- 10/21/2012 11:20:56 PM
That's true in large part because most data users and professionald lack foundation knowledge. Most of the concepts and terminology are ad-hoc and there is poor understanding of levels of representation, logic, data model and so on.
For example, data model is used interchangeably with logical model. But the concept of a data model was invented by Codd to represent essentially a mechanism by which we convert business models to logical models: structure, manipulation and integrity (see my reply to Beth). The relational model is a data model; there were claims that the hierarchic (graph) model and the network (CODASYL) model are data models, but they are insufficiently defined and lack a theoretical foundation equivalent to the RM. The so-called object model and XML model are trying to bring those back and regress us to those failed approaches. Most of the NoSQL people ignore the necessity of a data model altogether and thing they can get away without an explicit one.
My point of the usefulness of structure is precisely targetted at that.
- 10/21/2012 11:06:21 PM
To give you an idea: business models are INFORMAL and expressed in business terms (customers, shipments, accounts, business rules, etc.). Logical models are the FORMAL database representations of business and are expressed in db terms (R-tables, rows, columns, integrity constraints, etc.)
- by SethBreedlove, Data Doctor
- 10/21/2012 9:58:09 PM
When it comes to models, what is one man's treasure is another garbages. Same as data models. One's business model doesn't apply to another and would be considered unstructured to another.
- by BethSchultz, Blogger
- 10/21/2012 9:00:26 PM
Thanks Fabian -- @mnorth's explanation helped me get my mind around business & logical modeling in regards to databases, and I'll await your further insight on E/R modeling, etc. And, of course, I'm interested in hearing more about your approach to business modeling.
- 10/20/2012 12:48:01 AM
Indeed, exactly right, see my reply to mdmconsult.
There is a lot of delusion going on in the BigData/NoSQL community that they can achieve the objectives of database management without doing the upfront thinking, modeling and database design. It is OK to try mine data and come up with ideas, but once you do, what then?
In this context I refer readers to my question in the post: Why does Twitter annotate with metadata and what does that mean? That explains why they're using mySQL and not NoSQL.
- 10/20/2012 12:41:02 AM
But an important point is that what you can do with the data is determined by how the data is structured and how accurately is represents the reality which your analysis si supposed to predict. Get the latter two wrong and the analysis will be difficult, impossible or wrong.
- 10/20/2012 12:37:30 AM
Don't worry, this is what this blog is about. You will learn.
Pls note the important distinction between a business model that comes first and a logical model that represents it in the database. If you don't have the former, what are you going to represent in the database?
You are right that that's what people should do, but many don't. That has to do with the lack of knowledge of and appreciation for data fundamentals, which is due to the almost exclusive emphasis on tools, a deficiency in the education system.
As to how to model: there are software tools, or you can do it with pen and paper. What is important si to know (1) the business (2) modeling well.
Hang in there.