LC Bibliographic Framework Transition: Eric Miller's update forum

The Library of Congress recently posted a webcast: Library of Congress New Bibliographic Framework Initiative: Update Forum with Eric Miller. This is LC's initiative to develop a replacement for the MARC format.

The new bibliographic framework transition initiative will be focused on the Web environment, Linked Data principles and mechanisms, and the Resource Description Framework (RDF) as a basic data model. Zepheira, led by Eric Miller, has been active in the development of Semantic Web and library standards as well as open source tools to support linked data technologies and library applications. Zepheira will provide both a model (or models) to serve as a starting point for discussion and an analysis of related initiatives. Miller spoke to Library of Congress staff with an overview of the project and a report of progress to date.

In this webcast, Eric Miller describes the transition away from MARC as an evolution, not a revolution. It's a very interesting talk on how Zepheira is developing a linked data model that "deconstructs" and repurposes the valuable entities, subjects, etc. that are already contained in MARC records.

My persistent question, which I haven't seen an answer to yet is: Why at such an early stage of data modeling are we narrowing in on Linked Data? Shouldn't we be looking at a variety of models for a MARC replacement?

Comments

[Irvin]

It _was_ an interesting talk. But there seemed to be subtle hints that there is another agenda at play, namely, that RDA and the FRBR model it's based on are seen as just too complicated and this is a way of starting over with something simpler (and cheaper) to implement. Eg the referencce to FRBR as a previous (failed?) attempt to model bibliographic data. Or am I reading too much into it?

[Mark R.]

FRBR is too complicated for most so he might be on to something if that is the case. The idea of adopting a data model to structure bibliographic metadata makes perfect sense, but RDA does not go far enough as there are no definition of fields and a field may include multiple pieces of information, e.g. extent of work.

Aside from linked data and maybe an XML model, what alternative models are there? The promise of linked data to reduce the burden on smaller libraries that cannot dedicate 100% of the time to original cataloguing and maintaining authority records is a welcome development, IMO. That being said, is RDF the only way to implement linked data? How will vendors of ILS support this change and how fast?

[Christine Schwartz]

Irvin,

I took a look back at the webcast's transcript where RDA was discussed during Q & A. Eric's answer to the RDA question seemed to suggest a movement away from RDA toward a simpler model that would allow interaction with other communities outside of libraries. Curious reply since the US RDA test called for a new format to replace MARC in order to realize the potential of RDA!

[Christine Schwartz]

Mark,

Thanks for the comment. As far as defined fields for RDA, you might want to check out the RDA element sets and value vocabularies: http://rdvocab.info/

[David M]

Linked Data is the only way forward to an integrated web of data. Not just between libraries, or a library to a search engine, but total integration across all sorts of media, databases and applications. It will become the standard like HTML has for linking documents.

I've worked on major data & systems integration projects for 20 years, in architecture, design and coding (higher education, investment banking, telecoms & internet co). The same problems get hit, over and over. Linked Data is the way to break data out of the fragile, domain specific information silos and constant format churn.

It's about going beyond fixed data structures, and getting to the core - the data and context (using proven technologies like URLs). The shared vocabularies provide the context which allow a vast realm of new possibilities for information processing and access.

It makes data a 'part of the web', instead of just getting web interfaces or APIs to systems 'on' the web.

Why does library data need to be part of the web? Partly because it drives access and usage which are the real value measures of collections. Today, we're only seeing the first tiny starts of the potential of Linked Data, but it's already delivering incredible new capabilities as demonstrated by projects like IBM Watson.

Libraries and librarians are in a unique position to be at the forefront of the next great wave of Internet advance. Their skills as natural ontologists, information sleuths and master curators are going to be more valuable than ever.

[Christine Schwartz]

Hi David,

Thanks for your post. I found it very helpful. I agree with you that linked data will be an important part of getting library data on the web. But I also think that XML metadata (which is already used extensively in libraries) has its place for hierarchical, complex bibliographic description of library resources.