Should cataloging change to make it easier to group bibliographic records properly?
Satham Sanghera, writing in the Financial Times, has an article today about the ubiquity of Donald Trump:
After Trump Tower, Trump Parc, Trump Place, Trump Palace, Trump Plaza, Trump International, Trump National, Trump Marina, Trump Taj Mahal, Trump World, Trump Ice and "Donald Trump, The Fragrance", it was inevitable, I suppose, that eventually we’d get Trump University.
Another thing that Mr. Trump has done is to put his name in front of the titles of a series of business books. Here's a screen shot from an experimental interface to Open WorldCat after typing 'trump':
Unfortunately, each of the books in Trump's series is typically cataloged with just Trump as the short title (245 $a). There is also an author/title authority record for Trump, Donald, ‡d 1946- ‡t Trump. ‡l Russian. This is virtually the same pattern we see for works like Shakespeare's Macbeth, which allows our FRBR work-set algorithm to ignore the wide variety of subtitles and pull many of the Macbeth's together as a single work. Trump's books, however, are clearly different works and should not be brought together.
The problem, from the algorithm's viewpoint, is that the main author and short title are the same for these different works. This causes other problems, for example when searching for Donald Trump in OCLC's cataloging interface Connexion, you get a long list, many of which are just identified with the title Trump:.
Here's a list of things we might do:
- Always use the full title for matching. This would mistakenly break up many thousands of works.
- Use the full title if the short title is a single word. This would break 'Macbeth'.
- Have a table of over-rides so we can ignore that particular author/title authority record. This might be the easiest to implement, but seems sort of messy.
- Add a rule that when the short title ends in ':', use the full title. This might work, but we'd have to try it and see what matches it affects. Also, punctuation use is notoriously inconsistent.
- Get catalogers to ensure that the main entry/title combination is relatively unique to the work. I think this has some merit, but sounds like an uphill battle.
- Add authority records for each book in the series. This would work and is consistent with the idea of extending the authority file to control the grouping of works. It would require a fair amount of work, though.
- Allow people to note problems like this in the (forthcoming) WorldCat Wiki, letting them to override the machine groupings.
Anyone think of other approaches?
--Th
It seems to me that these titles might behave better if 'Trump' was considered a series title. This would shift the subtitle to primary position and theoretically clear up the confusion.
Posted by: F T K | September 15, 2005 at 12:42