Chatlog 2010-08-25

See original RRSAgent log and preview nicely formatted version.

Please justify/explain all edits to this page, in your "edit summary" text.

14:08:19 <RRSAgent> RRSAgent has joined #rif
14:08:19 <RRSAgent> logging to http://www.w3.org/2010/08/25-rif-irc
14:08:19 <AxelPolleres> let;s start...
14:08:19 <AxelPolleres> topic: http://www.w3.org/2005/rules/wiki/RIF_In_RDF#Importing_RIF_into_RDF 
14:08:34 <AxelPolleres> I think that we need to define the whole thing as follows:
14:08:41 <AxelPolleres> Maybe that could work: Any RDF doc that uses URLs in the rif:namespace, should be viewed as a pair (G,R) where G is an RDF graph and R is a RIF ruleset.
14:09:12 <AxelPolleres> the separation works as follows:
14:10:09 <AxelPolleres> R is built by extracting (in a consuming manner) any RIF document encoded in the RDF graph. 
14:10:38 <AxelPolleres> (following http://www.w3.org/2005/rules/wiki/RIF_In_RDF)
14:10:47 <AxelPolleres> G is the set of remaining triples
14:11:32 <AxelPolleres> Te semantics is the one of (R1, G) accordign to SWC
14:12:00 <AxelPolleres> where R1 is defined as R plus the following imports statements:
14:12:20 <AxelPolleres> Imports (G)
14:13:47 <sandro> messing with me mic...
14:15:32 <AxelPolleres> will continue typing thoughts in the chat...
14:15:57 <AxelPolleres> so we stopped at Imports(G)
14:37:24 <AxelPolleres> is RDF KR or a DB...
14:42:51 <AxelPolleres> lets assume G.rdf
14:43:13 <AxelPolleres> R.rdf is a graph encoding a RIF ruleset a la sandro
14:43:27 <AxelPolleres> my proposal for SPARQL would mean I can do
14:43:40 <AxelPolleres> SELECT ... FROM G FROM R ...
14:44:00 <AxelPolleres> SELECT ... FROM G FROM aux ....
14:44:15 <AxelPolleres> where aux has a single tripl 
14:44:18 <AxelPolleres>    R rif:usedWithProfile P .
14:45:26 <AxelPolleres> I consider the former simpler ;-)
14:46:43 <AxelPolleres> Imports (G RDF)
14:48:03 <AxelPolleres> R union { sandro knows axel }
14:55:00 <AxelPolleres> http://lists.w3.org/Archives/Public/public-rif-wg/2010Jan/0015.html
14:56:22 <sandro> I would change "...such R1 is an IRI referring to a RIF document and P is a URI ..." to
14:57:10 <sandro> "... such R1 is an IRI referring a RIF Document (which might be in XML or encoded in RDF, R1 being the label on the root node), and P is a URI ..."
14:58:06 <sandro> axel: this needs to be in SWC
14:58:22 <sandro> sandro: people who implement usedWithProfile will implement it no matter where the text is.
14:58:53 <AxelPolleres> G = ({},G}
14:59:05 <AxelPolleres> s/=/is considered as/
15:00:51 <AxelPolleres> assume R, G1, G2 in my triplestore
15:01:21 <AxelPolleres> assume R1, R2, G1, G2 in my triplestore
15:01:28 <AxelPolleres> G1, G2 are datagraphs
15:01:41 <AxelPolleres> R1, R2 are encodung different subsets of RDFS
15:01:47 <AxelPolleres> entailment rules
15:02:20 <AxelPolleres> R1,R2 are graphs encodung rulesets R1' R2'
15:03:24 <sandro>    < R1' rif:usedWithProfile rif:rdfsimple >
15:03:31 <sandro>     R1' rif:usedWithProfile rif:rdfsimple }{
15:03:37 <sandro>   {  R1' rif:usedWithProfile rif:rdfsimple }
15:04:11 <AxelPolleres> <G1>
15:04:12 <AxelPolleres> ...
15:04:19 <AxelPolleres> <R1> is a URI
15:04:48 <AxelPolleres> <R1> rif:usedWithProfile rif:rdfsimple
15:07:19 <AxelPolleres>  sandro knows axel 
15:08:30 <sandro> <R1> rdf:type rif:AssertedRuleset
15:09:05 <AxelPolleres> SELECT ... FROM G FROM R ...
15:09:06 <sandro> used-with-profile ===  is asserted, and has this profile
15:15:43 <AxelPolleres> SELECT ... FROM G FROM aux ....
15:21:37 <sandro> possibly with two RIF entailment regimes -- one for "direct" one for "rdf-based".
15:22:16 <sandro> axel doesn't want to see infrastructural triples in the final result.
15:22:39 <AxelPolleres> ... which is only possible with a kind of "direct semantics"
15:23:03 <sandro> can just be addressed by naming R in the query, of course.
15:23:23 <sandro> (which is, alas, out of scope for the current WG)
15:23:53 <sandro> (parameterized inference)
15:24:46 <sandro>  compromise-1: axel ignores that extra infrastructure triple
15:25:04 <sandro>  compromise-2: sparql just does the direct-semantcs style.
15:27:34 <sandro> (diverging)
15:31:52 <sandro> sandro: you're proposing a mandate that, if the guy in the middle does RIF processing, he has to throw out the rif triples....
15:32:26 <sandro> axel: you can just carry them along, still
15:32:35 <sandro> sandro: but you said they had to be removed...?
15:33:16 <sandro> axel: load the same into to two different named graphs, with different entailment regimes.
15:36:55 <AxelPolleres> Enc( Imports R )
15:38:08 <AxelPolleres> SELECT ?S ?P ?O from R 
15:39:42 <sandro> axel: I want to get the slots which are entailed by R.
15:39:48 <AxelPolleres> expected behavior for me is getting back the entailed slots as RDF
15:40:19 <sandro> sandro: I would give that PLUS the encoding of the ruleset in RDF, PLUS the usedWithProfile triples.
15:40:19 <AxelPolleres> expected behior of sandro is getting back all those plus triples encodung R
15:41:05 <AxelPolleres> this essentially is thegist of RDF-based semantics of RIF-in-RDF vs .direct semantics of RIF-in-RDF
15:42:11 <sandro> sandro: this is kind of just Answer Only About Inference....
15:42:24 <sandro> sandro: (give me back ONLY the inferred triples.)
15:42:34 <AxelPolleres> http://www.w3.org/TR/sparql11-entailment/
15:45:19 <sandro> owl ontology:   p rdfs:range c
15:46:12 <AxelPolleres>  \exist p^-.owl:Thing \subseteq c  
15:46:17 <sandro> does that ontology entail itself?     I believe it does, that even OWL DL is monotonic, on it's synteax.
15:49:13 <AxelPolleres> encoding OWL in RDF is like encoding RDF in XML ... it introduces ambiguities, because it\s a syntactical thing
15:51:24 <sandro> sandro: I don't like having triples disappearing from KB when you turn on RIF entailment.
15:53:02 <AxelPolleres> ACTION: axel to write down owl direct vs OWL RDF test case and confirm with Birte
15:53:03 <trackbot> Created ACTION-1045 - Write down owl direct vs OWL RDF test case and confirm with Birte [on Axel Polleres - due 2010-09-01].
15:53:25 <sandro> test case:       no bindings for ?s ?p ?o  on    p rdfs:range c    under OWL Direct Semantics
15:55:21 <sandro> sandro: I'm 97% sure that for any RDF graph G, OWL entails G (for each OWL semantics).
15:55:33 <sandro> sandro: I'm 97% sure that for any RDF graph G, G OWL entails G (for each OWL semantics).
15:56:38 <sandro> sandro: so if SPARQL is going its own way from that, then it can with RIF, too.
15:57:08 <AxelPolleres> ACTION: sandro to write down test cases
15:57:08 <trackbot> Created ACTION-1046 - Write down test cases [on Sandro Hawke - due 2010-09-01].
15:57:41 <AxelPolleres> ACTION: to write down "RIF-direct semantics" proposal for RIF-in-RDF 
15:57:41 <trackbot> Sorry, couldn't find user - to
15:57:59 <AxelPolleres> ACTION: Axel to write down "RIF-direct semantics" proposal for RIF-in-RDF.
15:58:00 <trackbot> Created ACTION-1047 - Write down "RIF-direct semantics" proposal for RIF-in-RDF. [on Axel Polleres - due 2010-09-01].
15:58:31 <sandro> action-1046?
15:58:31 <trackbot> ACTION-1046 -- Sandro Hawke to write down test cases for SPARQL's RIF Entailment and RIF TCs for RIF-in-RDF usesWithProfile -- due 2010-09-01 -- OPEN
15:58:31 <trackbot> http://www.w3.org/2005/rules/wg/track/actions/1046
15:59:24 <AxelPolleres> http://www.w3.org/2009/sparql/docs/tests/data-sparql11/ here are the test cases so far
15:59:51 <AxelPolleres> we have for SPARQL (heavily under construction, but I think there are already some for entailment)
16:20:08 <sandro>    1.   RIF_Frame_Document_Described_But_Not_Used
16:20:08 <sandro>      a  - output contains frame?    sandro:no axel:yes
16:20:08 <sandro>      b  - output contains description?  sandro:yes axel:no
16:20:08 <sandro>    2.   RIF_Frame_Document_Described_And_Used_Same_Graph
16:20:08 <sandro>      a  - output contain frame?    sandro:yes  axel:yes
16:20:09 <sandro>      b  - output contains description?   sandro:yes axel:no
16:20:11 <sandro>      c  - output contains usedWithProfile  sandro:yes axel:?
16:20:13 <sandro>    3.   RIF_Frame_Document_Described_And_Used_Different_Graph
16:20:15 <sandro>      a  - output contain frame?    sandro:yes  axel:yes
16:20:17 <sandro>      b  - output contains description?   sandro:no axel:no
16:20:19 <sandro>      c  - output contains usedWithProfile  sandro:yes  axel:?
16:20:37 <sandro> something like that.....
18:11:20 <AxelPolleres> AxelPolleres has joined #rif
18:11:35 <AxelPolleres> AxelPolleres has joined #rif
18:53:30 <AxelPolleres> AxelPolleres has joined #rif
19:17:26 <AxelPolleres> AxelPolleres has joined #rif
20:31:04 <AxelPolleres> AxelPolleres has joined #rif
20:32:21 <AxelPolleres> rrsagent, make records public
# SPECIAL MARKER FOR CHATSYNC.  DO NOT EDIT THIS LINE OR BELOW.  SRCLINESUSED=00000117