Library of Congress Classification improvements, part 1

We've pushed out the first step in a move to better classification on LibraryThing. This includes two phases related to the Library of Congress Classification, also called "LC Classification" or "LCC."

The LCC is the main shelf-order classification system used by academic libraries, as Dewey is used by public libraries. (There are exceptions in both directions, of course.) It's not as straightforward in conception and operation as Dewey, but it's powerful and flexible.

Phase one of this is to get the top two levels (or classes) of LCC into your catalog. Phase two will give you LCC pages and other tools that show a much larger slice of the full system.

Here's the LC's outline of the system: https://www.loc.gov/catdir/cpso/lcco/

You can find the LC Classification view here:



Here's what it looks like: (click to see it)



There's a button there to see two levels:



There's a field "LC Classification wording" available under settings

https://www.librarything.com/settings/styles#classification

The "LC Classification wording" field shows you the LC Classification system to many levels below 2. But that's the only place to see it—for now. And it doesn't link to anything.



In addition to adding LCC, we've also run scripts to produce work-level LCC codes for more works. So about as many of your books should have LCCs as have Deweys now. LCC's now cover 4,799,854 works, or 83% of the books on LT.

Thanks for this!!!

Should we post weirdness here? If so...

Thanks. Looking.

Maybe(?) worth mentioning: A Style apparently must include the "LC Classification" field in order for the "LC Classification wording" field to be populated.

>4 timspalding: Now Invalid LC Classification is doing the same thing. It wasn't when I posted >3 lesmel:

>3 lesmel:

Fixed.

>7 timspalding: Nope. Now, when I click "No LC Classification (359)" I get "Invalid LC" books and there's no "Invalid LC" in the LC Classification list

On the new page, I'm seeing 83 works with No LC Classification, but when I click the link I see only 6 works with Invalid Classification.

Nocode vs. noclass vs… I don't know. Give it 5m and it should be good. Thanks!

Edited. Done.

One volume of Madeline L'Engle's memoirs is in BV, and only one (out of 5, all of which are similar, and all of which are about Christianity).

Wow, I have more than a thousand works that have no LC classification. That shows my library is not very academic, doesn't it? Haha.

The number you might actually care to know is the five invalid works. Grave Peril and Summer Knight by Jim Butcher and The White Dragon by Anne McCaffrey are parts of series. There's also Princess Addison Gets Angry by Molly Martin and Cherokee Words With Pictures by Mary Ulmer and G. B Chiltoskey. I don't if LT can do anything to validate their classifications.

Weirdness.

Two books about Illinois law are classified as "Law of Idaho", and one about a child murder in Tuscany is classified as "Law of Kosovo"!

I'm a fan of LCC and this is a great feature; thank you! I'm looking forward to playing with it.

I appreciate that when I have LC Classification selected, it only gives the numbers for the specific Collection I've selected--so it's the same behavior as Tags, Authors, Series, etc. If you could make Genre work the same way as all those other options (and now the same as LCC), I'd be even happier.

(Regarding Genre ---- This is like the 4th time I've mentioned it, Tim. Could you please just tell me, "No, that's not how Genre is going to work, just live with it," so that I can stop asking? I realize I'm not on Talk that often anymore, but I've been an LT member for 14 years, and it's weird to post something multiple times and get literally zero replies. Does no one else notice that the Genre drop-down behaves differently than every other option, or is it really only me?!?)

>13 lilithcat: Ha. Check the authors. Sometimes codes like K 123 sneak in when the K is the author's last name.

>14 HeathMochaFrost:

Yes, I'll take a look at that problem.

>15 timspalding:

Hartman, Cohn, and Guarnieri

More oddities.

I have two books listed under ‘Law’, one is Famous Trials, OK but the other is Penguin Science Fiction

Another odd categorisation is The Supernatural Omnibus listed under ‘Education’.

Angela’s Ashes is apparently ‘History of the US’

Just poking around a bit for now. But, yes, I do notice several odd classifications.

>15 timspalding: >16 lilithcat: KKH is both the Law of Italy and the Law of Kosovo, not just the Law of Kosovo as shown in the LT LC Classification Wording.

>19 NinieB:

Thanks. And, weirdly, the short version on the LC site has only Law of Italy. (https://www.loc.gov/aba/cataloging/classification/lcco/lcco_k.pdf)

(For investigative purposes see their larger PDF for this area (https://www.loc.gov/aba/publications/FreeLCC/LCC_KJ-KKZ2021TEXT.pdf, page 168). It says Kosovo is KKH 9901–9999.

The larger trick here is that these abbreviated schedules, which cover some 8k nodes, are abbreviated. The real schedules have more than 500k nodes. The problem is that things get very sparse very quickly. LCC is already a bit much for most LT catalogs; trying to go to 500k deep would just be silly. But it's worth us talking about, especially those with experience with LCC.

>19 NinieB:

Thanks!

Further checking reveals that KFI is Law of Idaho, but also Illinois, Indiana, and Iowa.

>17 abbottthomas: Angela’s Ashes is apparently ‘History of the US’

Yeah. The LC itself has it at E184.I6 M117 1996. Strictly speaking 184.I6 it should be "Irish." But it's currently sticking it under African Americans. I've got some investigation to do as to why.

>21 lilithcat:

Yeah. And it's right on the abbreviated LCC schedules, but not in our data. I'll need to dig deep to figure out why.

Cataloging Calculator is your friend: https://calculate.alptown.com/

>12 aspirit:

Wow, I have more than a thousand works that have no LC classification. That shows my library is not very academic, doesn't it? Haha.

I had almost 500, but a hundred or so shouldn't have one (78 rpm records, e.g.). The remainder, in my library which has very little fiction, were mostly:

1. Pamphlets or vanity-ish books from very small presses (in my case, generally related to folk songs), e.g. Let's Stand Together /// Stanly Has a Lynching /// Floyd Collins of Sand Cave: A Photographic Memorial

2. Foreign books, e.g. A History of the Isles: Twillingate, New World Island, Fogo Island and Change Islands, Newfoundland and Labrador /// Favorite Bush Ballads

There are also a few Middle English books I'm surprised to see omitted, e.g. Sources and Analgues of Chaucer's Canterbury Tales /// Historical Poems of the XIVth and XVth Centuries. The latter of those was at least edited by an American. :-)

Still, I think this is off to a much better start than genres. :-) To me, this looks pretty good overall. For one thing, it has a music category. :-)

LCC can be weird sometimes.

For example, what is shine your icy crown doing under L898 (L. Education > L7-991. Education (General) > L797-898. Educational exhibitions and museums)

>12 aspirit: >25 waltzmn: I have over 2K - quite a lot of it fairly obscure European stuff though. So that's not all that surprising.

I have a bunch of works I manually entered LC numbers for years and years ago (mostly copied from the copyright page) that are showing as invalid. To be fair, I was mostly interested in getting them to show up correctly in the sort (and they do) but what is it validating on? Is it just that I used lower case? Is there something else it's checking for?

I like that it has an 'invalid' group though, that's very helpful! Can we have one for Dewey?

>24 lesmel: Maybe not my friend. The page freezes.

>26 AnnieMod: I was going to say that might be the cutter for Lovelace; but the cutter would be L684. Of the member libraries I can see, it looks like the data came from Amazon. The LCC is green text, so it looks like at least one of the three private libraries are the source of that call number.

OCLC Classify (http://classify.oclc.org/classify2/) will usually provide one or more LCC numbers used by libraries for your book.

I would not have thought that Flatland would be under science (QA699 .A13)

>30 lesmel: Yeah - that was my first thought as well when I saw that L but as you say, it will be L684.

Amazon as a source does not surprise me - the book is pseudo-educational (it is not but some of its promotional materials are kinda ambiguous) so them tagging as Education at some point won't surprise me...

Heres's another weird one science: https://www.librarything.com/work/14112676/book/145945428. The video of Be my Valentine Charlie Brown.

My weird one is The falcon and the snowman under Military Administration/ Intelligence.

>35 jjwilson61: That isn't weird since the book is about espionage/spies. The other LCC it tends to be is E743.5 (United States--Twentieth century--Political history--Un-American activities)

>32 al.vick: Flatland under QA for Math makes a lot of sense, since it basically is about math (& it was often used as a teaching text.)

DG is City in your 2 level description but it should be History of Italy according to the link in the first post.

Folks, you do realize none of this data is new, right? It's new presentation, but if you have a bad value in your catalog, go fix it, don't blame Tim. I've had the LC column for years.

>37 melannen: maybe, but it is really fiction. A story.

>39 lorax: Wasn't sure if he wanted us to change things now or not. Sounded like he was working on pulling in new data, and I think most of my data in this field is green "calculated" data anyway. So I wasn't sure what the best thing to was...Report or fix...

>39 lorax: How do we correct the LCC for our works? I must've missed those instructions.

There is a field in the edit book section.

>43 al.vick: And that feeds into the new features, I presume? Okay. Thank you.

>31 NinieB: Thank you for the link. The OCLC site works better for me.

Class numbers look like a matter of debate-- as what I'm seeing on OCLC is different than what's in green font for some of the valid LCC works I tested, and similar books within a series have different classifications-- so I'm deciding right now to step away and go back to ignoring the field in book details.

Happy cleanup to the helpers already familiar with the LCC.

aspirit (#42):

Same way you edit any other data for your books.

This is another place where not being able to properly delete green calculated data in one's own catalogue is rather inconvenient.

I can imagine users wanting to remove inaccurate LCC data from their books, but not feeling comfortable adding alternative LCC information instead -- looking it up is potentially arduous, and I doubt many users are intimately acquainted enough with the system to attempt a classification of their own.

>44 aspirit: As with 99% of cataloging, call numbers are left to "cataloger's discretion" -- basically, where do you want a particular item shelved physically in your library? This is why the same title can end up with different call numbers.

Shouldn't the second level of the outline contain the single-letter classifications also? At the top level, the letter just refers to "every classification that starts with D", for example (and means "History of Europe, Asia, Africa and Oceania"); but the next level down, just plain-old D means something more specific (General histories of the world; & General histories of Europe). This is the case for most of the letters (plain M is for music scores, for another example)

As it is, if I want to know how many I have in plain-B, I have to add up all the other numbers from BC, BD, etc. and subtract them from the total at the top level.

There are a couple of semi-official extensions of LCC that I think would be good to treat as valid:

1) NH classification for photography, so that photography books class with art, instead of with Engineering/Technology (in TR, as the official LCC has it).
This schedule is maintained by the Art Libraries Society of North America:
https://arlisna.org/publications/arlis-na-research-reports/publications-archive/...

2) FC classification for Canadian history. Created & maintained by Library & Archives Canada:
https://epe.lac-bac.gc.ca/100/200/301/lac-bac/class_fc_index-ef/e/n03/FC_Class_2...

>40 al.vick: It's presented as a story but its intent is to teach about dimensions in an entertaining way. Definitely Maths

Three cheers to Tim and his team!!

The first thing I did on taking over the Railway Society's librarianship was to ditch the card catalogue, which hadn't been maintained in years and was an incomprehensible mess, and start from scratch on LT.
Then the previous librarian came out of his frailcare (seniors home) and spent several hours bending my ear about how the way he used Dewey meant all the books ended up in one of about six classes (all too clear from his catalogue, sadly). So my second major change was to change to LCC. I could get about half the books classified from LT or its source libraries, but had to work through LoC's pdfs for the really obscure stuff in our "archive" collection (instructions to staff etc.).

Tim, you have just made my life significantly easier and our library classification that much more accurate. And I now have a brief note for our magazine. Thank you, baie dankie, ngiyabonga and the same in the other 8 official languages!

My "invalid" LC numbers are all in black font, so that's probably old data from either an import or my own choice? (I have a couple where I know I entered top-level letters and stopped; that's on me and I can see why it's not really "valid".)

Does fixing numbers help the script at all, or just my own catalog? If I delete invalid data, might I get the calculated data automagically?

(EDIT: yup, deleting "invalid" data allows whatever's on the Work level to flow to my copy. Which, in my 2 test cases, was calculated data. Yay.)

>37 melannen:, >40 al.vick:, >50 Foretopman:

It has a non-trivial element of social criticism as well. It was written for reading and thinking more than shelving.

>53 paradoxosalpha:

I now wonder if anyone’s ever written a book with the deliberate intent of making it hard to shelve.

>54 AndreasJ: Well, there are examples of chained books...

https://en.wikipedia.org/wiki/Chained_library

>54 AndreasJ: Not written, perhaps, but designed. I have at least one coffee-table book which I remember when it was published that the publisher said they had designed so it would always lie so that one's guests would see it, because its dimensions made it too difficult to shelve. It's something like 40x40 cm. All books that size are of course hard to shelve, but in this particular case, the publisher admitted that was the intention.

>32 al.vick: I haven't checked yet, but I imagine Lewis Carroll's "Euclid and his modern rivals", a book on mathematics but written in the form of a play, may have suffered something similar.

>53 paradoxosalpha: Well the high school geometry teacher who confiscated it from me for reading in class would definitely agree with the people who say it's not a math text. :D

“Ontology” is my middle name.

MARC records…

Some books have been bound in sandpaper out of hostility to shelving.

I have 3 books supposedly with an invalid LC classification, where that classification was found via http://classify.oclc.org/classify2/ so presumably exists in a library somewhere (its listing 40+ holdings for most of them).

(R6120.H66, BZ2301.G7, NH32.B95)

I'm sorry. I understand that some of my library is too obscure to have any LC classification. But way too many of the ones that have are absurd.

>61 tallpaul:
1) the R number is too high; R ends at 920
2) there is no BZ
3) NH, see my msg 49 above

I've whittled away at my invalid LC classification - many of these had Dewey numbers in the LC filed or other obviously incorrect text.

My last remaining issue is Joe Cinque's Consolation - https://www.librarything.com/work/453267/book/157633455
LOC Catalog entry says: KUA3.7.S56 G37 2004

When I enter this as the LC classification this work still shows as invalid. Any ideas?

Is there an easily accessible master list of LC classifications (deep) on LT here, or elsewhere?

>64 rodneyvc:: See:

http://classify.oclc.org/classify2/ClassifyDemo?wi=57345384 or
HV6535.A83
Social pathology. Social and public welfare. Criminology--Criminology--Crimes and offenses--Crimes against the person--Homicide. Murder, etc.--Region/Country--Other regions or countries, A-Z

http://classify.oclc.org/classify2/ClassifyDemo?wi=976429693
PN1997.2
PN1997.2.A-Z Drama--Motion pictures--Plays, scenarios, etc.--Individual motion pictures--Motion pictures produced 2001- . By title of motion picture, A-Z

>65 Crypto-Willobie: Two recommendations:

http://classify.oclc.org/classify2/ -- search for a title/author, see what call numbers are in use
https://calculate.alptown.com/ -- look up LC Call Numbers, Cutters, etc

>64 rodneyvc: The number you have should be valid. KUA3.7 is for individual criminal trials in the Australian Capital Territory

>67 lesmel:
Thanks, lesmel!

>68 Proclus: Thanks. I thought if I used the classification direct from the LOC catalog it should be treated by LT as valid.

>70 rodneyvc: I think "valid" depends on what Tim et al at LT are using for the classification tables. If they are using the current tables, some classifications are no longer used or get broken up for granularity -- thus invalid, technically, but still valid when the title was first classified.**

Also, Australia National University uses:

Joe Cinque's consolation by Garner, Helen
KM542.1.G37 2004, LAW LIBLess
HV6535.A83 C35 2006, LAW LIB

** I am a librarian; but I'm not a cataloger. Most of my cataloging knowledge is 20+ years out of date.

>71 lesmel: Thanks - HV6535.A83 C35 2006 did the trick!

> bnielsen

Some of the links people included about LoC are up above, they're very helpful.

I'm seeing Canadian authors writing in English clustered around PR9199.3. Canadian authors writing in French or other languages are going to be elsewhere and very well may be clusters of other English-language writers elsewhere. Isn't going to perfect but I'm finding it useful.

Edit: Involved adding LoC classification as a column in my library then sorting by it, that alphabetizes the codes.

>73 Settings: Thanks! Not sure if this is helpful for me. It gives me tons of books with no LoC classification and a lot of "T. Technology...", but yes, there's something usable here. Looking for Danish authors:

cat /tmp/lt.rdb | perl /tmp/row LC_Classification mat '/^PT8175/' | perl /tmp/column Primary_Author | perl /tmp/headchg -del | sort -u | sed -e 's/\(.*\), \(.*\)/\2 \1/' | cat -n
1 Johannes Allen
2 Karen Blixen
3 R. Broby-Johansen
4 Cecil Bødker
5 Albert Dam
6 Tove Ditlevsen
7 Peter Freuchen
8 Gunnar Gunnarsson
9 Martin A. Hansen
10 William Heinesen
11 Angelo Hjort
12 Jørgen-Frantz Jacobsen
13 Erik Aalbæk Jensen
14 Johannes V. Jensen
15 Hans Lyngby Jepsen
16 Frank Jæger
17 Harald Kidde
18 Hans Kirk
19 Tom Kristensen
20 Jens Kruuse
21 Martin Andersen Nexø
22 Niels E. Nielsen
23 Lise Nørgaard
24 Ib Spang Olsen
25 Leif Panduro
26 Nis Petersen
27 Robert Storm Petersen
28 Aksel Sandemose
29 Ole Sarvig
30 Hans Scherfig
31 Peter Seeberg
32 Tage Skou-Hansen
33 Finn Søeborg
34 Knud Sønderby
35 Villy Sørensen
36 Knud H. Thomsen
37 Tage Voss
38 Gustav Wied

And looking for Canadians finds these three:

cat /tmp/lt.rdb | perl /tmp/row LC_Classification mat '/^PR9199.3/' | perl /tmp/column Primary_Author | perl /tmp/headchg -del | sort -u | sed -e 's/\(.*\), \(.*\)/\2 \1/' | cat -n
1 Margaret Atwood
2 Yann Martel
3 Michael Ondaatje

Only problem is that I have over 3000 authors listed (and that's only the primary authors) :-)

ETA: and 4338 books with LC_Classification and 4000 without.

Another fun time sink :-) Here's some random notes:

Tim linked to this https://www.loc.gov/catdir/cpso/lcco/ which is sort of helpful for humans, but less so for computers (who can buy a subscription services).

So I can take one of my books which in the export file have LC_Classification Z103 something. But to find out that Z103 is Cryptography. Ciphers. Invisible writing
I need to read that Z102.5-104.5 = Cryptography. Ciphers. Invisible writing
which is much harder for a computer (i.e. I'll need to write a script to figure that out.)

Another book Matematiken genom tiderna has LC_Classification Zl QA 15
Should I just ignore the Zl or does it mean something?

>22 timspalding: Yeah. The LC itself has it at E184.I6 M117 1996. Strictly speaking 184.I6 it should be "Irish." But it's currently sticking it under African Americans. I've got some investigation to do as to why.

Update: I banged my head up a lot trying to understand a problem with LCCs like "E184.I6 M117 1996."

The problem is/was that, from a computer's perspective I6 could either be part of the class, or an author cutter. And even if it was the former, because there was no space before the decimal, there are a lot of LCCs that look like

E184 .I6 M117 1996 or
E184 I6 M117 1996

Where the I6 means "Irish." I finally asked about this on Twitter, and you can read it here:
https://twitter.com/librarythingtim/status/1428854979658407940

Anyway, my current plan is to go back to the source data and get the $a and $b for everything I can--which will be most LCCs on LT, but not all. This will allow me to know definitively what's class and what's item. Then I'm going to write a best-guess algorithm, to see if I can usually guess what's class and what's cutter without the $a and $b. Then I'll run that on any that are in doubt.

The result will be that LCC is going to get split up into two fields, LCC class and LCC item. This will allow me to distinguish between E184 books that are about I6 or by authors cuttering to I6.

>76 timspalding: The simple answer is that if there are two cutters, the first one is part of the class and the second is the item.

A new CK field was introduced also for canonical LCC, but unlike canonical DDC, clicking on the link leads to homepage, rather than the relevant section of LCC. Bug?