The DeweyBrowser is unutterably cool. The transcendant decimality of it! Having everything in tens makes for a very clean layout. Art and Ross have done something similar for LC Classification. Taking it beyond HTTP, they’ve implemented Georgia Tech’s catalogue as a WebDAV repository. It puts the catalogue, in the form of a directory structure containing MODS records, on your desktop. Finally, Martha Yee (COinS: ) showed how to do some of this magic within the OPAC, exploiting existing MARC semantics to treat your records as if they’re FRBRized when they’re not.

It illustrates the value of metametadata, something I’ve been thinking about in relation to our digitization project lately. The attaching of a classification to an item is so much more useful if that classification is itself embedded in a predictable way in a systematic, navigable structure of other classifications. The more metadata you have about the classification, the more you can do with it. But the embedding must be active: the reader must be able to make use of it effortlessly.

Some people can get along fine without this kind of stuff. In a recent posting Stefano Mazzocchi, interested in the Semantic Web and related technologies, argues for letting the structure emerge from the data and cites Google’s PageRank and the new tagging systems as examples. That’s all very well if your data is intrinsically interesting; but when it consists of bibliographic records, it’s just too meager to generate any very useful structure. No one has a good system for relevance-ranked searching of MARC records. And there are too many records with not enough readers to tag them effectively.

What we do have are LCSH headings. These allow us to group records on the same subject, and also to link to relevant external resources (in the form of subject searches against other repositories that use LCSH). We can get more benefit out of them if we pay attention to the metametadata: if, for example, we note which ones are geographic headings and enhance them with GIS data (as I’ve discussed before

  • still working on that one!). This allows users to navigate our records in a new space, defined by a new set of metadata, but without our having to touch the individual records.

A similar requirement that has come up recently is the need to link from our records to a new historical encyclopedia being developed by a partner institution. We don’t want to look at thousands of records to determine which ones need to link where; instead, we’ll trust the LCSH headings we’ve already assigned, and link them appropriately. The individual record screen will gather these external links and display them appropriately.

In a nutshell, it’s easier to enhance metadata with metametadata than it is to enhance data with metadata. The effects for readers can be just as dramatic.

Bruce D’Arcus points to ways for us to integrate this sort of thing directly into users’ citation databases. His CiteProc package, which stores MODS records in an eXist database and uses XSLT to format them for presentation, can incorporate a web service that fetches and formats citations from a remote source. There’s no reason why such a service couldn’t enhance the records as well, using the subject headings as keys. Back in my pre-library-school days, this might have been enough to get me to understand LCSH properly.