The shape of EPUB to come, #1

Moderator: Our newest contributor is a familiar name, Hadrien Gardeur—co-founder and CTO of the Feedbooks site offering e-books for many devices and in many formats. Feedbooks supports standards for both e-books and the Semantic Web. A hearty welcome to you, Hadrien! – D.R.

EPUB as a standard—it’s actually three standards, OPS, OPF and OCF—is a real step forward for e-books.

I really like how flexible the OPF standard is: with proper fallbacks you can very easily support all sorts of devices, and extend the document. But there’s still room for improvements, and the best way to support EPUB is to discuss how things could be better. For example, how about improved means of linking to other books? Or adding new capabilities to travel guides displayed on a GPS-enabled device?

In the official document for OPS there’s a section named “Future Directions”:

“This specification is designed to take advantage of current practices while preparing for future developments. Although details of subsequent versions of this specification remain to be determined, it is the expectation of the Publication Structure Working Group that continued evolutionary development will occur. The themes driving the creation of version 2.0 of OPS are: standards compliance (e.g. full namespace support), accessibility support, support for any XML document typesupport for a wide-range of XML document vocabularies, enhanced navigation support, and improved content presentational fidelity.

Other themes deemed important for future versions include: more rigorous separation of content and presentation, greater accessibility, better support for international content, Reading Device-specific presentation control and/or Reading Device profiles, enhanced support for inter-Publication linking, layering and managing markups (e.g. inking, highlighting, notes) within Publications, application-specific markup (e.g. math, chemical), multiple reading orders, and support for active content (e.g. multimedia, scripting), all while maintaining alignment with relevant standards. Additionally, maintaining backward compatibility to this version of this specification ought to remain a high priority. Future directions can be tracked at http://www.idpf.org.”

For my first post, I will focus on how we can annotate markup with semantics, and why this is useful in a book through two examples:

1. linking to another EPUB document
2. a travel guide on a GPS-enabled device

Right after Adobe released Digital Editions 1.5 beta 1, I had a little chat with Peter Sorotkin from Adobe and the Digital Editions team over at MobileRead. You’ll notice that in every book on Feedbooks we link to additional resources: other books from the same author at the beginning of the book, recommendations at the end of the book. I wanted DE to handle these links properly and directly download and add these books to DE’s library instead of going through the browser.

Here’s what Peter told me:

“Yes, it occurred to us before that this would be a useful feature. Here is what I think:

I would not want to download a book if I already have it. Also, an inter-book link may point to a specific section of a target book. So I think we need more information than simply a book download link. Ideally, there should be a reference file inside epub container that would describe target book metadata, so you could match it in your library if it is there, download link(s), so that it can be downloaded if it is not, and a reference to a specific section of the book (if needed). A quotation from the target book would be nice as well, so that we can attempt navigation even if user has different edition by just searching for the text.”

That’s a very interesting point of view, and it’s similar to what IDPF defined in its future directions: “enhanced support for inter-Publication linking”.

But instead of a specific document that would strictly reference other publications, I really believe that we should use something much more powerful: RDF.

Aside from reference to another document, there’s thousands of ways we could process semantic information to create an enhanced experience for the end-user.

Now the real question is: should we add inline semantic markups (RDFa) or should we reference all this information in a RDF file inside the container ?

From a content producer perspective, I believe that it’s much easier to use RDFa, although it could be possible to pre-process the XML file, parse the RDF and create a separate file.

From a reading system perspective, parsing a separate file is easier, although it shouldn’t be too difficult to parse RDFa either (it’s not that much of a resource hog, compared to CSS and pagination).

From the end-user perspective, there are three ways you should be able to interact with these sets of metadata:

1. click on a link defined as a RDF resource
2. the reading system could highlight some elements, add an icon, and let the user select an action
3. at the end of the flow, or at the end of the book, a human-friendly list of the RDF resources

Let’s see through two examples how this could work in a book…

Inter-Publication Linking

There’s a few ways we could identify this type of resource:

1. using an identifier (ISBN for example)
2. with DublinCore (title, creator etc…)
3. using an URI

The best solution would be to use a mix of all three. Here’s a basic example:

<rdf:rdf xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#">
    xmlns:dc="http://purl.org/dc/elements/1.1/"
    xmlns:dcterms="http://purl.org/dc/terms/">   <rdf:description rdf:about="urn:ISBN:0143039717">
    <dc:creator>Jack London</dc:creator>
    <dc:title>The Iron Heel</dc:title>
    <dc:language>en</dc:language>
    <dc:date>1908</dc:date>
    <dcterms:hasformat>
      <rdf:description rdf:about="http://www.feedbooks.com/discover/epub/2381">
        <dcterms:format>application/epub+zip</dcterms:format>
      </rdf:description>
    </dcterms:hasformat>
  </rdf:description>
</rdf:rdf>

In this example, the subject is the ISBN of the book. We could therefore link a footnote about this book using urn:ISBN:0143039717. The reading system could check the metadata to see if the book is already available on my computer, download the book on Feedbooks or redirect me to an online shop.

It would be interesting too to redirect the user to a specific point in the book. I can’t express this sort of information using DublinCore, but another ontology could be used for this (although it’s very hard to reference to a specific point of the document: using the name of the flow is a very bad idea, DOM ID won’t work that well either, I think that Peter is right when he said that you should include a quotation. This solution would work across multiple editions of the book, although it wouldn’t work with multiple translations of a book).

A travel guide on a GPS-enabled device

Now let’s see another example of what we could do using RDF in EPUB. I’ve just downloaded a travel guide on my GPS-enabled smartphone. I’m visiting Paris, I’ve just spent the whole day visiting le Louvre, and I’d love to have something to eat/drink. Flipping through the pages of the travel guide (it’s in EPUB), I’ve just found a place named Angelina: they have the best hot chocolate in Paris. Hopefully, my file is tagged with GeoRSS, I can click on Angelina and select the right action:

• tell me how I can go to Angelina from my current position and how long it’ll take
• or display Angelina on the map

Conclusion

Let’s add something as powerful and flexible as RDF in the next generation of the EPUB standard. With the right implementation on the reading system, the potential is unlimited. Instead of using a new type of document strictly limited to inter-publication references, it would be very easy to explore the possibilities of RDF/RDFa.

Technorati Tags: IDPF,International Digital Publishing Forum