Re: MVD projection algorithm.

On Wed, 14 Jun 2000 10:51:24 -0700, admin _at_ comp.databases.theory wrote:

>
>The only way to implement what you want is with
>Eigen vectors and would probably take longer to
>do than is feasible for your immediate needs
>
>Tin Blaskovic <tblaskov_at_jagor.srce.hr>
>
>> Hi all!
>>
>> After taking my database design class at the university, I
>> decided to make a small program that would do what most of
>> written exams for such classes ask for - namely that would
>> implement normalization algorithms. Since my class only
>> covered upto BCNF in detail, I was lacking some algorithms
>> for higher order normal forms. After some exhausting
>> Internet surfing and bugging some people through email,
>> I'm left with only one pretty vital algorithm to account
>> for before I can really complete that program.
>> The algorithm I need is that mentioned in the subject,
>> i.e. the algorithm for finding multivalued dependency
>> projections on attribute subsets. I have searched the
>> Internet and all literature that I can get my hands on
>> for this and I didn't find it. If any of you kind souls out
>> there has that algorithm in some book or his lecture notes,
>> I'd appreciate it very much if you could post it or email
>> it to me. I intend to release the program as PD, and if
>> it means anything, you'll get a honorable mention in the
>> credits section, that's all that I can offer :-)
>>
>> TIA,
>> Tin.

Tin,

I've sometimes found that the theory behind an approach lies in an unrelated field and goes under another name entirely, or is considered to be so obvious that information only appears in footnotes.

It's also possible that the theoretical approach does not give results in a reasonable time, and/or adhoc methods are faster and invariably used.

You might want to check out some source code to reverse engineer approaches used: for example, msql, mysql, postgres sql, I believe provide source.

Thanks. Take care, Brian Inglis Calgary, Alberta, Canada

-- 
Brian_Inglis_at_CSi.com 	(Brian dot Inglis at SystematicSw dot ab dot ca)
				use address above to reply