Talking Around The University of Maine at Augusta: A Twitter Mention Graph

Like many institutions of higher education these days, the University of Maine at Augusta communicates about its accomplishments and keeps track of the work of others using the social media service Twitter. In its communications, UMA traces the paths of the community that surrounds it.

Unlike the social media platform Facebook (oriented toward friend and family relationships) or Pinterest (devoted to the sharing of images), Twitter acts like a news clipping service of sorts. Limited to 140 characters of text, Twitter posts are like headlines in a newspaper, with links to web pages containing more information. Making headlines social, Twitter posts can mention other Twitter accounts that are relevant to the story. By tracking those mentions, we can find communities of posters who find one another’s work relevant.

To generate the social network graph you see below, I’ve searched through all Twitter posts made this year by the university’s official account, @UMAugusta, and identified all of the other Twitter accounts that @UMAugusta has mentioned. In a second step, I looked at the records of each of the Twitter accounts @UMAugusta mentioned and found out whether and how often they referred to one another. The result, formally speaking, is a level 1.5 ego network. In the graph below, Twitter accounts are indicated with labeled dots; in the parlance of social network analysis, these are called “nodes” or “vertices.” The larger a dot is in the graph, the more often it is mentioned by other Twitter accounts. Mentions between Twitter accounts are indicated with curved lines, which network analysts refer to variously as “lines,” “arcs,” “edges” or “ties.” The darker a line is, the more often mentioning occurred between two Twitter accounts.

Who Mentions Whom? A social network of mentions over Twitter surrounding @UMAugusta from January to October 2014

To highlight structure in the network of mentions surrounding @UMAugusta, I identified five clusters of Twitter accounts who mentioned one another especially often. These clusters are color-coded in the network graph above. Because the identification of clusters of conversants was driven by data, not by pre-conceived notions about which accounts might “naturally” be grouped together, it is curious to see how particular clusters focus on particular domains. Some patterns:

  • The dark green cluster in the lower-right of the graph consists strongly of offices and officers connected to student life and services at the University of Maine at Augusta.
  • The dark blue cluster in the upper-left of the graph is anchored around newspapers and newspaper reporters of central and southern Maine — the Portland Press Herald, the Kennebec Journal of Augusta and the Morning Sentinel of Waterville. These three newspapers are not simply tied by geography, but are also published under the aegis of the MaineToday Media company; @centralmesports is a joint outlet of the Kennebec Journal and Morning Sentinel. Other central Maine institutions — Colby College and the Holocaust & Human Rights Center — are also featured in this cluster.
  • The light green cluster in the lower-left of the graph features strong representation in the arts, with the 5 Rivers Arts Alliance, Harlow Gallery, photographer Jill Guthrie, and The Band Apollo included.
  • Immediate substantive commonalities in the red upper-right cluster, including my own account, the Maine State Library, the Maine Humanities Council and a edu-metrics website NerdScholar are elusive. We are tied to one another because of our mutual communications across disciplinary boundaries.
  • The light-blue cluster at the bottom of the graph is a remainder category, consisting mostly of Twitter accounts that UMA has mentioned but that do not mention other accounts often.
  • Finally, although these clusters identify groups of accounts that communicate more often internally, connections between clusters are frequent, indicating that most of the accounts mentioned by the University of Maine at Augusta are part of a broader community.

Data mining and visualization for this graph of the @UMAugusta network were carried out using free and open source NodeXL software.

Building Offline Community to study Online Community: the Social Media & Society Conference

Attending academic conferences can feel a bit like living in a retelling of Goldilocks and the Three Bears. A conference that’s too small can leave you feeling underfed. On the other hand, a conference that’s too large can be overwhelming, intimidating and even alienating. A conference on a highly particular subject may be quite useful if you select just the right one, but may be completely useless if you’re even slightly off the mark. The presentations at an overly general conference may lack those crucial connections that stimulate career-changing “aha!” insights. If you’ve been to enough conferences, you probably know what I mean.

How rare, and therefore how precious, is the conference that hits the Goldilocks sweet spot in between these distasteful extremes. The 2013 Social Media & Society International Conference was that conference for me. Gathering and connecting presentations on the causes, kinds and consequences of online social connection, #SMSociety13 managed to be more than simply the sum of its individual presentations. Researchers across diverse fields of social science, humanities, business and computer science shared distinctive approaches and concerns regarding the same substantive subject, which meant that we all had some basis for understanding but also had something to learn:

Topics of discussion at #SMSociety13, the 2013 Social Media and Society Conference

Attendance numbered in the sweetly moderate middle between a hundred and two hundred, providing a critical but collegial mass of thinkers who began conversations during one set of presentations and continued them across others. How do we bridge (or barricade) the quantitative-qualitative divide? How do we know who is “really” speaking in an online environment, and how do participants manage the online presentation of self? What are the ways in which online interaction leads to offline action? As we ran into one another again and again in various combinations, these questions carried over into the late night at a pub and over danishes in the morning, with an aggregate from far-flung places becoming a quirky community.

Photos from the 2013 Social Media and Society Conference at Dalhousie University in Halifax, Nova Scotia

The Social Media & Society International Conference meets again at Ryerson University in Toronto on September 27-28, 2014. Got a paper or panel in mind? Submit through this link: I’d love to see you there. Abstracts are due April 18. Poster proposals are due May 23.