The choice of an RDF schema for library data is strongly influenced by the audience for the data, which is another way of saying the use case.  Library data are intended for various audiences and three in particular stand out:  libraries cataloging in RDA, cultural heritage organizations in general, and library users.  Here are thoughts about these audiences and the schemas best suited for them.

Libraries Cataloging in RDA

For more than a century, libraries have shared the expense of cataloging information resources by sharing metadata.  This has evolved from the LC printed card program to the online bibliographic utilities of today.  For cost reasons, sharing of metadata will continue to be a crucial part of the library environment in a linked data world.

RDA is the current cataloging standard in the majority of U.S. libraries and in many other libraries around the world.  Opinions vary on the suitability of RDA and what direction it should take in the future.  But the fact remains that RDA is the standard and will remain so for some time.  It follows that libraries need to be able to exchange cataloging metadata in RDA among themselves and with their bib utilities and other partners.

This strongly implies that the RDF schema used for exchange in this use case should fully support RDA.  In other words, it seems a substantial waste to invest time making the distinctions provided by RDA but then to discard them because the RDF encoding does not support them.  Viewed in this light, the best current schema to exchange data among libraries cataloging in RDA is RDA/RDF.  Of course, RDA/RDF is not entirely ready for this use:  for example, properties of the Item class need to be developed to handle holdings information.

Cultural Heritage Organizations

Cultural heritage organizations (libraries, museums, etc.) collect and curate many of the same types of materials, but have different approaches to describing them.  In other words, they do not use the same descriptive standards but do have a need to share information with each other.  BIBFRAME has staked out this middle ground:  “ [it] balances the needs of those recording detailed bibliographic description, the needs of those describing other cultural materials, and those who do not require such a detailed level of description“ [1].  It is designed broadly to support a number of descriptive standards and avoids supporting any single standard in depth:  “it generally aims to be independent of any particular set of cataloging rules” [2].  BIBFRAME is well suited for exchange among cultural heritage organizations.

Library Users

Most users will not consume library linked data directly, but through search services and applications provided by libraries and their vendors, and by third parties.

Services and applications developed by libraries and their vendors will be designed by many different institutions and people, and it is hard to say what the best schema is to provide the underlying architecture for such services and applications.  Since this development is under the control of libraries, they could potentially use BIBFRAME, Schema.org, RDA/RDF, or something different.  I would hazard a guess that all of the above will be chosen by one institution or another.

Third-party (non-library) search services and applications are another matter.  At the current time, Schema.org is the frontrunner:  it has strong advantages of simplicity and support by major companies in the industry.  It is clear that RDA/RDF will never be accepted:  it is too complex.  The jury is out on BIBFRAME, but in my view its prospects are not good.  BIBFRAME lacks industry support, is relatively complex, and is late in getting started compared to Schema.org.  The fact that BIBFRAME comes out of the library community is also not in its favor.

Conclusion

No single schema is best for all of the audiences for library data.  The good news is that, in a linked data world, we do not need to use a single schema.  We can use multiple schemas, employing them where they are best suited.  As long as library data are maintained in a sufficiently fine-grained way, they can be converted to whatever format is required for the purpose at hand.

[1] http://www.loc.gov/bibframe/faqs/#q02

[2] http://www.loc.gov/bibframe/faqs/#q04