RIF telecon Jan 19 2010 -- 19 Jan 2010

<sandro> Harold, any response on the wiki problem yet?

<ChrisW> list agenda

<sandro> oh, sorry, didn't do the zakim agenda thing.

Admin

<AxelPolleres> hi!

<Harold> I am in another meeting; cannot join on the phone until in up to 60 mins.

<ChrisW> previous minutes: http://lists.w3.org/Archives/Public/public-rif-wg/2010Jan/att-0003/05-rif-minutes.html

yes, I can

<ChrisW> Scribe: StellaMitchell

<ChrisW> PROPOSED: Accept last meeting minutes

cw: any objections to above minutes?

<ChrisW> RESOLVED: Accept last meeting minutes

cw: any agenda ammendments?

csma: move xml syntax issue later?

sandro: and looks like Axel would like to talk about importing rif from rdf

axel: I can only stay for 20 minutes

Liason

axel: I think we should define a mechanism to import rif documents from rdf
... this would be useful for SPARQL
... we want to query RDF+ RIF combinations from SPARQL
... and because of the way SPARQL is defined, querying a RIF document that imports RDF (what we already define) does not work well

cw: you are suggesting defining RIF imports from RDF graphs?

axel: yes
... ....but we need to define the semantics of the import

cw: where do you think this would be documented?

axel: ideally, swc, but it's late for that so maybe a WG note

daver: suggesting something with SPARQL entailment regimes

axel: dave's suggestion is a minimilistic option, but not ideal because you can't refer directly to the RIF ruleset

<sandro> axel: Dave's approach would not allow the imported ruleset to be named in the graph

sandro: I support what axel is suggesting. Another way to do this would be to have a way of expressing rif in rdf (we are very close to that already).
... the SPARQL entailment regime idea would not provide support for non-SPARQL users who also want this capability
... I've been asked a number of times how to use RIF from RDF

cw: how close are we to having an RDF syntax of RIF

sandro: I think we may just have to define a namepace. we are very close.

cw: you mean we'd have to address the 1 or 2 places where rif xml is not striped?

<sandro> sandro: yes, like var.

sandro: yes, I've meant to do this, but have not had time

cw: any volunteers to work on this?
... I also get these questions

sandro: I can take an action for this

<ChrisW> ACTION: sandro to document an rdf syntax for rif [recorded in http://www.w3.org/2010/01/19-rif-minutes.html#action01]

<trackbot> Created ACTION-968 - Document an rdf syntax for rif [on Sandro Hawke - due 2010-01-26].

csma: why do you say that var is not striped?

<Harold> Var is a 'leaf class'.

sandro: we didn't name the property.
... we only have literal content for var and const; const has an obvious mapping

Actions

<Harold> <Var>PCDATA</Var> uses XML's PCDATA, which can be mapped to an RDF property.

<sandro> So, in RDF, that would look like <Var><name>PCDATA</name></Var> (where "<name>" might be "<varname>" or .... something else. RIF doesnt say.(

csma: I think actions 965 to 967 are obsolete
... we rolled back resolution and now need new actions

actions 965 to 967 have been obsoleted

close action-964

<trackbot> ACTION-964 Update public comment list closed

close action-961

<trackbot> ACTION-961 Check base64Binary case http://www.w3.org/2005/rules/wiki/Builtins_Binary closed

action 960 is continued

<trackbot> Sorry, couldn't find user - 960

<Harold> Rather than having another level of role tags within the Var class, I guess <Var name="CDATA"/> would be better for the purpose of RDF mapping.

continue 959

close action-958

<trackbot> ACTION-958 Draft response to IH closed

continue 952

continue 951 (very close to done)

continue 940 and 941, will draft replies now

Harold, what the status of reply to Alex Riaz?

continue 880

continue 850

Public Comments

csma: we have 3 that need replies and there is a new public comment?

<csma> http://lists.w3.org/Archives/Public/public-rif-comments/2010Jan/0000.html

<Harold> Alex Riazanov action is ongoing

stella: I think we decided (long ago) to provide RDF/XML versions of imported documents
... is there a recommended tool to convert turtle to RDF/XML?

(for test cases)

csma: I talked to this commenter about his issue and suggested he post to public list

cw: I think that's up to implementations. There is only one normative RDF syntax right?

sandro: now there is RDFa
... I'm not sure RDF that says there is only one normative syntax

cw: I think RIF doesn't need to take a stand on what RDF syntax needs to be supported, in imported documents

csma: so an implementation can claim to support RDF imports even if it supports only its proprietary syntax?
... another possibility would be to require a syntax indicator to go along with import statements
... so that consumers can check before processing

<ChrisW> ACTION: csma to draft response on public comment JA [recorded in http://www.w3.org/2010/01/19-rif-minutes.html#action02]

<trackbot> Created ACTION-969 - Draft response on public comment JA [on Christian de Sainte Marie - due 2010-01-26].

cw: do we have actions covering all unanswered public comments?
... yes, looks like we do

PRD

csma: I sent email summarizing the issue about refraction
... I tested a few systems, but don't have a license for Jess

I think we have 3 issues:

scribe: 1. the modify in clips is really a retract followed by an assert
... 2. in prd we look at state only after each action block. but the update of the agenda in most systems looks after each change...including intermediate states
... 3. currently the state of a rule instance is characterized by binding of rule variables, but this is not always adequagte, specifially when there is a disjunction in the condition of the rule and the disj doesn't contain any rule varialbles

So my proposal is that we change the PRD spec in three ways

scribe: 1. change definition of rule instance
... 2. we consider refraction with respect to all the states of the fact base, i.e. after each atomic action rather than after each action block
... (above 2 are fairly minor)
... 3. change modify so that it is not an atomic action, and if we do this we don't need modify in prd any more

<AdrianP> I would propose we keep modify with an atomic semantics

<AdrianP> Clips can represent their modify as retract+assert

<AdrianP> in RIF

cw: are these 3 points co-dependent

csma: yes
... modify_noloop cannot be implemented in clips

<Gary> I don't care about interoperating with clips.

<Gary> We need to consider some concrete test cases to better understand the issues here

changhai: one possibility is to keep modify as it is now and remove modify_noloop test case

<AdrianP> right, that is a problem of the Clips semantics which does not support atomic modifies

csma: we can't keep modify as it is because that would mean we have something in spec that cannot be implemented

changhai: it can be implemented by some

cw: we are discussing whether we need to change the semantics for CLIPS

csma: gary, how is modify implemented in Jess?
... in terms of agenda

gary: it's complicated.

cw: can jess implement modify_noloop

gary: no, but for a different reason that clips cannot

cw: would the jess problem with modify_noloop require a prd fix?
... different from the one csma proposed?

gary: i don't completely understand csma's proposal yet.

cw: gary, what changes to you have in mind?

gary: not completely sure yet, but related to existential variables

csma: no, taking bindings into account does not change anything with respect to clips and jrules

gary: for jess, it would

csma: details of behavior of different rule engines
... let's discuss more by email to clarify these situations

cw: if semantics needs to change, we need to do another last call and we would want to do that as soon as possible

gary: discussing slots and clips, and not happy with the way clips behaves there

<AdrianP> Jess modify changes the slot values of facts already in working memory

cw: csma, what granularity were you talking about wrt facts?

csma: atom in the rif sense

cw: so, csma will send some examples to gary?
... and is another telecon needed?

<ChrisW> ACTION: csma to send some examples of the failure case to gary [recorded in http://www.w3.org/2010/01/19-rif-minutes.html#action03]

<trackbot> Created ACTION-970 - Send some examples of the failure case to gary [on Christian de Sainte Marie - due 2010-01-26].

<AdrianP> yes, Tuesday would work for me

<ChrisW> ACTION: csma to schedule a PRD telecon next week [recorded in http://www.w3.org/2010/01/19-rif-minutes.html#action04]

<trackbot> Created ACTION-971 - Schedule a PRD telecon next week [on Christian de Sainte Marie - due 2010-01-26].

gary: csma, if you could come up with simple rif test case, positive and negative cases relating to the issues you were talking about

Implementations

csma: yes, by end of week and prd telecon next week

cw: status of implementations, any news?

csma: jose maria said they are currently implementing DTB and asked whether they should publish as a service or as a library
... this should be available within a couple weeks
... they are planning a complete implementatin of dTB
... will be open source and in java
... jos sent an email about implementations

cw: yes, that STI is not going to do anything soon
... anything more on ontobroker?

csma: unlikely that they would do more than they alreay have by the end of CR

cw: no progress on fuxi implementation

csma: we'll be doing additional work on jrules

cw: vampire/eye

harold: alex r is interested but not progressing as fast as planned

gary: oracle progressing, dtb is time-consuming

mk: one fld implementation has been sent to rif mail list

mdean: we have a silk dialect that is implemented but doesn't get have a semantics specified
... and something (logic programming dialect) else

XML Syntax of Import

<ChrisW> ACTION: harold to update core, bld, and fld xml schema to reflect resolution on imports [recorded in http://www.w3.org/2010/01/19-rif-minutes.html#action05]

<trackbot> Created ACTION-972 - Update core, bld, and fld xml schema to reflect resolution on imports [on Harold Boley - due 2010-01-26].

SemTech

sandro: would anyone be available to be on a RIF panel at semtech?

<csma> maybe

<Gary> not me

sandro: so far Sandro, Paul Vincent

<DaveReynolds> I'm a maybe

mdean: I'm planning to be there

cw: I'm planning to be there, but not definite

sandro: sandro, paul, mike d definite, several maybes

AOB

Test Cases

<MichaelKifer> I have to go, sorry

rdf_subclass5 may have sparked this

cw: love the builtin_strings testcase

<ChrisW> http://www.w3.org/2005/rules/wiki/Builtins_String

cw: any other comments on this case, we reviewed once

<ChrisW> PROPOSED: approve Builtins String

<ChrisW> +1

<Gary> +1 it works for me

<ChrisW> RESOLVED: approve Builtins String

<AdrianP> +1

gary: builtins_time....there is a typo
... relating to daytime vs. datetime
... and two typos in xml
... relating to commas or semicolons
... typo: search on xs:daytime

change to datetime

yes

xml is generated by a sool

and the xml validates by the schema

xml: ," (twice)

I will look into it

look for commas in the PS

<ChrisW> PROPOSED: accept Builtins_Time (modulo a few ximple typo fixes)

<AdrianP> +1

<Gary> +1

<ChrisW> RESOLVED: accept Builtins_Time (modulo a few ximple typo fixes)

<DaveReynolds> 0 (just because I haven't checked!)

stella: should we remove forall?

<ChrisW> PROPOSED: accept Builtins_boolean

<DaveReynolds> +1

<Gary> +1

<AdrianP> +1

<ChrisW> RESOLVED: accept Builtins_boolean

AOB

cw: any other business?
... prd telecon next week

<AdrianP> buy

- DRAFT -

RIF telecon Jan 19 2010

19 Jan 2010

Attendees

Contents

Admin

Liason

Actions

Public Comments

PRD

Implementations

XML Syntax of Import

SemTech

AOB

Test Cases

AOB

Summary of Action Items

Scribe.perl diagnostic output