Friday, September 12, 2008

nsdl metadata registry

This afternoon a group of us had the opportunity to sit down with Jon Phipps, implementer of the NSDL Metadata Registry.

I knew that such a thing existed. I understand RDF. I know about SKOS. I hadn't really given a lot of thought as to how to best take advantage of it.

Today, I had one of those skies are opening and angels are singing from on high moments. RDF can be used to model relationships between concepts and potentially enforce them through schemas. This can obviously be applied to improve discoverability when a hierarchical taxonomy is employed. Then my LC colleague Clay Redding said that he was experimenting with multiple schemas and managing additional local alternative labels in addition to authoritative preferred labels. And then Jon and Ed Summers mentioned the potential for this tool to map across schemas. My a-ha moment was understanding the potential for formalized mappings across metadata schemas to improve discoverability within and across collections described with hetreogenous taxonomies and vocabularies.

I remember using Chenhall's Nomenclature in records for ethnographic objects where we recorded every level of the hierarchy in its own field -- It was madness. I remember when we were in the early days of the AAT, busily submitting new terms and building the hierarchies, our dream was searching for "case furniture" and getting results with bookcases, chests, desks, wardrobes, and every semantic child where "case furniture" never appeared in the record.

I remember some research at USC in the late 1990s about thesaurus-enabled searching. OCLC's Metadata Switch project has done some work in cross-schema mapping. I know this is very difficult to accomplish. Today was the first time I saw a tool that might make the conceptual mapping simpler. But not simple. This is a potentially massively overwhelming task if it can't be done programmatically to a large extent.

I'm coming late to the party, but now I'm really intrigued by what might be accomplished in this arena.

1 comment:

dchud said...

I started to mention this the other day, but to close that loop, maybe you'd be interested in taking a look at the UMLS and research into cross-vocabulary search with its Metathesaurus.