A mouthful’O buzzwords, eh? Bear with me…

Much has been said about tagging and tag clouds; it seems there’s no shortage of these simple ideas mostly because they provide new ways to get a glimpse of information that is traditionally difficult to understand. After all, tag clouds provide ways to make unanticipated connections with information that’s not always obvious. Indeed, you can glance at a tag cloud and get a sense of what the content details might be about - exceptions might include poor tagging practices or time centricity (e.g., JetStream)

At MyST we decided to add a “topic cloud” feature to our corporate blogsite services but with a clear focus on findability. We noticed that most tag clouds worked on the premise that given a specific tag - show the associated information items (i.e., blog posts, or whatever). But that’s quite limiting - some of the best discovery experiences I can recall relate to discoveries of things that are “nearby”. A GPS tool can take you to a very specific position, but as an avid hiker, it’s what’s across the valley that suddenly becomes an interesting place to traverse to.

Topic Cloud was designed as a combination topic search engine and tag cloud rolled into one. Our customer still believe the UI is clunky (and it is), but it attempts to push the envelope because it merges ideas like search, cloud scaling, and topic families into one user interface.

Cloud Scaling

We thought of this idea because a tag cloud is contextual - at any moment you might be at the top of the cloud, or somewhere inside the cloud. When inside the cloud, the context is less ambiguous and the cloud should be as well - hense the term “cloud scaling”. For example - here’s a link to the top of the cloud for our own blogsite.

http://blogsite.com/topics

It’s big and not as helpful as the subcloud for “google” (assuming we value information about Google).

http://blogsite.com/topics/google

All outward indications are that we’ve written stuff about Google, but let’s say we’re looking for specific references to Google ads, but we’re uncertain how our authors might have tagged those instances.

Tag Families

Understanding what’s also close by is an extremely important aspect of search because we all tend to describe things differently. This is where tag families come in handy - they expose other more discrete tags based on containership of other more ambiguous tags. In this regard, Topic Cloud is extremely powerful because it automatically determines containership relationships.

There’s a lot to this Topic Cloud idea and in coming posts I’ll expose how we achieve these capabilities as well as discuss some of the native elements concerning how we use XML to fabricate and render clouds.

While Topic Cloud has been in use for more than a year across many corporate blogsites, we really haven’t explored the possibilities and obvious improvements, so I’d love some feedback on this. I’ll give you a good dose of the programming architecture in a future article. In the meantime…

• Topic Cloud Technical Brief
• Topic Cloud FAQ