andimou / bibliotheekgegevens Goto Github PK

We still need to decide on a prefix for the URLs for every resource, do we use existing URLs that may point to existing data (eg: We could use Bibnet to access information about a book using its ISBN number), or are we just going to use a single prefix for every resource that potentially does not actually point to anywhere, but keeps everything within the same "namespace"?

To help figure this out, I made a list of the different resources that are found in the library data and their unique identifiers that could potentially be used:

Book Description/Record/Item

Describes the "abstract" concept of an item that can be borrowed from the library (so mostly books, but also CDs, DVDs, etc...).

• BBnr Unique identifier used in the database internally (I think?)
• ISBN Probably a better choice, because its used universally?
• ISSN Describes a collection of items, not so useful for individual items, but might be worth considering for other purposes.

Conclusion: Use ISBN, pay attention to ISBN-10 vs ISBN-13.

Physical books/Sample/Item Copy

Describes a physical copy of something that can be borrowed from the library (eg: a softcover version of a book)

• Barcode The barcode that's on the book

Person/Borrower

Describes a set of anonymised people who fit a certain profile that borrowed something from the library.

• code A hashcode of some kind that (should) uniquely describe people who fit in this profile

Borrowing

Describes the fact that some Borrower borrowed an ItemCopy from the library.
I'm not sure how this will be incorporated into linked data, but it can be uniquely defined by the following identifiers:

• id Just a numeric id we assign ourselves
• (barcode, borrower, from_date, loan_period) The 4-tuple (barcode of the physical copy, id of the person who borrowed it, start date of the borrowing, number of days the item can be loaned) should uniquely define a borrowing.

Conclusion: Prefer the latter, unique identifiers are not very descriptive. Also try to keep the URIs as short as possible, so if you can drop loan_period, drop it.

Reservation

Describes the fact that some Borrower reserved an ItemCopy from the library.
Again, not too sure how to incorporate this into linked data.

• id Just a numeric id we assign ourselves
• (barcode, borrower) Seems to be the only information present for a Reservation - This will probably not be enough to uniquely identify it.

I've been looking into some existing vocabularies for book data and http://bibliontology.com/specification seems to be mentioned the most when it comes to library data. Unfortunately it does not completely cover all aspects of the data and lacks some things we might need (like the separation of book descriptions and physical books, a way to classify a resource as a CD/DVD).

For general information Dublin Core can be used.
I also noticed there was some price information about the books in the data, and since a library could be seen in a commercial context, should we also consider GoodRelations?