Areas for Discussion
Content vs Carrier
Not currently differentiated in schema. Discussions already in progress on public vocabs list.
Need to separate out the definition of content type and carrier type from the discussion about enumerated lists of types.
[kcoyle] 1) we should probably express this in terms of "things vs. concepts" to avoid library jargon.
2) carrier/thing will have a large overlap with "product" -- yet the product sub-scheme seems rather under developed in schema.org. Note, however, that "ISBN" is given in schema.org as an example of a product identifier.
For a use case for the library and publishing domains, see Use Cases#Use case: Content/carrier categories and RDA/ONIX Framework
TV & Radio proposed a PublicationEvent class which could for basis for a book/article.
For a use case for the library domain, see Use Cases#Use case: Publication event in library applications
schema:Book has 'isbn' - what about other identifiers and identifier attributes such as issue date, issued by, issued to, etc. See http://www.w3.org/community/schemabibex/wiki/Identifier.
How to 'simply' represent FRBR-ish relationships between creative works in a way that the wider web would understand
[kcoyle] I'd vote for NOT representing FRBR, per se. There is another question, however, and that is of dealing with duplication -- the many instances of the same bibliographic data coming from many different libraries. Amazon manages to cluster "duplicates" (probably using ISBN) and offer them as a single set. Library data will have many items that do not have an ISBN; is there anything that can be done to make clustering of duplicates easier?
[graham bell] Amazon almost certainly does not use the ISBN to create it's clusters (since each product in the cluster has a separate ISBN, and a separate ASIN, Amazon's internal ID). It is almost certainly simply matching title and contributor name(s). Assuming error-free metadata, this is obviously fairly accurate, but subject to error in high-profile cases (eg Q&A = Slumdog Millionaire) because metadata submitted to Amazon does not often include former or alternative titles. The standards-based approach is to use ISTC (http://www.istc-international.org) to create clusters. Though generally termed a work identifier, in FRBR terms, ISTC is an expression ID since, for example, translation or adaptation create new expressions and invite assignment of new ISTCs. However ISTCs are not yet widely used.
For a use case for the library domain, see Use Cases#Use case: RDA relationship designators unconstrained by FRBR entities
Skos and Schema.org
This is a proposal to extend Schema.org with Skos classes and properties to provide an annotation framework to organizations who publish controlled vocacabulary, thesaurus, glossary, lexicon... (i.e. BnF with Rameau (http://data.bnf.fr/12273666) or MetadataRegistry for RDA concepts (http://metadataregistry.org/concept/show/id/706.html)). Use cases can be : facilite discovery and access to controlled vocabulary content to any one who needs to use concept definition and identification to annotate it's own content, b) facilitate "human" alignment between a concept and its various definition and relationships depending on the editor's point of view, c) hightlight the work of web site publishers who are kind enough to give a clear defition of the concepts they use in their web site, giving a domain lexicon or glossary. Find here last version of the proposal :[]