I’ll be updating the DH@Stanford graph for the upcoming DH11 Conference. Below is the previous iteration and explanation.
Below is an attempt to map out the various individuals, projects and institutes involved in the digital humanities at Stanford. Connections are based primarily on descriptions of project participants on affiliated websites. The network maps below are static exports created from Gephi graphs and, along with presenting an interesting overview of the ongoing digital humanities work here at Stanford, problematize the concept of the Digital Humanities and highlight the need for metrics that go beyond discrete self-identification. Each image below links to a browsable network diagram:
Communities are determined using the Louvain method to measure modularity. For instance, the latest DH@Stanford graph has grown quite complex:
There are now 429 nodes with 556 edges detailing the projects, people, organizations and materials taking part in (broadly defined) Digital Humanities activities at Stanford. Analysis of the community of these nodes demonstrates three categories of module. The first is of the large, self-identified and self-contained projects that are, primarily, the sole “digital humanities” activity that a linked scholar is known to take part in. These include the Spatial History Project, the Lit Lab and Parker on the Web. These large, homogeneous entities could easily be evidenciary artifacts, based on incomplete knowledge of the full activities of their participants and the existence of greater connectivity between projects and scholars that is unreflected in the data.
The medium-sized modules include particularly active individuals (such as Paula Findlen or Dan Edelstein) as well as small archives or major, identifiable subcomponents of larger projects that did not appear earlier (supporting the truism that better data collection will reveal a more clear picture of the objects being represented).
Finally, the smallest modules include a hodge-podge of individuals, small projects and, especially, outside contributor-host university pairs. Many of these will fall out with better detail, but some of these smaller modules are accurate representations of small projects in need of integration into other projects or interested scholars who with whom collaboration may be a real possibility.
Visualization using network diagrams, while aesthetically pleasing and offering new analytical functions, reveals several problems:
Lack of Coverage – After building a draft of this map based on the project and institution pages across Stanford, I’ve already been confronted by projects and people that should be represented on the map but are not. By all means, if you know of a person or project, institute or other object that should be represented on this network, just leave a comment below.
Lack of Definition – The projects and people above have self-identified as being part of the Digital Humanities, but the definition of the Digital Humanities is contested and the result of such contention is that various people and groups may not consider themselves to be part of the same network, or the reverse.
Lack of Metrics -As it exists currently, this network map bases all measurements on number and direction of connections. This preferences self-identification of multiple connections with multiple projects and projects connected to many participants. For this map to move to a more useful space, there need to be agreed upon metrics for project and participant success. To differentiate between a highly productive digital humanities project and a project with a great name and a great list of collaborators that has failed to produce anything, we need to distinguish between the type and quality of the possible products.
How, for instance, do we measure and compare the product of Arcade, with its high visibility on the Web and large number of blog articles, with a project that publishes a large database, an interactive site or a traditional monograph? The inclusion of Product nodes (to complement the existing Person, Project, Institute, et cetera nodes) would better measure influence and quality of connections and allow for a more meaningful map.
As with missing nodes and edges, please feel free to leave a comment about any other aspect of this map that is particularly interesting, problematic or otherwise notable.