16:09:28 RRSAgent has joined #simile 16:09:44 em-lap has changed the topic to: simile 2003-06-06 PI telecon 16:14:22 marbut has joined #simile 16:14:32 hi, folks 16:14:46 I don't know whats happened to Mick and Rob, I've tried phoning them both at their desk. 16:14:57 I've also tried SMSing rob. 16:14:57 hi mark; thanks for the update 16:15:04 I'll go try phoning his mobile now? 16:15:09 thanks 16:15:11 yes please 16:15:18 the musak is driving us nuts :) 16:16:11 Is there a mute "music"? 16:16:22 I didn't listen to the intro! 16:16:32 they are supposed to call in now, apparently they get talking to Dick Lampman (head of HP labs) 16:17:45 JasonKinner has joined #simile 16:18:10 shall I act as scribe? 16:18:15 EricM, PaulS, KevinS, Jason, MarkB, Rob, Mick 16:18:25 Yes please MArk - I have network unreliability 16:18:45 (wireless lan) 16:18:52 JK: Discussion on the list of the utility of Harmony versus ABC 16:19:06 mickBass has joined #simile 16:19:11 I'll try adding things as I can 16:19:13 em-lap: I'm happy with JK's proposal as a stake in a ground, we don't have to conform to all the Harmony stuff 16:19:25 but we can use it as a bas for now. 16:19:33 RT: I'd agree with that. 16:20:23 MB: Conerned at complexity 16:20:41 MB: e.g. URN:1 is not adding much 16:20:56 MB: Why do we need it (complexity)? 16:21:36 JK: What changes do you propose? 16:22:02 MB: Action vs Events 16:22:30 MB=MarkButler 16:22:56 Bass: Talked to Carl Logoze 16:23:12 Bass: Action =undertaken by individual 16:23:19 mickBass: The key difference of actions versus event is action is done by an individual, which causes a transition point, could cause several 16:23:23 transition points 16:23:45 so if I run this tool it can cause lots of things to occur, so a modification event could be caused by lots 16:24:01 of different tools. E.g. was it caused by batch import or by an individual user. 16:24:25 RT: I think some more concrete examples would be useful, so I can see how single events can be fired off by an action. 16:24:49 JK: Multiple actions can be part of a single event, you can subdivided events if you want. I'll go to migration 16:25:18 If you had a migration event, you have multiple actions that take place, then you need to store the content in the new location etc etc 16:25:34 An event encapsulates a segment of time during which some change occurs. 16:26:00 em-lap: So the question depends on the level of modelling, so models can support both 16:26:30 em-lap: If we need the level of complexity, but if we are introducing a level of complexity, then I'm slightly cautious 16:26:49 In reading over the proposal, I started to think of examples of use cases. 16:27:25 em-lap: MacKenzie, have you had a chance to catch up on the extension schema stuff. 16:27:30 MS: No, not yet. 16:27:53 em-lap: We are trying to identify a generic enough model that will solve the requirements we have. So the specifics of the 16:28:13 model depends on the requirements. You supplied a use case to Jason a couple of days ago, so perhaps it might 16:28:34 make sense to go into that use case? 16:29:02 Mick bass I want to check process before we go there. One is complexity vs simplicity, the other is desirability to leverage previous 16:29:24 work where the thinking may be more mature. These are two different things to be consider. 16:30:04 zakim, who is speaking? 16:30:04 sorry, em-lap, I don't know what conference this is 16:30:10 damn... 16:30:46 RRSAgent, pointer 16:30:46 See http://www.w3.org/2003/06/06-simile-irc#T16-30-46 16:32:46 q+ 16:34:26 JK: Proposes std ABC classes and properties but use models of dspace objects with type info 16:34:28 X-Archived-At: http://www.w3.org/mid/00a901c32b9b$aa75fbd0$0400a8c0@STARGATE 16:35:12 JK: anonymous events and anonymous actions, typed objects but untyped actions or events 16:35:14 JK: Like example 1 - type info on actions and events 16:35:49 JK: only introduce new types for representing existential facets of actuality: i.e. just the container for an object's metadata 16:36:43 JK: For the current history system, vanilla ABC is perfect 16:37:20 My only concern is from MacKenzie: how do you identify bitstreams which have had this kind of modification on them 16:37:37 JK: We need to decide whether subproperties or subclass 16:37:44 q- 16:38:03 JK: hasPatient is too general 16:38:09 KS: What kind of changes? 16:38:26 MS: The purpose of the system is to support collection managers e.g. to record format migration 16:38:47 So you might need to find out all the images that were migrated because something went wrong. 16:39:10 KS: So arguably if we've serialized enough of the data under the patient, then an appropriate join would be possible? 16:39:26 JK: It might also be possible to retrieve based on the agent of the action. 16:39:35 mickBass: That's how I would have expected it. 16:40:26 MS: I thought my withdrawal example was getting at is changes in policies are also events we need to know about? 16:41:04 JK: There are few ways we can get at that information, e.g. look at the sequence of information when that object was withdrawn, 16:41:22 we might subclass haspatient, so we could query for that kind of modification 16:42:05 mark: I'm happy, I just want to understand it. 16:42:32 AndyS: I'm happy with the proposal. I'll incorporate that in the design document. I'll also try to address MacKenzie's issue. 16:42:57 (Sorry) JK : I'll incorporate that in the design document. I'll also try to address MacKenzie's issue. 16:43:07 (AndyS remark ends at first full stop) 16:43:32 JK: I don't know about existing data, 16:43:52 JK: There may not be enough data in the existing models to migrate them to the new model. 16:44:07 em-lap: So when might we see collections of data that conforms to the new model. 16:44:27 JK: There should be some code producing something in two weeks after this document closes. I also want 16:44:40 to confirm that no-one has concerns about the way dspace objects are modelled. 16:44:54 So in 2 weeks we should have some sample data to work with? 16:45:09 em-lap: So in the past there were concerns with history data being private? 16:45:43 mickBass: there was a proposal in the statement of work meant to ameliorate this. It said we need some kind of test jig or generator so 16:45:55 that we can get data without having to wait for a real dspace system. 16:46:28 JasonKinner: I would say it will be an additional week, week and half to produce a test jig 16:46:43 If I wanted to be safe I would say before july 4 16:47:24 mickBass: I will defer to Jason and Rob to close on the representation of dspace objects 16:47:33 JasonKinner: is rob here next week? 16:47:52 Rob: I have some concerns about the design, but in the interest of making progress I'm happy to defer on them 16:48:01 JasonKinner: we should plan on connecting on Monday 16:48:15 mickBass: can we close on this topic and move on. 16:48:42 mickBass: we have confirmation with all of the PIs for July 23rd and 24th. So lets confirm those dates, and also 16:49:06 make some comments on venue. Some we could use Endicott (?) house at MIT? 16:49:27 Mackenzie: But that's hard for people who are travelling here? 16:49:44 mickBass: So the intention was that people could stay there. 16:50:18 mickBass: I wanted to get the HP team engaged in a structure for the outcome of the plenary 16:50:47 plenary desired outcomes: 16:50:52 mb: 2 aims for simile 16:51:07 mb: 1) interesting SemWeb techniques/technologies. Can apply? 16:51:20 mb: 2) Use Cases, from PIs. Can we solve them? 16:51:28 mb: potential approach: 16:51:54 mb: a year of prototyping, follow by a integration/implementation 16:52:05 mb: need feedback on that approach 16:52:46 mb: summarizing, stage 1 - research, prototyping, small scale, not integrated. stage 2 - integrated system development 16:53:28 em: generally support, only caveat directed at funding aspect. Good to see near-term demos/deliverables 16:53:45 +q 16:55:02 q? 16:56:30 kenzie: to some degree we are time-competitive with mods to current DSpace using alternative approaches 16:57:12 em: can discuss benefits of SemWeb approach to help sell 16:57:41 mickBass: one of the pieces of feedback that I heard was two fold. When you are considering implementation / integration 16:57:58 you usually don't get it right first time, so it might be helpful to have some overlap 16:58:24 my observation is there would actually be some strong benefits to having that as early as possible 16:58:42 so its better to have some glimmering of an end to end approach, so that people can see where we are heading 16:58:56 rather than having a disparate collection of small prototypes 16:59:21 paul: one of things that I raised in the HP meeting is to see a systems diagram of how the prototypes fit together so that we can 16:59:27 see how they fit togther 16:59:34 and where the holes are. 16:59:41 mackenzie: yes it does 17:00:25 mickBass: so a big chunk work at the plenary will be about getting things to happen with the hires 17:01:05 Another chunk is to do with architecture / integration alternatives and when end to end versions begin to appear 17:01:34 also we need some consensus on the approach for interleaving of the tail end of the prototype work / the start of the implementation work 17:01:59 mickBass: so I propose as a next step, Rob, Mark and myself put together an agenda to bring back to the PIs for comment 17:02:06 Eric, Mackenzie: yes 17:02:28 mickBass: Can we spend a couple of minutes on Oscom and WWW2003. 17:02:55 mickBass: I only managed to attend a couple of talks. The SIMILE presentation was well retrieved, but there was too much material there. 17:03:14 I got to do a rerun at Cornell, and it went a lot more smoothly the second time 17:03:43 em-lap: At WWW2003, I found it to be an incredibly productive and exciting conference. 17:04:12 em-lap: Every web conference has a theme, if last years started Semantic Web this was full blown. 17:04:26 My presentation talked about a variety of SWAD activities include SIMILE. 17:04:56 For example the people involved in CIDOC stuff, Martin Duerr (?) 17:05:31 mackenzie: what about Mark's question/observations re: CIDOC? 17:05:45 Martin is interested in setting up a SW / Digital Library at an upcoming DC conference 17:06:18 Overall the conference was SW oriented, lots of people walked away seeing the benefits of SW, so special thanks to Andy who helped there. 17:06:25 Lots of interest in SIMILE. 17:07:11 andyS: last year was SemWeb vision. This year it was becoming tangible, with more clear functional decomposition and concrete explorations in particular areas 17:07:12 AndyS: compared with Hawaii, where the theme was the SW vision, this time it was the SW becoming tangible, e.g. seeing tools, so getting a sense its going to happen 17:07:25 mickBass: shall we close on this topic 17:07:39 mickBass: thanks to MacKenzie for distributing a potential job posting. 17:09:12 who has the log? 17:09:42 Log (URL) is on email already! 17:09:55 http://www.w3.org/2003/06/06-simile-irc 17:38:22 em-lap has left #simile 18:19:37 mickBass has joined #simile 18:31:03 mick has joined #simile 18:32:11 mickBass has joined #simile