The FRBR Blog

Earlier this week OCLC announced xOCLCNUM, which is like xISBN except that instead of giving it an ISBN and getting back related ISBNs, you give it an OCLC number and get back related OCLC numbers.

Timothy McCormick included this in his e-mail to the code4lib mailing list about it: “ISBNs have been assigned since 1970, to most but not all books published. OCLC numbers are assigned whenever a record is added to WorldCat, OCLC’s global union catalog. These records cover a large portion of all books, old and new, held by any library in North America and, increasingly other regions worldwide (most recently, National Library of China: see http://www.oclc.org/news/releases/20085.htm). So the coverage range of OCLC numbers is, not surprisingly, far greater than that of ISBNs: in WorldCat, for example, around 100 million OCLCnums compared to about 20 million ISBNs.”

Very useful.

Thom Hickey, of OCLC fame, posted FRBR and Uniform Titles on his blog Tuesday.

AACR2 lists four uses for uniform titles, but the most common is to group items that appear with multiple titles under a single heading. Works such as Don Quixote that are published in multiple languages and under hundreds of different titles benefit from this. Unfortunately, when trying to group manifestations into works, uniform titles do not always correspond to what anyone would consider a work.

Xiaoming Liu posted xISBN JSON Callback and a GoogleBooks Demo last week.

JSON output is native JavaScript, and callbacks are particularly useful for use with web service requests in client-side JavaScript. see: http://xisbn.worldcat.org/xisbnadmin/doc/api.htm#callback.

I don’t see the Google Books magic in the Google Books demo that he mentions, but that’s probably some temporary thing or something on Google’s end. I noticed that happen when I did some Google Books availability testing of my own. Give it a try and see if it works.

You can also get the xISBN output in tab-delimited format, perfect for shell hacking. Liu gives an example which I tweak slightly:

$ GET "http://xisbn.worldcat.org/webservices/xid/isbn/0596002815?method=getEditions&fl=*&format=txt" | cut -f 2,5
177669176 54619668 55847258 79871142	Learning Python
44960325	Learning Python
41466161	Learning Python
156890981 182576260 190774334 213482782	Learning Python.
44124125	Introduction à Python
173084464	Learning python.
69532105	Python

(GET is from Perl’s libwww-perl module, and I always use it and fetch in preference to curl.)

OCLC’s Xiaoming Liu sent word to the xidentifier-l mailing list on 11 February about some new features to xISBN. The xISBN API documentation has been updated to include it. You can now page through results at increments of your choosing. Along with the other metadata available, you can now ask for Library of Congress Control Numbers and cities of publication, and it’ll tell you if it knows. Liu gave this example which asks for only LCCNs and cities. And xISBN now returns full author names, as shown in this other example. Check the API for more — these are just recent additions.

ThingISBN Adds LCCNs, OCLC Numbers from last week explains how thingISBN, LibraryThing‘s xISBN equivalent, can now also return LCCNs and OCLC numbers if you ask if, as in this example.

The Library of Congress Control Number is the unique identifying number assigned to new things acquired by the Library of Congress. It was in the news lately because of the new LCCN permalink service, a very nice way of getting reliable access to the LC’s bibliographic metadata.

The OCLC number is an equivalent number assigned by OCLC for things in WorldCat. About a year ago the numbers, which go up by one each time something new is added, hit 100,000,000.

With these new services it’s easier to acquire useful metadata in a nice easy-to-use format. You can take an ISBN, find out more ISBNs, ask LC for what it knows about them, and so on. All three services will be useful for FRBRization projects. You may wonder why there are so many unique identifying numbers for books, but that’s just the way big databases work, and LC and OCLC are so enormous and copied that other people care about the control numbers they assign.

Tim Spalding, of LibraryThing renown, posted While You Were Sleeping, thingISBN Is Getting Better. thingISBN is LibraryThing’s equivalent to OCLC’s xISBN: give either one an ISBN and it will give you back a bunch of ISBNs that are other manifestations of the same work. thingISBN’s coverage and comprehensiveness is improving, as Spalding shows. It’s all done by LibraryThing users clumping things together, and it’s free.

Eric Hellman, an OCLC chap, sent ISBN Linking to Wikipedia to web4lib on Monday.

Last week, we quietly announced the addition of a Wikipedia article lookup facility to OCLC’s Worldcat xISBN service. The xISBN service groups ISBNs into works based on the entire Worldcat data set. This means that the service will connect Wikipedia citations to any of the ISBNs associated with the cited work. xISBN is an XML-based “grid service” designed to help libraries plug into the data without worrying too much about the wiring.

For more information on using the Worldcat xISBN service to get Wikipedia articles, see http://xisbn.worldcat.org/xisbnadmin/doc/api.htm

Here’s an example of how to access the service: http://xisbn.worldcat.org/webservices/xid/isbn/9789574760442?method=getEditions&format=xml&library=wikipedia&fl=*

The example is a Chinese expression of Alice in Wonderland. xISBN looks up ISBNs of related manifestations — in this case expressions of the book in the original English and in other languages, and Martin Gardner’s annotated version, which is actually a work of its own, a derivative one, but let’s not quibble — and then looks through Wikipedia to find which pages mention any of those ISBNs.

This is pretty neat. All you know about the Wikipedia entry is that it mentions a given ISBN, so there’s some mystery about it. Is it a page about the author, the book, the publisher, something mentioned in the book, or what? The relation is unknown. You can deduce, or infer, or possibly both, some details, however, based on, for example, some of the structured or semi-structured information on the Wikipedia pages.

The Annotated Alice doesn’t have one of Wikipedia’s infoboxes done for WikiProject Novels, but David Lodge’s Small World does. Structured user-contributed metadata! Talk about it!

Now, xISBN doesn’t match the ISBN of my item of this work with any others, so this isn’t the best example, but it’s late and I have to get up early, so it’s hard cheese for you, chum. Imagine it did, and that some of the Wikipedia pages linked therein were to pages with infoboxes and structured metadata, and from them you could infer relationships between all the various FRBR entities involved.

And once this kind of thing gets started, and if LibraryThing’s metadata (see Small World at the ‘Thing) were involved, then there would be more reason for sites to encourage people to add (semi-)structured relations between things, which could then be mined, which would enrich things, which would then feed back and encourage the specifying of more relations.

FRBR is an entity-relationship model. The entities we know well, except for some debate about the details of expressions. The relationships are the really fascinating part. We can grab a lot of them from MARC records, but the more tools to get them other sources, especially sources built by regular people, the better.

You know about xISBN, the OCLC service that takes in an ISBN and gives back a list of ISBNs of other manifestations of the same work.

Now, meet xISSN: “To use the service, you submit an ISSN embedded in a URL to the xISSN server, and the server returns a list of associated ISSNs and relevant metadata. ISSNs for different editions of the same serial are grouped together. An ISSN group may also have historical relationships with other groups.”

The xISSN API looks nice, but you need a working WorldCat Affiliate ID to use it, so I can’t put up any nice sample links here.

Updates have been a little slow recently but I should be back on schedule now. I’ll catch up on all the exciting FRBR-related news of the last few weeks.

Xiaoming Liu sent some mail to the xISBN mailing list t’other day announcing some improvements.

Here’s a demo of xISBN. Just go there and press the button. (Excellent choice of manifestation and work, says I.) The “author” and “publisher” attributes in the results are new.

Liu included a link to these xISBN statistics: 14,940,275 ISBNs have been grouped into 11,840,205 work sets.

In 2001 in The Concept of a Work in WorldCat: An Application of FRBR (353 KB PDF), Bennett, Lavoie, and O’Neill (all of OCLC) said: “The average work in WorldCat has approximately 1.5 manifestations, indicating that for the most part, works in WorldCat are small, single-manifestation entities. More than 25 million of the 32 million works in WorldCat (78%) consist of a single manifestation. Ninety-nine percent (99%) of all works in WorldCat have seven manifestations or less, and only about 30,000, or 1% have more than 20 manifestations.”

Liu posted numbers showing what they know now about works and manifestations:

• Text showing how many records have a given number of manifestations
• Graphical respresentation of the above

This is only manifestations with ISBNs, and the earlier analysis looked at all manifestations in WorldCat. Liu said on the mailing list, “Our result shows that average work has 1.26 manifestations. More than 86% works consist of a single manifestation. And more than 99.5% of all works have seven manifestations or less.”

Amigos Library Services is new to me, but they’ve been around for quite a while as a group of libraries in the southwestern United States helping each other. Yesterday they announced FRBR-Applied WorldCat Comparison and said, “Individual subscribers now have the option of applying elements of the FRBR algorithm to your WorldCat comparison via the FirstSearch Administrative Module.” I’m not quite sure what that means, but there you have it.

An announcement was sent to OCLC’s xidentifier-l mailing list about some updates they made to the Library Lookup bookmarklet, which uses xISBN to make it so that if your web browser is displaying something about a book, you can press a button and see if your local library has a copy. Pretty handy. If you’re looking at a book at Amazon, knowing you can have a copy (though perhaps a different manifestation) of it waiting at your local library branch in a couple of days will make you rethink whether you really need to buy it.

I haven’t tried the new system, and there was some discussion about it on the list; it was reported that the current system is not the final one, so if you’re going to do any serious work with it, get on the list and ask.

Ultimately, what we want are tools so that whenever someone mentions a book, CD, DVD, work, expression, whatever, we will have simple tools that will let us find, identify, select, and obtain an item of some suitable expression and manifestation of that work. (Perhaps the exact manifestation mentioned.) “The Three Musketeers is one of the greatest adventure novels ever written, you say? Let me press a button and see my options. Ah, new Pevear translation from Viking, available for sale for $24.99 but my library has it and I can pick it up right now, or I can get the Richard Lester movie on DVD in the shop on the corner.” This tool is one step towards that.