« Going LOCAH: a Linked Data project for JISC | Main | SLOODLE gets further funding from the JISC »

July 16, 2010

Finding e-books - a discovery to delivery problem

Some of you will know that we recently ran a quick survey of academic e-book usage in the UK - I hope to be able to report on the findings here shortly. One of the things that we didn't ask about in the survey but that has come up anecdotally in our discussions with librarians is the ease (or not) with which it is possible to find out if a particular e-book title is available.

A typical scenario goes like this. "Lecturer adds an entry for a physical book to a course reading list. Librarian checks the list and wants to know if there is an e-book edition of the book, in order to offer alternatives to the students on that course". Problemo. Having briefly asked around, it seems (somewhat surprisingly?) that there is no easy solution to this problem.

If we assume that the librarian in question knows the ISBN of the physical book, what can be done to try and ease the situation? Note that in asking this question I'm conveniently ignoring the looming, and potentially rather massive, issue around "what the hell is an e-book anyway?" and "how are we going to assign identifiers to them once we've worked out what they are?" :-). For some discussion around this see Eric Hellman's recent piece, What IS an eBook, anyway?

But, let's ignore that for now... we know that OCLC's xISBN service allows us to navigate different editions of the same book (I'm desperately trying not to drop into FRBR-speak here). Taking a quick look at the API documentation for xISBN yesterday, I noticed that the metadata returned for each ISBN can include both the fact that something is a 'Book' and that it is 'Digital' (form == 'BA' && form == 'DA') - that sounds like the working definition of an e-book to me (at least for the time being) - as well as listing the ISBNs for all the other editions/formats of the same book. So I knocked together a quick demonstrator. The result is e-Book Finder and you are welcome to have a play. To get you started, here are a couple of examples:

Of course, because e-Book Finder is based on xISBN, which is in turn based on WorldCat, you can only use it to find e-books that are listed in the catalogues of WorldCat member libraries (but I'm assuming that is a big enough set of libraries that the coverage is pretty good). Perhaps more importantly, it also only represents the first stage of the problem. It allows you to 'discover' that an e-book exists - but it doesn't get the thing 'delivered' to you.

Wouldn't it be nice if e-Book Finder could also answer questions like, "is this e-book covered by my existing institutional subscriptions?", "can I set up a new institutional subscription that would cover this e-book?" or simply "can I buy a one-off copy of this e-book?". It turns out that this is a pretty hard problem. My Licence Negotiation colleagues at Eduserv suggested doing some kind of search against myilibrary, dawsonera, Amazon, eBrary, eblib and SafariBooksOnline. The bad news is that (as far as I can tell), of those, only Amazon and SafariBooksOnline allow users to search their content before making them sign in and only Amazon offer an API. (I'm not sure why anyone would design a website that has the sole purpose of selling stuff such that people have to sign in before they can find out what is on offer, nor why that information isn't available in a openly machine-readable form but anyway...). So in this case, moving from discovery to delivery looks to be non-trivial. Shame. Even if each of these e-book 'aggregators' simply offered a list1 of the ISBNs of all the e-books they make available, it would be a step in the right direction.

On the other hand, maybe just pushing the question to the institutional OpenURL resolver would help answer these questions. Any suggestions for how things could be improved?

1. It's a list so that means RSS or Atom, right?


TrackBack URL for this entry:

Listed below are links to weblogs that reference Finding e-books - a discovery to delivery problem:


Here, we're endeavoring to list all e-books in the catalog. (We're getting there). (They are not neccesarily in OCLC despite being in our catalog).

So you could potentially search by title/author in the catalog. Or by ISBN, if a) the ebook has an ISBN (not all do), b) you know the ISBN (xISBN is one answer), and c) the ISBN is actually in the catalog record (probably, if it has one, but not neccesarily).

In some cases, the eBook catalog record will have the print ISBN on it too, so that search is even easier, search by print ISBN, limit by electronic format. IF the ebook is in the catalog and has the print ISBN on it. (Catalog searches of this sort will become a lot easier when we overhaul our public catalog with Blacklight).

Additionally though, this is a problem my Umlaut 'link resolver' service tries to solve. Umlaut is essentially a 'known item' service provider -- give it a specific item (by title/author, ISBN, whatever), it'll try to find ALL the copies we have, and other services.

It works... so-so. It doesn't yet use xISBN automatically, but some day I'd like it to. It's success is constrained still by the quality of our internal metadata, and by it's own algorithms attempting to deal with varying quality metadata balancing 'precision' (minimize false positives) and 'recall' (minimize false negatives).

But it sometimes, I believe often, works. Check it out one succesful example:

It is odd. I'm wondering if they are looking at the business model in the wrong way and by forcing you to create (free!) account with them they have better opportunities for monitoring what you are looking at?

Dawsonera are very explicit they that are a platform 'for use in libraries', which is a stance on a business approach.

I'm wondering with the others if this reflects poor search / cataloguing platforms available (http://blogs.semantico.com/discovery-blog/2010/07/why-recycled-journals-systems-don%E2%80%99t-work-for-books-and-reference/) rather than an explicit approach to hide information, but i think it is definitely a) a good question to be asking and b) a great service to deliver :-)

Ebrary offers http://www.titlepreview.com to provide the community the ability to search and preview the first few pages of each top level chapter for any title offered for sale by ebrary without having to sign in.

All titles offered for sale or subscription are available from that site. You may also search by print isbn or e-isbn to find titles.

Finally, if you are an ebrary customer, you may also request activation of the "list" feature by sending an email to support@ebrary.com One of the benefits of this feature is that you may import title lists or lists of isbns into ebrary and if ebrary has the title, then those titles will be imported to your "list". From there you may purchase or subscribe to to titles on the list.

The comments to this entry are closed.



eFoundations is powered by TypePad