Datatyping

Since the WG apparently will be putting thinking caps back on regarding 
datatyping, I though I'd show the way I'd like the new datatyping to work:

I submit the graph:
http://robustai.net/mentography/jennyAge10.gif
which hopefully is consistent, yet includes the dreaded situations.

In this graph, note the following::

1) There are two kinds of literal nodes:  TypedLiteral(s) and 
LexicalNodes.  
2) TypedLiteral Nodes can be subjects.
3) LexicalNodes cannot be subjects - as in M&S.
4) TypedLiteral Nodes gain their semantics from the triples in the 
graphs (not node labels).
5) All nodes in that graph are tidy - there are no Bnodes in that graph.
6) Identity is defined on all nodes.
7) Equality is defined on LiteralNodes but not on LexicalNodes.

In that graph, Jenny's age is not identical to John's age, yet I would 
like the MT to entail that equality.
In that graph, John's age is identical to the title of movie B, yet I 
would like the MT to entail that John's age is not equal the title of 
movie B.  

That graph is over on the Public-CMaps server, so anyone who is running 
that fine graphing tool provide by UWF [1], is welcome to add to , 
and\or correct it. The project is SemanticWeb and the name of the graph 
is jennyAge10.

[1] http://cmap.coginst.uwf.edu/

Seth Russell
http://radio.weblogs.com/0113759/

Received on Tuesday, 24 September 2002 08:44:32 UTC