Latest SQL Server Analysis Patch

Does anyone knows what is the latest SQL Server 2000
Analysis Services patch and how can I get it (Web address)
Thanks.That I know of? 8.0.921.0. http://support.microsoft.com/?kbid=838004
(Sorry, you can't download the patch without talking to Microsoft Product
Support Services. Follow the instructions in the article.)
Aaron Bertrand
SQL Server MVP
http://www.aspfaq.com/
"James" <anonymous@.discussions.microsoft.com> wrote in message
news:1002a01c40d33$7c0a4ec0$a101280a@.phx
.gbl...
> Does anyone knows what is the latest SQL Server 2000
> Analysis Services patch and how can I get it (Web address)
> Thanks.|||Thanks for the info.........
Is there a website where there is the list of all patches?
Thanks.

>--Original Message--
>That I know of? 8.0.921.0.
http://support.microsoft.com/?kbid=838004
>(Sorry, you can't download the patch without talking to
Microsoft Product
>Support Services. Follow the instructions in the
article.)
>--
>Aaron Bertrand
>SQL Server MVP
>http://www.aspfaq.com/
>
>"James" <anonymous@.discussions.microsoft.com> wrote in
message
> news:1002a01c40d33$7c0a4ec0$a101280a@.phx
.gbl...
address)
>
>.
>|||I don't know of an official one. I try to keep this one up to date:
http://www.aspfaq.com/2160
Aaron Bertrand
SQL Server MVP
http://www.aspfaq.com/
"James" <anonymous@.discussions.microsoft.com> wrote in message
news:b8ee01c40dbe$b20e5e90$a001280a@.phx.gbl...
> Thanks for the info.........
> Is there a website where there is the list of all patches?
> Thanks.
>
> http://support.microsoft.com/?kbid=838004
> Microsoft Product
> article.)
> message
> address)

Latent Semantic Analysis (LSA)

Does SQL 2005 offer any tools to perform Latent Semantic Analysis on large data sets? Say I have millions of daily search queries and I'd like to link queries to one another based on semantic content with a goal of mapping them to larger "categories" . One way of doing this is to compare word frequency and proximity to construct a semantic "weight space". This is the approach generally used in LSA.

Anyone using SQL 2005 for this? Anyone see a way it could be done?

We do have the text mining transforms in Integration Services for determining words and phrases from a corpus of documents. We don't have anything specific to LSA, but if you have categories and run a logistic regression you may be able to pull the coefficients to get the information you need. For an example, see http://www.sqlserverdatamining.com/DMCommunity/TipsNTricks/1509.aspx. If all else fails, you can write custom algorithms that plug in to the architecture using either C++ or any .NET language.|||

Had a look at the Text Mining tutorial. It appears to me that the term extraction and term lookup transforms destroy the notion of proximity between the words in the phrase (seems to just create a frequency count of terms). I believe proximity is the key to deriving semantic meaning using techniques like LSA. I might be missing something, though. Can you see a way of using the transforms while maintaining this info?

FYI, here are some good LSA (LSI) links:

http://lsi.research.telcordia.com/lsi/papers/JASIS90.pdf

http://en.wikipedia.org/wiki/Latent_semantic_indexing

|||

You're right, the transforms do eliminate any proximity, except that our transforms extract key term and phrases. That is, it will extract "Data Mining" as a single entity rather than necessarity seperating into "Data" and "Mining". You can specify the maximum phrase length in the options for the transform.

I could imagine that you could break up these phrases with a post process and indicate their proximity - I don't know if that would help though - you would still lose the proximity between the phrases an dothe rowrds and phrases.

Larger Scale SSAS 2005 deployment configuration options...

I'm wondering if there is anything like

Microsoft SQL Server 2000 Analysis Services Operations Guide.doc

Microsoft SQL Server 2000 Analysis Services Performance Guide.doc

for SSAS 2005. I'm working on a 22 Gig cube (that is size with no aggs) with a 32-Gig RAM Server and want to do more tuning - I should be able to get the whole thing in RAM, no? Shouldn't that make it really fast? I'm particularly interested in all of the Analysis Services Properties for SSAS.

What does "-3000" mean for several of the properties? I was thinking that raising many of the max RAM settings could be good, but if they are based on percentages, then that would scale with the server and wouldn't need tuning. There seems to be extremely little info. I see the basic info on 3 or 4 of the RAM settings with the percentages, I want much more than that.

I've looked around and read/skimmed 3 books, but few go into how to deploy to 32 Gigs of RAM / 8 processors and get the best use of the system...

So I'm looking either for direction on a paper or book I've missed or for specific help - both are good. Feel free to mention property changes you've implemented and that worked for you here!

Some information for you:

Server configuration properties docuement is found here: http://download.microsoft.com/download/4/f/8/4f8f2dc9-a9a7-4b68-98cb-163482c95e0b/SSASProperties.doc

Project REAL implementation references have a lot of useful information: http://www.microsoft.com/sql/solutions/bi/projectreal.mspx

Edward.
--
This posting is provided "AS IS" with no warranties, and confers no rights.