Sql Server 2008 “Contains” search searching for numeric value returns all records with a different number of the same length?

Question

I have a search where users can enter any value, sometimes these are legitimately just numeric. When searching for

Select * FROM Updates WHERE contains(Remarks, '"10010234331"')

it will return unrelated results that have numbers in of the same length, ie one match for the above is

"PO input differently on orders, refs are: 

1001024894 
10010248940"

As you can see, the search is not a substring of either of these. Any ideas how to tell it to not just guess?

Answer 1

Right.

As unbelievable as this sounds, it turns out that if the full-text index "language for word-breaker" is set to anything other than "English", the full-text index will only index the first 5 numbers of a number string. This means that the above is returning as a match because they both start "10010".

"French, "German", "Hebrew", even "Neutral" was returning incorrect results, only "English" returns only matches for the whole string.

create table wtFTI( taskid int not null  ,  remarks text, constraint   [PK__wtFTI] primary key(taskid))
insert into wtfti
values(3513792, 'Remarks: 1001019658  was cancelled 26/08')
GO

CREATE FULLTEXT CATALOG [TaskRemarks]WITH ACCENT_SENSITIVITY = OFF

GO

CREATE FULLTEXT INDEX ON [dbo].[wtFTI] KEY INDEX [PK__wtFTI] ON ([TaskRemarks]) WITH (CHANGE_TRACKING = AUTO, STOPLIST = SYSTEM)
GO
ALTER FULLTEXT INDEX ON [dbo].[wtFTI] ADD ([remarks]LANGUAGE 'French')
GO
ALTER FULLTEXT INDEX ON [dbo].[wtFTI] ENABLE
GO


Select * FROM [wtFTI] WHERE contains(Remarks, '"1001019000"')

ALTER FULLTEXT INDEX ON [dbo].[wtFTI] DROP ([remarks]) 
GO
ALTER FULLTEXT INDEX ON [dbo].[wtFTI] ADD ([remarks] LANGUAGE 'English')
GO


Select * FROM [wtFTI] WHERE contains(Remarks, '"1001019000"')