The default Solr stopwords.txt file is empty, so SOMEBODY created that
non-empty stop words file.
The StopFilterFactory token filter in the field type analyzer controls stop
word processing. You can remove that step entirely, or different field types
can reference different stop word files, or some field type analyzers can
use the stop filter and some would not have it. This does mean that you
would have to use different field types for fields that want different stop
word processing.
-- Jack Krupansky
-----Original Message-----
From: Stavros Delisavas
Sent: Thursday, October 17, 2013 3:27 AM
To: solr-user@lucene.apache.org
Subject: Re: Local Solr and Webserver-Solr act differently ("and" treated
like "or")
Thank you,
I found the file with the stopwords and noticed that my local file is
empty (comments only) and the one on my webserver has a big list of
english stopwords. That seems to be the problem.
I think in general it is a good idea to use stopwords for random
searches, but it is not usefull in my special case. Is there a way to
(de)activate stopwords query-wise? Like I would like to ignore stopwords
when searching in titles but I would like to use stopwords when users do
a fulltext-search on whole articles, etc.
Thanks again,
Stavros
On 17.10.2013 09:13, Upayavira wrote:
Stopwords are small words such as "and", "the" or "is",that we might
choose to exclude from our documents and queries because they are such
common terms. Once you have stripped stop words from your above query,
all that is left is the word "wild", or so is being suggested.
Somewhere in your config, close to solr config.xml, you will find a file
called something like stopwords.txt. Compare these files between your
two systems.
Upayavira
On Thu, Oct 17, 2013, at 07:18 AM, Stavros Delsiavas wrote:
Unfortunatly, I don't really know what stopwords are. I would like it to
not ignore any words of my query.
How/Where can I change this stopwords-behaviour?
Am 16.10.2013 23:45, schrieb Jack Krupansky:
So, the stopwords.txt file is different between the two systems - the
first has stop words but the second does not. Did you expect stop
words to be removed, or not?
-- Jack Krupansky
-----Original Message----- From: Stavros Delsiavas
Sent: Wednesday, October 16, 2013 5:02 PM
To: solr-user@lucene.apache.org
Subject: Re: Local Solr and Webserver-Solr act differently ("and"
treated like "or")
Okay I understand,
here's the rawquerystring. It was at about line 3000:
<lst name="debug">
<str name="rawquerystring">title:(into AND the AND wild*)</str>
<str name="querystring">title:(into AND the AND wild*)</str>
<str name="parsedquery">+title:wild*</str>
<str name="parsedquery_toString">+title:wild*</str>
At this place the debug output DOES differ from the one on my local
system. But I don't understand why...
This is the local debug output:
<lst name="debug">
<str name="rawquerystring">title:(into AND the AND wild*)</str>
<str name="querystring">title:(into AND the AND wild*)</str>
<str name="parsedquery">+title:into +title:the +title:wild*</str>
<str name="parsedquery_toString">+title:into +title:the
+title:wild*</str>
Why is that? Any ideas?
Am 16.10.2013 21:03, schrieb Shawn Heisey:
On 10/16/2013 4:46 AM, Stavros Delisavas wrote:
My local solr gives me:
http://pastebin.com/Q6d9dFmZ
and my webserver this:
http://pastebin.com/q87WEjVA
I copied only the first few hundret lines (of more than 8000) because
the webserver output was to big even for pastebin.
On 16.10.2013 12:27, Erik Hatcher wrote:
What does the debug output say from debugQuery=true say between the
two?
What's really needed here is the first part of the <debug> section,
which has rawquerystring, querystring, parsedquery, and
parsedquery_toString. The info from your local solr has this part, but
what you pasted from the webserver one didn't include those parts,
because it's further down than the first few hundred lines.
Thanks,
Shawn
