Is taxonomy anti-intellectual?

The work of developing a taxonomy is no small task and by no means something to take lightly. This is a highly intellectual exercise that is of extreme importance in relation to findability of our information within our information stores. Yet many organizations, though not intentionally, tend not to address this aspect of information properly. In some cases, lack of flexibility is cited as being the reason.

If the perceived purpose of taxonomy is to provide structure to the unstructured, helping us gain and maintain control over our information, does it not then stand to reason that taxonomy creates an environment that is anti-intellectual? 

Does a structured classification framework with controlled vocabularies that limits users abilities to apply metadata descriptors to information go against the growing trends for use of folksonomy and the concept of collaborative tagging across an enterprise for the purpose increased flexibility in information sharing?

What do you think?

Bob Larrivee - AIIM

Comments

[Ted]

Great topic. One point of view depends on the flexibility and implementation of the taxonomy - i.e. how "collaboratively" can it actually grow and evolve. A community-driven taxonomy may be quite an intellectual exercise, and with proper governance never fully cross the threshold to a "folksonomy". Another point of view is the one about "if an intellectual is being intelligent, and no one's there to hear, read or understand him/her, are they really intellectual?" Controlled vocabularies may limit topic and descriptor variances, but vocabularies are necessary for communicating and providing evidence of "intellectualism" (vs. rumor-mongering).

[Steve]

Perhaps it depends on the environment. Here in South Africa when working with government you want to be as prescriptive as possible and therefore a controlled vocabulary approach works well, hinting the user at how he should tag items.
But in highly charged private enterprises this often creates great frustration.
Back to government, a key challenge is the ability to train the users to regularly re-assess and update the controlled vocabularies.

I am a bit of a newbie - what do you mean by folksonomy? Unstructured tagging of content with keywords?

[Simon Smith]

Hi Bob, apologies for the delay in commenting, I have been on leave.
As mentioned by Ted, this is a good topic for discussion. I feel that there is much talk on folkonomies as being the only way to go in terms of sharing and disseminating content to a wider audience. That said I believe a problem with this approach would be that it doesn’t always suit organisational needs and can cloud or clutter up the ability to find content.
I think that a more controlled taxonomy can actually enhance the "findablity" of content, Public Sector Office is a good example of this. That said and again in agreeance to Teds point, a user enhanced taxonomy, where properly regulated can and does aid in content discoverability.

In summary, I believe there will always be the need to apply more localised terms or descriptors to a taxonomy but always with control being exercised .

[Wayne Brooks]

Introduction of a taxonomy does not necessarily limit user application of metadata descriptors, as there is a school of thought that suggests that taxonomies by definition are inclusive of metadata, such that controlled vocabularies would be a findability enhancer as opposed to a findability inhibitor.

[Michael Elkins]

There are times when a more structured environment is called for and times when a bit of flexibility is required. Flickr is an example of the benefits of giving the user the ability to tag their content as they choose. The tagging is driven by the dynamics of the social network. On the other hand, social tagging for a large scale engineering project such as an oil platform could spell disaster and drive up project costs considerably.
Each industry has a language of its own which should be incorporated. Consider how the healthcare industry benefits from SNOMED.

[akshivalik]

Yes it is true that taxonomy limits the end user, however most subject matter experts can provide a useful baseline. In case of scientific content, the categorization process is well thought and debated. For example. we would not want anyone to change the taxonomy for animal kingdom.

To your point, most product classification has a subjective element to the classification process and every user may have a taxonomy that suits them best. This is where tagging (folksonomy/social-tagging) is useful and can be used to assist the official taxonomy.
On hp.com, I always have problems using their taxonomy, they divide printers in two groups for office-use and for home-use, we all know that some printers could fall in either category. Unfortunately, they don't cross-list products in both categories. In such a case tagging which provides a more mesh type navigation is a better approach. Amazon does a good job by providing both a taxonomy and allowing users to tag items as well.

Another soultion is to have multiple taxonomies, each catering to a specific need. At Dupont.com, all products are categorized in three taxonomies depending upon the usage.

[Karen Coghlan]

Organizations that devise a very high level taxonomy to describe their information assets according to their major business functions will improve the understanding of all employees. This not only enhances findability it also provides a handy orientation guide to new employees -- so that's what we do and why we create all that information.

Yes, this is an imposed structure that limits a personal view of the information landscape but shouldn't everyone working within an organization share a common view of the most basic business activities?

I agree with Wayne concerning the use of controlled vocabularies especially in complex work environments where it is fairly important to avoid misunderstandings -- like health information.

Keeping it light and flexible is quite a balancing act but well worth it in the end which is why it is always good to use all the findability tools available, time and resources permitting.

Information that is facing the public is another matter. Social tagging and faceted classifications might provide a little structure and a lot of flexibility. It is not easy to draw the line between helping and hindering but perhaps a bit of choice doesn't hurt?

[Shawn]

Now on to my second organization and second chance to develop a taxonomy, I'm thinking of taking a different approach...perhaps to make it a more intellectual approach but only time will tell... In place of the output of the taxonomy being primarily around the configuration of folders (which obviously includes things like classification, permission, metadata, etc.) I'm thinking to design this taxonomy as a series of interrelated metadata. Consider that a user will file a document 4 levels down a tree. The folders and subfolders may start as (1) Function (2) Subfunction (3) Year (4) ID No. The user filing in subfolder #4 has informally 'tagged' that document (or record) with the titles of the parent folders. Secondly the taxonomy may only make sense to one part of the organization, and thereby increasing training costs to have the other parts of the organization understand the overall structure and why it is used. Developing the taxonomy as a series of interrelating metadata (supported by a strong metadata model) may allow for us to display the structure in a variety of views which is most logical to our end users. Some may see subjects, some may see org chart, some may see function, some may see frequently used items. It can essentially allow for a dynamic folder structure view that will make the most sense to the corporation as a whole and to the end user as a single employee, thereby increasing overall user adoption.

[Jerry M. Withers]

Whatever you choose to call them, both a structured taxonomy and a 'folksonomy' approach have value to specific audiences depending on the knowledge domain, application, and the search or finder navigation tools available. A structured taxonomy provides a better approach to designing formal navigation aids, such as menus, indexes, site hierarchies, etc. A folksonomy is very useful for a domain that lacks formal structure, or is creating a classification structure incrementally. I have found a blend of both approaches to be useful for most knowledge management applications, despite the immaturity of the tools available for implementing both approaches at the same time. One of the critical success factors is the ability to adequately specify a topical index to content. Another is the ability to appropriately tune the results returned from the search engine, giving more weight to the structured keywords. It may be obvious to most of us, but another critical success factor is having "any" type of effective classification of content. Whichever approach is chosen, the most critical success factor is one or more knowledge managers with the skills needed and time available to adequately tweak these formal or informal structures to suit the domain and its users.