Full text query contained only ignored words

Hello All,
Please explain this Fulltext behavior to me.
We have an asp application that is using sql that returns the above error
when you search for the word can't. If you search for "don't" or "Doesn't"
or pretty much anything else with an apostrophe, results are returned.
The same thing happens when you search for a decimal number. If I search
for 4.0, I get the above error. If I search for 4.00.0 I will get results.
Thanks in advance!
Jake
"GitarJake" <spamaintme@.oz.net> wrote in message
news:C0vRe.101151$Ep.81169@.lakeread02...
> Hello All,
> Please explain this Fulltext behavior to me.
> We have an asp application that is using sql that returns the above error
> when you search for the word can't. If you search for "don't" or
> "Doesn't" or pretty much anything else with an apostrophe, results are
> returned.
> The same thing happens when you search for a decimal number. If I search
> for 4.0, I get the above error. If I search for 4.00.0 I will get
> results.
Check the noise.X file in your FTDATA\Config folder (where X is the
extension related to your word breaker language - eng for English, enu for
US English, etc).
4 is in the list. I'm guessing that when 4.0 is passed FTS treats it a
number and so finds it in the noise word file, whereas when 4.00.0 is used
it's obviously not a number and so does a search for it as is.
Can is a word in the noise word list too. I'm guessing FTS splits can't into
can and not internally, and errors because can is in the list.
What you could do is empty the noise word file - leave only a single line in
it with a single space on it. Then run a full population. This has the
effect of not erroring on any words (although some characters that are used
as word breakers will still throw the noise words error, you'll have to
filter these out in your query builder).
Dan
|||You might want to review the following
http://www.indexserverfaq.com/noise.htm
Hilary Cotter
Looking for a SQL Server replication book?
http://www.nwsu.com/0974973602.html
Looking for a FAQ on Indexing Services/SQL FTS
http://www.indexserverfaq.com
"GitarJake" <spamaintme@.oz.net> wrote in message
news:C0vRe.101151$Ep.81169@.lakeread02...
> Hello All,
> Please explain this Fulltext behavior to me.
> We have an asp application that is using sql that returns the above error
> when you search for the word can't. If you search for "don't" or
"Doesn't"
> or pretty much anything else with an apostrophe, results are returned.
> The same thing happens when you search for a decimal number. If I search
> for 4.0, I get the above error. If I search for 4.00.0 I will get
results.
>
> Thanks in advance!
> Jake
>
|||Hi Daniel,
Thanks for responding.
I have removed "can" from both the fulltext noise.enu word list in
ftdata\config; and the Indexing noise.enu. The wordbreaker in FTI is
neutral. I completely deleted the FT catalogs and manually recreated and
repopulated them. The results were the same.
Then I deleted the contents FTI's noise.enu except for a space and now, the
search simply does not return results. It seems like I can search for the
word can't on any other search engine but ours. Below is the query we are
passing using just the word *can't*. Can you tell if there is anything
within the query that might be better able to produce results?
select ARTICLE_ID, ARTICLE_STITLE, ARTICLE_LTITLE, ARTICLE_SUMMARY
,ARTICLE_STATUS, ARTICLE_APPROVED, DEPART_REF, USER_REF_OWNER,
USER_REF_EDITOR, ARTICLE_REVIEW_DATE, ARTICLE_EXPIRE, RANK,
PUBLIC_DIVISION_REF From ARTICLES as A, containstable(ARTICLES, *, '
"can''t" ') as AST where (LANGUAGE_REF= 1 and ARTICLE_DISABLE_SEARCH <> 1 )
and AST.[KEY] = A.ARTICLE_ID order by RANK desc
MIS query: "can''t"
TIA
Jake
"Daniel Crichton" <msnews@.worldofspack.co.uk> wrote in message
news:%23BYyg8vrFHA.3096@.TK2MSFTNGP15.phx.gbl...
> "GitarJake" <spamaintme@.oz.net> wrote in message
> news:C0vRe.101151$Ep.81169@.lakeread02...
> Check the noise.X file in your FTDATA\Config folder (where X is the
> extension related to your word breaker language - eng for English, enu for
> US English, etc).
> 4 is in the list. I'm guessing that when 4.0 is passed FTS treats it a
> number and so finds it in the noise word file, whereas when 4.00.0 is used
> it's obviously not a number and so does a search for it as is.
> Can is a word in the noise word list too. I'm guessing FTS splits can't
> into can and not internally, and errors because can is in the list.
> What you could do is empty the noise word file - leave only a single line
> in it with a single space on it. Then run a full population. This has the
> effect of not erroring on any words (although some characters that are
> used as word breakers will still throw the noise words error, you'll have
> to filter these out in your query builder).
> Dan
>
|||To test this I added "Can't you see?" to the end of a book title in the pubs
database. I then created a new catalog on the title column, ran a full
population, and then ran this:
select * from titles where contains(title,'can''t')
and up popped the row I'd added the phrase to. I then tried this:
select a.* from titles a ,containstable(titles,*,'can''t') b where
a.title_id = b.[key]
select a.* from titles a ,containstable(titles,*,'"can''t"') b where
a.title_id = b.[key]
and again, I get the same row back from both versions. So it's not a problem
with the apostrophe, nor an issue with the double quotes.
The only other thing in your query that will prevent the row from appearing
will be if ARTICLE_DISABLE_SEARCH = 1 or LANGUAGE_REF <> 1. Have you tried
running the query with these clauses removed just in case?
Dan
GitarJake wrote on Sun, 4 Sep 2005 12:33:07 -0400:
[vbcol=seagreen]
> Hi Daniel,
> Thanks for responding.
> I have removed "can" from both the fulltext noise.enu word list in
> ftdata\config; and the Indexing noise.enu. The wordbreaker in FTI is
> neutral. I completely deleted the FT catalogs and manually recreated and
> repopulated them. The results were the same.
> Then I deleted the contents FTI's noise.enu except for a space and now,
> the search simply does not return results. It seems like I can search for
> the word can't on any other search engine but ours. Below is the query we
> are passing using just the word *can't*. Can you tell if there is
> anything within the query that might be better able to produce results?
> select ARTICLE_ID, ARTICLE_STITLE, ARTICLE_LTITLE, ARTICLE_SUMMARY
> ,ARTICLE_STATUS, ARTICLE_APPROVED, DEPART_REF, USER_REF_OWNER,
> USER_REF_EDITOR, ARTICLE_REVIEW_DATE, ARTICLE_EXPIRE, RANK,
> PUBLIC_DIVISION_REF From ARTICLES as A, containstable(ARTICLES, *, '
> "can''t" ') as AST where (LANGUAGE_REF= 1 and ARTICLE_DISABLE_SEARCH <>
> 1 ) and AST.[KEY] = A.ARTICLE_ID order by RANK desc
> MIS query: "can''t"
> TIA
> Jake
> "Daniel Crichton" <msnews@.worldofspack.co.uk> wrote in message news:%23BYyg8vrFHA.3096@.TK2MSFTNGP15.phx.gbl...