Digitisation of museum collections promises to transgress geography and physical space by dismantling institutional and even national boundaries and opening-up a possibility to achieve a truly data based overview of composition of national and transnational cultural memory. National mass digitisation projects, at leasts in their intention, would allow to get an insight into "memory mountains" and "memory valleys" (Zerubavel) or dense and empty historical periods both on nation and potentially transnational level. It would allow to get an insight into how, e.g. the Baltic sea was represented in photography of 1930s? Or what kind of material objects are available from 16th century across Europe? Or which century or decade are most represented in museum collections and which are blank spots? At the same time, digitisation being not entirely technical but also political process, the new digitised databases have their own digital barriers of data aggregation. Accessibility barriers, disparate infrastructures, and data formats serve as impediments to full utilisation of the digitised cultural memory aggregations.
The paper will compare two publicly available national data aggregations – national catalogue of museum collection of Latvia (NMKK) and digitised database of museum collections of Finland (FINNA) – and analyse the available metadata and images, using both a conventional quantitative data analysis tools and machine learning algorithm (VGG19). How data of each of these collections can be accessed? What ways of aggregation they allow or lack? With the focus on these questions the paper will contribute to deeper understanding of cultural memory in digital era.
