What is Homeland Security? Citation Networking and the Mapping of a New Discipline
Homeland Security is a new concept in the academic world. Even as a federal agency it’s relatively new, having been created 12 years ago in response to the September 11 terrorist attacks. Its formation from the consolidation of 22 other agencies meant that it immediately inherited numerous non-homeland security responsibilities, and the process of sorting out how everything should work is ongoing.
In this rapidly evolving environment, academic institutions have had the difficult task of determining exactly what is entailed in “homeland security” as a discipline. Training a new generation of professionals with the skills necessary to successfully implement (and create) policy is critical, as the DHS workforce is both older than the national worker’s average and more likely to retire earlier. Deciding what literature and coursework should be emphasized in educating the next generation (both in undergraduate and graduate programs) is a challenge that is complicated by the newness of the discipline as well as the changing nature of the department and the work.
I believe one tool that may be useful in understanding what homeland security practitioners find valuable (both in terms of subject matter and methodology) is citation network analysis. An offshoot of social network analysis, it looks at which authors and documents cite one another and then uses various mathematical techniques to visualize relationships between them. Similar tools have been useful for decades in link analysis and social network analysis, but more powerful computers and new algorithms have greatly expanded the utility of this approach as far more connections can be processed than ever before (often millions at a time). This has led to a dramatic increase in the use of citation network analysis, as documented in Figure 1 at left.
Citation networking can be used at varying levels of abstraction. “Real” relationships can be identified by mapping which scholars have cooperated in co-authoring papers, or by showing what papers have been cited (and presumably been read and digested) by specific authors. Networks that display these relationships are generally going to be quite sparse, and are likely to miss many quite similar academic works in drawing the connections. By moving into co-citation (linking works that were cited by the same authors) one takes a big step towards network density and, perhaps most importantly, creates new (non-existent in the real world) connections that provide researchers with mapped theoretical links between authors and papers that might otherwise never be noticed. This, of course, has its drawbacks as well in that some of those connections are likely to be almost random in nature, indicating relationships that don’t really exist. If one reads too much into connections that may be insignificant, or assumes that connections imply stronger relationships than they actually do, there is a danger of overestimating the importance of documents that have broad exposure but lesser relevance. (see figure 2 below)
As an emerging discipline, and especially as a new subject area that tends to overlap heavily with other existing disciplines (e.g. law enforcement, national security, emergency response, political science), someone wanting to outline the research priorities in homeland security would be well served to examine the homeland security academic literature and examine the patterns that emerge. If there are few citations in common between the papers, one might decide that the discipline is still developing, at best, or too dispersed to even qualify as a single discipline at worst. But if clear clusters materialize out of the data, it should be a relatively simple task to identify what subjects, papers, authors and research tools are most important to the literature under examination and draw boundaries around just what “homeland security” is, at least in the existing literature. This has been done by a number of others for different disciplines; for a particularly interesting example, see Scott Weingart’s “Networks Demystified” that uses these tools to define the literatures of philosophy of science and history of science.
I recently decided to undertake such a project myself to see if any interesting patterns might be identified in the homeland security literature. Unfortunately I hit the first roadblock almost immediately: the Web of Knowledge, which indexes citations from academic journals and makes this type of research possible (without countless hours manually entering dozens of citations each from thousands of articles) only indexes one journal related to homeland security, and it is focused largely on emergency management. I continued anyway by searching all the other journals that have been indexed for the keyword “homeland security” and came up with about 2500 articles, which between them listed over 100,000 citations. I must note here that many of these articles may have mentioned homeland security only in passing; there is no way to know without reading them. But hopefully most were relevant, in some regard, and this type of analysis depends on irrelevant relationships being overwhelmed by the thousands of relevant ones. In any case, the breadth of topics covered by these articles was incredible. They included biology, chemistry, and political science articles that somehow related to homeland security.
I entered the citations of those articles into a directed network table and ran it through some various networking software tools. Most algorithms did not yield illustrative results, but after a few iterations I came up with some clustering that recognizes 12 different types of relationships, most highly interconnected with others. (See graphic below.) Unfortunately, simply by looking at these I have not yet been able to discern what these clusters mean. This may be due to a high number of irrelevant relationships, or the sparseness of articles on homeland security in other hard or soft science journals leading to weak relationships being amplified by the networking software.
My final conclusion is that not enough data currently exists (more accurately, that the most relevant existing data has not yet been indexed—and such indexing, particularly for co-citation networks, would be incredibly time-consuming) to offer a confident assessment of whether the emerging homeland security discipline is dense enough to be considered an academic discipline, and how its various facets interact. If the Web of Science were to begin indexing homeland security journals, or if a researcher had the time and expertise to be able to better identify articles relevant to the homeland security project and then include those in a co-citation network, I believe that much insight could be gained as to whether homeland security might soon be identified as an independent academic discipline of its own.
For more by this author, consider checking out the following:
Yep, the CIA Really Did Train Terrorists in Florida
Security, Liberty and Architecture: Creating Safe — and Safe-Feeling — Public Spaces
