Can a group grow a classification?

The tricks is: Can a group grow a classification, or is strong central control necessary? Can someone build "social software" (features not technology) that threads the needle between order and chaos?

Worst idea: You dump Cutter into a wiki. People expand and change it constantly. Books classed X on Monday (and shelved there) are classed Y on Tuesday. One day PHP is under programming languages, the next it's under web development. Cats lie down with dogs. It rains blood.

The "physicality" of classification systems is something tag enthusiasts diss. But it's a necessary in the library. You can't change classifications constantly.

Some solutions to the control/chaos issue:

1. Forking. Except in very limited circumstances, this isn't a solution. It's a problem.
2. "Stages" of review and approval. What can minimize "drag" while keeping what's good about classification?

I'd be interested to hear what librarians and catalogers think. Is an open classification doable? This is me throwing out an idea. Who can catch it?

Can a non-librarian ask how an online database benefits from an inventory control system?

Mostly it doesn't. The idea would be develop a system which physical libraries would use.

I love the idea of open classification. But, I really think that LCSH is the way to go since so many MARC records exist with these subject headings already. Data-mining and connection-making will be much, much richer.

I feel strongly that LCSH should be available to the American tax payer in electronic format. However, LOC ain't playing that game. Here is a recent email exchange I had on the topic:

==========================

Dear Bill:

Your message was referred to me for comment. Library of Congress subject headings are available for free to the public in electronic format at . The value-added online service, Classification Web , is created and maintained using subscriber money (not tax dollars). Without subscriber funding, Class Web would not be available.

It is true that GPO explored FDLP distribution of LCSH in PDF. Based on user feedback it was determined that access to LCSH via LC's online authorities service (URL above) was seen by testers are being much more useful and user-friendly.

I hope this answers your question. If you have follow-up questions please don't hesitate to contact me directly.

Bruce Johnson

Question History:

Patron: To the Digital Reference Team:
Why aren't Library of Congress Subject Headings available to the public in electronic format. Aren't they funded by taxpayers? Aren't they free of copyright since they are government publications? Aren't they available in PDF format via the e-LCSH that was piloted with Federal depository libraries?

Librarian 1: I am referring your questions to the Library's Cataloging Policy and Support Office (CPSO). You will receive a response to your questions from CPSO staff.

Sincerely,

Peter Armenti
The Digital Reference Team
Library of Congress

Loche A McLean

Nov 14
The Library of Congress Subject Heading records are available for free at:

http://authorities.loc.gov/

The creation and upkeep of the individual records are funded by Congress. The formatting and distribution of the printed Library of Congress Subject Headings are funded from sales under the Act of June 28, 1902 (chapter 1301; 32 Stat. 480; 2 U.S.C. 150).

The subject heading records are protected by copyright outside the U.S., not within the U.S.

The PDF version from the Government Printing Office was done as a one-time experiment. Results showed the printed version was preferred by a great majority of users and the effort required to index and produce a yearly PDF version was not cost effective.

Loche A. McLean
Cataloging Distribution Service
Library of Congress
202 707-1285

I wonder about how cataclysmic this cats lying with dogs thing would have to be.

What if the classification system did not depend on physical labels on book spines, but used any identifier on the book (ISBN, title, color, weight) to connect to a database (LibraryThing) which incorporated a rapidly-evolving classification system (WikiCutter) to determine where the book should next be re-shelved? The physical book wouldn't always have to go back in the same place in order to always be findable and the classification scheme would be better prepared to handle social revolutions.

LT could even use its "shelf view" to provide an easy way to locate the unlabelled book. (It's next to the big red one on the shelf with all the little yellow ones)

Developing a system that physical libraries would have available is definitely one use for a new system. But a classification system can be, and often is, the basis for a subject catalogue. This is especially true outside the US. The biggest example that comes to mind is the huge British National Bibliography (BNB) arranged by Dewey. (or at least it used to be, i haven't seen it in a while).

The subject catalogue of the Engineering Societies Library in NY (a couple of million records or so) used to be a classified catalogue (arranged by the Universal Decimal Classification, similar to Dewey be adapted and expanded for more in-depth indexing). Also the Crerar (sp.) Library in Chicago used to have a Dewey-arranged subject catalogue.

And the small special geology and water resources library where i worked as an LIS student back in the 80s also had a classified subject arrangement (also by UDC because they had in-depth coverage of the earth sciences).

Alpha access to a non-online classified catalogue is by an index of subject terms/headings referring to the class number for each term/heading. In an online environment one can easily have both an alpha and a classed subject catalogue.

What about multiple subject concepts? Several subject class numbers can be assigned to any item. One just needs to make sure the bib record indicates clearly what class no. is being used for the physical location.

Note: Sometimes the class system is used only for the catalogue, and items are shelved according to some other system whether a subject or a non-subject system. That's the case with my catalogue which i've arranged by Dewey -- or Dui lol -- but my physical library is not arranged that way at all -- currently i'm using the "pile" system, as in piles of books on the floor :)

Edited to add this geekish note: Some years back when i had a library job where i was dealing with the OCLC MARC format of that time in addition to whatever 080 or 090 one used for the main class no. (normally the call number), they also had, in the 600s (subject headings block) a repeatable 699 field -- specifically for adding added class numbers to a classified catalogue. The field also had a subfield for an alphabetical term.

(This is all just thinking out loud here, trying to make sense of the topic as I go along, probably more for my benefit than anyone else's....)

There are two issues to separate that I think some posts have mixed: one is *subject headings* and the other is *call numbers*. Subject headings (as LCSH) are not really at issue here I don't think; books can have any number of LCSH subject headings (I think 10 is the practical limit), and they don't necessarily relate to the physical location of a book on any shelf; they are "access points" in cataloging lingo. We could talk about that, but that's another topic.

The main issue here seems to relate to *call numbers*, which have as their primary function the location of a physical book on a shelf (so you can call for it). In modern times, call numbers have usually embodied a subject-aspect, and so it's been necessary to select one primary subject that each book is "about" so it can be grouped with like books on the shelf. But there, even LC often provides an alternative LC call number (often for individual numbers of analyzed serials), for those libraries who want to shelve things differently from the LC itself.

The MARC format seems to have fields for LC call numbers, Dewey call numbers, Universal Decimal call numbers, and various others; see fields 050-099 here:

http://www.itsmarc.com/crs/Bib1468.htm

There doesn't seem to be a pre-established slot for a Cutter call number, but there are plenty of empty slots for local use. So what we're talking about is really just creating another field in LT for "Cutter call number" -- in my catalog view I could even display three parallel columns, one with LC call number, one with Dewey, one with Cutter (and LT could in principle pull several other types out of the underlying MARC records, too).

Then the question becomes, how to populate this field? If the "wisdom of crowds" model is followed, that presumes thingamabrarii will assign them; I think that's assuming a lot. Tim seems to suggest that there might be some kind of automated mapping from, say, the LC schedules to the Cutter schedules, yes?

Are any of the libraries that still use Cutter available via Z39.50? If so, are their Cutter numbers embedded in their MARC records? If so, perhaps that could be a primary source for seeding LT with Cutter numbers that then could be "greened"-in where they are absent.

(A bit disorganized, this post. I'm thinking it through.)

Boston Athenaeum still uses Cutter and is available via z39.50 at www.bostonathenaeum.org:7090/Voyager.

Here is a record I just retrieved (98 = cutter call number):

000 00693cam a2200229I 45
001 282933
003 OCoLC
008 720726s1966 nyu b 000 0 eng
010 __ |a 65028600
019 __ |a 304141 |a 5076793 |a 169483
035 __ |a ocm00366305
040 __ |a DLC |c WSU |d OCL |d SER |d OCL |d YNG |d JOC |d OCL |d HIR |d SER |d BAT
049 __ |a BATT
050 0_ |a PR6039.H52 |b Z78
082 __ |a 821.912
098 11 |a VEP |a .T3594 |a .zm
100 1_ |a Moynihan, William T.
245 14 |a The craft and art of Dylan Thomas |c by William T. Moynihan.
260 __ |a Ithaca, N.Y., |b Cornell University Press |c 1966
300 __ |a xvi, 304 p. |c 23 cm.
504 __ |a Bibliographical footnotes.
600 10 |a Thomas, Dylan, |d 1914-1953.

I don't see why if one can make an open computer operating system, why one couldn't make an open classification system. I think the question is finding the most viable system to work with.

>8 mckinney:

Good example. OK, so they are using MARC 098 for the Cutter call number. If LT decides to do something along these lines, it would be good to make LT data follow this format in some way, for better interchange and universality.

09x fields:

http://www.itsmarc.com/crs/Bib0740.htm

Based on the Winke article, it looks like at least a couple of libraries are updating the Cutter schedules on their own. It might be worth while to see if they would be willing to donate their notes and information to the cause.

Yeah. Unfortunately, they're working independently, so they've surely done semi-incompatible things.

I connected with him by emal. I'm going to give him a call tomorrow.

T

If LoC is public domain anyway, can't we borrow elements off it to update Cutter?

Not being a professional cataloger, maybe I'm not understanding, but I found the LC subject authority records available and searchable online. This should be enough for individual librarians to be able to catalog a particular title that LC has chosen not to catalog, I would think.

What am I missing?

I ran over to the library school library. We have 4 copies of Rules for a dictionary catalog--none of which are circulating. The opac says that we have v.1 of the expansive classification, but on the shelf we only have v.2. What a mess! It is a good thing that you can read it (Rules for a dictionary catalog) online, although I still prefer reading print!

From what i remember reading about the 2nd edition of Bliss is that one of the things it intends to address is the divide between "subject headings" and "call numbers", or, more accurately, classification numbers.

(To digress before getting to my main point, call numbers are class numbers used to physically locate an item rather than simply locate it in a classed subject catalogue.)

One the one hand we have "subject headings" or thesaural "subject terms".

On the other hand we have "classification numbers" (or alphabetic or mixed alphabetico-numerical codes).

They both represent the exact same thing -- a subject concept.

How they differ is how they represent the concept, i.e., the "language" they use. The alphabetical term represents the concept within the concept of the alphabetic language. A classification code represents the concept within the concept of a conceptual system created by someone (hopefully with some knowledge of the subject concepts).

It just so happens that numbers or codes are also convenient as physical addresses for filed or shelved items, hence their popularity as call numbers.

> 8 and > 10, using MARC 098 for the Cutter call number

Just to mention it: from the Winke article, p. 125: "... EC notation, which is placed in the MARC field 084$a and identified by the source code 'cutterec0 in 084$2 ..."

Yes. I saw. I need to see if LT has any so far. Probably the Athenaeum books have it.

Dystopos:

It strikes me that the problems here are, in no particular order:

a) it adjusts your circulating copies, but not the ones that only get taken out once in ten years;

b) additional effort needed to shelve - rather than just check the spine, you need to query the db & give it a slip of paper or something with the code, which will slow matters up. You'd also need to adjust shelving a lot more often than you do now, as certain sections expand at the expense of others;

c) you wouldn't know, reliably, where things are, and have to look them up anew each time. Big, big timewaster there.

A is probably the killer - after a year of this and one or two big "reclassifications", you might find a third of your books on some topic are in one room, and the other two thirds (unused) are in another, whilst that two thirds has had a new migration of books intermingled with *them*, etc, etc. Not to mention having to move all the signs around!

But the thing is, grand conceptual changes in classification, the sort of thing this would be good with, aren't that common. Once you've stabilised out the old hangovers - doing things like fixing Ancient History or Religion, and giving computing a decent section of its own, to name the classic holes in Dewey - you're left with relatively minor incremental developments, fiddling subsets of specific fields, so on and so forth. Major, major changes - like computing above - really don't happen that often.

And a slightly outdated classification system, on the whole, is probably about as useful for most readers as a perfectly up-to-the-minute one. As long as it was written with the same worldview, you're okay - the last decade is good, the last few years more than enough.

(I still use DDC-20 at work, complete with USSR headings and no Internet - but we manage with a few small patches. Improvements would be nice, but our readers can survive without - as long as the system's consistent and well-labelled, it doesn't have to be conceptually pure)

Does anyone know about the pricing and licensing issues from the inside?

I think it would be a terrific idea with the right technology structure behind it. A wiki I agree would be no good, but perhaps some kind of collaborative site where members can suggest alterations/new categories and others vote for/against those changes.

With regards to the classification being changed "by the masses" regularly, well perhaps there could be a regular schema "e-published" every few months or so - so whilst the latest version would always be visible online, Libraries participating could decide if they wanted to take the e-published schema on board or not whenever it comes out.

I think (being a professional cataloguer) people should have to prove their credentials before getting the privilege of fiddling with it.

I think that I would like to be able to fiddle with it though I am not a professional.

I do however understand the reason behind that condition and think that I would rather be only able to make suggestions then to have it be messed up by people who have not put some thought into what is being done.

If a new classification were to be done, the wikipedia model would probably be too chaotic ti be useful.

But how does the open-source process for software creation work? I am not at all a programmer but it seems to me in order to work software can not be written chaotically. And programs like Linux etc. obviously work.

So what is their process and how could it be adapted to developing a classification?

Well open-source software isn't just developed in one way - there are a multitude of ways.

The way Linux is developed (or at least the kernel) is that there is a benevolent dictator with a number of lieutenants. Anyone can submit a patch (or a whole new module) to the kernel. Whether it gets in or not is up to that group. Linus Torvalds has on a few occasions grudging accepted the majority view even though h personally disagrees with it. It works mainly because Linus is one of us, a software developer, and has our trust. Whether it would work in this case is debatable as a lot of the energy that whizzes about LT are from people who are liberated by being in control and not subservient to a librarian who knows best.

Also on LT are plenty of people who are well-versed in their particular area of expertise (for example computer science and computer programming) it would seem foolish to throw any help these people might offer back into their faces.

I believe they used "subject specialists" when developing LC

BSD is evidently similar except there are elected committees at the top instead of Linus.

I just started a new thread about consideration of broadening the scope beyond just Cutter, altho what i said there could just as easily have gone here. In any event, what Andyl said about the way Linux is managed seems like a possible model. And re Katissima's post, certainly one wants people who are knowledge about the areas of knowledge they are undertaking to classify.

As i think i mentioned in the other thread, nearly the whole classification has been published in multiple volumes, and what hasn't been finalised is available in penultimate draft form. (For more info, see links to the Bliss Classification Association website in the other thread.) Each volume is very in-depth, more so than either LCC or DDC, and can serve as a special subject classification for the subject of that volume. For example, Class H (Human Biology, Health and Medicine) is basically as detailed as the MeSH Thesaurus (not the Nat. Library of Medicine Classification which is not nearly as in-depth as Bliss 2nd ed. class H). Subject experts certainly were involved in developing the various areas of the classification.

I plan to join the Bliss Classification Association (BCA). I cant do it via the web, i have to mail an international money order, and i'll have to wait til after Christmas. But anyway, in an email to them i mentioned that i thought a major lack was not having a one-volume briefer-level version of the whole classification. Not only would this be good for special libraries to classify stuff outside their core subject area, but it would be good for smaller general libraries, plus a good marketing tool for those who wanted to get a feel for the whole classification.

The BCA puts out an annual or semiannual bulletin which includes updates to the classification.

I am intrigued by the possibility that that LT users such as ourselves who are interested in the classification issue might be able to join forces with the BCA