Re: Use of the term "hierarchy"

From: Marshall Spight <marshall.spight_at_gmail.com>
Date: 25 Aug 2005 21:55:39 -0700
Message-ID: <1125032139.295191.210140_at_g44g2000cwa.googlegroups.com>


Kenneth Downs wrote:
> Marshall Spight wrote:
> >
> > Sure. We've been discussing this a lot. There was even a thread
> > I started about a month ago called (IIRC) Three Kinds of Logical Trees.
> >
> > Since then, I've been referring to these as homogeneous trees, and
> > static heterogeneous trees. There are also dynamic heterogeneous trees,
> > like parse trees.
>
> I did not catch in that thread the specific example i gave of the employees.
> Should I reread the thread?

Perhaps so.

I said:
  "homogeneous tree"
    All nodes are the same type, tree has varying structure   [...]
  Examples:
  homogeneous: org chart. Every node is a person record, but the   structure of the organization may be of whatever form.

You said:
  ... employees and their supervisors. This is   usually presented as a hierarchy of like items.

I said "tree"; you said "hierarchy." I said "org chart"; you said "employees and their supervisors." I said "homogeneous"; you said "like items."

It seems like we're talking about the same thing.

My hope is that we can discuss a taxonomy for kinds of trees, because the original thread wasn't very satisfying. I believe you identified two different kinds. One is the fixed-type, varying structure kind, which we both exemplified by the org chart (although you didn't use that exact term) and the other is the fixed structure hierarchy, which you exemplified by Jobs/Orders/Order lines and I exemplified by Customer/Account/Invoice/Line Item.

I believe there is a third kind, in which there are varying kinds of nodes, and the structure is likewise dynamic. I think this is what you were talking about when you proposed different node types for vice presidents, directors, etc. The example I used was a parse tree, but it's the same idea.

To my mind, SQL handles the static structure type quite well; ML would be the canonical best-tool-for-the-job for the dynamic mixed-type tree, and OOPLs do quite well for homogeneous type tree.

Why the deuce can't we get a decent language that has the best tools for all three kinds; that's what I want to know!

Marshall Received on Fri Aug 26 2005 - 06:55:39 CEST

Original text of this message