RT @ALA_ACRL: Follow all the #acrl2011 action with the handy Twitter widgets on ACRL Insider! http://bit.ly/fbgT6J
RT @JohnsHopkins: New interface launched for searching JHU libraries collection. http://jhu.md/eVStjO
RT @ernestopriego: RT @SimonTanner New Head of Digital Humanities at King’s College London Andrew Prescott @Ajprescott http://bit.ly/ffKVwP
Examining the digital humanities community via #DHapi
Contributors: Katy Berube and James Neal
Goals of SNA Analysis:
The purpose of this analysis was designed to examine relationships as well as the primary social structure of the participants and discussants of a two-day workshop on developing APIs (Application Programming Interfaces) for the digital humanities sponsored by the Maryland Institute for Technology in the Humanities (MITH) on February 25 and 26. Participants and discussants of this event actively used the hashtag #DHapi in tweets on the social networking and microblogging service Twitter. The data used in this analysis is based on tweets containing the #DHapi hashtag used to communicate and identify messages related to the event. This analysis was designed as completion of the Social Network Analysis assignment for Dr. Derek Hansen’s class LBSC708P - “Communities of Practice” taught at the University of Maryland College of Information Studies, Maryland’s iSchool in the Spring semester of 2011. The findings may be of use to the digital humanities community at large, MITH, and the National Endowment for the Humanities who funded this event through the Office of Digital Humanities Start-Up Grants program.
Key insights expected to be gained from this analysis include:
A. What is the social structure of the #DHapi related discussion?
B. Who are the important discussants and participants?
C. What is the nature of the #DHapi conversation?
The data used in this analysis was drawn from Twitter data downloaded by Professor Derek Hansen and distributed for use to Katy Berube and James Neal. The network examined is directed, weighted, and unimodal. While the unimodal nature of the network is designated, it should also be noted that the network also includes organizations as well as individuals that comprise the vertices (e.g. @UMD_MITH = the Maryland Institute for Technology in the Humanities, @zotero = Zotero, the open source browser add-on for bibliographic data management, and @HASTAC = a digital humanities “network of networks” (Wikipedia) ). The tweets examined containing the hashtag #DHapi are dated from February 24, 2011 - March 1, 2011.
The data sets used in this analysis as identified by the visualizations below are:
B. #DHapi - Betweeness Centrality
C. #DHapi - Circle
Overall graph metrics:
Graph Metric Value
Graph Type Directed
Unique Edges 2822
Edges With Duplicates 255
Total Edges 3077
Connected Components 1
Single-Vertex Connected Components 0
Maximum Vertices in a Connected Component 154
Maximum Edges in a Connected Component 3077
Maximum Geodesic Distance (Diameter) 5
Average Geodesic Distance 2.087199
Graph Density 0.124862066
This visualization is an unfiltered raw display of the entire #DHapi data set containing all 154 vertices and 3077 edges. The edges represent follow, mention, and reply to relationships. It is depicted using the Harel-Koren Fast Multiscale graph. “‘Raw’ visualizations like this require refinement to display useful insight” (Hansen et al, 94) While this initial visualization may not tell us much about the participants and discussants or the nature of the conversations of #DHapi, it does immediately suggest a tight-knit group or community, perhaps a collection of subgroups, as well as several sets of disconnected individuals. Ways to refine this raw visualization would be to use the “Find Cluster” function of NodeXL in order to identify “pockets of densely connected vertices that are only sparsely connected to other pockets.” (Hansen et al, 93) Given the density of the vertices in this visualization, clustering would be extremely useful. Another refinement to this raw visualization that would make it more useful would be to use the new NodeXL feature, Group Layout, which places clusters in its own treemaped region. (Smith)
Given the density of this network and the 3077 edges, it would be helpful to perhaps to create a different graph that shows the follower/following relationships separately from the conversations (mentions/replied to).
#DHapi - Betweeness Centrality
This visualization displayed as a grid is designed to identify which individuals are most central to the community and/or conversation using the hashtag #DHapi based on measures of centrality and number of Twitter followers.
The size of the vertices indicates the betweeness centrality score. The larger the image, the higher the betweeness centrality score. The top most between contributors to #DHapi were: @james3neal, @umd_mith, @digitalhumanist, @zotero, @hastac, @dancohen, @wonderchook, and @georgeonline. An insight based on betweeness here is that the “connectors” and “influencers” for the #DHapi event come from the most obvious examples, to more nuanced and subtle ones. The obvious: connection and influence by @umd_mith, @digitalhumanist, @james3neal - the event was sponsored by MITH (@umd_mith); the organizer was David Lester (@digitalhumanist); and James Neal (@james3neal) assisted in organizing the event. Slightly obvious but more of a nuance: Dan Cohen, not a participant and not in attendance at the conference (albeit he is a leader in the digital humanities community with a large number of Twitter followers), nevertheless was able to bring awareness to #DHapi with this subtle tweet (note how Cohen characterizes using the #DHapi hashtag “to watch” the event):
Where an image (square) of a vertex is displayed it identifies individuals/organizations with more than 500 followers on Twitter. Individuals/organizations with less than 500 followers are displayed as discs without an image.
Given the emphasis placed on the size, shape, and quality of containing an image, those vertices represented by images that are the largest represent the most important event participants and discussants.
The edges in this visualization indicate mention relationships where vertices mention #DHapi and each other in the same tweets. The Twitter practice of retweeting is most prominently displayed here. For example, the image (extreme right) of @micahvandegrift - MLS student at Florida State University, is connected to @GeorgeOnLine - assistant professor of English at the University of South Carolina Upstate, UMD alumni and former member of the MITH staff, by virtue of the fact that Micah tweeted a retweet of a modified tweet by @miriamkp - a Mellon postdoc at Emory University library and conference participant, of a tweet by George. This simple illustration demonstrates the complexity of social networking and influence. While Miriam does not appear as a square image in the visualization (because she has <500 followers), two top betweeness participants/discussants (Micah and George) are connected through her. Furthermore the message about a shared Google Document regarding crowd-sourced transcriptions reached all of the people following George, Miriam, and Micah as well as anyone who might have been following the #DHapi hashtag on Twitter. This microcosm and small example illuminates some of the dark spaces and areas of the earlier raw visualization. Think of what might be gained from further exploration of Twitter follower/following/mention/replied to relationships via edges in NodeXL.
#DHapi - Circle
This circular visualization is of the “top” 25 participants/discussants of #DHapi based on number of followers. Image size is determined by the number of followers. A limit was placed depicting only those individuals/organizations with more than 1,200 followers. The circle was meant to represent the self reflexive nature of the digital humanities community. The edges represent all the possible relationships. An insight is that there exists a well connected and fairly commonly shared bond amongst #DHapi participants/discussants with more than 1,200 followers. Each of the individuals/organizations in the circle represent active, vocal and engaged participants in the digital humanities communities. The circle is instructive in that it contains a well represented sample from various aspects of the digital humanities community including organizers of the #DHapi conference, #DHapi conference keynote speakers, organizers of other digital humanities conferences, digital humanities software platforms, digital libraries, leaders of other digital humanities centers, leaders of MITH, and members of the MITH community.
The most salient aspect of this visualization is the presence of the figure who exists outside of the circle with no edges connecting her to the core group. She exists because of her Twitter number of followers and the fact that she mentioned #DHapi in a retweet of @JannaQ. She is Michelle Rafter @michellerafter, a journalist from Portland, Oregon. This retweet was Michelle’s only instance of mentioning #DHapi.
The presence of Michelle Rafter as an outlier of the core digital humanities community is worthy of note of how a Twitter hashtag has the capability of spreading far and wide outside of its primary or target audience even for such a specialized and esoteric an event as digital humanities and APIs.
This final image is a screenshot from Twitter that reflects some of the reality that we hoped this Social Network Analysis would reveal. The tweet is from Mashery, a business that is a leading provider of API management services who sent two representatives to cover the #DHapi workshop and hosted a happy hour on the first night of the event. In the tweet you see the indication of “cc” to “copy” two of the top betweeness centrality leaders @james3neal and @GeorgeOnline as well as the use of the #DHapi hashtag. As James Neal will testify #DHapi was indeed “a great time” and was a vivid portrayal of a community of practice using the nuances of Twitter technology, among others and the hashtag to document, communicate, and share information about its community, its practice and its culture.
Digital Humanities Center Funded at Emory | Emory University | Atlanta, GA http://t.co/FhAsTAj
Listening to almost exclusively Branford Marsalis since attending the concert on last Saturday. Good to have him back in my camp.
About 2 and a half hours of pure bliss for me tonight - @bmarsalis and @T_Blanchard performances at @kencen - THANK YOU!