Lists Home |
Date Index |
> Garshol seems to be quite in favor of his Topic Maps in his
>paper "Topic maps, RDF, DAML, OIL:A comparison" (available
> However, after looking into their diagram representations in
>light of Semantic Network, I found it is indeed that Topic
>Maps(TM) inherently precedes RDF family(RDF+DAML+OIL) and RDF is,
>at heart, a relative lower-level knowledge representation language
>with at least one level lost.
> So what's RDF's value in the effort to Semantic Web?
hehehe - I'll pass on that question...
from the Summary of Garshol's doc:
[[[In RDF, the URI of the resource identifies it. In topic maps, the topic
may have a URI that points to the resource that is the topic, and it may
have any number of URIs that point to resources that explain what the topic
In a very real sense this makes RDF substantially lower-level than topic
maps, and means that it is impossible to make sense of RDF models without
knowledge of their schema to the extent that one can, for example, display
the data in a user-friendly fashion. With topic maps, however, this is
possible, because of the higher level of abstraction.]]]
At heart RDF is a lower level that Topic Maps, but RDFS allows for the
creation of levels such as that found in TMs. I don't believe the "higher
level of abstraction" here is correct - a bnode can quite happily take the
role of topic, giving the same "level of abstraction". My take on this would
be that TMs have additional semantics hard-wired, which makes it easier to,
e.g. display the data in a user-friendly fashion, but that such semantics
can easily be given to RDF documents if required.
[[[DAML and OIL use the same data model as RDF, which they do not enrich,
and to which they contribute very little in the way of semantics, and so no
further consideration needs to be given to them.]]]
This seems to suggest something of a misunderstanding of the layering of
RDF. DAML and OIL don't enrich RDF per se, but they do provide considerable
semantics on top of RDF. I'm not familiar enough with TMs to really say
whether DAML+OIL require further consideration in this particular context,
but I can say for certain that further consideration is most certainly
worthwhile in the context of the Semantic Web. I'm willing to bet that Aaron
& EM's comments included something along these lines. (deadlines prevented
inclusion in the doc). How *does* one do FOL with TMs, btw?
I do think it's a good paper, and it's great to see TM Semantic Web work - I
think it's unrealistic and probably unhealthy to think of the SW entirely in
terms of one model/format. TMs bring a different perspective to the table,
and as Lars and others have demonstrated, RDF & TMs can coexist happily.
btw, I'm rather surprised that XTM didn't come up in the recent "XML'ers do