Chatlog 2012-04-11

From RDF Working Group Wiki
Jump to: navigation, search

See panel, original RRSAgent log or preview nicely formatted version.

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

14:09:34 <RRSAgent> RRSAgent has joined #rdf-wg
14:09:34 <RRSAgent> logging to
14:09:36 <trackbot> RRSAgent, make logs world
14:09:36 <pchampin_> Guest: Tom (tbaker) Baker
14:09:36 <Zakim> Zakim has joined #rdf-wg
14:09:38 <trackbot> Zakim, this will be 73394
14:09:38 <Zakim> ok, trackbot; I see SW_RDFWG()11:00AM scheduled to start in 51 minutes
14:09:39 <trackbot> Meeting: RDF Working Group Teleconference
14:09:39 <trackbot> Date: 11 April 2012
14:19:35 <AndyS> AndyS has joined #rdf-wg
14:28:09 <danbri_> danbri_ has joined #rdf-wg
14:45:28 <gavinc> gavinc has joined #rdf-wg
14:57:15 <Zakim> SW_RDFWG()11:00AM has now started
14:57:15 <pchampin_> pchampin_ has joined #rdf-wg
14:57:21 <Zakim> +??P0
14:57:26 <AndyS> zakim, ??P0 is me
14:57:26 <Zakim> +AndyS; got it
14:57:34 <AndyS> trackbot, start meeting
14:57:37 <trackbot> RRSAgent, make logs world
14:57:39 <trackbot> Zakim, this will be 73394
14:57:39 <Zakim> ok, trackbot; I see SW_RDFWG()11:00AM scheduled to start in 3 minutes
14:57:40 <trackbot> Meeting: RDF Working Group Teleconference
14:57:40 <trackbot> Date: 11 April 2012
14:57:56 <AndyS> zakim, this is 73394
14:57:56 <Zakim> ok, AndyS; that matches SW_RDFWG()11:00AM
14:58:07 <Zakim> + +1.540.898.aaaa
14:58:09 <ivan> zakim, dial ivan-voip
14:58:09 <Zakim> ok, ivan; the call is being made
14:58:11 <Zakim> +Ivan
14:58:20 <NickH> +44.203.318.0479 isn't Zakim anymore!
14:58:30 <NickH> it is some American company
14:59:08 <Zakim> +??P3
14:59:20 <NickH> Zakim, ??P3 is me
14:59:20 <Zakim> +NickH; got it
14:59:21 <tbaker> tbaker has joined #rdf-wg
14:59:27 <NickH> NickH: still here
14:59:31 <NickH> arg
14:59:39 <davidwood> Zakim, who is here?
14:59:39 <Zakim> On the phone I see AndyS, +1.540.898.aaaa, Ivan, NickH
14:59:40 <Zakim> On IRC I see tbaker, pchampin_, gavinc, danbri_, AndyS, Zakim, RRSAgent, swh, MacTed, LeeF, mischat, ivan, Arnaud, davidwood, manu1, manu, yvesr_, NickH, trackbot, sandro, ericP
14:59:52 <davidwood> Zakim, aaaa is me
14:59:52 <Zakim> +davidwood; got it
14:59:58 <Zakim> + +1.707.861.aabb
15:00:00 <NickH> Zakim, mute me
15:00:00 <Zakim> NickH should now be muted
15:00:04 <gavinc> Zakim, aabb is me
15:00:04 <Zakim> +gavinc; got it
15:00:39 <PatH> PatH has joined #rdf-wg
15:01:04 <Zakim> + +1.781.899.aacc
15:01:45 <davidwood> tbaker, please try again
15:01:55 <Zakim> + +1.408.996.aadd
15:02:13 <Zakim> +??P8
15:02:15 <Arnaud> zakim, aadd is me
15:02:15 <Zakim> +Arnaud; got it
15:02:25 <Zakim> +??P13
15:02:28 <pchampin_> zakim, ??P8 is me
15:02:29 <Zakim> +pchampin_; got it
15:02:42 <tbaker> zakim, ??P13 is tbaker
15:02:45 <Zakim> +tbaker; got it
15:02:51 <ivan> zakim, who is here?
15:02:55 <Zakim> On the phone I see AndyS, davidwood, Ivan, NickH (muted), gavinc, +1.781.899.aacc, Arnaud, pchampin_, tbaker
15:03:03 <Zakim> On IRC I see PatH, tbaker, pchampin_, gavinc, danbri_, AndyS, Zakim, RRSAgent, swh, MacTed, LeeF, mischat, ivan, Arnaud, davidwood, manu1, manu, yvesr_, NickH, trackbot, sandro,
15:03:08 <Zakim> ... ericP
15:03:10 <Zakim> +??P14
15:03:12 <sandro> zakim, aacc is Sandro
15:03:16 <Zakim> +Sandro; got it
15:03:16 <swh> Zakim, ??P14 is me
15:03:24 <Zakim> + +31.20.598.aaee
15:03:26 <Zakim> +swh; got it
15:03:28 <Guus> Guus has joined #rdf-wg
15:03:37 <Zakim> +PatH
15:03:44 <ivan> zakim, aaee is Guus
15:03:55 <Zakim> +Guus; got it
15:03:58 <Zakim> + +1.781.273.aaff
15:04:09 <MacTed> Zakim, aaff is OpenLink_Software
15:04:19 <Zakim> +OpenLink_Software; got it
15:04:30 <AZ> AZ has joined #rdf-wg
15:04:39 <MacTed> Zakim, OpenLink_Softwareis temporarily me
15:04:41 <MacTed> Zakim, mute me
15:04:46 <pchampin_> scribe: pchampin
15:04:47 <MacTed> Zakim, OpenLink_Software is temporarily me
15:04:48 <MacTed> Zakim, mute me
15:04:51 <Zakim> I don't understand 'OpenLink_Softwareis temporarily me', MacTed
15:04:56 <Zakim> sorry, MacTed, I do not know which phone connection belongs to you
15:05:00 <Zakim> +MacTed; got it
15:05:01 <davidwood> PROPOSED to accept the minutes of the 4 Apr telecon:
15:05:01 <davidwood>
15:05:06 <Zakim> MacTed should now be muted
15:05:06 <pchampin_> topic: last week minutes
15:05:27 <PatH> zakim, mute me
15:05:32 <Zakim> PatH should now be muted
15:05:38 <Souri> Souri has joined #RDF-WG
15:05:40 <pchampin_> RESOLVED to accept the minutes of the 4 Apr telecon
15:05:44 <davidwood> Action item review:
15:05:44 <trackbot> Sorry, couldn't find user - item
15:05:44 <davidwood>
15:05:44 <davidwood>
15:05:54 <pchampin_> topic: action item review
15:06:15 <Zakim> +??P25
15:06:29 <gavinc> Andy's review is complete. And we asked Pierre Antoine not to do it yet
15:06:41 <AZ> zakim, ??P25 is me
15:06:41 <Zakim> +AZ; got it
15:06:50 <Zakim> + +1.603.897.aagg
15:07:15 <AndyS> q+
15:07:15 <Souri> zakim, aagg is me
15:07:16 <Zakim> +Souri; got it
15:07:32 <AndyS> q-
15:07:44 <Zakim> +LeeF
15:07:56 <AndyS> My review is gone-as-much as i can - need to do grammar 
15:09:31 <AndyS> close ACTION-157
15:09:31 <trackbot> ACTION-157 Review Turtle LC draft by April 9 closed
15:09:52 <Zakim> + +1.617.324.aahh
15:09:56 <pchampin_> david: Tom Baker is with us today
15:10:05 <ericP> Zakim, aahh is me
15:10:05 <Zakim> +ericP; got it
15:10:09 <pchampin_> ... he is involved in the RDF for Library group
15:10:19 <ivan> s/is/was/
15:10:23 <pchampin_> ... and came to us to discuss a use case he has for named graphs
15:10:36 <ivan> s/Library group/Library Incubator group/
15:10:47 <pchampin_> ... may be an important one to include in our short list
15:11:06 <pchampin_> topic: named graphs
15:11:23 <gavinc> Functional Requirements for Bibliographic Records (FRBR)
15:11:24 <davidwood> s/Library Incubator group/Library Linked Data Interest Group/
15:11:40 <pchampin_> tbaker: distinction between general properties of a resource (title, subject)
15:12:04 <ivan> s/Data Interest Group/Data Incubator Group/
15:12:08 <pchampin_> ... and very specific properties (e.g. the fact that it's missing a page)
15:12:49 <PatH> zakim, unmute me
15:12:49 <Zakim> PatH should no longer be muted
15:13:05 <ivan> zakim, mute me
15:13:05 <Zakim> Ivan should now be muted
15:13:07 <pchampin_> ... FRBR has different levels, from very generic to very specific (Work/Expression/Manifestation/Item)
15:13:17 <pchampin_> ... semantized as four disjoint classes
15:13:52 <pchampin_> ... problem: in the end of the day, I'm describing a book, a single thing, not 4 distinct entities
15:14:34 <pchampin_> ... if I'm describing a bible, some properties will apply to the Work or the Expression
15:14:41 <pchampin_> ... while some others will apply to the Item
15:14:47 <ivan> -> Functional Requirements for Bibliographic Records
15:14:48 <Zakim> -gavinc
15:14:55 <pchampin_> ... but I don't want to have to specify to which it applies for every property
15:15:39 <pchampin_> ... idea: to use named graphs
15:15:44 <Zakim> +gavinc
15:16:09 <pchampin_> ... you have 4 chunks of  statements : 
15:16:15 <ivan> -> Tom's email with the details
15:16:32 <pchampin_> work-level, expression-level, manifestation-level, item-level
15:16:50 <pchampin_> ... work-level, expression-level, manifestation-level, item-level
15:17:14 <pchampin_> ... each maintained at different level (national library for the work-level, personal? for item-level)
15:17:33 <pchampin_> ... better solution than to copy-paste info from high-level to low-level?
15:18:33 <pchampin_> ... refering to "frames" as a mean to merge information from different levels; is that the same as a graph?
15:18:53 <PatH> q+
15:18:54 <davidwood> q?
15:19:00 <gavinc> q+
15:19:57 <pchampin_> david: where do you see the line between what should be handled by RDF itself, and what should be handled by a higher level application?
15:20:51 <ivan> ack PatH 
15:20:59 <pchampin_> tbaker: the notion of "frame" seems to be close to the notion of named graph, hence relevant for RDF
15:21:05 <davidwood> tbaker: Hopes that RDF will support named graphs and merges of named graphs.
15:21:08 <davidwood> ack PatH
15:21:30 <pchampin_> ... then the way item-level inherits manifestation-level but not the other way around would be application specific
15:21:49 <pchampin_> pat: why named graphs? why not simply graphs?
15:22:35 <pchampin_> tbaker: provenance is an important feature for the library world
15:24:10 <pchampin_> pat: then what do you want to name? a static graph or a "box" whose content may change in time?
15:25:09 <pchampin_> tbaker: that would rather be the "box"
15:25:17 <pchampin_> ... although there would be some versioning issue,
15:25:42 <pchampin_> ... you would want to be able to access a previous version of a given description level
15:25:46 <davidwood> ack gavinc
15:26:15 <tbaker> URL of the FRBR use case mentioned by Gavin?
15:26:18 <pchampin_> gavin: I hope that some of our use cases match, as they come from an implementation of FRBR in RDF
15:27:35 <tbaker> +1 for dynamic
15:27:44 <pchampin_> gavin: in that implementation, the naming *had* to be dynamic (i.e. name "boxes" rather than static graphs)
15:27:57 <pchampin_> ... as different organizations maintained different levels
15:28:30 <pchampin_> ... and they had to reference someone else's description, despite the fact that this description might evolve
15:28:40 <tbaker> q+ to ask if Gavin has proposed something quite similar to this concept?
15:29:35 <pchampin_> gavin: not everything true about the work has to be true about the item
15:29:58 <davidwood> ack tbaker
15:29:58 <Zakim> tbaker, you wanted to ask if Gavin has proposed something quite similar to this concept?
15:29:59 <pchampin_> ... e.g. the title of the item may include the words "2nd edition", while the title of the work does not include them
15:30:01 <PatH> q+ for my third q
15:30:42 <pchampin_> tbaker: this is an interesting question to the library people
15:30:51 <pchampin_> ... which use case are about FRBR, precisely?
15:31:30 <pchampin_> gavin: the term FRBR does not appear explicitly, but several of them come from that implementation
15:31:40 <pchampin_> ... an example is: using the subject of the graph as the name of the graph
15:31:40 <PatH> tbaker says +1 for dynamic but he also wants strict provenancing for old versions, so there has to be some static stuff under the hood.
15:32:29 <pchampin_> tbaker: this notion of primary subject is an important question for us
15:32:29 <tbaker> q-
15:33:05 <PatH> i will stay on the q until this topic is done.
15:33:12 <ericP> q?
15:33:13 <pchampin_> david: we've been discussing a way to associate information to the graph itself, rather than the graph content
15:33:23 <Zakim> -NickH
15:33:25 <pchampin_> ... would that address this question about the primary subject?
15:33:33 <ericP> q+ to present two representations <>
15:33:35 <davidwood> ack PatH
15:33:35 <Zakim> PatH, you wanted to discuss my third q
15:33:37 <pchampin_> tbaker: yes, that would be helpful
15:34:01 <ericP>
15:34:11 <gavinc> +q to answer why you need named graphs
15:34:50 <gavinc> -q
15:35:07 <pchampin_> eric: what is the motivation again for having separate graphs rather than one big graph?
15:35:14 <pchampin_> ... (see URL above)
15:35:28 <pchampin_> tbaker: the motivation is to have the different graphs maintained by different people
15:35:40 <pchampin_> ... and different provenance information associate with them
15:37:18 <MacTed> descriptionOfWork generated by, transcribed by, input by...
15:37:18 <MacTed> work generated by, edited by, printed by...
15:37:18 <MacTed> regarding "primary subject", these may be useful predicates -- foaf:primaryTopic , foaf:topic
15:37:28 <ericP>
15:37:51 <davidwood> ack ericP
15:37:51 <Zakim> ericP, you wanted to present two representations <>
15:39:24 <gavinc> btw there should be relationships between work1G expression1G manifestation1G and item1G
15:39:30 <gavinc> otherwise I have no idea why you'd bother ;)
15:39:43 <ivan> q+
15:39:51 <davidwood> Social convention is allowed :)
15:41:23 <MacTed> context!
15:41:28 <pchampin_> eric: where would I put information stating "Moby Dick is a bad whale"?
15:41:33 <ericP>
15:42:04 <pchampin_> tbaker: this is the kind of question that would generate a lot of discussion in the library community
15:42:13 <ericP> gavinc, can you give me a relationship to add to the trig?
15:42:16 <pchampin_> ... everyone agrees that it is good to have different levels
15:42:23 <PatH> tom just made the case for rdf :-)
15:42:24 <ericP> (between work1G and expression1G)
15:42:34 <pchampin_> ... but there is disagreement about where to draw the line
15:42:47 <gavinc> ericP: will help ;) 
15:42:57 <davidwood> ack ivan
15:43:59 <PatH> q+
15:44:04 <pchampin_> ivan: in eric's example, the graphs are mostly not relating to each other
15:44:16 <pchampin_> ... so something must be missing
15:44:36 <mox601> mox601 has joined #rdf-wg
15:44:58 <pchampin_> ... how do you expect the properties to "fly" from work to item, e.g.,
15:45:11 <pchampin_> ... do you expect a generic mechanism to take care of this for you?
15:45:13 <gavinc>
15:45:24 <gavinc> Example FRBR relationships
15:45:25 <pchampin_> ... or would that be FRBR-specific rules?
15:45:38 <MacTed> Zakim, unmute me
15:45:38 <Zakim> MacTed should no longer be muted
15:46:19 <MacTed> there are triples left out of the "Many Named Graphs" example
15:46:33 <pchampin_> tbaker: re. FRBR-specific rule, same answer as to David's question sooner.
15:46:34 <MacTed> they *are* there in the "One RDF Graph" ... and they *should* be in the second as well
15:46:36 <gavinc> properties frbr:realization, frbr:realizationOf etc
15:48:14 <MacTed> q+
15:48:41 <MacTed> q+ to say inference is commonplace for e.g., subclassOf
15:49:38 <pchampin_> tbaker: the "inheritance" rules would be FRBR-specific; but what can be generic is how to merge different kinds of informations stored in different named graphs
15:49:46 <ericP> copies the subject relationships from the turtle (but no graph relationships)
15:50:04 <tbaker> Ivan: we have to be able to express somehow that we take the union of those four graphs and operate rules on the union of the four graphs.
15:50:23 <pchampin_> ivan: from your example, I understand that we need a mechanism to merge the content of several graphs together
15:50:42 <pchampin_> ... so that we can then do something with that union of graphs: like rules, RDFS, etc...
15:51:17 <tbaker> Pat: If I understand FRBR, can see that there need to be links starting from item and going to more general... A chain from the item back up to the work.  What I don't see is any particular need for links going down the hierarchy.
15:52:24 <gavinc> PatH, realization, realizationOf
15:53:36 <davidwood> q?
15:53:40 <pchampin_> tbaker: one would want to be able to follow the relations in both directions
15:53:40 <davidwood> ack PatH
15:54:04 <Zakim> +Arnaud
15:54:13 <gavinc> PatH,
15:54:29 <danbri_> folks, sorry I can't join the call. I wrote a giant something on FRBR a year ago; ... short version: FRBR is about describing mass-produced entities, and the link to our notion of 'class' needs examining.
15:54:43 <pchampin_> pat: why not using class-instance relation in the FRBR hierarchy instead of specific relations?
15:54:49 <davidwood> Thanks, danbri
15:54:58 <pchampin_> ... e.g. Items are instances of the corresponding Expression
15:55:11 <ericP> -> with some N3 handle inference
15:55:18 <gavinc> +1 danbri
15:55:53 <Zakim> -Arnaud
15:56:00 <davidwood> ack MacTed
15:56:00 <Zakim> MacTed, you wanted to say inference is commonplace for e.g., subclassOf
15:56:25 <tbaker> +1 danbri's post on public-lld
15:57:00 <pchampin_> macted: the fact that FRBR does not model it as class-instance does not prevent one's specific ontology to add that relation and use it for inference
15:57:26 <pchampin_> s/corresponding Expression/corresponding Manifestation/
15:57:55 <pchampin_> tbaker: there has been counter-proposals to the 4 distinct levels
15:58:00 <PatH> +1 also danbri which even has a T-shirt design
15:58:05 <pchampin_> ... someone made a proposal with only 3 levels
15:58:18 <pchampin_> ... music catalogs have a different classification
15:58:46 <davidwood> q?
15:58:56 <danbri> (tshirt example ... that's as close as I've come to finding a role for OWL in my life :)
15:59:02 <tbaker> q+ to ask about next steps
15:59:36 <pchampin_> macted: what about containers?
15:59:54 <pchampin_> tbaker: this is another big discussion
16:00:04 <davidwood> ack tbaker
16:00:04 <Zakim> tbaker, you wanted to ask about next steps
16:00:25 <Zakim> -Arnaud
16:01:20 <pchampin_> david: I propose to put tom's use cases in our short list
16:02:01 <sandro> Pat: What Tom seems to need is the ability to get at graphs, at bits of graphs, and do reasoning with those bits of graphs        [ scribe attempt ]
16:02:26 <davidwood> …also to attach metadata to graphs
16:02:43 <PatH> Toms use case needs named graphs to support provenance information, metadata, and inference access to the contents of the named graphs, all combined in application-specific ways. 
16:03:40 <ericP> <> uses graphs to perform inferences performed by ?p ?o in <> 
16:03:43 <tbaker> q+ to suggest putting it in wiki with Eric's example
16:04:04 <pchampin_> sandro: sounds very general, difficult to say if it is addressed or not
16:04:14 <pchampin_> pat: this is more of an aggregate use case
16:04:17 <davidwood> ack tbaker
16:04:17 <Zakim> tbaker, you wanted to suggest putting it in wiki with Eric's example
16:06:06 <pchampin_> david: could Tom and Eric formulate these use cases on the wiki?
16:06:40 <tbaker> q+ to quickly respond to Pat
16:06:49 <ivan> q+
16:06:53 <ivan> ack tbaker 
16:06:53 <Zakim> tbaker, you wanted to quickly respond to Pat
16:07:05 <gavinc> There is AN ontology which may be horribly wrong
16:07:25 <davidwood> ACTION: EricP to work with Tom Baker to add the FRBR use case to 
16:07:25 <trackbot> Created ACTION-162 - Work with Tom Baker to add the FRBR use case to  [on Eric Prud'hommeaux - due 2012-04-18].
16:07:45 <pchampin_> pat: my concern is: should we focus first on RDF or on the ontology? but let's go on RDF
16:09:02 <davidwood> q?
16:09:02 <pchampin_> tbaker: focusing on the RDF and named graphs will help keep the core of the ontology simple [scribe hopse he got it right]
16:09:08 <davidwood> ack ivan
16:09:33 <PatH> This may be a first, discussing ontology standards while emptying a cat litter tray.
16:09:33 <pchampin_> ivan: I don't feel that we are so far off to address Tom's use cases
16:10:12 <PatH> +1 to ivan on union issue.
16:10:20 <gavinc> Need to be able to construct unions too!
16:10:32 <PatH> i feel jubilant.
16:10:48 <tbaker> I feel jubiliant that Pat feels jubilant
16:10:48 <pchampin_> ... if we agree that the default graph of a dataset would always be the union of all the named graph, then we might have what Tom needs
16:11:36 <sandro> q
16:11:38 <sandro> q+
16:12:08 <AndyS> Maybe it is not feature of the TriG exchanged, but is done with the named graph once received.
16:12:11 <pchampin_> ... there should be no semantic problem; it's the syntactical details that need to be worked out
16:12:20 <ericP> -> persuant to ACTION-162
16:12:51 <AndyS> Maybe it is not feature of the TriG exchanged, but is about what is done with the collection of named graphs once received. -- requirement no dft graph in this case.
16:12:54 <tbaker> david: Proposal already addresses metadata on graphs (Ivan's formalization of Sandro's as tweaked by Pat).  The deltas are in relation to where we want to draw the line on inference.
16:13:01 <PatH> we never say X is always Y if we can give users an easy way to say this particular  X is a Y
16:13:25 <pchampin_> david: the remaining question would then be: where do we draw the line for inference, right?
16:13:30 <tbaker> Ivan: We draw the line on inference.  Ontology-based inference based on FRBR is out there, we should not touch, but in my proposal, inferencing done only on the default graph.
16:14:01 <tbaker> ... This is not the way tom wants to use them.  Rather: take the union and do inference on that.  Additional trick needed.
16:14:01 <pchampin_> ... with a mechanism to "quote" some of the graphs (i.e. keep them out of inference)
16:14:24 <davidwood> ack sandro
16:14:25 <tbaker> david: In order to facilitate that inferencing, Tom's proposal may be missing a layer.
16:14:30 <sandro> sandro: How about we have Merge-Default-Graph-Datasets and Explicit-Default-Graph-Datasets, and use Trig {} to serialize the difference?
16:14:43 <pchampin_> david: it seems to me that Tom's proposal is missing a layer, though
16:15:09 <davidwood> Does the inferencing requirement suggest the need for a graph that holds named graphs (single-level nesting)?
16:15:10 <tbaker> Sandro: As Ivan puts it: in SPARQL world, two types of datasets.  Never been explicit about ... . 
16:15:14 <PatH> I have to go very soon.
16:15:25 <pchampin_> sandro: in the SPARQL world, there has always been two kinds of dataset
16:15:27 <AndyS> q+
16:15:35 <davidwood> ack AndyS
16:15:50 <pchampin_> ... need a way to make this explicit in Trig
16:16:10 <PatH> I got to go, sorry.
16:16:17 <Zakim> -PatH
16:16:18 <Zakim> -gavinc
16:16:36 <tbaker> Thank you all for the great input - looking forward to follow-up!
16:16:54 <davidwood> Thanks, tbaker!
16:16:56 <AndyS> Thanks Tom.
16:17:46 <davidwood> adjourned
16:20:04 <ericP> tbaker, take a peek at <> ?
16:20:59 <Zakim> -swh
16:21:15 <tbaker> ok - do we want to fold in some of the text from
16:21:57 <mischat> mischat has joined #rdf-wg
16:24:55 <Zakim> -Guus
16:25:17 <Zakim> -AndyS
16:25:20 <ericP> tbaker, yeah, i linked to it as a placeholder. it should absolutely recapitulate everything that folks need to swap in the decisions parameters
16:25:55 <davidwood> pchampin, please see the directions at the top of