March 3, 2005 at 11:10 am
Dealing with Full Text indexing, i have a strange issue. The index is set to Neutral and when i do a contains or containstable and search for dates in this format '1-Feb-2005' which exists in html form that is the FullText column. Anytime the day number is less than 10 i get results like i did '*-Feb-2005' a few docs rank higher that don't have '1-Feb-2005' but '2-Feb-2005' or '*-Feb-2005' . Now if i change it to '10-Feb-2005' I get the results that is expected. Has anyone encountered this situation? Any help would be appreciated. Thank you.
Jim Kitterman
March 3, 2005 at 11:39 pm
Jim,
The most important think to understand but interputing "funky" SQL FTS results to know what OS platform your SQL Server is installed on, primarly because the word breaking dll that is used by SQL Server 7.0 and 2000 is dependent upon the OS-supplied wordbreaker dll. Could you reply with the full output of -- SELECT @@version -- as this will provide the necessary information.
Additionally it is important to know what datatype (text, ntext or image) you have your HTML form stored in for your FullText column. Could you provide the sp_help <your-table-name> output as well?
Finally, it is helpful to know exactly what is the text or punctuation (html metatags) that are surround or are in contact with your search phrase '1-Feb-2005' and as your using the Neutral "Language for Word Breaker" have you removed any noise words, single letters or single digits from noise.dat (the neutral noise word file)?
With this information, I should be able to explain the "funky" FTS results.
Thanks,
John
SQL Full Text Search Blog
http://spaces.msn.com/members/jtkane/
John T. Kane
March 4, 2005 at 9:26 am
Microsoft SQL Server 2000 - 8.00.760 (Intel X86) Dec 17 2002 14:22:05 Copyright (c) 1988-2003 Microsoft Corporation Standard Edition on Windows NT 5.0 (Build 2195: Service Pack 4)
The column is of type TEXT. The tag adjacent to the text opening <font size="3"> tag. I have not modified the Noise word file yet.
sp_help Column info:SX_OCR_Text text no 16 no (n/a) (n/a) SQL_Latin1_General_CP1_CI_AS
Thanks for the link to the BLOG it will be invaluable to me!
JimK
March 4, 2005 at 11:03 am
I checked the noise file and I think I found the culprate. I removed the line with the numbers. 1 2 3 4 5 6 7 8 9 0. Repopulation and testing seems to have fixed the results.
Thanks for your help John.
JimK
Viewing 4 posts - 1 through 3 (of 3 total)
You must be logged in to reply to this topic. Login to reply