Chatlog 2010-02-24

See original RRSAgent log and preview nicely formatted version.

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

<sandro> Meeting: SPARQL Entailment Regimes
<sandro> chair: bglimm
<sandro> scribe: Axel
15:02:38 <RRSAgent> RRSAgent has joined #sparql-ent
15:02:38 <RRSAgent> logging to http://www.w3.org/2010/02/24-sparql-ent-irc
15:02:51 <bglimm> agenda: http://www.w3.org/2009/sparql/wiki/Agenda-2010-02-24
17:02:17 <kpbrown> kpbrown has joined #sparql-ent
17:02:20 <kpbrown> kpbrown has left #sparql-ent
17:30:06 <Zakim> Zakim has left #sparql-ent
17:40:21 <ivan> ivan has joined #sparql-ent
17:40:28 <Zakim> Zakim has joined #sparql-ent
17:46:03 <bglimm> Hi Ivan
17:46:10 <ivan> hi birte
17:46:21 <bglimm> can we set the link to the agenda as chat topic?
17:46:29 <bglimm> http://www.w3.org/2009/sparql/wiki/Agenda-2010-02-24
17:46:32 <ivan> of course
17:46:35 <bglimm> I didn't manage to do that
17:46:45 <ivan> ivan has changed the topic to: http://www.w3.org/2009/sparql/wiki/Agenda-2010-02-24
17:46:54 <bglimm> :-)
17:46:55 <ivan> ivan has changed the topic to: agenda: http://www.w3.org/2009/sparql/wiki/Agenda-2010-02-24
17:47:32 <ivan> zakim, room for 10 for 90 minutes?
17:47:33 <Zakim> ok, ivan; conference Team_(sparql-ent)17:47Z scheduled with code 26631 (CONF1) for 90 minutes until 1917Z
17:47:44 <ivan> here we are:-)
17:47:58 <ivan> I asked for 90 minutes if we round out (and we still have 10 minutes)
17:48:44 <ivan> ivan has changed the topic to: Zakim code 26631, agenda http://www.w3.org/2009/sparql/wiki/Agenda-2010-02-24
17:49:19 <bglimm> :-)
17:50:32 <bglimm> I've updated the wiki agenda now
17:58:37 <Zakim> Team_(sparql-ent)17:47Z has now started
17:58:44 <Zakim> +bglimm
17:58:45 <AxelPolleres> AxelPolleres has joined #sparql-ent
17:58:53 <AxelPolleres> hi all
17:58:57 <bglimm> Hi
17:59:01 <bglimm> The access code is 26631
17:59:02 <AxelPolleres> chime?
17:59:07 <bglimm> Not yet
18:01:44 <AxelPolleres> will sandro join?
18:01:52 <bglimm> He said so I think
18:01:55 <ivan> zakim, dial ivan-voip
18:01:55 <Zakim> ok, ivan; the call is being made
18:01:57 <Zakim> +Ivan
18:02:41 <Zakim> +AxelPolleres
18:03:42 <sandro> sandro has joined #sparql-ent
18:03:48 <sandro> zakim, what is the code?
18:03:48 <Zakim> the conference code is 26631 (tel:+1.617.761.6200 tel:+33.4.89.06.34.99 tel:+44.117.370.6152), sandro
18:03:51 <sandro> rrsagent, pointer?
18:03:51 <RRSAgent> See http://www.w3.org/2010/02/24-sparql-ent-irc#T18-03-51
18:04:11 <Zakim> +Sandro
18:04:15 <sandro> zakim, who is on the call?
18:04:15 <Zakim> On the phone I see bglimm, Ivan, AxelPolleres, Sandro
18:05:33 <sandro> (axel to make sure anything important gets recorded, but not really scribe.)
#18:05:35 <AxelPolleres> topic: general entailment regimes issues
18:05:50 <AxelPolleres> inconsistency handling is not clear
18:06:55 <AxelPolleres> sandro: at least, is there a protocol way to signal inconsistency, even if not required
18:07:03 <AxelPolleres> ... seems to me that we want to have that
18:08:13 <ivan> q+
18:10:04 <chimezie> chimezie has joined #sparql-ent
18:10:13 <AxelPolleres> birte can you paste the C2 variation again?
18:10:28 <Zakim> + +1.216.445.aaaa
18:10:31 <bglimm> Wait a sec
18:10:35 <AxelPolleres> guess that's chime?
18:10:42 <chimezie> Zakim +1.216.445.aaaa is me
18:10:44 <chimezie> yes :)
18:10:47 <bglimm>     Current C2 in PWD:
18:10:47 <bglimm>     (C2) Each variable x that occurs in the subject position of a triple in
18:10:47 <bglimm>          BGP is such that sk(μ(x)) occurs in sk(SG).
18:10:48 <bglimm>     Proposed C2:
18:10:48 <bglimm>     (C2) For each variable x in V(BGP), sk(μ(x)) occurs in sk(SG) or in
18:10:48 <bglimm>          Vocab.
18:12:16 <AxelPolleres>  rdf:_1 .... rdf:_n ?
18:13:28 <AxelPolleres> Vocab is a meant here to be a finite subset of the RDF, RDFS, OWL vocabularies
18:13:34 <bglimm>  Here vocal is defined as the reserved vocabulary for the entailment
18:13:34 <bglimm>     regime (e.g., the RDF vocabulary for RDF entailment) minus terms of the
18:13:34 <bglimm>     form rdf:_n with n in {1, 2, …}.
18:15:24 <AxelPolleres> this condition is not about inconsistencies, is it?
18:15:49 <AxelPolleres> ivan: we don't want to restrict, where we don't have to.
18:17:51 <AxelPolleres> my problem is ... we cannot reach the infinite wisdom... anyways
18:21:17 <AxelPolleres> if we want to keep answers finite... we have to drop intuitiveness at some point :-(
18:24:00 <AxelPolleres> still la bit puzzled about the role of data values in C2 
18:24:29 <AxelPolleres> take the decision on C2 on email...
18:25:25 <AxelPolleres> birte to please explain the role of data values in C2
#18:25:40 <AxelPolleres> topic: shall we mark rif "at risk"?
18:26:25 <AxelPolleres> ivan: rif not being rrec shouldn't be a problem, confident that it will be ready there, afraid that it would be interpreted in underied way, if we mark it
18:26:31 <AxelPolleres> birte: ok
18:31:08 <AxelPolleres> I think we can define SPARQL wrt a RIF/RDF combination... but not really how you get to the scoping graph?
18:31:32 <AxelPolleres> we can just say what the scoping grpah for some RIF/RDF combination is
18:33:27 <AxelPolleres> we can sort out how we get to the correspondoing RIF/RDF combination later on.
18:34:07 <ivan> q+
18:34:25 <AxelPolleres> birte: at some point it was discussed whether we'd have an entailment regime per ruleset.
18:34:28 <AxelPolleres> q+
18:34:34 <ivan> ack ivan
18:34:42 <AxelPolleres> ivan thinks that this is useless..
18:34:59 <bglimm>  Here's somebody from RIF suggestion that: http://lists.w3.org/Archives/Public/public-rif-wg/2010Jan/0019.html
18:35:05 <AxelPolleres> q+ to state that this point is void, if we have solved "rif:imports"
18:35:18 <AxelPolleres> ack AxelPolleres
18:35:18 <Zakim> AxelPolleres, you wanted to state that this point is void, if we have solved "rif:imports"
18:36:43 <AxelPolleres> sandro: the case that is still open is... 
18:37:09 <AxelPolleres> ... how does the client state which ruleset?
18:37:27 <AxelPolleres> FROM mygraph ...
18:37:36 <AxelPolleres> where mygraph refers to a ruleset
18:38:05 <sandro> don't we need:   QUERY ... FROM ... WITH <entailment-regime or ruleset > ....
18:39:04 <sandro> ahhhh,   FROM <ruleset>
18:39:55 <bglimm> Select ... FROM data.rdf FROM rules.rdf WHERE { ... }
18:40:22 <sandro> so this needs RIF-in-RDF for FROM.
18:41:01 <sandro> ivan:  the rules.rdf would be an rdf graph with the rif:import triple in it.
18:41:10 <AxelPolleres> the current definitions (if we don't want to interfer with SPARQL 1.0 ) we can't do FROM <ruleset>
18:41:16 <bglimm> yes that would work
18:41:34 <AxelPolleres> ... but we can do FROM <... rif:imports  ...>
18:42:49 <bglimm> it should probably called rif:useRuleSet or something like that
18:43:04 <bglimm> or even sparql:useRuleSet
18:44:08 <AxelPolleres>  sparql:useRuleSet
18:44:51 <sandro>  conclusion: if necessary, we can do rif-import under spaql
18:45:01 <AxelPolleres> conclusion is... if all goes wrong, we can do� "rif:imports" ourselves.
18:45:57 <AxelPolleres> I think we can handle bnodes.
18:46:24 <AxelPolleres> chime suggested to stick with *strongly safe* core?
18:47:30 <bglimm> if p(?n) then p(?n+1).
18:47:30 <bglimm> p(1).
18:47:31 <bglimm> and the query
18:47:31 <bglimm> p(?x).
18:51:01 <AxelPolleres> sandro: why the finiteness restriction in first place?  
18:51:05 <AxelPolleres> can we change that?
18:51:42 <AxelPolleres> ... no implementation crucially relies on that.
18:52:17 <AxelPolleres> ...if it's just not intuitive, we may want to question that.
18:53:19 <sandro> axel: simple way to restrict finiteness:    things in answer must occur in data.
18:55:38 <sandro> sandro: that's a non starter -- you couldn't get results from a computation.
18:56:07 <sandro> axel: you could allow computations to recursive level N from the data.
18:57:03 <sandro> sandro: just say it has to be any finite number of recursions.
18:57:11 <AxelPolleres> sandro: how if we only restrict that any implemnentation has to limit the recursion depth.
18:57:16 <AxelPolleres> ... but not how?
18:58:42 <AxelPolleres> p(1)
18:58:56 <AxelPolleres> q(x+1) :- p(x).
18:59:10 <AxelPolleres> r(x+1) :- q(x)
19:05:43 <AxelPolleres> sandro: suggests... just leave behaivor of SPARQL undefined whenever the closure of R is infinite?
19:07:23 <AxelPolleres> chime: suggest to standardise a possibly boring finite subset 
19:07:35 <AxelPolleres> sandr: suggest to just don't worry about finiteness
19:07:43 <AxelPolleres> (two quite opposite positions)
19:09:16 <sandro> sandro: use case, define an odd() predicate, then query for { ?s ?p ?o; ?o a oddNumber }
19:09:41 <Zakim> -bglimm
19:09:52 <bglimm> sorry dropped out
19:11:33 <Zakim> +bglimm
19:11:48 <ivan> q+
19:12:06 <AxelPolleres> ivan is still on the q
19:12:50 <sandro> axel: a use case for RIF in SPARQL with strongly-safe core == subsets of RDFS-entailment, ad hoc OWL-RL entailment.
19:13:46 <AxelPolleres> ivan: what comes to my mind... did we have in SPARQL a way to include a time/resource limit?
19:14:01 <AxelPolleres> sandro: we only have result limit
19:14:08 <sandro> sandro: I think it only has LIMIT; with entailment, yes, it would be nice to have TIME limit, too.
19:14:08 <bglimm> resource limit: LIMIT N
19:15:03 <AxelPolleres> not for this round, I guess.
19:15:50 <sandro> sandro: maybe you can throw this in the SPARQL protocol, as an ignore-if-you-don't-understand 
19:16:32 <Zakim> -Ivan
19:16:32 <bglimm> bye
19:16:33 <Zakim> - +1.216.445.aaaa
19:16:33 <AxelPolleres> thanks birte...
19:16:35 <Zakim> -Sandro
19:16:35 <bglimm> adjourned
19:16:46 <Zakim> -bglimm
19:16:48 <Zakim> -AxelPolleres
19:16:49 <Zakim> Team_(sparql-ent)17:47Z has ended
19:16:51 <Zakim> Attendees were bglimm, Ivan, AxelPolleres, Sandro, +1.216.445.aaaa
19:17:10 <AxelPolleres> rrsagent, make records public
19:17:31 <sandro> rrsagent, pointer?
19:17:31 <RRSAgent> See http://www.w3.org/2010/02/24-sparql-ent-irc#T19-17-31
19:19:20 <sandro> ewwwww.    commonscribe doesn't want to handle this irc log being on a strange channel.
19:21:55 <bglimm> hm, what's so strange about it. Should I pick a different one for the next teleconf?
19:22:57 <sandro> it's just looking for #sparql.    maybe easier to use that, next time, since there's no conflict with the WG at that time.
19:23:34 <bglimm> ok. Whoever is not interested in entailments can just ignore us ;-)