Re: Thesaurus Problem

On Aug 28, 5:35 am, "Hilary Cotter" <hilary.cot...@xxxxxxxxx> wrote:
You don't need to rebuild your catalogs, you do need to restart MSFTESQL,
but don't need to reboot.

What language are you querying in and what language is your content it? Is
it binary content?
--
RelevantNoise.com - dedicated to mining blogs for business intelligence.

Looking for a SQL Server replication book?http://www.nwsu.com/0974973602.html

Looking for a FAQ on Indexing Services/SQL FTShttp://www.indexserverfaq.com"Organic Man" <davidmcmil...@xxxxxxx> wrote in message

news:1188272200.321569.78670@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

On Aug 27, 10:49 pm, "Hilary Cotter" <hilary.cot...@xxxxxxxxx> wrote:
Did you save your thesaurus file with Unicode encoding? After SP 2 it
must
be saved as a Unicode file.

--
RelevantNoise.com - dedicated to mining blogs for business intelligence.

Looking for a SQL Server replication
book?http://www.nwsu.com/0974973602.html

Looking for a FAQ on Indexing Services/SQL
FTShttp://www.indexserverfaq.com"Organic Man" <davidmcmil...@xxxxxxx>
wrote in message

news:1188250168.087879.56150@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

I am trying to get the full-text search thesaurus to work in Sql
Server 2005 Express edition. I live in the USA so I assume the
tx.ENU.xml is the appropriate file to modify. I used Notepad to
modify the tx.ENU.xml file and saved as unicode:

<XML ID="Microsoft Search Thesaurus">

<thesaurus xmlns="x-schema:tsSchema.xml">
<diacritics_sensitive>0</diacritics_sensitive>
<expansion>
<sub>littre</sub>
<sub>leydig</sub>
</expansion>
<replacement>
<pat>NT5</pat>
<pat>W2K</pat>
<sub>Windows 2000</sub>
</replacement>
<expansion>
<sub>run</sub>
<sub>jog</sub>
</expansion>
</thesaurus>

</XML>

I closed Management Studio and reopened to allow MSFTESQL service to
restart. Then I ran these queries:

SELECT FullDocNo
FROM FullDocuments
WHERE CONTAINS(SectionText, 'littre') *** returned 3 rows ***

SELECT FullDocNo
FROM FullDocuments
WHERE CONTAINS(SectionText, 'leydig') *** returned 169 rows ***

SELECT FullDocNo
FROM FullDocuments
WHERE CONTAINS(SectionText, ' FORMSOF(THESAURUS, littre) ') ***
returned 6 rows ***

Thus the Thesaurus does not seem to be working since it should have
returned at least 169 rows. I rebooted my entire system to make sure
Sql Server is starting fresh.

Any help in sorting this out will be greatly appreciated.

Hi Hilary,

Yes, I saved as unicode and double checked to be sure.

Are there any specific instructions for restarting MSFTESQL service,
other than closing and reopening Management Studio?

Does the full-text catalog need to be updated?

I have tried several things; stillno luck so far;

1. I am querying in English. To be safe I made the changes to the
ts.ENS.xml, ts.ENG.xml, and ts.ESN.xml files.

2. The datatype for the SectionText column that I am searching was
ntext. I inherited this db which was originally created in Access and
migrated to Sql 2000. So I changed the datatype to nvarchar(max) to
make it conform to modern standards. Still no luck.

3. I created a new SQL Server 2005 db with full-text search enabled
and moved all the data over from the old one which I believe is still
SQl 2000.

4. I wondered if case mattered in the xml file (since Leydig is in
caps in SectionText), so I altered the xml like this:

<XML ID="Microsoft Search Thesaurus">

<thesaurus xmlns="x-schema:tsSchema.xml">
<diacritics_sensitive>0</diacritics_sensitive>
<expansion>
<sub>Leydig</sub>
<sub>leydig</sub>
<sub>littre</sub>
</expansion>
<replacement>
<pat>NT5</pat>
<pat>W2K</pat>
<sub>Windows 2000</sub>
</replacement>
<expansion>
<sub>run</sub>
<sub>jog</sub>
</expansion>
</thesaurus>

</XML>

5. I wondered if this could be a limitation of Sql Server 2005
Express, so I moved everything over to my Vista machine running Sql
Server 2005 Developer Edition. No luck.

6. I wondered if there was something problematic with my choice of
search terms so tried new ones by changing xml and sql like this:

<expansion>
<sub>vagus</sub>
<sub>pneumogastric</sub>
</expansion>

SELECT FullDocNo
FROM FullDocuments
WHERE CONTAINS(SectionText, 'vagus') *** returned 213 rows ***

SELECT FullDocNo
FROM FullDocuments
WHERE CONTAINS(SectionText, 'pneumogastric') *** returned 514 rows
***

SELECT FullDocNo
FROM FullDocuments
WHERE CONTAINS(SectionText, ' FORMSOF(THESAURUS, vagus) ')
***returned 213 rows ***

So it still does not seem to work.

In case you are wondering, the db contains old antiquated medical
terms that are unfamiliar to modern clinicians and researchers but are
useful when interpreting old medical manuscripts.

I am out of ideas for now. Any thoughts on what is wrong?

.

Relevant Pages

  • Re: Thesaurus Problem
    ... Looking for a SQL Server replication book? ... be saved as a Unicode file. ... FROM FullDocuments ...
    (microsoft.public.sqlserver.fulltext)
  • Re: Thesaurus Problem
    ... you are searching in British or International English its tsENG.xml. ... Looking for a SQL Server replication book? ... FROM FullDocuments ...
    (microsoft.public.sqlserver.fulltext)
  • Re: Thesaurus Problem
    ... sp_configure 'default full-text language' ... RelevantNoise.com - dedicated to mining blogs for business intelligence. ... Looking for a SQL Server replication book?http://www.nwsu.com/0974973602.html ... FROM FullDocuments ...
    (microsoft.public.sqlserver.fulltext)
  • Re: SQL 2008 Full text indexing
    ... that is a major issue to integrate full-text engine into SQL Server. ... One of the side effects is performance improvement since full-text querying is included as operators in Query Execution Plan. ...
    (microsoft.public.sqlserver.fulltext)
  • Re: Thesaurus Problem
    ... be saved as a Unicode file. ... Looking for a SQL Server replication book?http://www.nwsu.com/0974973602.html ... FROM FullDocuments ... Are there any specific instructions for restarting MSFTESQL service, ...
    (microsoft.public.sqlserver.fulltext)
Loading