Paul Everitt, with the assistance (in a comment) of Richard Volpato, draws attention to an email from Steve Newcomb in which he discusses reification in Topic Maps and RDF.
The example he uses is the representation of the statement,
"Tom buttered the bread."
If Tom and The Bread are nodes, an arc between the two can represent the fact that Tom buttered The Bread. However, to say anything about the buttering (it was done with a knife, on Friday) requires that the buttering itself be a node.
I can't help feeling that allowing late reification, as in RDF, is a serious problem when knowledge representations are supposed to be passed round, merged, modified, and so on. For that, I need to be able to track provenance, and the sensible way to do that is to make assertions about elements of the graph; I need to be able to say, "
He concludes by implying that the understanding of these differences in reification (early in Topic Maps vs. late in RDF) expose how TMs and RDF are complementary. That I don't fully understand yet.
Comments