ALA TechSource Logo
 
curve Home spacer Publications spacer Subscribe spacer Blog spacer About  
    

Continuing the Conversation: New Models of Metadata

Submitted by Daniel A. Freeman on May 18, 2011 - 2:50pm

workshops

We just wrapped up the second session of the ALA TechSource Workshop Using RDA: Moving into the Metadata Future with Chris Oliver, Karen Coyle and Diane Hillmann. Karen Coyle led today’s session, New Directions in Metadata. Here are some questions that came up based on what we discussed today, as well as some resources from the event. Whether you attended or not, feel free to chime in via the comments area with questions or comments--Chris, Karen and Diane will be part of the discussion as well!

  • If you have a Web of data, do you still have bibliographic records?
  • How would we deal with copyright if we make our data available in this way, especially with the recent restrictions that OCLC has tried to place on their records?
  • Does the long history of library use of controlled vocabularies give us a leg up on moving to linked data? Or is it still a problem in making them machine-readable?
  • Do you think the trend toward goverment data could resonate into the library world. Would LC records be considered government data?
  • Is there any limitation with links only going in one direction? Would you need to create two links to express a relationship--one for the relationship and one for the inverse relationship?
  • Is there a danger of creating false or weak relationships between things?

You can also chime in on Twitter. The hashtag is #libdata

Karen’s page of links: http://kcoyle.net/presentations/links.html

The preliminary readings for this workshop were:

Karen Coyle: Understanding the Semantic Web: Bibliographic Data and Metadata, Chapters 1 and 2 http://alatechsource.metapress.com/content/g212v1783607/?p=b4700bc9fec34b12a3f42a94a9fd9d4f&pi=0

Diane Hillmann, Karen Coyle, Jon Phipps and Gordon Dunsire: RDA Vocabularies: Process, Outcome, Use http://dlib.org/dlib/january10/hillmann/01hillmann.html

(Webcast) Barbara Tillett: What RDA Is and Isn't http://www.loc.gov/bibliographic-future/rda/trainthetrainer.html

RDA Prospectus: http://www.rda-jsc.org/rdaprospectus.html

(Presentation with Slides and Notes) Tom Delsey: Moving Cataloguing into the 21st Century. http://tsig.wikispaces.com/Pre-conference+2010

RDA Scope and Structure http://www.rda-jsc.org/docs/5rda-scoperev4.pdf

Bowen, Jennifer and David Lindahl. RDA and the eXtensible Catalog. http://hdl.handle.net/1802/14588


Karen’s Slides:
New Models of Metadata


Comments (5)

Question: "Is there a danger

Question: "Is there a danger of creating false or weak relationships between things?"

In the world we're talking about, 'Anyone can say anything about any topic', so the notion of point of view is very important. This is something we're not used to in LibraryLand--we have the idea that catalogers should be objective when creating data, not expressing a particular point of view. The critical need is to know 'who says?' when evaluating statements, and both the W3C and DCMI are looking into how to ensure that linked data statements come with provenance to answer that question (among others). This is a much more 'open world' way to look at data, one that does seem a bit dangerous for those of us from a more closed environment. One of the skills we'll need then, in this new world, is the ability to evaluate this information, and to choose what works for us and our applications.

Question: Is there any

Question: Is there any limitation with links only going in one direction? Would you need to create two links to express a relationship--one for the relationship and one for the inverse relationship?

One could create relationships going both ways and identify them as reciprocal, which would allow them to be used for inferencing. This would also allow applications to create these for you (the Open Metadata Registry does this with 'broader' and 'narrower' relationships in the SKOS vocabularies). So there's not just one way to do it, but it's definitely one of the things we should be taking advantage of when we think about RDA implementation!

Question: "Does the long

Question: "Does the long history of library use of controlled vocabularies give us a leg up on moving to linked data? Or is it still a problem in making them machine-readable?"

Clearly our experience with community created and maintained vocabularies and authority files SHOULD give us a leg up on moving to linked data, and I believe that ultimately we will see that benefit becoming real. At the moment, we're largely relying on the current managers of those assets to make that leap of faith and technology, but the process has been slower than many of us would like. I'll talk a bit more about these issues next week in the third (and final) portion of the webcast.

"If you have a Web of data,

"If you have a Web of data, do you still have bibliographic records?"

You still have bibliographic descriptions. In other words, you can still gather, display, offload, or share a particular set of data that you consider to be a single description. There could be a standard set of fields that the library community treats as a "record." At the same time, if someone has an application that only uses, say, the date of publication and the size of the book, that is perfectly legitimate.

In the data world today we (and just about everyone else creating data) pre-determine the contents of a record. Because of the pre-determination, we all have to use exactly the same record. In the linked data world what makes up a logical "set" is up to the application that wants to use the data. In an environment where it is easy for applications to draw on data from a variety of sources, libraries can pull their bibliographic descriptions from the same pool used by authors creating citations in a document, or reading groups making up their reading lists. Serious academic use of bibliographic data can ignore the cover art and promotional materials that will be vital for online bookstores.

It's not that you can't do this today, but it is difficult because the different bits of data are in separate databases and don't use the same record formats. With a large pool of data that is intercompatible and homogeneous in its data format, these different "views" are a natural outcome.

I have some links that

I have some links that respond to the question about OCLC and its record use policy. At the recent OCLC members council meeting there was a session on linked data including the issue of opening the data for linking on the web. In Karen Calhoun's remarks, she stated that OCLC is considering opening the data for use following the Open Data Commons license ODC-BY, which allows re-use of the data with attribution (similar to a Creative Commons CC-BY license, if you are familiar with that).