I'm finding it difficult to comment on the encyclopedia without a printed
copy, but here are a few typos and a few thoughts. I hope there will be
more as I work with it this term.
1. The link attached to "list of all the entries" in intro.htm goes to
enc-contents.htm, but that file doesn't exist. At first I thought it
might just be index.htm, but index.htm doesn't have the list down the
left side which is described for enc-contents.htm (.) This file would
be a very useful additional device for navigating the encyclopedia.
2. Under C. in the $relation entry, 3rd line, change "their" to "there"
3. The confusion is probably mine, but I find the terminology in the
discussion under the entry for "relation" confusing in the B. part.
_in-links_ and _out-links_ are called relations, but then in the example
_John is the husband of Mary_, "husband" is called the relation.
4. To make matters worse, in part C., relations are called "nodes"
(shouldn't a relation be a link between nodes?) and grammatical
functions are brought in. The point made with them is clear (that
they are organised in an isa hierarchy), but the relationship between
function and relation is not.
These are all minor matters, but I would like to review my understanding
of these basic WG elements to see if I'm off-base in some way. I set aside
in-links and out-links because I don't understand how they fit in yet.
I'm not sure how useful it is to think of the mathematical or set-theoretical
definitions of relation and function, but if we recall these (a relation
is a set of ordered pairs or a mapping between two sets A & B subject to no
other restriction than that each pair (a, b) is specified as in the relation
or not in it, and a function is a mapping between two sets such that no
element in the domain (A) is mapped into two different elements in the range
(B)), it seems that most WG syntactic dependency relations are functions:
a verb can only have one object, etc. but for semantic relations this may
not be so. It is for "husband" (except for the Nayar) in most societies,
but isn't for "wife" in most African societies, etc. And "isa" is a relation,
not a function.
One could speak of an isa relation as a relation with a label:
a subject isa complement OR
isa(complement, subject) OR
isa(complement) = subject
and couldn't one, dispensing with the triangle, write an arrow-headed
arc with a label "isa" from complement to subject?
I think it is best not to do this, but to continue to use the triangle
as the "label", but perhaps it is important to note that all relations
(and functions are just a particular kind of relation) can be notated
aRb or R(a, b) or R(a) = b.
Thanks,
Chet
|