On 21 Sep 2000, at 21:43, David Norris wrote:
> > Take a look at the following two "# Search words:" headers from searching
> > two indexes. Any idea what "precd" is and why it's there?
> precd = precedes
> Looks odd to me. I would expect the two queries to look alike. Perhaps
> Jose can shed some light on it.
> > # Search words: keywords = ( " sodium in that food from " )
> > # Search words: ( keywords = ( ( sodium precd in precd that precd food
> > precd from ) ) )
Dave is right. It means precedes.
This is why...
The search string is preprocessed: First, I locate the phrase
delimiter ". Then I change " sodium in that food " by
sodium precd in precd that precd food. Why I am doing this odd
thing? Because the phrase search is implemented like an and
search. So I have defined the internal operator precd. Thus when
the code finds this new operator it issues and "extended and": it
does a "normal and" with the position counters of the words have to
be consecutives (more or less).
If you do not like this message, we can do two things:
1- Preserve the search string in the search function of swish-e and
show the original one. Anyway, at this moment the original string is
never written because the stopwords are removed from the query.
Of course, if you do not have stopwords or phrases in your query
the search strings should match.
2- Since your cgi script has the original search string, you can use
Let me know what you think
Received on Fri Sep 22 08:40:36 2000