I have been pondering knowledge hierarchies for a number of years. I originally drafted these slides one night at home while I was working as Knowledge Architect at Intuit. They never went anywhere, but I still think there are some interesting ideas, and I'd love to get your feedback.
Cheers!
Christine

I logged into my Yahoo! account earlier and was presented with the splash page, including the "trending topics" widget in the upper right. This is basically a ranking of the top searches in the last 'x' hours, presumably a day as is the case with Yahoo! Buzz. One of the rising searches was "stent," the #1 spot was "Bill Clinton." It's not an unreasonable leap that folks were wondering why President Clinton was in the hospital today getting stents - what the heck is a stent?

On any site, Yahoo or Twitter or a news provider that provides a tag cloud or "trending topics" of some kind, there is distortion in the visualization of what's hot. Noise in the stream. Concepts actually related by association, lemmatization, typographic errors and such can take up many spots on a trending topics list. If in doubt, check out any number of Wordles to see the same concepts represented singly, plurally, with misspellings and gerunds and more. This can make it difficult to identify what the masses are really interested in.

One way to remove the clutter is to link the data. Semantically align the concepts and their syntactic variations so that only the preferred term appears in the trending topics list, while the alternate and closely related terms appear only on closer examination of that topic. This allows for a separation of the wheat from the chaff, leaving analysts and curiosity seekers a clearer view of what stories are of interest, what memes are growing. Highly focused sites would need to determine the best settings for relationship proximity to serve up the "right" granularity of topics to their audience.

I believe this will improve analysis on many fronts - SEO, media coverage, data/content acquisition and analysis prioritization, brand analysis and more. Clarity on the front line, with deeper analysis and linking at a secondary level will help focus and refine results, and most importantly make better use of people's time and attention.

I am looking forward to speaking at Enterprise Data World next month in San Francisco. My great friend and colleague Kevin Lynch and I will be updating our well-received "Six Weeks to the Semantic Web" presentation from a few years back. Hope you can join us!