Visualizing Visions | Exploring the Rijksmuseum dataset

| Floor Koeleman discusses her PhD project Visualizing Visions, which focuses on the use of digital tools for art historical research |

As an art historian I look at the past, but today I will look at the future. What innovative methods and techniques can be used for art historical research? This question is central to my PhD research at the History Department of the University of Luxembourg.

Other academic areas have embraced digital tools to the extent that there is no need to put the extra emphasis on their use. There is no such thing as Digital Mathematics, Digital Biology or Digital Economics. Is this because they don’t use digital methods? The opposite is more true. Using and developing high-tech, cutting edge technology is a key part for their scientific process and progress made. For scholars in these fields, explorative research driven by digital innovation has been an important and accepted methodology. Yet, the use of digital tools for art historical research is still considered a distinct and maybe even obscure sub-discipline: digital art history.

I focus on the following three aspects in my research.

  1. Experimenting with cutting edge digital tools and methodologies.
  2. Collecting and processing digital data and sources.
  3. Exploring how to visualize complex outcomes in an engaging and interactive way.

Let me explain by using my project Visualizing Visions as an example. There is an ongoing trend to digitize museum collections. A good example is the Rijksmuseum, whose digitization project started in 2007 and covers their complete collection containing over one million items. Many of them are already extensively annotated with title, artist, dating, size of the work etc.

When we look specifically at the digital paintings collection of the Rijksmuseum, we can filter out about 1800 paintings. Studying this subcollection through data and tools together with a computer scientist, we identified a number of themes we found interesting from both an art historical and a digital perspective. This resulted in a website consisting of 5 stories where the visitor can connect to artworks in the Rijksmuseum interactively. You can explore the results yourself on www.visualizingvisions.com.

Even though art history is a topic of general interest, a great deal of knowledge is hidden away in highly specialised literature. The Web, on the other hand, offers the opportunity to reach a much larger audience. When we look at The Art Gallery of Jan Gildemeester Jansz by Adriaan de Lelie, on the picture above, we can see a different time where people gather to discuss and enjoy paintings. And that is certainly what I want to achieve; reaching as many people as possible and get them excited about art.

One way to describe my research process can be found in Ben Fry’s Seven Stages of Visualizing Data.

  1. Acquire – Obtain the data, whether from a file on a disk or a source over a network.
  2. Parse – Provide some structure for the data’s meaning, and order it into categories.
  3. Filter – Remove all but the data of interest.
  4. Mine – Apply methods from statistics or data mining as a way to discern patterns or place the data in mathematical context.
  5. Represent – Choose a basic visual model, such as a bar graph, list, or tree.
  6. Refine – Improve the basic representation to make it clearer and more visually engaging.
  7. Interact – Add methods for manipulating the data or controlling what features are visible.

Step 1 through 5 can be considered common in the classical research process and is not necessarily linked to a digital research approach. Only in step 5 (Represent) we start to visualize. It is in step 6 (refining data) and 7 (making data interactive) that we need digital tools by definition.

An indispensable tool for this is D3.js, a JavaScript library for visualizing data using common web technologies. When using D3.js, once we get to the point of representing our data, we get steps Refine and Interact for free. Ideally you start this part of the process with pen and paper and for the final visualizations you can move over to D3.js, which is relatively easy to use, very accessible to our target audience because you can publish it on the Web, and can lead to wonderful and creative results. There is currently not really anything better for our purpose.

But beware, trying something new comes with snags! Acquiring data means something different for a computer scientist than for an art historian. It is not self-evident that art historical data is digitized, so you might have to do this yourself. When data is available in a digital format, such as the Rijksmuseum dataset, this does not automatically mean it is reliable. Fact checking is often necessary. In our case we viewed and stored all data relating to the digital paintings collection, obtained from the Rijksmuseum API in JSON format, in a relational database. We could then query this data through SQL statements in order to discern patterns.

Going through the seven stages is an iterative process. You will most likely go through them a few times in a nonlinear way, returning to previous stages when necessary. This is a different way of working from what is generally taught at Universities, and requires you to step out of your comfort zone.

I have a background in art and product design, as well as in academic art history. Experimentation means something completely different for both disciplines. At the Design Academy Eindhoven (the Netherlands) we would do all kinds of experiments on material, form, and expression, not knowing what the outcome will be. This process of trial and error, or research by doing, is essential to learn new skills and to discover what does and does not work in practice. The scientific experiment, on the other hand, assumes an expected outcome (hypothesis) to be tested for validity. I see great potential in the design methodology for academic research. This is not only a fun way of working, but also challenges you to do something new and be open to new ideas, whether they are digital or not.

During my undergraduate studies at the Radboud University Nijmegen (the Netherlands) I completed a number of projects that brought me to where I am now. For my bachelor’s thesis, for instance, I used Microsoft Excel in combination with Delpher to research wallpaper advertisements in their collection of digitized Dutch newspapers. During my research internship at the State Hermitage Museum in St. Petersburg, I used OCR technology to translate Cyrillic literature on the Russian-Dutch artistic and cultural relations at the time of Peter the Great. And finally, for my master’s thesis I created a dataset with historical sites from the former Roman Empire displaying post-Pompeian wall paintings, and linked it to available public resources such as Getty TGN & AAT, Iconclass, and Schema.org.

To stay informed you can keep an eye on my project website and personal blog.

Floor Koeleman is a PhD Candidate in (digital) Art History

Top image: The Art Gallery of Jan Gildemeester Jansz, Adriaan de Lelie, 1794 – 1795, oil paint on panel, h 63.7cm × b 85.7cm. Rijksmuseum Collection