[MarkLogic Dev General] Determining stems for proper nouns?

David Sewell dsewell at virginia.edu
Sat Mar 17 16:44:17 PDT 2012


Some more experimentation does indicate that there is an internal 
dictionary of proper nouns used to determine stemming. Looking for 
capitalized words ending in "s" that are treated as plurals and stemmed 
without terminal "s", i.e.

cts:words("a", "document")[matches(., '^[A-Z][a-z]+s$')][cts:stem(.) ne .]

yields (for my data) around 1100 entries, from "Abbotts" to "Aztecs" 
through "Methodists" and "Unitarians" to "Yankees", "Youngs", and 
"Zanes".

Whereas the list of capitalized words ending in "s" that stem only to 
themselves,

cts:words("a", "document")[matches(., '^[A-Z][a-z]+s$')][cts:stem(.) eq .]

runs to 26,000 entries, most of them common nouns, but many unusual or 
misspelled proper nouns from "Aanabaptists" to "Zweerts". The poor 
Quakers were just plain overlooked, I guess. :-)

Thanks, Mike and Geert, for suggestions on refining search results.

DAvid

On Sat, 17 Mar 2012, Mary Holstege wrote:

> On Fri, 16 Mar 2012 19:46:48 -0700, David Sewell <dsewell at virginia.edu> wrote:
>
>>
>> If I had some clue as to the set of words like "Quakers" and "Whigs"
>> that do not stem to singular nouns, I could create a custom dictionary
>> to handle such cases. Are MarkLogic's decisions here based on an
>> internal dictionary? algorithms? both?
>
> Both, but unfortunately I can't tell you what is in that dictionary,
> or the exact circumstances under which the rules get applied because
> stemming is licensed from Inxight (or one of their successors or
> assigns) and we don't have a lot in the way of details.
>
> I think your options are to either run a little test over the word
> lexicon to determine which words need special handling in a custom
> dictionary, and maybe repeat this experiment from time to time to
> see if it needs adjustment, or to accept the lack of precision
> and run case-insensitive.
>
> Sorry I can't be more help here.
>
> //Mary
>
> Mary Holstege
> Principal Engineer
> Mark Logic Corporation
> _______________________________________________
> General mailing list
> General at developer.marklogic.com
> http://developer.marklogic.com/mailman/listinfo/general
>

-- 
David Sewell, Editorial and Technical Manager
ROTUNDA, The University of Virginia Press
PO Box 400314, Charlottesville, VA 22904-4314 USA
Email: dsewell at virginia.edu   Tel: +1 434 924 9973
Web: http://rotunda.upress.virginia.edu/


More information about the General mailing list