Artificial Intelligence and the Preservation of Historic Documents

Abstract

In recent decades, digitization has been presented as an important strategy both for the preservation of historic documents and for giving increased access for researchers to such materials. In the Norwegian context, this has not only implied the digitization of printed matter but also the digitization of audiovisual material like photography and analog tape recordings. From a technical perspective, there are of cause difficulties in digitizing such a variety of material when considering the diversity of media formats dating back to the nineteenth century. However, from the archival community criticism has been raised not only about the quality of the work but also the concerning the selection process, the organization of the material, and the collection of metadata. The National Library of Norway, especially, has attempted to avoid the problem of selection by attempting to digitize all of visual, sonic and audio-visual culture heritage. But this has created even greater challenges for the organization and registration of metadata. This is an issue that is apparent to the National Library, so it is looking into the possibility of using artificial intelligence – learning algorithms – to organize the material. The key issue is that the preservation of historic material is also dependent on the preservation of context and on metadata enabling us to interpret and understand the material at hand.