I just read a paper by Clay Shirky titled "Ontology is Overrated: Categories, Links, and Tags." It is a wonderfully clear paper explaining the fundamental difference between the creation/management of taxonomies and the organic growth of "folksonomies."
The paper motivated me to jot down thoughts that have been rattling in my head for the past few months. I have been trying to determine whether and to what extent aspects of a folksonomic approach could be used in the context of the hospital.
Consider two cases:
1. A hospital web site offering health information content, and
2. That hospital's intranet
Health Information Content:
Though I am excited by organic, individual tagging tools there are situations where introducing this can be problematic. Health information content, in particular, seems to be a strong candidate for more centralized, regulated classification and tagging for a few reasons:
(a) Allowing organic tagging could increase the probability of inaccurate tagging and, given the nature of the content, this could have truly harmful repercussions,
(b) Even though the existing classification / browsing schemes may not meet the widest range of user expectations, these are schemes that have been engrained in our consciousness (e.g., Body Location, disciplines). Altering these may cause more disruption than continuing with the imperfect classification.
I know this is not an either/or proposition. Both systems could co-exist with an appropriate level of oversight to prevent error and with a responsive adjustment of classifications.
My current opinion is to NOT introduce organic tagging and classification tools for health information content (other than, perhaps, offering users the ability to self-select "content bookmarks" for future recall). An analysis of search logs, user interviews and user observation can give us a pretty solid picture of possible tags and alternative terminology that we can fold into the system. "Folksonomy" occurs, but it is mediated.
Intranets:
It seems that internal systems can benefit from both methods as a result of two oh-so-entrenched properties in corporate organizations: hierarchies and the need for cost efficiencies.
Given that a company (and in particular, a hospital) may have fairly entrenched and solid hierarchies for their organization and documents, it would be best to faithfully present that as a browsing scheme. Employees are expected, as part of their jobs, to become familiar with this. Being nice to the user is not as pressing, it would seem, as in the consumer-facing side.
At the same time, these corporate hierarchies tend to bury related items in separate branches and are not sensitive to actual usage patterns. An employee may need access to multiple resources that are filed away in separate "locations." We could spend the time and resources to identify these inter-relationships and implement ways of making this explicit on the intranet.
However, given the compulsive need for "cost cutting and efficiency," creating a system that allows individual users to tag and associate as they see fit seems to decentralize and minimize the creation and management of such a system. This would allow for idiosyncratic associations and quick recall. Thus the move toward "personalized portals."
-----
Let me phrase the issue in a more philosophical way. In Shirky's paper, he notes that the different approaches to this issue tend to be informed by differing conceptions between:
1. Those who see the world as having clearly articulated entities independent of our ways of conceptualizing them, and
2. Those who see the nature of entities as resulting from our own process of sense-making.
See "Semantic Challenges to Realism" for more.
The table of elements may be ideal since its categorization scheme relies on the independent facts pertaining to the number of protons in the nucleus. However, categorizing massive amounts of photographs may not be served best by a rigid taxonomy.
Where does health information fall? Even if I believed that health information is also, to some degree, relative to the sense-making activities of individuals, I would still opt for a stable and regulated categorization of the content. We can monitor the terminological and categorization tendencies of our user base and adjust.
To do otherwise would simply be too risky a proposition -- risky in ways that are more tangible than most other content domains. The philosophical principles take a back seat when following them produces a significant risk of detrimental pragmatic consequences.
