More on taxonomy and context

A reader asked for a deeper explanation of self-fulfilling taxonomies. Lets assume that we are a consulting firm that focuses on medical benefits. As part of a retainer, clients get access to a membership-driven website that maintains a collection of white papers, medical journal articles, and various industry driven content. Four levels of navigation are created based on a well established medical index. Each level of the hierarchy possesses labels and medical definitions. In most cases, web designers call the project done.

Now, we will also have a keyword search that allows the user to search through the article index for specific searches. To support this, each piece of content uploaded will require 10 keywords to be tagged to the article. Where things get interesting however is that we will also associate the keywords to the medical article node (or branch) under which the article is categorized. In the beginning of this website, life will be pretty mundane. However, consider the site after 500 articles have been uploaded. If we take a look at the taxonomy with the associated keywords, and then rank the keywords of each node based on frequency of use, several things will begin to happen.
1.) You will have at the very top, common keyword descriptors which probably will match the existing labels and medical definitions.
2.) As you get into the second tier of keywords, you now have a rich vocabulary describing various aspects of the topic (node) which can be searched against as well.
3.) When you examine the top 30% of the keywords used, you will also be able to identify current trends relative to the subject matter.

BUT WAIT, we're not done yet.
The great challenge in on-line content (as well database rentals, etc.) is stickiness: getting the individual to explore related content. Now take the rich indexes you have built, pull them apart, and find the keyword relationships laterally across the index. Let's invent a word for this example. "Progenesisatrophy" (That was fun!). Progenesisatrophy is associated to 5 different nodes in your navigation. Now, this is NOT about associating the articles (i.e. Amazon, other shoppers bought X, Y & Z). This is about associating topics to one another. Once you have identified the patterns, you can;
1.) Create a hard navigational structure based on the findings,
2.) Provide the category associations at the content level, and
3.) Include the related taxonomy inside the search results.

Coming next: Context.
Your homework assignment. Travel to www.computerworld.com and explore their navigation and then I will tell you why.