Use Case - Support multiple digital formats for books

[Natkeeran]

Islandora 7.x tightly couples the Content Model with Digital/Content Format. A book can be in many digital formats (example), including EPUB, PDF, plain text and other formats. Current release tightly couples the book with the paged content view and the archival viewer. For example, out of the box, it does not support a pdf book format (without generating the large image derivatives). Large Image derivatives can be problematic, because they are resource intensive to create, store and serve.

In the Book/Newspaper case, the content model should not be tied to the viewer/paged content. The content model should be abstract and support other digital formats such as pdf or text.

It would be good to describe what are the best practices in building/relating Content Models with Digital Formats with Metadata Profiles with Viewers/Displays (and workflow).

Related use case: https://github.com/Islandora-CLAW/CLAW/issues/688

[dannylamb]

Yes... wow... now I see those things have been conflated and we have an opportunity to detangle them. Fantastic use case @Natkeeran 👍 👍 👍

[ajs6f]

:+1: x 1000. Don't make my metadata depend on file formats!

[mjordan]

Totally. Another example of this binding in 7.x is that a PDF is not as much a "content model" in 7.x as it is a "MIME type model". A lot of different types of content can be represented as a PDF file - an image, a document, a book, a newspaper issue, etc.

[Natkeeran]

@mjordan

Yes, MIME type model is a good way to put it. Islandora Scholar (Citation, Thesis Content Models) uses pdfs, even though it does not use PDF content model.

End users currently create a collection containing multiple content models or a custom module to include multiple type of MIME types in a single collection. The difficulty is pdf metadata profile does not always work for a book.

[rosiel]

During today's CLAW call, we (folks on this thread) discussed using taxonomy terms to "tag" items with types, without having to create separate bundles. For a Bibframe example, a "Agent" bundle object could be a Family, Organization, Person.... That way - if Context is powerful enough - we could include "inheritance" as in, "If the object is tagged as Agent or any subclass/taxonomic-child of Agent, then do this..."

Bibframe makes a distinction between the intellectual Work being described, the Instance (form/edition) and Item (specific file copy). That's one way to isolate this - you might have a Book (Work), which can be available via Printed Edition(Instance), epub (Instance), PDF (Instance), or set-of-122-tiffs-with-structural-metadata-to-order-them (Instance). [Sidenote: The Bibframe AV Assessment[1] notes that while Instances (especially digital ones) are often made up of sub-parts like multiple files, or multiple bitstreams with different encodings, bibliographically that's just one "thing" and description of its various parts is better suited to preservation schema like PREMIS.]

So we could really step back and assign metadata profiles depending on the type of Work (monograph, manuscript, artwork, photograph, performance, ...)

[kstapelfeldt]

688