents per hour to 41.
The Arolsen Archives — International Center on Nazi Persecution maintains the world’s most comprehensive archive on the victims and survivors of Nazism. Part of UNESCO’s Memory of the World, the archive includes information on some 17.5 million people from the various victim groups. These documents are valuable as historical records but also an important source of knowledge for society today.
Through its #everynamecounts project, the center is digitizing and indexing its 110 million documents and objects, including camp registrations, transfer lists, prisoner lists and family member inquiries. The task is massive in scale and complexity. When relying solely on manual indexing, a volunteer transcribes about four documents per hour, a rate that could take decades, if not longer, to digitize the collection.
Working with Accenture, volunteers and historians, the center is dramatically accelerating the speed of digitization, indexing the documents more rapidly than it ever has before.
Thanks to human-machine collaboration, the Arolsen Archives can enter an estimated 40X more documents per hour into a searchable database.
Accenture’s Solutions.AI is a collection of artificial intelligence solutions designed to unlock new efficiencies and growth, enable new ways of working and facilitate game-changing innovation. To support human volunteers, the Arolsen Archives is tapping into artificial intelligence, optical character recognition and other emerging technologies within Solutions.AI for Processing.
Through this human-machine collaboration, the organization is entering an estimated 40X more documents per hour into a searchable database.
The Arolsen Archives and Accenture prioritized digitization of documents considered too difficult for humans to read and transcribe. That difficulty may be due to a variety of factors, including weathering, illegible entries and/or inaccuracies. AI technologies — with continual instruction by humans — are enabling faster, more accurate digitization of these documents.
AI reviews documents
Solutions.AI reviews the documents and assigns a confidence level to each field (e.g., last name, religion, region, etc.).
AI assigns confidence levels
Solutions.AI assigns a “high” confidence level to documents that it can read easily.
Humans review selected documents
Humans provide feedback on high-confidence documents to help the AI better interpret lower-confidence documents.
AI learns and improves
Feedback from volunteers and historians continually improves the accuracy and speed of document preservation.
"Our work is to make sure victims’ stories can be told, so we can be vigilant when we witness intolerance."
— FLORIANE AZOULAY, Executive Director – Arolsen Archives, Germany
Housing the world’s most comprehensive archive on the victims and survivors of Nazi persecution, the Arolsen Archives have been recognized by UNESCO’s Memory of the World program. With nearly 110 million documents and objects in the collection, the center is a vital source of knowledge and information regarding the stories of millions.