Re: Sensible and NonsenSQL Aspects of the NoSQL Hoopla
Date: Fri, 6 Sep 2013 14:02:19 +0200
Message-ID: <5229c44b$0$1680$e4fe514c_at_dreader35.news.xs4all.nl>
On 2013-09-06 11:22:21 +0000, vldm10 said:
> Dana srijeda, 4. rujna 2013. 18:35:36 UTC+2, korisnik Jan Hidders napisao je:
>
> Hi Jan,
>> Codd explicitly referred to first-order logic in his work, so the >> link>> with Frege was clearly made. A reference to Tarski would have >> been>> nice, though. :-) The clear link notwithstanding, the relational >> model>> cannot be meaningfully described or thought of as a copy of >> FOL, either>> bad or good, since it has a very different purpose and >> makes entirely>> different claims about its purpose and utility. To >> confuse the two is>> misunderstanding entirely what database theory and >> database models are>> about.
>
> I think it is necessary to clearly specify the model and exactly what
> people are doing when they use the model. Given that this is a very
> important and broad topic, I'll simplify this, in order to get a better
> explanation what the model is. Unfortunately on this topic has no
> literature.
> When a man does some complex projects in the real world he uses the
> model. Models are used extensively in architecture, engineering,
> databases etc.. Schematically, this can be represented as the following
> triplet:
>
> (*) Man Model RealWorldProject
> Here, the model could be anything. But the most important and most
> powerful type models are mathematical models. For example, the model
> may be a partial differential equation that describes some of the
> technical-physical project. The model can be Boolean algebra that
> models the design of electronic circuits. Obviously this type of model
> is pure mathematics, these models are not a technical or special models
> flavored with mathematics. Many database theorists for many parts of db
> theory claim that it is math. I also think it is math, in fact I think
> that's part of the new math that has a great future.
If you define the Relational Model as just the datamodel, then you would be right that there was not much new there. But Codd was not just proposing a data model, but rather a set of engineering principles for building effective generic DBMSs. The data model is only a part of that, and the new insight and/or claim is here that this is in practice an effective model for the intended purpose. This is not a mathematical claim but an engineering claim. So even though it has strong links with certain mathematical theories, to think of the thing itself as a mathematical theory is to entirely miss the point.
- Jan Hidders