15:59:03 RRSAgent has joined #annotation 15:59:03 logging to http://www.w3.org/2015/11/04-annotation-irc 16:04:24 PaoloCiccarese has joined #annotation 16:05:38 takeshi has joined #annotation 16:06:23 davis_salisbury has joined #annotation 16:06:36 present+ davis_salisbury 16:06:56 Present+ Rob_Sanderson, Frederick_Hirsch, Paolo_Ciccarese 16:07:04 trackbot, start telecon 16:07:05 Present+ Tim_Cole, Benjamin_Young 16:07:06 RRSAgent, make logs public 16:07:08 Zakim, this will be 2666 16:07:08 I do not see a conference matching that name scheduled within the next hour, trackbot 16:07:09 Meeting: Web Annotation Working Group Teleconference 16:07:09 Date: 04 November 2015 16:07:19 Present+ Doug_Schepers 16:07:27 Present+ Takeshi_Kanai 16:07:32 Chair: Rob_Sanderson, Frederick_Hirsch 16:07:39 Present+ Benjamin_Young 16:07:52 Present+ TB_Dinesh 16:08:00 Agenda: https://lists.w3.org/Archives/Public/public-annotation/2015Nov/0005.html 16:08:06 Present+ Tim_Cole 16:08:16 fjh has changed the topic to: https://lists.w3.org/Archives/Public/public-annotation/2015Nov/0005.html see members list for logistics 16:08:29 present+ shepazu 16:08:38 Present+ Paolo_Ciccarese 16:08:57 i I have called in, sorry, new computer is wonky 16:08:59 Jacob has joined #annotation 16:09:10 Present+ Davis_Salisbury 16:09:16 Topic: Agenda Review, Scribe Selection, Announcements 16:09:23 having trouble with it, sorry. 16:11:47 I can give it a shot 16:11:59 scribenick: davis_salisbury 16:12:22 Getting started 16:13:08 Frederick sent an amended version of the minutes 16:13:20 q+ to ask about web platform minutes 16:13:26 ... that Ben is a co-editor and not a chair 16:13:33 ack fjh 16:13:33 fjh, you wanted to ask about web platform minutes 16:13:59 Are there additional minutes from the Web Platform meeting? 16:14:22 They should exist and will be tracked down/ 16:14:40 No more questions about the minutes. 16:14:41 proposed RESOLUTION: Minutes from TPAC approved: http://www.w3.org/2015/10/25-annotation-minutes.html http://www.w3.org/2015/10/26-annotation-minutes.html 16:15:19 proposed RESOLUTION: Minutes from TPAC breakout approved, following change to clarify editor/chair status http://www.w3.org/2015/10/28-annotation-minutes.html 16:15:52 Hi :) Ok now I have to remember what I said. Oh dear... I think the basic idea was to encourage you to find a balance between giving the majority of users (implementors, spec readers, annotators, ...) some sensible default general purpose vocabulary, while also leaving open the option for those "in the know" to make the informed decision to switch to a different vocabulary that more closely suits their needs. At schema.org we're work[CUT] 16:16:09 trying now, just cut and paste those two above? 16:16:22 http://www.w3.org/2015/10/25-annotation-minutes.html http://www.w3.org/2015/10/26-annotation-minutes.html 16:16:31 Web Platform WG minutes: http://www.w3.org/2015/10/21-webapps-minutes.html#item06 16:16:54 TPAC was very productive with alignment and where things are going 16:17:11 .... Monday mostly talked about FindText 16:17:26 .... on Tuesday we talked mostly about the model and Protocol 16:17:52 .... on Wedneseday the break out was fairly well attended, roughly 20 attendees 16:18:05 .... good feedback generally. 16:18:53 .... Monday specifically discussed some internationalization issues 16:19:08 q+ to ask whether regex vis edit distance is an open issue or was resolved at F2F, what should we know about this? 16:19:29 .... also with the Web Platform group discussed the complexity of the FindText API and how to address those concerns 16:20:19 RESOLUTION: Minutes from TPAC approved: http://www.w3.org/2015/10/25-annotation-minutes.html http://www.w3.org/2015/10/26-annotation-minutes.html 16:20:50 Doug says that there was not concrete feedback as they had just read the spec before discussing it 16:20:56 ack fjh 16:20:56 fjh, you wanted to ask whether regex vis edit distance is an open issue or was resolved at F2F, what should we know about this? 16:21:00 .... Doug will follow up and seek more feedback 16:21:05 do we have an open github issue on regex and findtext 16:21:28 .... one concrete thing they did give feedback on was serialization. 16:21:32 streaming seems to always be a consideration 16:22:14 .... they are open to the idea of FindText API generally but whether they are open to this one specifically remains to be seen. 16:23:00 RESOLUTION: Minutes from TPAC breakout approved, following change to clarify editor/chair status http://www.w3.org/2015/10/28-annotation-minutes.html 16:23:40 Doug has already asked for Dom to have the issues for FindText reflected in the mailing list, which is done 16:24:14 .... regarding the whether or not we are going to keep the edit distance, Doung does not think that it was seriously looked at by the browser vendors 16:25:02 .... Doug will follow up with them about which items were serious and which were not. General takeaway was that at a distance was a possible blocker but more feedback is necessary 16:25:47 Another issue the addition of a paramater that would allow for regions of content to be speced 16:26:01 .... it would be good to have an issue to cover this issue 16:26:35 oops. good to create an issue in re media fragments and regions of content 16:26:58 .... while it is not listed as a specific deliverable it is still related to Robust Anchoring and should perhaps be considered 16:27:10 q+ to ask about Dublin core vs schema.org when appropriate 16:27:14 ..... i.e. find text is not the only thing discussed 16:27:17 ack fjh 16:27:17 fjh, you wanted to ask about Dublin core vs schema.org when appropriate 16:27:40 s/it would be good to have an issue to cover this issue// 16:27:47 s/oops. // 16:27:59 Tuesday: group talked with the Social Web WG members present 16:28:21 .... that discussion revolved somewhat around parts of Activity Streams, like collections 16:28:33 .... and how we can use the collections formalization 16:28:50 .... main issue is timing, and their Charter lasts until the end of next year 16:29:04 .... it is not clear that their work can be referenced by us 16:29:13 .... in time for our Charter. 16:29:51 The proposal at the moment is that we should, at the end of the calendar year. accept the status of Activity Streams 16:30:31 .... we should reference this for now but in the future make a slight change to the context and refer to the full ontology once it is (hopefully) produced 16:31:00 There was also some discussion around the Protocol and paging, particularly with respect to the overlap with list and choice 16:31:16 .... and how playlists for annotations would work etc. 16:31:36 .... there was discussion around Dublic Core vs. schema.org 16:31:55 ..... everyone up the chain was present, including Ivan and Tim BL 16:31:57 q+ was discussion about moving away from dublin core? 16:32:06 q+ 16:32:12 ack fjh 16:32:52 Step 1: can we reference schema.org, which we can 16:33:16 .... we still have the question about whether to use both but there is no resolution 16:34:18 Also talked with Dan Brinkley, and the ordering available via schema.org was too complex, better to use Activity Streams 16:34:49 People who want to propose changes should create issues in github with rationales, otherwise use the original issue 16:35:27 Doug spoke with Dan about this, and he believes that what we should do is not pick a winner but instead leave open the possibilty of allowing both 16:35:53 .... it is an interesting approach i.e. not referring to one schema but allowing people to have a choice of vocabularies 16:36:14 .... a link to the discussion can be posted to track the discussion 16:36:34 q+ 16:36:43 q+ to ask about issue of syncing annotations and offline 16:36:43 Other topics on Tuesday were logistics related, notably: do we need another fact-to-face meeting 16:37:05 ..... tentative proposal was yes, sometime in March or April in Europe. 16:37:09 ack TimCole 16:37:47 Tim: required clarity in re paging in Protocol or also in how we model it 16:38:14 The particular issue discussed was in order to have ordering you need to have a list item that asserts the position in the list 16:39:24 Tim: in some of the examples on the site, a list is provided with a property as to whether it was ascending or descending, which was confusing 16:39:56 .... there were essentially two different methods shown 16:41:12 ack fjh 16:41:12 fjh, you wanted to ask about issue of syncing annotations and offline 16:42:05 For about an hour we tried to talk through some of the open issues, and Issue 21 was specifically discussed 16:42:12 Could someone explain exactly what the itemListOrder predicate accomplishes? Is this something a client is expected to consume and exploit to apply some sorting algorithm on the client-side? Doesn't that conflate content and presentation? 16:43:19 ..... a few comments were added [sorry having trouble keeping up with this one] 16:43:54 How can a third party merger two conversations? Strong use cases for having an equivalence notion between annotations 16:44:35 which ATOM item is this? is this a layer of functionality that this group needs to work on, or is it for v.next or later 16:44:42 The discussion in the hallway was to suggest that an annotation came from a particular URI 16:45:22 Right now the use cases were deemed strong enough and should be added back 16:45:37 ..... and Rob thinks that they would not require that much work 16:46:08 Doug understood that the group considered that a UUID could be used to resolve this via being created on the client side 16:46:45 The proposal is essentially that we allow a property on an annotation called via that would be a use of the IANA via operation 16:47:01 .... the value of that property could be an IRI that was a UUID 16:47:28 q? 16:48:09 q? 16:48:16 The containing element is a little bit weak possibly, but the overall semantics seem correct. We should have a longer discussion leading to a former proposal for a future call. 16:48:56 The other logistical question: what can we get done before the end of our Charter. Also, by when do we need to to do this 16:49:35 We should of course focus on the model and FindText and try to get all of these in a solid state by April so that overall timing works out 16:49:49 .... officially the Charter ends in October 16:50:23 .... but we want to make sure that we get what we want into the process by April so that we do not run the risk of having the Charter shut down and our hard work wasted 16:50:42 Frederick Ivan and Rob have been discussing how to make progress 16:50:55 q+ 16:51:11 .... how do we make a sustainable march forward with recommendations and work to get what we want done 16:51:19 ack shepazu 16:51:21 We should put this on the agenda for the next call 16:51:49 Doug: the impression was that this group would be re-chartered, and that that would be the expectation 16:52:17 .... the idea was that we were in good shape and that the idea was overall to re-charter instead of getting an extension 16:52:38 The discussion was around the risk of not getting re-chartered 16:53:01 rrsagent, generate minutes 16:53:01 I have made the request to generate http://www.w3.org/2015/11/04-annotation-minutes.html fjh 16:53:15 q+ 16:53:20 ack TimCole 16:53:23 There is every intention to re-charter but was must show that we are making good progress or we run that risk, i.e., we need to make sure we prove that we are doing good work and making progress 16:53:38 Tim: how much do we need to get done to reach the stage in regards to testing? 16:53:41 Present+ Jacob_Jett 16:54:04 We talked about it, and there were a lot of changes to the data model, but are we close enough to being stable for testing? 16:54:32 This was discussed somewhat at TPAC but no real resolutions were reached 16:54:54 ..... we must have a full test suite at least but we don't necessarily need it to go into CR 16:55:12 .... as long as we are solid then we will have the first few months of CR to work on it 16:55:35 ..... we don't want to go into CR without a plan though as we don't want to get to CR and risk not getting the test done 16:56:22 For testing we don't necessarily need a big bang i.e. some things can be done now that we know are fairly stable. Anyone chomping at the bits to do testing should go ahead and do it 16:56:32 ..... any progress will help going into CR 16:56:43 s/For testing/fjh: For testing/ 16:57:04 q? 16:57:08 tantek has joined #annotation 16:57:11 Doug: there was a discussion about client-side testing, and a testing methodology, that would be easy to maintain. Hopefully this will be started sooner rather than later. 16:57:46 s/i I have called in, sorry, new computer is wonky// 16:57:54 s/having trouble with it, sorry.// 16:58:00 s/I can give it a shot// 16:58:16 One more thing: we had a slot for some implementation demonstrations. Takeshi showed off some progress for enabling annotations on a reader. Benjamin did a demo as well, some others 16:58:22 s/Getting started// 16:59:12 Hopefully we are now up to speed in re the TPAC discussions and we can now make progress on the issues 16:59:15 s/They should/shepazu: They should/ 16:59:56 i/Agenda Review/note - all comments from Rob unless noted/ 17:00:30 rrsagent, generate minutes 17:00:30 I have made the request to generate http://www.w3.org/2015/11/04-annotation-minutes.html fjh 17:00:49 Thanks, I will try to work on it for next time. 17:01:07 tbdinesh has joined #annotation 17:01:26 Topic: Other Business 17:01:44 Thanks for scribing Davis 17:01:49 Topic: Adjourn 17:01:54 rrsagent, generate minutes 17:01:54 I have made the request to generate http://www.w3.org/2015/11/04-annotation-minutes.html fjh 17:51:01 tilgovi has joined #annotation 18:24:22 timbl has joined #annotation 18:32:26 tantek has joined #annotation 18:57:20 kevinmarks has joined #annotation 19:29:36 Zakim has left #annotation 19:55:14 tilgovi has joined #annotation 19:56:59 azaroth has joined #annotation 20:20:57 I don't want to clutter up issue #96 (https://github.com/w3c/web-annotation/issues/96) with a meta-discussion 20:21:21 So maybe someone here can explain to me the value of adding things to the data model that are optional and elsewhere modeled 20:21:36 Why do we say anything about creation time at all, for instance? 20:23:05 Or, as sort of an interesting inquiry, what would be the most minimal data model possible for annotations that left it open to implementors to choose from other vocabularies as needed and evolve best practices outside the web annotations data model spec document 20:41:05 m4nu has joined #annotation 21:41:28 tilgovi: I think a good bit of these things are for "JSON developers" 21:41:44 RDF people obviously already know they can model the entire world, given the appropriate vocabulary ;) 21:42:03 for the JSON folks, we're giving them "prebaked" mappings to URLs that have meaning--whether they know it or not ;) 21:42:48 if that target audience knew there was such a thing as a "vocabulary" (and that they were in effect limitless), then we'd really only be specifying a very few things 21:43:27 ...and perhaps azaroth's new "RDF-y" model document will focus on those bits...and the JSON-LD focused one I'm making will simply call out these additional keys in usage scenarios (as it mostly does now...) 21:52:02 I think I am +1 on splitting model and vocabulary 21:52:17 Maybe even a best practices or example library separate from both 21:52:46 the JSON-LD doc is 1 of (probably) a handful of "serialization" docs...technically 21:52:54 but in this case we plan to lead with that do--for most developers 21:53:04 and reference the model and/or vocabulary docs from that 21:53:13 Because it's a little strange to me to see the data model document cluttered with creator, creation time, generator, etc and see discussions on the issues about updated time, license and more 21:53:28 Or maybe, it's not strange 21:53:30 but it is noisy 21:53:47 it obscures the simplicity of what annotation may be at its core 21:53:50 The model would still need to reference those things, but separate from the annotation specific ontology 21:53:57 IMO 21:54:15 Here is what we define: (bla bla bla) ; Here are other things that we recommend you use too: (bla bla bla) 21:55:06 But if the community turns around and says screw dublin core, we're all in on schema.org, that shouldn't affect what we've done beyond updating the context for JSON-LD 21:55:17 and from the serialization document perspective, I'd like to hand people a "what we expect you to record" sort of document 21:55:25 even if...at it's core...there's not very much 21:55:58 And the counterpoint is that if you don't say "you should give this information" then no one will, or there won't be sufficient convergence to make the annotations useful in practice 21:56:28 E.g. if there isn't an agreed upon way to assert the creator of an annotation, then spam filtering, reputation modeling, author search (etc) all go out the window 21:56:42 and those are valuable to enable from v1.0 I think 21:56:51 ...or all become extremely custom 21:57:02 and if that's "just JSON", then we've no sure way to map it back to anything meaningful 21:57:48 Yup :) The goal is interoperability of systems, not defining a custom format. We just have to make it look like we're defining a custom format some of the time :/ 21:58:55 21:59:06 hehe 21:59:18 tilgovi: you're spot on that there is (deliberately) not much "going on" in this model 21:59:24 and that the core bits are Selector's and such 22:00:15 bigbluehat: so ... annotates ... 22:00:46 I agree with Ivan and Jacob that it doesn't directly help anyone. 22:01:04 OTOH from a *protocol* perspective, re-invigorating the rel type would be valuable 22:01:26 "if the community turns around and says screw dublin core, we're all in on schema.org, that shouldn't affect what we've done beyond updating the context for JSON-LD" 22:01:28 yeah. it's mostly the rel value I think has...value 22:01:46 And if we just didn't talk about these things because their tangential to some more core aspect of annotation then there wouldn't be a need to update any context 22:01:51 *they're 22:02:17 except that if we don't define them in a context, then every other developer will have to too 22:02:18 tilgovi: True enough, but see the text wall regarding interop 22:02:32 the tl;dr is that "graph know how" becomes a requirement to successfully be that flexible 22:02:40 ...which is why RDF exists in the first place 22:02:50 If we're opinionated that there should be a way everyone does something, we should say so. 22:02:57 and why JSON can safely be considered insufficient for modeling the world's hive mind ;) 22:03:04 I'm not sure we really know what our interop goals are. I mean, is being able to, say, sort the annotations by creation time something that we really care about interop on? 22:03:27 Yes 22:03:28 it was for RSS and Atom fwiw 22:03:37 And harvesting. And search. 22:03:51 regardless of that use case, I think we're (still) not wanting to require understanding RDF. 22:04:02 Otherwise you would get all of the annotations in a container in some random order when you ask for them via the protocol 22:04:18 if folks knew that (in its fullest) and had the proper tooling in their server and client, then all these specs (AS2, etc) would be *much* smaller 22:04:19 I don't see the problem with that for any use case I have 22:04:29 Especially if the annotation is separate from the body 22:04:48 The body we're not specifying its shape and as a user I may really want to be sorting on the creation time of the body 22:04:56 tilgovi: I will say that getting a date-sorted list back from http://hypothes.is/api/search is pretty handy :) 22:05:06 Not when some crawler indexed my comment as an "oa:commenting" annotation 22:05:19 tilgovi: true that. 22:05:43 How would the crawler know if the annotation is new or stale? 22:05:48 although, that's oa:generated 22:05:57 not oa:created 22:05:57 but the door's been (deliberately) left open to *anything* being a body of annotation 22:06:04 and not all those things have created time 22:06:08 known...or recorded 22:06:15 I guess what's really stressing me out is that I don't see the applications that create annotations 22:06:29 ? 22:06:35 you built at least 2 of them :) 22:06:38 I see the applications that create bodies which may or may not explicitly reference targets and annotations as objects that a crawler might infer for some purpose of normalization and indexing 22:06:54 Yeah, but Hypothesis co-creates the annotation and the body 22:07:00 Hypothesis hasn't even disentangled the two, yet 22:07:04 right, and most people will do the same 22:07:04 Because it has zero compelling reason to 22:07:22 I think that shims will create just annotations 22:07:33 And the target used to be part of the same object with the body and all the other metadata but I separated that partly just to align with the spec as a gesture of interest 22:07:36 however, I'd still like to be able to say: {"body": "http://twitter.com/....blah", "target":"http://w3.org/"} 22:07:52 For example, something that mapped youtube comments into oa:Annotations wouldn't create either the target (the video) of the comment 22:08:19 And for the foreseeable future, there'll be a lot of those systems out there 22:08:25 we need a format (and have one, I think) that can be used to map existing resources into an "annotation space" 22:08:34 which is not something we'd have the other way 'round 22:08:46 along the same lines as ... that system tantek used for facebook / twitter from blog post 22:08:56 bridgy 22:08:58 voila 22:10:10 To me, so long as we don't *require* features, and each feature can be justified and demonstrated (a requirement for TR anyway), having them in the specs is better than silence 22:10:15 As you can just ignore them 22:10:26 Don't need created time, then don't put it in 22:10:30 :) 22:10:36 Enjoy! 22:10:41 will do.tnx! 22:10:56 one parting thought 22:11:05 tilgovi: it would be good if you narrowed in on the *required* bits 22:11:12 and let us know what we've over spec'd there 22:11:23 +1 22:11:30 k...i'm done for now. :) 22:11:32 l8rs! 22:12:36 The only things I care about are selecting and relating. I've been told that relating is off the table, so all I care about are the selectors. 22:14:11 I'm pretty sure that if I were to write any of the applications I have in my head that could be considered related, somehow, to "annotation", I would be adding unmotivated complexity by instantiating an oa:Annotation class at all. 22:14:44 All of your use cases have exactly one body and target? 22:14:47 Unless it's just functioning as a simpler way than a named graph to hold some provenance about a relation. 22:15:09 And the creator of the relationship is always the same as the creator of the body? 22:15:29 Well, I don't actually like the body multiplicity. 22:15:33 So 1->n 22:15:50 No, definitely not that the creator of the relationship is necessarily the same as the creator fo the body. 22:16:05 That's why I said "unless it's just functioning ... to hold some provenance about a relation" 22:16:13 But if I can't actually state the relation than that's right out. 22:16:16 Ahh, I see 22:16:26 *then 22:17:00 Yeah, Annotation mostly functions in that role -- a node to hang provenance about the relationship(s) off -- and indeed without explicitly recording a particular predicate 22:17:19 Yep. And that's what's so strange to me. 22:17:25 It's a place to put metadata about a thing but not the thing itself. 22:17:30 The thing itself is absent. 22:17:39 I totally sympathize, btw, as I tried to promote that pattern several times in OAC and CG days 22:17:50 I can say why, the motivation, I have this metadata but not what it does. 22:17:59 Yep 22:18:31 Didn't we have a proposal, maybe from Ray, about ... I can't remember the term .... the intention of the annotation to induce some mutation or change 22:18:39 Who: creator. When: created. Why: motivation. What: body ??? target 22:19:11 Like some expected result of processing it. 22:20:16 Because I totally empathize with that 22:20:35 And boiling it down to some very simple RDF building block 22:20:58 I often find myself imagining that what I want out of annotations is to say, "Here I am, here's when I said X, and my intention is that you adopt X into your knowledge graph" 22:21:00 Yup. The building block that we ended up choosing was skos:Concept, and thus oa:Motivation 22:21:16 What about a graph body? 22:21:21 Yeah 22:21:26 Graph body makes tons of sense to me. 22:21:31 We left it behind, didn't we? 22:21:41 with a specific motivation so you can reliably process it 22:22:02 We took named graph bodies out of scope because they're not standardized beyond JSON-LD 22:22:18 But if the graph is serialized, then it's just like any other body 22:22:28 I'm not sure I would care about the motivation, specifically because, as has been said in the recent discussion, the motivation is vague and tries to model knowledge that is private to the user. 22:22:59 eg: {type: Annotation, body: {"content": "