One of the challenges with folksonomy-like, user-generated skills and interest tags is that interests and skills that are very closely related don't get clumped together. So the shared-interest interest graph gets fragmented and it's harder for people to find and coordinated with each other.

You're probably already thinking about something like this, but just in case not, it would be great to have something like:

* Members list their interest and skills tags as they do today (and maybe with some new community organizing processes to encourage filling this out)

* AI synthesizes/aggregates these disparate tags into a higher-level ontology (note this aggregation process could be triggered automatically by the system after certain fixed periods of inactivity in new tag submissions, or could be triggered by group admins)

* The results are automatically displayed over the skill map (possibly with a zoom/un-zoom tool that expands and collapses the ontology's hierarchy). This might be set by the admin function or members might be allowed to do this real-time themselves when view the map.

Note: something like this would allow for other use cases too. For example, with something like this, it becomes much easier to display a kind of heat map or sorted directory of members that makes shared interests with particular members much more obvious.

Related: