Source: #DAH symposium and Network Visualizations

This week followed a somewhat different structure, as we had a “Special Day” on Monday, when the Digital Art History conference was being held at Duke. Unfortunately, I missed the only portion of discussion/workshop this week, which was held on Wednesday. This is too bad; as I probably could have used some additional discussion time on our readings about data in Digital Art History. Fortunately, I was able to attend the last two presentations in the #DAH conference at Duke. Ingrid Daubechies, a Professor of Math at Duke, has been working with museum curators and conservators by developing algorithms that go so far as to detect the wood grain on frames that show up on x-rays, in order to remove that shape and texture for more precise cleaning and conservation. C. Griffith Mann of the Metropolitan Museum of Art closed out the presentations with “Museums in a Digital World: Engaging Audiences in the Collection”, which was an overview of his work with digital tools for museum visitors at the Cleveland Museum of Art, like an interactive digital wall. Its nice to see the connection between the behind-the-scenes metadata from the museum database and the end result for a playful, interactive museum going experience.

Our readings this week were themed by data. I would like to learn more about using textual data and using machine-readable text. I took English professor/Digital Humanist/blogger Ted Underwood’s advice[i] and checked out JSTOR’s Data for Research API, which is a free tool that gives access to 7 million journal articles. The really cool thing about this API is that it includes additional tools beyond search and machine-readable text: information about word frequencies, citations, topic modelling, and visualization tools.

This past Friday I plugged in over 100 relationships (for now sticking with mutual relationships) for my Kress project. I chose 3 relationship types (married to, related to, worked with) and 3 main entities between Artists (Designers/Engravers); Authors (Humanist writers/Translators/Editors); and Printers (Printing Houses/Publishers/their spouses). I included spouses (read: wives), as these women are often the link between generations of printing houses in Basel. There was intermarrying between the major competitors from the beginning of the printing boom in this city – my aim was to show the use of marriage as a business tool to benefit multiple families. While this intermarrying did not form some kind of formal conglomerate, once I saw just how many women I was listing who were connected as a daughter or wife or in-law to various printing houses, I speculate that this intermarriage at least fostered healthy competition in Basel, and strengthened the city’s role as a Northern European center of publishing. It possibly brings up a potential research area of women’s work in these printing houses: were they supportive housewives or did they play an active, if unacknowledged role in the business, in creating partnerships between artists and engravers, in negotiating contracts, proofreading, physically running the presses?

A network visualization is a great way to tell a story – like other data visualizations we talked about in class, I think there is potential for misleading the viewer if they are not properly described and contextualized. For example, “Martin Luther” as a node is rather small in my Google Fusion table. One could mistakenly read this as a mistake for such a key figure of the Reformation in the 16th century. However, the story I am telling is of networks in Basel only, and it appears that as far as working with Basel publishers, Luther stuck with only the Curio family, and then specifically its patriarch Valentin Curio. Some nodes are also larger due to a high number of connections through marriage and bloodlines, while not reflecting a high production rate or anything like that. In playing with this network visualization, which I will share in its final stage, has been really instructive in the choices we make with data. Even my fairly simple scheme (3 entities, 3 relationship types, only mutual relationships) led me to think critically about just how exhaustive I want to make my chart, or how exhaustive I am able to make my chart, given the data available to me is certainly incomplete. It also reinforces the importance of expertise in a specialization of art history – I’m looking forward to running this list of relationships by the Art Historian I’m working with – while I am working more from “raw data” here than deep historical knowledge, Miranda will certainly be able to brainstorm more publications and partnerships that she has come across in her research.

Below is my network visualization as-is from Google Fusion charts. I’m going to play with other visualization apps like Palladio. The way I organized my spreadsheet did not separate by gender, and I want to make that a focus on my chart, maybe just by color-coding.


[i] Ted Underwood, “Where to Start with Text Mining,” The Stone and the Shell.