Re: morphing links.

On Tue, Jan 28, 2003 at 11:07:27PM -0500, Bob DuCharme wrote:

> Could you type out a few XML elements to show us examples of some of these 
> genealogy elements and the kinds of linking and/or link typing you'd like 
> to see, using any element and attribute names you like?
> 
> These kind of boundary-pushing use cases are the best input of all.

Here's a similar scenario, taken from the Dialog Mapping community,
which uses the IBIS grammar to map discussions.  IBIS is a simple
grammar consisting of questions, ideas, and arguments.  Ideas link to
questions, arguments link to ideas.

Arguments can be pro or con, but that is specified by the link type,
not by the node type, which is simply "argument."  The reason for this
is that an argument may be both pro or con, depending on its context.
The same node may support one idea, but refute another.

When displaying an IBIS map graphically, it's useful to be able to
display the nodes as pro or con, rather than simply as arguments.  The
software must obviously be aware of the link types in order to do
this.  This implies a tight coupling between content and the linkbase,
but that makes sense because the linkbase defines the structure of the
"document," the document being the map in this case.

This is the reverse of Douglas's original posting.  Douglas described
a scenario where the link type changed based on the node types.  This
is a scenario where the node types (or at least the perceived node
types) change depending on the link types.

I don't think you need a special "link language" to deal with these
issues, but I do think it's good to consider scenarios where the
content and linkbase are tightly coupled.

-Eugene