Mr Dörk, you work in the field of visualisation research. In your lecture on The Visualisation of Cultural Data, you showed us the kind of experimental approaches to digital collections that are possible. How did you discover cultural data as a research subject and as material for visualisation?
Marian Dörk: In fact, I have always been interested in data which is more than just figures, though not necessarily from the cultural sector. In the field of information visualisation, we deal with complex and multidimensional data spaces and datasets, such as text corpora, social relationship networks, objects or collections. My current focus on cultural data arose from the VIKUS – Visualizing Cultural Collections project, which I lead at the Potsdam University of Applied Sciences.
What intrigues you about working with cultural data?
In comparison to the visualisation of other data, cultural data is very rich. On the one hand, it is about metadata: information about objects for which the temporal and spatial dimensions and the context are significant. On the other hand, the objects themselves, i.e. the artefacts in a collection, can display their own visual, aesthetic or material characteristics. These visualisations raise new and different questions which are of interest to researchers in the fields of cultural studies and humanities, as well as visualisation.
What exactly is visualisation research? When was this discipline established?
Visualisation has existed for centuries: ever since we started recording information and data, this also occurred visually. Since the early 1990s at the latest, there has been a computer science based research community which focuses on computer-aided data visualisation. There is a distinction between scientific visualisation, which is the visualisation of scientific data, in particular for natural sciences, and information visualisation (known as Infovis), which is the visualisation of abstract data and relationships. What is more, visualisation researchers refer to and have been influenced by the many predecessors in the fields of cartography, statistics, graphic design, town planning and cognitive psychology.
Essentially, visualisation research is about making large volumes of complex data viewable so that they can be analysed and new patterns and references can be identified. This visualisation is computer based and uses algorithmic processes. It enables researchers to interact with the data.
What role do visualisations play in terms of collection contexts and the reception of cultural collections?
Currently they play a minor role. Institutions generally focus on presenting their collections in a physical space. However, together with visualisation experts, cultural institutions are increasingly exploring the possibilities of digital research and communication methods such as visualisation.
What ideas, expectations and wishes do libraries, archives or museums approach you with?
When we collaborate with an institution, there are two areas visualisation can contribute to. The first is generally an internal one: the curators want to gain a better understanding of the collection they work with. Through visualisation, the range, dimensions and contexts of the collection can be perceived in a new way. The second aspect is an external function: the desire to make increasingly digitalised collections available for re-use in an exciting, insightful and perhaps also entertaining or educational way.
In recent years, cultural institutions have invested extensively in digitalisation, both in the creation of digital content and in digital indexing. Visualisation is linked to a desire to make these treasures, which are in principle lifted, accessible to all and available to many different user groups.
Different overviews and levels of detail in the visualisation project ‘Past Visions Penned by Frederick William IV’ by Katrin Glinka, Christopher Pietsch and Marian Dörk in cooperation with the Prussian Palaces and Gardens Foundation Berlin-Brandenburg and Programmfabrik. © VIKUS
And are you generally able to fulfil these wishes? Or are there limitations?
Yes to both questions. There are also limitations, but I consider them to be productive; they are interesting friction. There is often the notion that there is an ideal visualisation, but this is a myth. Data is always multi-facetted and there is always more than a single graphic or visualisation is able to present. I argue that if you want to approach a complex theme or collection, there is no such thing as one visualisation. Our capacity for perception is also a limitation. Pledging to create an overview of millions of data points does not yet mean that we can actually understand them or recognise something meaningful. There are human limitations, limitations in our ability to read and interpret things. Not every viewer can create meaning from and interpret what they see.
So the perfect visualisation is an impossible promise. But can you tell us what, in your view, makes a good visualisation?
Visualisations do not work well if they are static. They should support interactive analysis and exploration in a way which does not only pass time, but which actually encourages cognitive processes. In this sense, a visualisation should be oriented towards the viewers’ questions. Of course, it should also be suited to the data and focus on the most important aspects.
A variety of perspectives are essential here: how can we incorporate multiple perspectives in terms of critical or feminist epistemology? And in practical terms, how can we create dynamic interfaces which allow us to switch between representations? There is still a lot to develop and a need for further research.
Can you tell us about a successful visualisation?
The Explore · Australian Prints + Printmaking project by Mitchell Whitelaw und Ben Ennis Butler at the University of Canberra is a good example. They present five different visualisations which brilliantly complement each other and show how a collection of historical prints can be made accessible from multiple perspectives. The aspiration is not to present a set of data in a conclusive way; instead, the composition invites the viewer to switch perspectives and thereby gain different impressions.
Is there such a thing as ideal data?
All available data is ideal data (laughs), because it is attainable data: data you can work with. That may sound banal, but it cannot be taken for granted. What is more, it should be available in a structured form so that the key aspects and dimensions can be extracted. This is both fundamental and the very first precondition. Finally, the quality of data is an important factor. If there is a variation in data quality and the use of data fields, this presents a particular challenge. Heterogeneous data is always both difficult and fascinating; and now I come to think about it, it is more often the rule than the exception. This raises the following questions: what do you do if essential dimensions for particular objects are not documented? What happens to these objects? Do they slip through the net? Do we only want the complete data sets? If so, what happens to the rest?
In addition to the data that is displayed, we should always scrutinise the missing data. Incidentally, from an information visualisation perspective, it is fascinating to look at data when the exact content is not yet known and this is what inspires data analysts. This raises the question of how such data can be represented before we actually know what is interesting about it.
Your lecture was presented a part of an MWW workshop in Weimar on the indexing and visualisation of writers’ libraries. The workshop focussed on bibliographical data. What opportunities do you see for this particular type of data?
Up to now, bibliographical data has generally been viewed from a search field perspective. This is what determines the conventional search mode. However, bibliographical data also provides the option of supporting other forms of access because it downright provokes other references from books and other entities, from authorship and co-authorship to keywords and related terms. Implicit references can also lead to links due to corresponding metadata, such as a similar period of production, or two unconnected authors whose work was published by the same publisher. Additionally, visualisation enables us to work on a completely different level if the information goes beyond bibliographical data and is able to incorporate complete texts, images or content analysis in addition to metadata.
On the one hand, we are used to looking at the pie charts and bar charts that anyone can create with a few clicks in a standard Office programme. On the other hand, a wide range of visual pageantry can be found on the internet. What role do good design and aesthetics play in visualisation?
Aesthetics has become a major theme in visualisation research. You could say that Edward Tufte, a great defender of purist information visualisation, propagated visual asceticism. In contrast, studies have shown that diagrams, in other words, static graphics, actually communicate information more effectively when they contain decorative elements: chartjunk.
And what is your personal opinion?
I deem design to be of particular importance, when we are not just dealing with quantitative data, but when we are focusing on objects which have an aesthetic of their own. If we want to visualise and convey a collection which has its own aura, spirit or atmosphere, then we should make an effort to preserve this to some extent. Up to now, most visualisation techniques and search systems have tended to be aesthetically agnostic.
What does that mean in practical terms?
Well, these systems are often oriented towards neutrality, clarity and, in line with Edward Tufte, minimalism which does not do justice to the particular qualities of the collections which I just described. Let us compare a digitised collection with an exhibition: a physical presentation in which aesthetics play a key role, from the colour of the walls, to the lighting and composition of the exhibits. All these crucial curatorial and aesthetic considerations are generally ignored in a digital representation.
However, just as in the fields of philosophy or art history, aesthetics is not a finalised concept, this is also the case for information visualisation, and visualisation researchers would not claim to be able to take on this task. Our current research in Potsdam is about developing collection interfaces which are both content appropriate and culturally sensitive.
If the cultural data fairy would grant you three wishes from cultural institutions, what would you wish for in the future?
Openness would be top of the list: in relation to communication and the desire to actually make content publicly and freely available so that experiments such as visualisation or data analysis would not be hindered by legal barriers. My second wish relates to access to data, as I mentioned earlier, for technical availability and structuring of data. And finally, I would wish for a desire to work with designers and visualisation researchers so that rather than being a process disjoined from institutions and curators, the communication and design of interfaces would be an interactive, interdisciplinary process. This is the only way to create outcomes which do full justice to collections. This is the triad I would wish for.
Dr. Marian Dörk is research professor for ‘Information Visualisation and Management’ at the Potsdam University of Applied Sciences. He leads VIKUS - Visualizing Cultural Collections, a three-year project funded by the Federal Ministry of Education and Research (BMBF). Along with his interdisciplinary team, he researches new methods of digitally presenting cultural heritage in an accessible way. He was invited by MWW researcher Lydia Koglin to present an evening lecture on ‘The Visualisation of Cultural Data: Experimental Approaches to Digital Collections’ in Weimar on 11th April 2016. This lecture was part of the two-day MWW workshop ‘Libraries within the Library: Indexing, Reconstruction, Visualisation’ in the scope of the MWW research project ‘Writers’ Libraries’.
Lydia Koglin is the ‘Digital Humanities’ research project associate at the MWW Research Association in Weimar.