Using data I described in last week’s blog post, I created a Google Fusion network visualization showing the connections between various subfields of machine intelligence research. I think looking at this visualization probably does far more for me (with the background research I’ve done on specific projects, scholars, and important dates connected with these research fields) than for anyone without any specific knowledge of this field. Because users of the visualization can’t click through to see other metadata I’ve associated with the nodes in the network, the information conveyed is rather superficial. Still, the network visualization helps to give a general sense of the important fields and their relationships to subfields in machine intelligence.

More on the art history side of the art-and-technology project I described last week, I’ve also generated an ImageQuilt using the ImageQuilt Chrome plugin. These particular images are all associated with Experiments in Art and Technology, a initiative of the 1960s and 1970s founded by Billy Klüver, an engineer at Bell Telephone Laboratories, his colleague Fred Waldhauer, and the artists Robert Rauschenberg and Robert Whitman. The group aimed to collaboratively produce art using new technology. A loose association of artists and technologists, artists including Jean Tinguely, Andy Warhol, Jasper Johns, and Yvonne Rainer were also involved with Klüver’s projects. The idea behind the E.A.T. collaborations was to allow for the creation of works that may not have been possible without the involvement of engineers, who would in turn be inspired by the artists to help shape the future of technology.1E.A.T. Experiments in Art and Technology. Edited by Sabine Breitwieser. For the Museum der Moderne Salzburg, 2015 Many of the images are from a 2015 exhibition, E.A.T. Experiments in Art and Technology at Salzburg’s Museum der Moderne. It’s an interesting way of bringing together some of the more canonical images of works from this movement, but again, I’m not sure how informative it is other than on a superficial level. I found it useful as part of my broader research efforts, but I wouldn’t incorporate it into a final version of a project.

Image quilt of google image search, "E.A.T. Experiments in Art & Technology"
Image quilt of google image search, “E.A.T. Experiments in Art & Technology”

Though neither of these visualizations were what I had in mind as I was plotting out my data last week, and though I’m not sure that they’re useful in conveying or clarifying much information to an outside observer, they’ve helped me to organize some segments of my research in these areas. I’ve also learned how to organize my data in Google Fusion tables as a result of inputting some elements of my research into a spreadsheet.

In the course of looking more into network visualizations and how they have been used in digital art history projects in order to determine what form (in an ideal world, with plenty of time) I’d like my visualization to take, I came across the Performance Artist Database, created by Matthew Miller in conjunction with his MFA thesis at Pratt. Focusing mainly on Fluxus artists, Miller organizes performance artists, events, and interactions through the use of quantitative analysis in order to explore the development of artistic movements and mediums in what is traditionally a difficult art form to preserve. Click ‘View Entire Network’ to see Miller’s network visualization of his research, searchable, and with nodes linked to his own metadata (dates, locations, and performances) as well as corresponding dbpedia items. It’s impressive, useful, accomplishes what he set out to do, and could serve as a starting point for further research in this area.

References   [ + ]

1. E.A.T. Experiments in Art and Technology. Edited by Sabine Breitwieser. For the Museum der Moderne Salzburg, 2015

Comments

Erin,

I had some similar thoughts as I made my own visualization, an ImageQuilt of Schwitters’ various works with typography. I found the exercise to be useful in helping me to think through the source material, and represented a visual organization of this thought process, but I’m not sure how helpful it would be to someone looking at it without much background knowledge, or even as a visualization in a report or presentation. The end product is an interesting looking block of images, and certainly more visually engaging than just a few photographs, but perhaps doesn’t do too much beyond that superficial level. The ImageQuilt app was valuable to me though as a tool, and the process of collecting and arranging images helped to generate some potential research questions for me.

I think this is true of many visualization tools we’ve thought about – they can be equally useful as tools for research as well as tools for generating visual illustrations of research. Often the visualization itself is a kind of byproduct of the research process – an artifact that is visually engaging but also demonstrates the work done by the scholar. This is definitely true of the Performance Artist database that you should. The network visualization both engages me with the research, but also gives me great insight into Miller’s research process. One of the keys here, as you note, is that this particular visualization readily links to the data behind the network. As a reader of this visualization, I can go back and forth between the network and the entries for specific artists, which helps me to learn more about individual artists and to also contextualize that information within the broader network of their peers, exhibitions, artworks, and so on. I do especially like the links out to open data repositories like dbpedia, which situates Miller’s project within a then broader network of information and research.

As we continue to think about visualization of research data and visual literacy required to make sense of these visualizations, I think it will be important to assess how the data behind the visualization is presented. These examples have helped me to think further about that.

Funnily enough, timing is everything and there does seem to be a synchronicity in art historical research on certain topics–I didn’t know about this Bell Labs group until I visited Mashup at the Vancouver Art Gallery a week ago (have ordered the catalog for Sloane Art Library) and saw one of Rauschenberg’s pieces (a large series of rotating plexiglass discs with printed translucent images that would overlay as you spun the discs by pressing buttons on a controller–influenced by Duchamp’s spinning wheels but predictive of video game interactivity in some ways) as well as images of his collaboration with Yvonne Rainier.

You and Colin both make me think of the aspect of working with images that my colleagues most often profess to miss, that of playing with slides on a light table to make arrangements conducive to thoughts related to research comparison and contrast methods. ImageQuilt is one of several light table mimicking tools in digital form. So is this one aspect of doing art history (compare/contrast) that isn’t really a methodology but is really so fundamental to art historical thinking that it is still seen as useful to preliminary research by younger scholars like you and Colin? Thinking all the way back to our first week and how Johanna Drucker talked about how the digital has the potential to change methodology of art history. Is image aggregating just digitized or digital then? And what can the digitized versions bring that light tables couldn’t (quantity of image data? speed? really, truly finally having all of the images you want to compare without having to compete with so-and-so professor who borrowed the slide of the Farnese Hercules?).

Hi Erin,

While the Google Fusion chart you posted may seem at first superficial, there is at least more information there for a user than first meets the eye. For example, when I hover over the nodes I do see that you have included directional information. I can see this chart being helpful for a user who perhaps has some prior knowledge, like maybe the level of research you have done, in trying to understand the relationships between these subfields. I would think that when you were creating this chart you had to really solidify your determinations of categories, of which subfield is nested under which other, or if they are simply related in both being examples of machine learning. Like Kelsey brought up in class, this is a very IO (Information Organization) stage of looking at data. When you mentioned metadata you have associated with the nodes in the network, is this metadata located in your Google Fusion table for your reference? Or somewhere else? I am looking forward to playing with Gephi in class to hopefully be able to link out from nodes to metadata or other web resources. If I’m understanding correctly from the example we looked at in class that you brought to our attention from the Fluxus network chart, Gephi has this, and other major design advantages over the simple/superficial Google Fusion.

css.php