2012年3月27日星期二

Full text search txt or doc files?

Hi,
I'm putting together a system and one of the requirements is to have a
searchable CV function.
I've got all the code to load the files on to the image fields, I've indexed
and got it mostly working.
Before I go to far down the road what is your opinion on having txt files
instead of doc files to search? The SQL seems to be more flexible on
searches rather than on the binary files and the index files themselves are
smaller.
My main concern is I'd have to do the text conversion automatically, any
pointers on this?
Does anyone have any views on the best way to go about this
Many thanks for any help you can give
Jim FlorenceHave you tried asking this in the sqlserver.fulltext news group?
As long as all expected binary types are supported by appropriate IFilters
on your server, using binary data will be just fine - AFAIK.
ML
http://milambda.blogspot.com/|||Thanks for the pointer, embarrassingly I didn't see it existed :0/
I just noticed on trying one query with "select like" it would only work on
varchar fields and failed on the blob, could well be mistaken though.
Thanks again
Jim
"ML" <ML@.discussions.microsoft.com> wrote in message
news:45DBD971-2EE7-42C5-AD3A-C3F542B4F8DA@.microsoft.com...
> Have you tried asking this in the sqlserver.fulltext news group?
> As long as all expected binary types are supported by appropriate IFilters
> on your server, using binary data will be just fine - AFAIK.
>
> ML
> --
> http://milambda.blogspot.com/|||The LIKE operatoron binary values works if the searched string is cast to
binary or varbinary, but only if the binary column contains a true
representation of the text, which is not how Word documents are represented.
What exactly did you try?
BTW: with FTI you really shouldn't use the LIKE operator - FTS functions are
more efficient in this case.
ML
http://milambda.blogspot.com/|||Thaks for thatr I knew I'd have missed something obvious.
I'm not entierly sure what the difference is between FTI and FTS, I thought
they were one and the same
Jim
"ML" <ML@.discussions.microsoft.com> wrote in message
news:29D91E26-BE4C-4832-B888-8FA47423C793@.microsoft.com...
> The LIKE operatoron binary values works if the searched string is cast to
> binary or varbinary, but only if the binary column contains a true
> representation of the text, which is not how Word documents are
> represented.
> What exactly did you try?
> BTW: with FTI you really shouldn't use the LIKE operator - FTS functions
> are
> more efficient in this case.
>
> ML
> --
> http://milambda.blogspot.com/|||Jim,
I am trying to do something similar. I have a list of .doc files that I am
trying to do i full text search on. I would really appreciate some help on
how to do this. Some sql code would be great. Thanks
qtrsteve@.yahoo.com|||Sorry for a late reply.
FTS = full-text search
FTI = full-text index
FTS uses FTI.
I was referring to FTS functions: contains, containstable, frettext,
freetexttable. FTS functions are far more efficient than the LIKE operator,
or at least the processing overhead of FTS is pretty constant compared to th
e
overhead when using the LIKE operator.
ML
http://milambda.blogspot.com/sql

没有评论:

发表评论