The July Communications of the ACM has an article on Collaborative Structuring: Organizing Document Repositories Effectively and Efficiently (full text for members only) by Harris Wu and Michael Gordon. The authors have developed a mechanism to merge personal structures into a larger consensus structure (or a "collaborative structure") that helps everyone navigate through a collection of documents.
This should sound familiar to people who have been looking at social bookmarking / collaborative filtering within organizations. It takes the idea of folksonomy to another level, relying on the structures individuals create rather than just the tags they use.
I think this may have particular relevance within intranets, as one of the concerns about social bookmarks / tagging within the enterprise is that the audience is not large enough to get useful filtering. If they add structuring on top of the tagging, then there might be some more value to pull out of the individual practices.
The struggle, however, is the same struggle of merging any hierarchies (or even merging personal tag clouds): terms mean different things to different people, people use the same terms to mean different things, people want different relationships between concepts. Fortunately, there is the central item of the "document," so advanced algorithms could filter through some of this meaning jungle to provide useful structure to the internal repositories.