Archive for the “LCSH” Category

Jeff Beall reposts some of the discussion on the  MODS mailing list on the recently published draft of MADS/RDF.  Beall wonders if the proposal, and the initial comments indicate that the entire semantic web effort is doomed. This may be an over-reaction.

I spent some time working on a formal analysis of the semantics of  pre-combined[1]/subdivided subject headings, and how they could be added in a downwards compatible way to SKOS. I had hoped to discuss the topic at the DC VOCAMP held at The Library  immediately after ISWC 2009, but there was no-one at the camp who was interested in library vocabularies, I let the matter drop[2].

Since support for pre-combined headings are one of the  major design goals of  MADS/RDF, I have been looking quite closely at the initial  draft.  I am not yet ready to publish detailed results, but I wanted to take this chance to make a few quick notes on the initial draft.

Minor Issue

There is a quick change that could make it easier for others to evaluate the proposed ontology.  Currently, the URI/IRI of the ontology is “″.  At the moment does not resolve as a URL.  Also, the name is slightly different from current usual practice.  Ontology URIs are typically of the form …/<Year>/<Month>/<Name>; the MADS IRI is in the form …/<Name>/<Year>/<Month>. This encourages some common tools, such as Protege, to abbreviate the ontology name as “11″.

Major issues

  1. The SKOS mappings are imported as part of the reference ontology; however due to problems in naming, this may not show up correctly.  It might be better to include the axioms in the mapping file directly in the main ontology.
  2. The SKOS mappings are incomplete; it is possible that concentrating  on narrowing the differences between SKOS and Standard(NISO Z39.19)/Library KOS semantics would give an end result that should be more useful for use by The Library.
  3. The current draft was created without support from an ontologist or owl expert. There are several problems in the modeling that result from this lack of resourcing.  For example, there are several places where a constraint or restriction is given in a comment that could readily have been represented in OWL, and many properties are not fully or properly specified.  There is a big difference between the ontology design and schema design, and without support resources, it is easy to fall in to some common anti-patterns.
  4. The current draft was created without ready access to Subject Matter Experts (SMEs) in the Cataloging COI.  As a result, the ontology may not adequately represent the problem domain, and may make it harder to extend so as  to support LC rules and policy on subject heading formation. When modeling any complex area, regular interaction with relevant SMEs is essential throughout the process, both for knowledge elicitation and model validation.  Early, regular review of the ongoing work  by MODS/MADS Editorial Committee may be helpful.
  5. The current draft does not define the semantics of pre-combined headings; there are different possible interpretations,  but none are discussed, and some inferences appear to be common across all readings.
  6. Given known  data errors in the syndetic structure of LCSH (see e.g. LCSH is to Thesaurus as Doorbell is to Mammal), it may better to devote resources to correcting these issues before deepening the semantics of exposed Linked Data.  This would make the LC vocabularies more reuseable in the web of data, as recommended in the Deanna Marcum’s  ”Response to: On the record: Report of the Library of Congress Working Group On The Future of Bibliographic Control“.

[1] Following Soergel, I prefer this term to “Pre-coordinated”

[2] I also worked on relating the to the implied ontologies inferable from Knowledge Organization Systems that used sub-types of BT relationship (BTG, etc) that are part of the international standards, and which were originally included in SKOS, but which were moved to an extension , then finally dropped and relegated to examples in the published recommendation.

Comments No Comments »

TThere’s a critical difference between subject based knowledge organization schemes created to organize books and other documents, (e.g. LCC, DDC, and LCSH), which are defined in terms of the things which a work is about, and rigorous, class based ontologies that are based on the properties of the things themselves.

Some relationships may be valid in terms of subjects, but may not apply directly to the things themselves. For example, a car wheels are part of cars, and cars are a type of vehicle. These relationships are always* true, and are examples of two different types of hierarchical relationships – partitive (is-part-of) (BTP) , and generic (is-a-kind-of) (BTG)

Always-aboutness is the combination of all the different types of hierarchical relationship – this is the Broader Term (BT) relationship. In terms of aboutness, we can easily say that everything about car wheels is always about cars, and everything about cars is always about vehicles. We can also directly infer that everything about car wheels must always be about vehicles. In other words, the always-aboutness relationship must be transitive.

Transitivity doesn’t hold when you mix different types of hierarchy – car wheels are not a kind of vehicle, but that has nothing to do with the transitivity of plain unqualified Broader Term.

When thinking about systems for organizing information, it’s absolutely critical to remember the difference between the document, and the thing the document is about.

Or in other words:

This is your Unicorn. This is your Unicorn on SKOS. 

Comments No Comments »

This is the latest version of the Doorbell -> Mammal graph; it shows the direct and indirect broader terms of doorbells in LCSH.

This incarnation of the graphic adds one new piece of visual information that seems to be very very suggestive. Dashed lines are used to indicate broader term references that have never been validated since BT and NT references were automatically generated from the old SA (See Also) links  in 1988.  Click to download the PDF version

Dashed doorbell graph

Comments No Comments »

By a curious coincidence, I just noticed that the earliest surviving entries in the LCSH Subjects files were created on January 27th, 1986. This date would have been the first Monday after the ALA Midwinter conference (Jan 18th-22nd ,1986).

From the time stamps, it seems as if the system became live for new entries on this date. Older entries weren’t added until Feb 11th.

Less than half of these converted entries have had their references evaluated. This is suggestive.

Reference evaluation (008/29)

Value Count Percentage
a (Evaluated) 45,495 38%
b (Not Evaluated) 46,387 39%
n (No References) 27,320 23%
Total 119,202 100%

[LCSH 150s, established, as obtained from of December 2006]
The oldest surviving record, for “Grouper fisheries”, was assigned LCCN “sh 86000005″.
You should have seen the ones that got away.

Comments No Comments »

When you look at the Library of Congress Subject headings as individual entries it’s almost impossible to understand just how confused much of the hierarchical reference structure has become.

I’ve written some code to generate graphical representations for the Broader Terms of entries in the LCSH. The starting term appears at the bottom of the graph; according to the rules, this term is a specialization of every other term on the graph. Top level terms are highlighted using double circles.

Layout and rendering is courtesy of the wonderful graphviz (AT&T Research). I have generated dot files for all entries in the LCSH; I need to set up an dynamic renderer so they can be viewed online, but a p7zip archive raw dot is available here. (5M compressed, 672M uncompressed)

Lets see what the LCSH has to tell us about Doorbells.

Doorbells are a Social science. Doorbells are Souls. Doorbells are even Ontologies – which would explain why Protege keeps beeping at me. But most of all, Doorbells are mammals.

Obviously this conclusion is absurd. Everyone knows that doorbells aren’t hairy. But where are the errors that lead us to this mistaken conclusion, and how can we start to correct them? That’s the subject of tomorrow’s post.


Comments 3 Comments »