Re: Ideas for World Hierarchy Example

dawn wrote:
> JOG wrote:
> > dawn wrote:
> >
> > > Neo wrote:
> > > > > > Marshall: Almost none of the data is encoded in a way a machine can do any useful semantic processing on it.
> > > >
> > > > > Dawn: when you click on words in a wiki, similar to "clicking" on a foreign key value within a database (an actual instance of a database), you navigate to another node (called a "document", think "record") that is set up as a tree (specified in xhtml, for example) with more foreign key values found by which you can find more documents (records). A wiki is a web. It can be modeled as a digraph with trees on the nodes. This is pretty much the model for many databases that are not RDBMS's by design (e.g. UniData, UniVerse, OpenQM, Revelation, jBASE, D3, Cache', UniVision)
> > > >
> > > > Would it be fair to say that RMDBs offer a higher degree of
> > > > systematicness but less flexibility?
> > >
> > > With the implementations of each model (intended model, no
> > > implementation is perfect), it would be fair to say that there are more
> > > constraints built into the dbms in the case of the RDBMS and more
> > > flexiibility with the di-graph.
> > [snip]
> >
> > It would be even fairer to say that a di-graph model cannot model n-ary
> > relationships,
>
> I don't understand why you say this. Can you give an example?

A graph is made up of a binary mathematical relation. Not all information exists in binary relationships, often being n-ary. I think we all agree there no? (Of course if you want to build an n-ary logical model out of a graph structure in your code, then sure, go ahead, it is irrelevant to discussion of the logical level. I /think/ Kent called this the pseudo-binary model.)

Let me also emphasize that my post was referring to neo's binary fetish, and not to your post at all - I sadly only see his words in quotations now.

> It is easier to implement n:m relationships between "entities" using the data
> modeling approaches for non-1NF dbms's than for 1NF, right?

1NF has nothing to do with my point about n-ary relationships. However I certainly find thinking in terms of propositions far easier these days than the OO mindset I was once stuck in, and I have found this transition extremely liberating.

>


> > whereas RMDB's can (although they are not the only


> > possible n-ary model). And fairer too to say that a binary model is

> > necessarily a weak subset of an n-ary model (e.g. hypergraphs).

> >

> > I cannot understand why with 40 years of research, papers, and

> > practical illustrations such as the replacement of navigational

> > databases, and now clear failure of the Semantic Web's di-graph data

>

I honestly know only of failures when it comes to graph data models, in all fields with which I am involved. And I certainly wouldn't consider Pick to be binary at the logical level of relationships which I was referring to, if that helps to align the vocab I am using.

> There are even more years of
> history we can look to for that. It might be the case that there are
> both bigger failures and bigger successes using the model I prefer,
> where there might be more of a cookbook approach not to fail with the
> RDBMS tools. But I have worked with both and there is a decided
> difference, where I know what the clear favorite is for me, even if you
> have a different preference.

>


> > Its almost as though there is a concious disrespect in some parts of

>

I was not referring to whatever businesses hodge-podged together before 1970 in order to make them a buck. I was referring rather to people who choose to ignore the hard work of those who have produced scientific research, with its solid theoretical arguments and valid empirical results, over the past 40 years.

Too may people seem to prefer to stand on the toes of midgets rather than shoulders of giants when it comes to databases.

> Fortunately over the years, much of what worked in the
> past as come into the SQL-DBMS tools, but we are still sitting here
> stuck with 3VL and 1NF in many production systems (with more still
> being written).

I have just posted in another thread as to my current view of 1NF so no point discussing it again here (where it isn't actually relevant either). Nor is 3VL relevant, but for what it is worth, I am in agreement with you as to the merits of 2VL. Regards, J.

> If our discipline had been more mature at the time, I
> would hope this wholesale disregard for what were the industry best
> practices at the time would not have been disregarded without adequate
> (or any?) emperical data suggesting they should be.
>
> cheers! --dawn
Received on Sun Jan 14 2007 - 06:02:09 CET