FAU researchers develop software for the digitisation of art treasures
Visitors strolling through an exhibition tend to catch only a small glimpse of the countless treasures that are stored in a museum. Treasures that need to be sorted, catalogued, presented. Where museums used to work with index cards, electronic databases are now becoming more and more common. FAU computer scientists are helping to bring our cultural memory up to speed for the digital future.
In fact, museums have worked with electronic databases to catalogue their objects for a long time. Researchers can search by place, year, name or key words, for instance. However, if a researcher wants to search the entire inventory of a museum or even several museums for all graphic works by one artist, things may become a little bit difficult already: if the works were tagged with such key words as ‘drawing’, ‘woodcut’, or ‘etching’, then a general search for the term ‘graphic art’ won’t return any results. This only works if the programme recognises how the individual terms are connected, and that is where conventional databases meet their limits.
The solution lies in a type of software that connects categories and properties in a sensible way and which allows several institutions to exchange data online easily: a digital depository of knowledge for museums, archives and other collections. This is exactly what Prof. Dr. Günther Görz from the Digital Humanities working group at FAU is working on together with the Germanic National Museum in Nuremberg and the Zoological Research Museum Alexander Koenig in Bonn: WissKI – short for Wissenschaftliche Kommunikations-Infrastruktur (Scientific Communication Infrastructure) – has been running since August 2009 and will receive funding from the German Research Foundation (DFG) for another two years.
The programme is based on the reference model of the International Council of Museums. A standard model has been in development there for ten years; it determines how categories of object descriptions relate to each other. The goal of WissKI is now to translate this semantic model into software. The advantage of this is that the data on objects from different museums can be imported very easily – no matter whether the information is from a museum of natural history, engineering or art. Scientists hope to be able to answer new kinds of questions faster: What do the plants and animals featured in the paintings of a certain era tell us about the climate at the time? How can the far-flung works of one artist be compiled digitally and thus the relationships between the different places in which they worked be visualised, for example through the artist’s journeys?
This becomes possible thanks to the complex structures hidden behind a seemingly simple form with fields such as ‘artist’, ‘date’ or ‘material/technique’. For instance, if Nuremberg is entered under ‘place’, the database gleans several pieces of information at once: the information carrier was created in a place, the place has a name, and the name is Nuremberg. Even text entered into a free field is only really free text at first glance: the system offers automatic indexing with a hit rate of about 80 percent. Thus, in a test using a self-portrait by Albrecht Dürer, ‘Albrecht Dürer’ and ‘Agnes Frey’, Dürer’s fiancée, were correctly identified as names. However, the programme also treats the little word ‘Self’ as a person – after all, there’s a portrait of them. This is why it is important to check the automatic index and correct the text where necessary.
An aspect of the system that works without human intervention, though, is the exchange of data sets among different museums or archives that use WissKI. With the click of a mouse, information from already catalogued treasures can be used in new databases. WissKI is currently designed exclusively as a kind of virtual research environment for researchers. ‘We do also envisage it being used as the foundation for online exhibitions or exhibition catalogues in the future,’ explains Prof. Görz.
But there are still plenty of things for the WissKI team to do before then. Currently the researchers are trying to get the software to answer not only simple questions regarding persons, places or properties of works, but also more complex questions that allow for conceptual conclusions – for example that terms such as ‘drawing’, ‘woodcut’ or ‘etching’ are subordinated under the umbrella term ‘graphic art’. ‘We want for knowledge that has been collected, whether it be in an object catalogue or a special research project, to be made more quickly and easily accessible to others,’ Görz summarises. A digital cultural memory for everyone.
Prof. Dr. Günther Görz
Phone: +49 (0)9131 85 29099