AI to support research on victims of Nazi persecution

A team from Accenture set out to develop, in 10 weeks, an AI solution that facilitates the extraction of information on the victims of Nazi persecution from documents taken from the Arolsen archives. This tool increases the precision of these compilations, such as increasing the speed of the process by 40, for an average of 160 acts processed per hour, explains the company.

To do this, Accenture employees took over an already existing AI solution, which relies on optical character recognition and machine learning technology. In practice, each document kept in the archives is examined, and these elements, such as the family name or the date of birth, are inserted into a database. The tool makes it possible in particular to recover archives that are particularly difficult and tedious to consult for humans, such as lists of prisoners and transfers with dozens of rows, or even concentration camp registers.

Data available online

Nevertheless, human oversight of the process remains important, both to ensure the accuracy of the transcription, but also to improve the AI. By correcting the information, the volunteers educate the machine – by pointing to characters or abbreviations that are initially unknown to it. A bit like we find it in the Captcha letters to be validated. To support this project, the Arolsen Archives rely on a participatory production model, #everynamecounts.

We are overwhelmed by the number of volunteers supporting the digitization of our archives”, notes Floriane Azoulay, director of the Arolsen Archives. ” Our collaboration with the Accenture team stands out. It is fantastic that there is now a digital solution to capture the contents of archives more quickly, making the most important information about the fate of victims of Nazi persecution searchable in our online archive. »

Example of a document processed by the Accenture AI solution as part of the #everynamecounts project of the Arolsen archives. Archive Arolsen

Since Accenture implemented the AI ​​solution in December 2021, the tool has classified more than 160,000 names of victims of Nazi persecution, extracted data from more than 18,000 documents, and grouped more than 60,000 items into groups. to improve identification and analysis. More than 950 people from Accenture have volunteered for the project to date.

READ: Fragmentary ancient texts finally revealed thanks to AI

The Arolsen Archive houses the world’s largest collection of documents on the Nazi persecution – 110 million digital records and objects, some of which fit into the program memory of the world of UNESCO – to keep alive the memory of the crimes of the German terrorist regime.

An essential part of the Archives’ work is to make these materials accessible to anyone who wishes to search for traces of victims and survivors of the Holocaust, the persecution of minorities and forced labour.

Credits: Arolsen Archives

We would like to thank the author of this short article for this amazing material

AI to support research on victims of Nazi persecution

You can view our social media pages here and other pages related to them here.