Re: [BIBFRAME] Instance of ISBNs in MARC records
So, do we have a design hypothesis to reflect upon here?
Since a null hypothesis is supposed to be expressed in terms of no difference between treatments, samples, populations, etc., a null hypothesis for ISBN utilization and prevalence ought to be:
There is no difference in ISBN assignment strategies, as indicated by the subsequent quantity of ISBNs assigned to specified types of BIBFRAME-referenced information products.
Over what would be be the case with internally-generated unique IDs assignment strategies, quantities assigned, etc. (Most design techniques provide for entity/object unique ID generation)
Were the ISBN null hypothesis supported, BIBFRAME, etc. designers could employ ISBN numbers rather than having to generate numbers internally, or move null hypothesis construction and testing on to alternative existing identification systems.
A qualitative glimpse of the percentages reported below (assuming that the inventoried records have internally-generated unique identifiers assigned to them) indicates that the null hypothesis is not supported.
So - what should the design response to the above suggested finding be?
Ron Murray
---------
On 5/20/13 11:49 AM, "Graham Bell" <[log in to unmask]">[log in to unmask]> wrote:
And this accurately reflects the raison d'être of the ISBN -- it is an identifier that identifies individual products. The EPUB, the PDF and the MOBI are all different products (even if they all have the same content), and may be available from different suppliers, priced differently, exclusively sold through different outlets, readable on different devices etc. Other identifiers (like the so far not widely used ISTC) can be used to collocate multiple products with essentially the same content.
And as Laura correctly points out, publishers assign ISBNs, primarily in order to facilitate trading of books among their trading partners. For physical books, this is a well-understood system. For e-books, not so much: some trading partners don't insist on ISBNs, others do. Others still can allocate ISBNs* if the publisher does not. And a few rely internally on proprietary identifiers†. Best e-book practice on this is pretty clear these days, but actual practice among publishers still varies.
Now for Schlomo's MARC records, it's quite understandable that some have no ISBN and others have multiple ISBNs, since the MARC record is not bound to a particular 'product' quite as closely as a ISBN. Having said this, I do find the nearly 40% without ISBNs a bit surprising -- are serial publications or other items that would not be eligible for ISBNs included in the sample?
Graham Bell
EDItEUR
* that is, the retailer can in theory allocate an ISBN if the publisher does not -- though in practice I don't think many retailers do this.
† like 'proprietary EANs' (special GTIN-13s reserved for internal use only), or ASINs within Amazon. Because of their proprietary nature, I suspect these aren't much value in a library record
EDItEUR Limited is a company limited by guarantee, registered in England no 2994705. Registered Office: United House, North Road, London N7 9DP, UK. Website: http://www.editeur.org
On 5/20/13 7:49 AM, Laura Dawson wrote:
Karen, at the ISBN Agency we've discovered that this is not consistent. Amazon doesn't supply an ISBN - the publisher does. And many publishers don't bother assigning an ISBN to their mobi files, because it's not a requirement at Amazon. BN/Nook assigns a proprietary EAN if no ISBN is provided by the publisher. Best practice dictates that different formats receive unique ISBNs, but many publishers don't adhere to that.
The number soars, but not consistently.
From: Karen Coyle <[log in to unmask]">[log in to unmask]>
Reply-To: <[log in to unmask]">[log in to unmask]>
Date: Monday, May 20, 2013 10:40 AM
To: <[log in to unmask]">[log in to unmask]>
Subject: Re: [BIBFRAME] Instance of ISBNs in MARC records
Another interesting piece of data:
For eBooks, if they are distributed by different vendors (Amazon v. iTunes, etc.), each VENDOR provides a different ISBN. So the exact same thing gets more than one ISBN. Also, every different eBook format from the same vendor is supposed to have a different ISBN: PDF, ePub, Mobi, etc. So as we go more digital, the number of ISBNs expands soars!
kc
On 5/20/13 7:05 AM, David Weinberger wrote:
Is this information from Mac Elrod and Shlomo Sanders the frequency of ISBNs publicly bloggable? It's very interesting.
- David Weinberger
On Mon, May 20, 2013 at 9:19 AM, Shlomo Sanders <[log in to unmask]">[log in to unmask]> wrote:
From a large sample of data that we get from publishers:
- Only 24.5% with one ISBN!
- 40% with multiple ISBNs!
- Largest group has no ISBN!
ISBN Count Percentage of records with X ISBNs
0 38.57%
1 24.49%
2 17.00%
3 5.88%
4 10.77%
5 2.14%
6 0.65%
7 0.09%
8 0.12%
9 0.24%
10 0.04%
Thanks,
Shlomo
Experience the all-new, singing and dancing interactive Primo brochure
-----Original Message-----
From: Bibliographic Framework Transition Initiative Forum [mailto:[log in to unmask]] On Behalf Of J. McRee Elrod
Sent: Sunday, May 19, 2013 20:34
To: [log in to unmask]">[log in to unmask]
Subject: Re: [BIBFRAME] Instance of ISBNs in MARC records
In SLC's database of over 300,000 records, over 169,000 have no ISBN, over 114,000 have one, over 19,000 have two, and the number of ISBNs ranges up to 36 in one record.
For us, Instances can't be based on ISBNs.
Total Records: 315389
020s Records
00 169021
01 114221
02 19745
03 7403
04 3238
05 1149
06 344
07 118
08 29
09 42
10 11
11 16
12 10
13 12
14 4
15 5
16 3
17 7
18 1
19 1
21 1
22 1
23 1
25 3
27 1
28 1
36 1
__ __ J. McRee (Mac) Elrod ([log in to unmask]">[log in to unmask])
{__ | / Special Libraries Cataloguing HTTP://www.slc.bc.ca/
___} |__ \__________________________________________________________