Blog Post

DML & Scholars are at the heart of HASTAC

DML & Scholars are at the heart of HASTAC

Click for a large, legible version!

In my continuing effort to understand HASTAC through data, as part of the NSF EAGER grant research project, I turned my attention to exploring Groups on Groups are a venue for collaboration across the network, a way to share ideas and information, a place to discuss shared interests. Groups can be private -- open by invitation only -- but the vast majority are public, and open to any HASTAC member who might wish to join.

Groups vary significantly in terms of membership and activity. The largest group, Scholars, has several hundred members, and has hosts highly-commented blog posts and Scholars Forums on a frequent basis. At the other end of the spectrum, there are many Groups with just a few members, often working around a single project, with only a brief period of activity.

However, all of the groups, when analyzed collectively, can tell us something about the structure of Much like this (fantastically illustrative) post by Kieran Healy on "Using Metadata to Find Paul Revere," we can use network connections between Groups to identify those with a central role in the network.

This is done by building a persons-to-groups matrix; that is, a matrix with one row for each HASTAC member and one column for each HASTAC Group. For each member, a 1 in a given column indicates that they are a member of the associated Group, and a 0 indicates non-membership. This matrix is interesting in itself, but allows (by a simple matrix multiplication), the creation of a square persons-to-persons matrix and a square groups-to-groups matrix. Each cell in the P:P matrix indicates the number of Groups shared by the "row member" and the "column member," while each cell in the G:G matrix indicates the number of members shared by each Group (again, the Paul Revere analogue is a handy explication of this very idea).

These two square matrices are adjacency matrices, amenable to social network analysis and graphing. In this post, my focus is on the Group-to-Group network. The network analysis does not itself necessitate any assumptions about the meaning of ties between each group, nor the magnitude--number of shared members--of that tie. However, I often think of these networks as being characterized by individuals' unique interests, or perhaps homophily, which suggests that Groups with similar membership (i.e. strong ties) are relatively similar in content or purpose.

Even a simple analysis of the Group-to-Group network is illustrative. All of the common network centrality measures suggest the same conclusion: The heart of is DML and the Scholars. By eigenvector centrality, four of the six most-central Groups are Digital Media & Learning Competitions 1 through 4, and one is the DML-related Badges for Lifelong Learning Group. Scholar Groups are also prominent in the network, representing the 4th9th18th20th, and 26th most central nodes.

The graph visualization, included at the top of this post, or at a separate URL here, is even more interesting and informative. Vertex colors indicate Group types: DML competition cohorts are red, DML winning projects are purple, Scholar classes and scholar groups are dark and light green, and member-created groups are blue. A community-finding algorithm (or just casual observation) identifies three major clusters: HASTAC Scholars (right), Early DML Competitions (lower left), and Recent DML Competitions (upper left). Obviously, there are connections between nodes across these three clusters, but the essence of a network cluster is essentially a tight community of nodes connected with similar neighbors.

There are other interesting "bridge" Groups -- such as those in a small group between the Earlier and Recent DML Competition, and those between the Recent DML Competitions cluster and the Scholars cluster. Badges for Lifelong Learning and Collaborative Data, for example, are evidently of interest to individuals who inhabit both the Scholars and DML cores. One of the things that stands out most significantly is the extent to which there are ties between nodes all across the entire network: the Visualizing Venice Group is connected with the Access + Digital Literacy Group, for example, emphasizing the extent to which HASTAC members really do compose an interdisciplinary alliance.

What other patterns do you see? Does this visualization fit your own mental model of HASTAC? What questions does this raise?


This material is based upon work supported by the National Science Foundation under Grant Number 1243622. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.


No comments