Re: The stupidest design I ever saw

Neo wrote:
> > "Formal Concept Analysis in Information Science" article by Uta Priss
> > http://upriss.org.uk/papers/arist.pdf
>
> Both the XML example given earlier which modelled vehicle
> classification as a hierarchy

I don't care about anything related to XML, unless my job requires it.

> and Formal Concept Analysis which also
> employs a classification hierarchy (see example in Fig 2) are flawed in
> that it can lead to situations where a thing classified by it immediate
> parent, isn't classified by the parent of the parent.

Wrong. The formal concepts lattice defines a partial order relation. If an object and a formal attribute are ordered by the lattice order relation then the object is in the "is a" relation to the concept defined by the attribute. There is no requrement for the attribute to be the immediate node on the leattice. Example: Garfield "is a" mammal.

> Take a look at all the links between various nodes in Fig 2 and
> determine if there is a consistent relationship between nodes taking
> into account direction of the link. Basically the most consistent
> relationship seems to be that a lower node is an instance/example of
> the upper node. You can ignore top and bottom nodes which represent
> Universal Concept and NULL respectively. So what is the relationship
> between mammal and cartoon? Is cartoon a mammal?

Both cartoon and mammal are formal attributes. If two attribute are in the lattice partial order relation, then one concept is a subclas of the other. Cartoon is a mammal in this example.. Received on Mon Apr 10 2006 - 03:04:42 CEST