how can rdbms convert information into knowledge?

In order to answer this, let's first define our terms. I went to 
http:\\www.dictionary.com and did a quick lookup of "information" and
"knowledge". The parts of the definitions the pertain to a question
about databases follow:

Information:
A collection of facts or data: statistical information.
Computer Science. Processed, stored, or transmitted data.

Knowledge:
The sum or range of what has been perceived, discovered, or learned.
Familiarity, awareness, or understanding gained through experience or
study.

Based on the definitions above, knowledge cannot be achieved without a
way to collect, manipulate and summarize information. A Database
Management System (DBMS) is, in its simplest terms, an organized place
to collect information. This  by itself does not get us to knowledge.
At this point, it is still just a big box of random information.

By accessing this data using a query language, like SQL, or a report
generator, we have the ability to organize, summarize and display that
information. This gets us closer, but still only provides half of our
definition of knowledge.

A Relational Database Management System (RDBMS) allows us to create
relationships between random groups of data. It allows us to
experiment with and interact with the information in different ways
and at a much deeper level than is possible with summarized lists.
This ability to experiment with, study and reorganize our information
can eventually lead to the experience and understanding we call
knowledge.

So, the answer is that an RDBMS cannot convert information into
knowledge, because it is not actually capable of "knowing" or
"understanding" what it contains. It is, however, an extremely capable
tool that we, as "aware" beings, can use to turn information into
knowledge.

- Hammer

---

Clarification of Answer by hammer-ga on 09 Nov 2002 06:00 PST

Oops! I got the slashes backwards in my link:

http://www.dictionary.com

---

Request for Answer Clarification by dd_friend-ga on 09 Nov 2002 18:57 PST

Dear Hammer Sir:
I appreciate your prompt answer.
Nevertheless Mr. curios7 rises important questions
that do deserve a clarification.
 I want to understand what is the normalization
imposed by rdbms and how it is different from - Bussines Objects, for
example.
An example of a specific task or problem - coped by both strategies,
would  help me a lot to understand the process.

---

Clarification of Answer by hammer-ga on 10 Nov 2002 06:12 PST

Normalization is about optimizing your database design to reduce
repetition of information, increase efficiency and improve data
integrity. It can be a good thing to do in general, but, in specific,
the ability to work well with your information should never be
sacrificed to normalization. What good is an extremely efficient
database that can't actually deal with your information? Here is a
tutorial on normalizing a relational database:
http://www.flws.com.au/showusyourcode/codeLib/code/normalize.asp

Curious7's comments depart from your question into the general realm
of acquiring knowledge. While it is true that a rdbms is not
appropriate for all information and all knowledge, that was not your
question. I do not agree with the statement that information contained
in a database is only useful to a database administrator. A good
database designer does not "shoehorn" data into a badly normalized
design. A good designer designs the database to fit the data, not
vice-versa. My answer to your question is offered, not as a
"theorist", but as someone who has been designing real world databases
for use by non-database administrators for the last ten years.

Bottom line: A database does NOT impose normalization. A database
designer imposes normalization.

- Hammer

How can you define knowledge so diffusely?
The sum or range ... (it is so wide it can be everything!!).
Something should be said about the relevancy of facts, information
or data that we use when we base our judgement on it.
Isnt that the way we differ between someone who knows from someone who does'nt?

My view is that the formal structure which an rdbms forces its
database designers to impose when "normalizing" recalcitrant data is
simply one way of organizing certain kinds of information.  The
process of shoehorning data into an rdbms rarely leads to knowledge or
insight that is useful to anyone who's not a database administrator.

On the other hand, more broadly useful knowledge often comes from
identifying patterns or trends within structured information, often by
asking and answering questions. The tools which best help people do
this (e.g., data marts, OLAP, MicroStrategy, Business Objects, SAS)
often have to implicitly or explicitly undo the data normalization so
carefully established to satisfy the rdbms theorists.

What's more, much (perhaps most) useful knowledge involves
unstructured information, approximate patterns, shades of meaning,
intuitions, estimates,  metaphors and nuances which fit poorly, if at
all, into an rdbms.  Using an rdbms for "knowledge management"
(whatever that means) is like the kid with a hammer using the world as
a nail.

If you want to convert information into knowledge, even in a narrow
domain, an rdbms typically needs other tools. And it certainly doesn't
perform the conversion for you.  As the domain broadens, and the
information becomes less structured, you are far less likely find
useful knowledge through an rdbms than though a tool like the
Web+Google.

- curious7

dd-friend,

I would echo hammer's statement here.

As a Computer Programmer/Systems Analyst for more than 15 years, I
have worked with and designed numerous different kinds of databases.
The Relational Database, in my opinion, is by far the most useful. It
allows the data to be stored more efficiently by eliminating most
duplication/redundancy (via the process of normalization, which breaks
information down into its basic elements). It also makes it fairly
easy to pull together the desired elements (such as name, address, all
orders/invoices, etc) using a small number of keys (such as customer
number or social security number).

A well-designed database will not require the data to be
"shoe-horned". If major "shoe-horning" is required, the database
design itself is seriously flawed and needs to be re-engineered.

A well-designed database, when accompanied by user-friendly, intuitive
query tools (such as SQL) that can be used by non-computer
specialists, provides a wealth of knowledge that is useful to the
manager, salesperson, information analyst or other person making the
query.

In my opinion, hammer has given the answers (and there are two of
them) to your question "How can rdbms convert information into
knowledge?":
- a well-designed rdbms provides a good, workable framework for
converting information into knowledge;
- however, the rdbms does not itself convert the information into
knowledge -- the query tools do that.

I hope that my comments assist you in better understanding the answer
to your question.

Regards,

aceresearcher

I came across the following site which tells about the concepts of
RDBMS, OLTP, data warehousing, and OLAP in simple terms. Will be
helpful in understanding the RDBMS role in information technology and
knowledge management.

http://philip.greenspun.com/wtr/data-warehousing.html

dd-friend: rac's link gives an excellent example of how designing a
database for efficient transactions (a common goal for an rdbms) can
conflict with designing a database for asking questions.  Questions
help turn information into knowledge.
   For Flash demos of Business Objects products, try
http://www.businessobjects.com/forms/demos_login.asp

Hammer:  I confess to departing from the question a bit to indulge a
pet peeve.  The "knowledge management" buzzword in the Subject also
seemed to warrant a broader view.
   I agree strongly that being able to "work well" with the data
should take priority over normalization.
   Like you, "I do not agree with the statement that information
contained
in a database is only useful to a database administrator."  I wasn't
trying to make that statement; only that the process of normalization
itself rarely yields knowledge as useful as do query tools/warehouses,
which often denormalize data to work well.
   I'm a bit puzzled by your statement that "A Relational Database
Management System (RDBMS) allows us to create relationships between
random groups of data."  We may _observe_ relationships, or
_formalize_ them when designing a database, but if we're creating
relationships rather than discovering them, it's called "social
science."  (Just kidding!)  Besides, if the groups are  random, can
they have relationships at all?

aceresearcher:  A person who considers SQL a "user-friendly, intuitive
query tool" must work with different users than I do, or maybe they're
an ace researcher.  Googling for "sql intuitive" turns up ample
results describing how SQL becomes intuitive if you just add "product
x."  Or you can read randomly from Joe Celko.....

Peace^^Paul