In this message, I want to discuss the logical types of links. It
seems to me that hyperlinks could encoding knowledge of several
different types. Three that come to mind are:
1 structural - relations between documents, e.g. document X cites
document Y; word X is defined in document Y; possibly extending to
the kind of argumentation links used in e.g. gIBIS, NoteCards (e.g.
claim X is refuted by argument Y)
2 presentation - the user should see X and Y at same time if possible.
3 construction - to build document X, include files Y and Z, and
run procedure W at display time.
It is not obvious to me that these three should be treated in the
same way. In particular, I doubt the need for presentation information.
The lesson of SGML is that documents should be marked with logical
structure, leaving presentation out.
It also seems to me that Raisch's list mixes several types, and
this is undesirable.
For example, I see the "annotation" link as encoding presentation
information. It differs from "replacement" only in that you see the
original and the target at the same time. (At least, that's how he
describes it. One could also define it as a kind of logical relation
- it is a comment on the target. But that's a different subject.)
A second example is that one might want to use an "Execution" link (or
its subtype "collection") to implement any of the first four types
(replacement, annotation, inclusion, expansion).
To continue the discussion, we might consider whether there are
other logical types for links, how to decompose Raisch's links
into these categories, and whether to encode presentation in links.