Tim,
It sounds like you want to be able to search on standard identifiers and
are frustrated that the Internet Archive's access doesn't allow it
(although it looks like they do have an ISBN search)? And I'm curious,
why would you want or need to pull down only records that have OCLC
numbers of ISBNs in particular? What is it you need to do that makes
only those records useful?
Because I have no imagination? :)
Levels of engagement with OCA:
We're a scanning participant (as are y'all). There are three levels of
opac engagement I can think of:
*Link to your digitized materials.
*Link to all materials that have been digitized that you have already
selected for your collection.
*Link to all materials that have been digitized.
The catalog:
Sure it's outdated and hopelessly linked to dead tree resources. But a
significant number of my patrons still use it. Faceted browsing only
encourages them. Will it go away? Probably. Will there be a whole lot
of people who use it to their benefit until it goes away? Probably. In
the mean time I'd just as soon let them have direct access to a digital
copy for those who want it and discover it in our opac.
What I want to do:
I'm aiming for the second level of engagement. We've got to do the first
for obvious reasons.
I think the third level is a nice idea. But it's going to involve:
*The Library having a discussion about what that means for our catalog.
*Importing full records. Based on what I've seen these records will
need to be massaged somewhat (not the least of which is to put an
856 into the participant provided marc records, and/or provide all
the opac useful data from those provided records into what the oca
stores and makes accessible via some soon to come API).
Both of these will take some time and significant effort. I'm hoping, as
well, that all this work can happen at a much higher level. OCA/Open
Library + OCLC work out some arrangement or something. Then any library
can download an "oca collection." I hope something like this kind of
upper level coordination happens, so that tens/hundreds of libraries
aren't out there trying to solve the "how do I put them all in my catalog"
issue and duplicating effort.
Why do the second level:
I think it will be good and timely bang for the buck. I'm estimating that
a fairly modest development investment can get me (and others) there
quickly and without having to do much navel gazing about the nature of
collection development/role of the catalog.
Caveat: I'm coming at this from a "what's available at oca, do I have it?"
rather than an "I've got it, does oca have it?" perspective.
Development involves finding a way to connect oca unique ids to
matchpoints in our catalog (that would be the oclc numbers, lc numbers,
issns, isbns). Searching those matchpoints in our catalog. Adding an 856
for matches. Lather, rinse, repeat (as their collection grows).
Will that offer perfect coverage in the catalog:
Nope. There is a whole lot of fuzz in these "unique identifiers." I
still think there's be a huge bang for the buck, though.
Why the desire for all the unique ids:
Because my experience with the catalog says that checking as many as
possible will get me the greatest recall and looking up a few doesn't cost
too much more than looking up one. More match points, more likely to get
best coverage.
Why not add an xISBN hook:
Trying for speedy development/lazy/hoping someone else would.
How long would such a service be useful:
No idea. But if five academic libraries could use it for 18 months, that
would probably be a pretty decent roi.
I emailed the list because I'm lazy and wanted feedback.
If someone else if offering this (soonish), I'd just as soon piggy back.
If there's a decent proxy for what I'd like (say, that third level gets
worked out in 3 months), I'd just as soon wait.
If I do do it, laziness leaves me wanting to prove that my investment
wasn't wasted effort. I.e. I do it the "best" way for the most
people (with the caveat that laziness may trump laziness should
feature creep get the best of things).
My questions remain (I guess):
Anyone know of something coming that will do what I want or be a good
enough proxy to avoid this? (nothing clear so far)
If not, would you like to have something like I describe? (some "yes")
Would having these features be enough?
search that enables combinations of:
oca unique id
oclc number
isbn
issn
lc number
participant's unique database number (e.g. iii bnum)
participant's alias (<scanningcenter>chapelhill</scanningcenter>)
addeddate (so you can get new stuff)
returns: oca identifier, other unique id(s)
And, if such a service were available:
How would you like it? I've not built a queryable webservice and am
going on record as ignorant. Is there a query language I should
lean toward? A return data structure that I should adopt?
All this stems from the my belief that what I can see of the architecture
indicates a split between what the participating library sends and what
the oca system uses. It appears that their index/record of record simply
ignores all those hooks people have been adding to bib records. If both
parts were wrapped and offered up with a Solr interface I could get on
with putting links into my catalog.
Still, they make both their record (with their identifier) available, and
my record (with the rest) available. So, like I said in an earlier post,
I'm glad for the opportunity to be frustrated.
Whew. If I'd been hacking instead of writing I'd have something to show
and y'all would be less bored.
Thanks!
-t
Like Karen and Bess and others have said, I recommend that you
coordinate this with the Open Library project. At the meeting last
Friday, it did sound like they would be interested in providing
identifier disambiguation types of service - give them an ISBN, and
they'll give you the records associated with it.
Also, there was discussion about building an Open Librar yAPI (to enable
some cool integration with wikipedia), and I suggested a that libraries
using an API would want the search results to include information about
whether the title has a digitized copy. So I would hope the service that
you're envisioning is something that would be provided by an Open
Library API (but we don't know when that might come about).
As OCA moves forward, folks may well be digitizing identical books. So
there may not be a one to one relationship between unique catalog
identifier, unique oca identifier, and isbn/lccn/oclc number.
-emily
----------------------------------------------------------------------
Date: Thu, 6 Mar 2008 08:47:04 -0500
From: Tim Shearer <[EMAIL PROTECTED]>
Subject: musing on oca apiRe: [CODE4LIB] oca api?
Howdy folks,
I've been playing and thinking. I'd like to have what amounts to a unique
identifier index to oca digitized texts. I want to be able to pull all the
records that have oclc numbers, issns, isbns, etc. I want it to be
lightweight, fast, searchable.
Would anyone else want/use such a thing?
I'm thinking about building something like this.
If I do, it would be ideal if wouldn't be a duplication of effort, so
anyone
got this in the works? And if it would meet the needs of others.
My basic notion is to crawl the site (starting with "americana", the
American
Libraries. Pull the oca unique identifier (e.g. northcarolinayea1910rale)
and
associate it with
unique identifiers (oclc numbers, issns, isbns, lc numbers)
contributing institution's alias and unique catalog identifier
upload date
That's all I was thinking of. Then there's what you might be able to do
with
it:
Give me all the oca unique identifiers that have oclc numbers
Give me all the oca unique identifiers with isbns that were
uploaded between x and y date
Give me the oca unique identifier for this oclc number
Planning to do:
keep crawling it and keep it up to date.
Things I wasn't planning to do:
worry about other unique ids (you'd have to go to xISBN or
ThingISBN yourself)
worry about storing anything else from oca.
It would be good for being able to add an 856 to matches in your catalog.
It
would not be good for grabbing all marc records for all of oca.
Anyhow, is this duplication of effort? Would you like something like this?
What else would you like it to do (keeping in mind this is an unfunded pet
project)? How would you want to talk to it? I was thinking of a web
service,
but hadn't thought too much about how to query it or how I'd deliver
results.
Of course I'm being an idiot and trying out new tools at the same time
(python
to see what the buzz is all about, sqlite just to learn it (it may not work
out)).
Thoughts? Vicious criticism?
-t
------------------------------
Date: Thu, 6 Mar 2008 11:05:41 -0500
From: Jodi Schneider <[EMAIL PROTECTED]>
Subject: Re: musing on oca apiRe: [CODE4LIB] oca api?
Great idea, Tim!
The open library tech list that Bess mentions is [EMAIL PROTECTED],
described at
http://mail.archive.org/cgi-bin/mailman/listinfo/ol-tech
-Jodi
Jodi Schneider
Science Library Specialist
Amherst College
413-542-2076
------------------------------
Date: Thu, 6 Mar 2008 08:32:43 -0800
From: Karen Coyle <[EMAIL PROTECTED]>
Subject: Re: musing on oca apiRe: [CODE4LIB] oca api?
We talked about something like this at the Open Library meeting last
Friday. The ol list is [EMAIL PROTECTED] (join at
http://mail.archive.org/cgi-bin/mailman/listinfo/ol-lib). I think of
this as a (or one or more) translate service between IDs. It's a
realization that we will never have a unique ID that everyone agrees on,
that most bibliographic items are really more than one thing, but that
since we have data about the bibliographic item we have many
opportunities to make connections even though people have used different
identifiers. So we could use an "ID-switcher" to move among data stores
and services. Is that the kind of thing you are thinking of?
kc
--
Emily Lynema
Systems Librarian for Digital Projects
Information Technology, NCSU Libraries
919-513-8031
[EMAIL PROTECTED]
