Source: DAH Post #10: Network Visualization and the Reduction of Data
Scott B. Weingart’s blog post on the basics of network visualization, “Demystifying Networks”, is a concise overview of the pitfalls digital humanists (and data scientists) can fall into when deciding to use networks to visualize their data. In addition to giving some background information on organization theory and the makeup of networks, Weingart places special emphasis on bias in presentation of data, and the various forms it can take.
Weingart is the digital humanities specialist at Carnegie Mellon University and a historian of science. His blogging style is an excellent example of personal/professional writing about research interests and important goings-on in the field of network analysis. Something to strive towards.
In “Demystifying Networks”, Weingart notes that though networks could conceivably be used on any project, that doesn’t necessarily mean they should be. In a similar vein, digital humanists who use a technology for a purpose other than that for which it was originally designed should be able to fully justify that usage. Weingart stresses that humanistic data are typically flexible and open to interpretation, while node types used in network visualizations are concrete. Attempts to fit humanistic data into network visualizations must acknowledge and contextualize any reduction of data that results.1Scott. B. Weingart, “Demystifying Networks”. www.scottbot.net, accessed April 6, 2016. jQuery(“#footnote_plugin_tooltip_1561_1”).tooltip({ tip: “#footnote_plugin_tooltip_text_1561_1”, tipClass: “footnote_tooltip”, effect: “fade”, fadeOutSpeed: 100, predelay: 400, position: “bottom right”, relative: true, offset: [10, 10] });
It was especially useful to read Weingart’s post in conjunction with taking a second look at “The Global: Goupil & Cie/Boussod, Valodon & Cie and International Networks” section of Pamela Fletcher and Anne Helmreich’s “Local/Global: Mapping Nineteenth-Century London’s Art Market”. Using the stock books of Goupil & Cie/Boussod, Valodon & Cie, held by the Getty Research Institute, Helmreich presents and contextualizes a network visualization of information about London’s role in the internationalization of the nineteenth-century art market. In discussing the research question, Helmreich notes that this type of study is not well-suited to traditional art historical methodologies involving close reading of a small dataset, but instead benefits from “distant reading”. Quoting Frank Moretti, Helmreich writes that a different framework ‘allows the scholar to look at “units that are much larger or smaller than the text.’ Moretti adds that ‘if we want to understand the system in its entirety, we must accept losing something,’ but justifies this loss by pointing out how distant reading holds the promise, by allowing a larger corpus than before to be studied, of producing analyses ‘that go against the grain of national historiography.’” In this example, Helmreich and Fletcher acknowledge a reduction of data (the “close reading” of artworks and associations) in order to clarify a larger picture of the London art market that is more appropriate and specific to their research question.2Pamela Fletcher and Anne Helmreich, with David Israel and Seth Erickson, “Local/Global: Mapping Nineteenth-Century London’s Art Market,” Nineteenth Century Art Worldwide 11:3 (Autumn 2012). jQuery(“#footnote_plugin_tooltip_1561_2”).tooltip({ tip: “#footnote_plugin_tooltip_text_1561_2”, tipClass: “footnote_tooltip”, effect: “fade”, fadeOutSpeed: 100, predelay: 400, position: “bottom right”, relative: true, offset: [10, 10] });
This acknowledgement is also pertinent to Weingart’s discussion of bias in another of his blog posts on network visualization, “#humnets, paper/review”. In that post, Weingart summarizes UCLA’s Networks and Network Analysis for the Humanities conference, describing the reaction at the conference to his talk on bias in network analysis. Noting that everyone present was well aware of the problem bias poses, Weingart asserts, “As long as we’re open an honest about what we do not or cannot know, we can make claims around those gaps, inferring and guessing where we need to, and let the reader decide whether our careful analysis and historical inferences are sufficient to support the conclusions we draw. Honesty is more important than completeness or unshakable proof; indeed, neither of those are yet possible in most of what we study.”3Scott. B. Weingart, “#humnets, paper/review,” www.scottbot.net, accessed April 6, 2016 jQuery(“#footnote_plugin_tooltip_1561_3”).tooltip({ tip: “#footnote_plugin_tooltip_text_1561_3”, tipClass: “footnote_tooltip”, effect: “fade”, fadeOutSpeed: 100, predelay: 400, position: “bottom right”, relative: true, offset: [10, 10] });.
Moving to this week’s example:
Since I used Google Fusion tables in my last network visualization test, I decided to give Gephi a try. Rather than focus on the tenuous and arbitrary connections between subfields of neural network research, this network uses data that is a bit more concrete. In keeping with the focus of the TimeMap from my last post, I’ve charted artists and engineers who were involved in LACMA’s Art and Technology Program and Experiments in Art and Technology (E.A.T), as well as two major E.A.T. projects, 9 Evenings: Theatre and Engineering and the Pepsi Pavilion at the 1970 World Expo in Osaka, Japan. This list of names, and therefore this visualization, is by no means comprehensive–the Pepsi Pavilion alone included contributions from over 75 artists and engineers. In the interests of time, I’ve included only the most active participants in both E.A.T. and the A&T program. As sources, I used:
Maurice Tuchman, Art & Technology; a Report on the Art & Technology Program of the Los Angeles County Museum of Art, 1967-1971. Los Angeles County Museum of Art; distributed by the Viking Press, New York, 1971.
Kathy Battista and Sabine Breitwieser, E.A.T. – Experiments in Art and Technology. Translated by Karl Hoffman. Köln: Verlag der Buchhandlung Walther König, 2015.
The text placement is, as we discussed in class, not ideal. And the other problems with this network are the same ones that Weingart attributes to any beginner in network visualization in his blog post: the fact that bimodal networks are difficult to work with, that Gephi works best with single node networks, that it measures centrality of nodes, and the size of those nodes is adjustable in the visualization in accordance with the narrative the user is striving to create. However, this network represents a start to a more comprehensive project exploring the connections between artists and engineers in the art and technology programs of the late 1960s and early 1970s.
1. ↑ Scott. B. Weingart, “Demystifying Networks”. www.scottbot.net, accessed April 6, 2016. 2. ↑ Pamela Fletcher and Anne Helmreich, with David Israel and Seth Erickson, “Local/Global: Mapping Nineteenth-Century London’s Art Market,” Nineteenth Century Art Worldwide 11:3 (Autumn 2012). 3. ↑ Scott. B. Weingart, “#humnets, paper/review,” www.scottbot.net, accessed April 6, 2016 function footnote_expand_reference_container() { jQuery(“#footnote_references_container”).show(); jQuery(“#footnote_reference_container_collapse_button”).text(“-“); } function footnote_collapse_reference_container() { jQuery(“#footnote_references_container”).hide(); jQuery(“#footnote_reference_container_collapse_button”).text(“+”); } function footnote_expand_collapse_reference_container() { if (jQuery(“#footnote_references_container”).is(“:hidden”)) { footnote_expand_reference_container(); } else { footnote_collapse_reference_container(); } } function footnote_moveToAnchor(p_str_TargetID) { footnote_expand_reference_container(); var l_obj_Target = jQuery(“#” + p_str_TargetID); if(l_obj_Target.length) { jQuery(‘html, body’).animate({ scrollTop: l_obj_Target.offset().top – window.innerHeight/2 }, 1000); } }