Buried in LC's Policy and Standards Division's Progress on the Recommendations Made in "Library of Congress Subject Headings: Pre- vs. Post-Coordination and Related Issues" (http://www.loc.gov/catdir/cpso/pre_vs_postupdate.pdf), is the statement at 3.f that they will try to
"Build more coded structures within LCSH to enhance the ability for systems to suggest terms. Code many subject authority records so that headings can be matched with an appropriate free-floating subdivision list to simplify the construction of subject heading strings and allow for automatic or computer-assisted assignment of subject strings and machine validation. The Class Web Coordinator will work with David Williamson to explore how to automate this. Use detailees to target new categories that are needed and to experiment with 'Plants' as a start.
"All of the subdivision authority records (MARC 21 tags 180, 185) include 073 fields, which contain the Subject Headings Manual instruction sheet number(s) to which a subdivision applies. Matching the subdivision to the headings requires that 072 fields, the other half of the 072/073 pair, be added to each subject heading record.
"PSD is beginning an experiment to determine the best way to approach the addition of 072s. One roadblock to the experiment is the number of exceptions to the general rules, even within an instruction sheet for one pattern heading. PSD has decided to approach the problem from two angles. In the first approach, a specialist will work with an existing pattern that contains a relatively small number of headings (Materials, H 1158) to determine whether the applicable authority records can be found easily, and whether the addition of the 072 to any of the records could be misleading (i.e., are there any exceptions to which any or all of the free-floating subdivisions for materials would not apply). The second approach consists of another specialist writing an instruction sheet for a new pattern (topical works on moving images) with computer processing in mind (i.e., with few or no exceptions included). The headings will then be extracted from LCSH and any unforeseen exceptions wi!
ll be accounted for.
PSD plans to use the results of this experiment to determine whether a project to add 072 fields to subject authority records can move forward at this time; and if so, what the procedures should be. If the experiment is successful, David Williamson, LC's Cataloging Automation Specialist, will be consulted to programmatically add the 072 fields to the selected subject authority records."
I have long thought that LC's current strategy of explicitly establishing pre-coordinated strings is like a short-sighted game of whack-a-mole. This approach of explicitly encoding the base headings would enable automatic validation of novel combinations and provide more flexibility for non-pre-coordinated applications and different types of browsing.
Right now, LC is creating records for "validation purposes" that explicitly establish whole strings (http://catalogablog.blogspot.com/2007/10/validation-records.html)
In this new approach, existing subdivision records already have a code saying which categories of base headings they can be used with. So "$x Diseases" (sh 99002330) has an 073 of
$a H 1100 $a H 1103 $a H 1147 $a H 1164 ?z lcsh
where H1100 is "classes of persons," H1103 is "ethnic groups" and so on.
The corresponding base heading records have nothing to tell you that this is a heading for a class of persons (infants) or an ethnic group (African Americans). LC is proposing to add this information to those records.
The edge cases would be really hard, but if it works, it would make for better automated validation of subject headings and prevent OCLC's control headings function from accepting things like "Hats $x Diseases."
So I'm really excited to see that they're trying this and hope the project is successful.
Kelley McGrath
kmcgrath_at_bsu.edu
Received on Fri Jun 04 2010 - 15:49:19 EDT