We are having a similar metadata problem at work. We are hoping to solve it by requiring users to categorize their content. We are providing them with a controlled vocabulary list and asking them to enter up to three subject codes. The goal is to have these categories drive a search engine and a browse by topic feature.

Of course, there is the problem of not being sure if they picked the correct subject codes. There doesn't seem to be any way to guarantee this without human intervention. Also developing a controlled vocabulary list that meets every person's needs is hard as hell.

One solution is to have global metadata, data that should be collected about every file(node) and local metadata, info that is specific to a section or content type. This allows us to fine tune our metadata without putting a huge burden on the user. Having global and local metadata also cuts down on the size of any controlled vocabulary list.

The sections of the Q&A section could be a good start for subject codes(global metadata). The goal of browsing by topic is a good one and it would compliment the Super Search nicely.

The work of the Dublin Core and the W3C (RDF) could provide some good ideas and blueprints for this topic.

Get Strong Together!!