Skip to main content.
home | support | download

Back to List Archive

Meatdata Searching: Hyphen gives bad results

From: Jens Olscher <JOLSC(at)not-real.doh.health.nsw.gov.au>
Date: Mon Oct 25 1999 - 03:39:11 GMT
Dear Fellow Swishers!

I know how to index metadata and how to generally search on them. But using some Dublin Core tags in our metadata created a problem for me that you might be able to help....hopefully!

Here is a sample tag of a document that gets indexed:

<META NAME="DC.Date.Created" SCHEME="ISO8601" LANG="en" CONTENT="1999-07-07">
<META NAME="AGLS.Date.ValidTo" SCHEME="ISO8601" LANG="en" CONTENT="2000-01-07">

Please note the hyphens in the dates, eg 1999-07-07.

If I search with this command: 
swish-e -w "DC.Date.Created = 1999-07*" -f index.ndx
the above file WILL NOT be returned. There will be NO RESULTS whatsoever with this search. 

If however I search using the much broader term: 
swish-e -w "DC.Date.Created = 1999*" -f index.ndx
the above document WILL be returned, but unfortunately together with 1000s others that have been created in 1999....

So my suspicion is that the hyphen screws up the results. Trying to escape it as in "1999/-07*" or "1999\-07*" or simply leaving it out as in "199907*" or even "1999*07*" all failed with no results returned at all (last one had too many results, same as 1999*).

Has anybody else encountered the problem - and perhaps solved? Could anyone offer a direction where I could poke around trying to solve this problem? It would be very very neat if swish-e could pinpoint which documents have been created last month or even exactly yesterday, using standard Dublin Core metadata!

If this problem has not and can not be solved with the current version, I would strongly suggest this to be on top of the bug fix list.

Thanks for any help offered!

Jens Olscher
Webmaster
NSW Department of Health
Received on Sun Oct 24 20:39:30 1999