bijan: Alpha release, temporary site ("Cool URIs may not change, but geeky URIs change all the time!"0
bijan: It actually reads and runs a reasonable subset of N3. I have some confusions about []/log:forSome, built-ins aren't built, er, in yet. Nor does it (yet) support any of cwm.py's cmd line interface. It does have a REPL, though. A partial one, at least.
bijan: It can handle rules12.n3, which isn't too shabby.
bijan: Performance looks pretty good, thus far. Parsing is a bit slow, and using rdf_db.pl to write each result (even when filtering) isn't very speedy, but you can construct very fast Prolog queries against the n3 store.
danbri_: Very cool! But have you considered rewriting this in Python?
bijan: No.
AaronSw: Sean has something similar in Python, though.
bijan: Well, GEEZ, TBL has something similar in Python too!
AaronSw: Silly bijan, his isn't a CLONE!
bijan: Annoying Aaron, the original is similar to its clones.
bijan: WOOHOO! Just got a hand-compiled, builtin-using rule to work.
bijan: Still havn't a clue about existential quantification...
bijan: It actually reads and runs a reasonable subset of N3. I have some confusions about []/log:forSome, built-ins aren't built, er, in yet. Nor does it (yet) support any of cwm.py's cmd line interface. It does have a REPL, though. A partial one, at least.
bijan: It can handle rules12.n3, which isn't too shabby.
bijan: Performance looks pretty good, thus far. Parsing is a bit slow, and using rdf_db.pl to write each result (even when filtering) isn't very speedy, but you can construct very fast Prolog queries against the n3 store.
danbri_: Very cool! But have you considered rewriting this in Python?
bijan: No.
AaronSw: Sean has something similar in Python, though.
bijan: Well, GEEZ, TBL has something similar in Python too!
AaronSw: Silly bijan, his isn't a CLONE!
bijan: Annoying Aaron, the original is similar to its clones.
bijan: WOOHOO! Just got a hand-compiled, builtin-using rule to work.
bijan: Still havn't a clue about existential quantification...
sandro_: Or perhaps to put it more clearly, the second case isn't done by having "f" be a function of the DENOTATED OBJECT, of course, but of the name.
bijan: Er...aren't they all functions on names, e.g., "X is a resource label"?
bijan: Does it matter? Rephrase the challenge so that 2 or 3 functions are permitted. Can they have the six properties?
sandro_: Demonstrably solved in http://www.w3.org/2001/12/logic-layering/
sandro_: (although perhaps not in a way anyone but me can read.)
bijan: I was going to ask for the cliff notes :)
sandro_: test-Conditional.rdf shows a FOL formula reified in RDF -- that it, all its structure is described in RDF. The names of the RDF nodes are unnecessary, but useful. extract-otter-formula is some perl code to turn a described FOL formula into on that otter can use. (RDF is a subset of FOL, so proving this for FOL proves it for RDF.)
bijan: The existence of an *f=quote" function still eludes me from this description.
sandro_: Agreed -- there is no f=quote -- the search for one was misguided.
bijan: Ah! That was a proof of the lack? I still don't quite see it (ok, I don't get it at all :)).
sandro_: (something about a hole in the ground. :-)
sandro_: I'm proving you can do reification along the lines of quoting, but you can't work from [MARY] back to ["MARY"], you have to reify BEFORE you take the quotes off.
sandro_: Jan's problem is trying to do it after taking the quotes off.
sandro_: which I agree is difficult and probably impossible.
bijan: Er..For the Record, still not following :) Which Mary, what quotes? Whose cheese?
sandro_: In the discussed article, Jan is trying to talk about talking about the assertion, "Mary had a lamb". (or maybe about the assertion that Mary had a lamb; the difference between these two makes all the difference.)
bijan: Hmm. In the aritcle, Jan is discussion two alternative reification tasks, non-quoting and quoting (of the statment, or of the stating). (Roughly.) He claims the latter isn't a possible reading of RDF reification (as it stands).
bijan: I don't see that one of the conditions is being able to get back to the stating, ergo, that doesn't seem a condition of RDF reification. Maybe I missed that bit?
bijan: Or is this the injective condition?
bijan: (I would need that explained, as it doesn't look that way to me.)
sandro_: okay, maybe I'm misreading the issue here. I agree "f=id" is the best way to interpret M&S Reification. I would furthur argue that M&S reification doesn't support blank nodes properly, so you need something more like in my example.