Artefacts at a click

FAU computer scientists program intelligent research software

What do the Metropolitan Museum of Modern Art in New York, the Max Planck Society and the German Archaeological Institute have in common? They are all now working with a pioneering software programme developed by scientists from the Department for Computer Science 8 (Theoretical Computer Science) at Friedrich Alexander University Erlangen Nuremberg (FAU) led by Prof. Dr. Günther Görz. This software makes it possible to exploit databases semantically. As a result, the search and research processes are significantly simpler and their results far more comprehensive. Whereas conventional search engines only search databases for a specified phrase – i.e. an exact string of characters – filtering out all other information that may be relevant to the given search term, with the software from Erlangen users can cut through colossal amounts of data logically and systematically. Now the University of Oxford is taking advantage of this know how, too. Using the programme developed at the FAU, the British University has developed the CLAROS search system. This tool links up databases on ancient art from across the globe, including from museums and art collections.

“Image you are looking for a photo of a vase portraying Hercules with spear in his hand,” Prof. Görz offers as an example of the Erlangen research software. “But you don’t know whether a vase with that motif exists or what it is called. So you simply describe to the computer the arrangement of the motif you are looking for using important characteristics.” CLAROS allows you to input a range of search terms such as “vase” or depictions such as “Hercules” into search categories including “location”, “time period” or “database”. The programme then provides a list not only of all images featuring your desired motif, but also additional photos of vases where Hercules is depicted with other weapons or tools. Alternatively, the user can upload an image into CLAROS or input a link to a photo on the internet that closely resembles the Hercules vase they are looking for. In this case, again, the programme searches for images that match the specified motif. “Procedures such as this are not technically possible with traditional search engines”, said Görz.

But that is not all that CLAROS can do. For instance, it can also find an ancient Greek inscription that has been split in two, with one half in an Athens museum and the other in a US art collection. Both halves have been digitally recorded and can be viewed on the internet. But the two halves are not mutually referenced and thus there is no complementary referencing. Moreover, the Greek museum’s database has been set up differently from its American counterpart and the description or location of the inscription has only been provided in Greek or English. If you were using conventional search engine, you would very quickly run into obstacles trying to piece together the two matching parts. This is because your traditional search tool only looks for identical strings of characters and there aren’t any in this case. CLAROS, on the other hand, is able to identify that these two artefacts form a matching pair.

How CLAROS works

CLAROS is based on the Conceptual Reference Model (CRM). Participating academics see CRM as a very general system composed of terms and characteristics that can be used to describe objects. The model is particularly suited to museum documentation because it facilitates universal access to and exchange of a wide range of data. CRM has gained recognition as the international standard for museum documentation.
For a long time, the idea behind this system didn’t get off the ground. To implement this construct into a computer programme, Prof. Görz and his team used “Web Ontology Language”, a logical programming language used for knowledge representation and processing on the “Semantic Web.

Many scientists are now using the Erlangen computer programme for their own projects and developing it further. The software developed by Prof. Görz is available free on the internet.

Further information is available at http://erlangen-crm.org

Further information:

uni | press service | research No. 27/2011 on 21.06.2011