Above is the tag cloud I ended up with. My approach to tagging on CiteULike was designed to be intentionally organic and sub-optimal, in an attempt to replicate natural folksonomy development. I avoided using the “view all tags” button when tagging articles in order to reflect the ways in which multiple users can create redundancies based on different perceptions of which components matter most when creating tags.
The most visible artifacts of that are the network-networks and practice-practices pairs; the circumstances under which those arose were, in review, cases where one article or group of articles dealt specifically with the singular. theoretical concept (in terms of a specific framework, often) while another dealt with more general references and direct applications in practice. Thus, they are not perfectly identical in intent, but to an outside observer that is not evident.
“Knowledge” is by far the most prominent tag. This is due to the core principle I retained throughout the semester in my tagging efforts, that I would orient everything around that single concept, branching off into more specific expressions and variations according to the focus of a given article-“information” when that was being dealt with as an explicit distinction from knowledge, for example. Predictably, “management”, “organization”, and “transfer” were three other major tags, as each represents a core expression of knowledge in the literature.
A number of infrequently-used tags also exist. These generally emerged when an article had an overwhelming focus on a specific idea to the point where it was about that in particular; the concept of knowledge “stickiness” is a good example of this type of tag.
In hindsight, this approach was too generalized and natural. My efforts reflected the natural formation of a folksonomy in form but not in depth, as a single individual is ultimately incapable of processing the same quantity of material which goes into the construction of a folksonomy. It would likely take several times this many articles and a much more comprehensive tagging effort (a minimum of 6-7 tags per article) in order to accurately reflect the patterns of such a system.
It also lacked the strong structure of a traditional organized vocabulary, or even of a culled folksonomy, being bound only by a single guiding principle rather than a set of criteria for what types of tags should be employed, which variation of a given tag should be used, &c.
For a project of similar scope and time span, in future I would use the following design, or set of requirements. First, I would develop a core set of five or six essential tags, and then limit myself to using no more than four or five for any single article. This would ensure a strong basis for comparison on key features while preventing over-generalization based on rationalizations that almost every article touches upon almost every core issue in some regard. Likely this would also contain a measure of reform in term construction, phrasing these core terms as “Knowledge_(#Area of interest#)” to eliminate the use of the plain “Knowledge” tag as a meaningless label applied to every article.
Following that, I would enforce a mandate on myself that I include at least three area-specific tags for each work dealing with the more detailed and specialized content of the piece. These tags would be bound to apply to no more than half of the total volume of articles and no fewer than two total. Additionally, I would allow for up to three article-specific tags per work, to cover important concepts not discussed anywhere else in the literature as well as authors’ unique terms and phrases.
Although this is rather more structure than would normally be seen in a folksonomy, this sort of approach is not quite a true folksonomy, as it consists of only a single contributing individual. As an aside, it might be interesting to construct an overall class folksonomy drawing from each participant’s tags to create a larger impression of tagging patterns.