16:04:43 RRSAgent has joined #simile 16:04:52 So I have 5 mins ... I'll go back to writing a new operating system then ... :-) 16:05:42 Andy, I was hoping you could explain a little about the four layers you describe in your comments to the relevant technologies piece 16:05:45 got it should start now 16:05:53 mickBass has joined #simile 16:06:39 Eric, Mackenzie, Mark, Mick, Kevin, Rob, Andy 16:06:42 Rob - OK. 16:07:49 four agenda items, plus one extra one from Mark 16:07:59 the first item is the relevant technologies document 16:08:17 Rob: I've added Kevin's changes into CVS, with Andy's comments I'm thinking of suggesting some changes 16:08:37 one thing, I was going to ask Andy to describe the structure he has at the top of his comments 16:08:57 I'm interested in whether Andy thinks this is a good structure for the document or something deeper? 16:09:25 AndyS: the layers try to split the system up into separate layers 16:09:48 there are some thing specific to the domain or application, but also some general things 16:10:00 Rob: so it's more a way of structuring the document 16:10:07 AndyS: yes. 16:10:35 I was looking at as a way of structuring the document, look at in the different lifecycles of content, ontology and metadata 16:10:37 Mick observes that Andy's comments reflect a potential structure / architecture 16:10:53 Rob: Yes I think there is a better way of making the document more navigable 16:11:18 I need to make my own comments as well. 16:11:36 AndyS: We can't track changes in the document, we want a way of doing that. 16:11:43 Andy: hard to understand where doc has changed 16:12:16 AndyS: what is the timescale here? 16:12:17 ROb: aiming for another rev incorporating Andy + Rob input, by Monday. 16:12:51 ACTION: Rob send email propose wed time for Andy, Kevin, Rob sync point 16:13:36 kevins2 has joined #simile 16:13:56 mark: agenda item 2 on use case restructuring 16:14:03 mark: feedback 16:14:08 mark: 1) too many use cases 16:14:14 2) some use cases too "specific" 16:14:24 mark: overlap across use cases 16:14:57 mark: have proposed generic use case "support for heterogeneous schemas and instances" 16:15:10 mark: will give overview 16:15:39 ... can decompose "supp for het schema & inst" into sub use cases 16:15:50 ... community approaches library, doesn't have schema or ontology 16:16:01 ... approaches library but have RDBMS (or implicit schema) 16:16:12 ... approaches library with XML (with or without schema) 16:16:23 ... approaches library with RDF (and maybe RDFS) 16:16:55 ... [missed a couple] 16:17:01 ... seek feedback 16:17:11 a) is generic use case a good idea 16:17:30 b) are the sub-use cases correct and appropriate? 16:17:46 kenzie: I think it's a good direction 16:18:12 ... because it allows us more flexibility. i.e. we might start with learning objects, but some communities might approach us with a substantial corpus of existing material and metadata 16:18:33 ... i.e. we could put other specific "instances" of "others as identified". 16:18:49 ... the scenarios and steps do cover the demands that we are seeing. 16:19:09 em: clarify email / document. are they the same 16:19:30 mark: sent an email, then have written a gif diagram attempting to break down components and relate to use case 16:19:49 em: I took away two themes 16:20:12 ... 1) community comes to SIMILE and says "help us to do these things, but we haven't thought about what metadata we need or how to manage it" 16:20:23 ... 2) we have metadata, but it's in a variety of different forms 16:20:36 ... subcase of 2 - we have metadata and it is already in RDF 16:21:09 em: this is the email that you were stepping us through? 16:21:11 mark: yes 16:21:26 em: I think doing high-level scenarios a good idea 16:21:38 ... major themes are the ones that DSpace is grappling with today 16:22:01 em: kenzie, please characterize breakdown: need a schema, or have a bunch of metadata please support 16:22:14 kenzie: more the latter. If they're in camp1 we just suggest dublin core 16:22:55 em: you're not running a "metadata consulting branch" that helps them figure it out... 16:23:13 kenzie: we can do that, but we try to lead them to an existing schema rather than create something new. 16:23:23 kenzie: Q - is there demand for a "roll your own" schema service? 16:23:53 em: is that somthing that you do and/or would consider, and is it something that we should be prioritizing? 16:24:04 kenzie: rarely have community say "I don't have any metadata, help me" 16:24:57 kenzie: hit 2 and 3 frequently. 2 most common - community has an ontology, but its not in XML or RDF. It's a CDIF, DTD, or database schema 16:25:10 ... next most common: have a schema, not in RDF. 16:25:30 em: I combine these together. Important question - do they have a formal model or not? 16:25:44 kenzie: as a practical matter, XML schema is usually easier... 16:25:50 em: because we have tools to deal with that 16:26:38 kenzie: if 1) and you don't have a schema, we'd need to hook you up with the right one. 16:26:57 mark: Q - have been looking at faceted search interfaces 16:27:04 ... are you guys familiar? 16:27:27 kenzie: familiar with faceted thesauri, but usually these were used in different way than what you were looking at. 16:27:40 kenzie: your's was more like boolean search? 16:28:00 ks: not boolean search, more like navigating (multiple) hierarchies that are constrained. 16:30:15 mark: are facets a type of schema? what's the relationship there 16:30:33 kenzie: they haven't been "thought about in that way". Philosophically they may be. 16:31:37 em: direct relationship - if you can describe facets and how they relate, use for classification, use for organization and presentation to the user. strong correlation. 16:32:28 mark; do users really want to navigate information like this? Is it a requirement here? Is this where schemas like DC are not sufficient? 16:33:33 em: DC not intended for this, so of course not sufficient. kenzie brought up mesh. if way stuff is organized can be represented via schema, then you could build generic faceted search/navigation interfaces, and drop in arbitrary particular facets. 16:34:40 ks: why DC not sufficient for faceted search? 16:35:39 in the past, facets are like attributes of one element, e.g. a subject heading, it had multiple attributes or facets associated with it 16:36:09 in another case, if you break it up into different elements it becomes more like DC, but it may be presented the same in the UI 16:37:23 em: in some cases user interested in search and browse at the same time 16:38:21 em: systems like brownsauce do a good job at this. 16:39:02 Mick are you still taking minutes? 16:39:26 AndyS: there is a split here about what you can use use cases for, e.g. is it for what we address or is it because we want to build on 16:39:42 if we are implicitly still targetting four or more use cases, that doesn't help 16:39:53 so I just wondered which one we are doing here? 16:40:00 mickBass: within the generic use case? 16:41:00 MacKenzie: The schema visualisation one is different, but you need that one? 16:41:36 mickBass: We have learning objects, VRA core, biomedical and schema registry and navigation is a functional component that is required 16:41:40 to implement the other use cases. 16:42:16 mickBass: there are multiple dimensions of focus here, e.g. focus on a specific end to end instance, another dimension of focus 16:42:38 is along the end to end decomposition, e.g. consider different approaches 16:42:52 those all represent different function points along the end to end use case 16:43:20 we have some discussion, so I'm hearing from Andy that is helpful, perhaps you can state your two alternatives 16:43:42 AndyS: I just want to understand where we are going with the use case, e.g. in UML you do use cases for two different things 16:43:50 so we need to decide which mode we are in 16:44:10 MacKenzie: to my mind, what we need to do is implement one of the specific examples 16:44:20 but in the framework of the generic use case 16:45:43 em: so under the category of picking one, MacKenzie are you willing to take an action item to get hold of some instance records for some of these 16:45:57 can you get ten, twenty, hundred, by the plenary? 16:46:46 at MIT, the ones we will have the most of are the Learning objects. Whether I can get my hands on them in the next couple of weeks, I don't know 16:47:02 With visual images, they are easier to get hold of, but they may not be the most helpful 16:47:21 So it might not be possible to have a lot for the plenary, but I can get some, does that help? 16:48:28 MacKenzie: I will endevour to get some for the plenary 16:48:44 kevins2: Can I make at filling in the 16 steps 16:49:07 As I understand it, Mark made a suggestion that instead of these four use cases, we would define a general use case, but 16:49:42 being amongst a less general use its less well defined, but we want things that are well defined, so instead of reducing them down 16:49:54 we try to learn more about them 16:50:33 em: the next step is the group focusses on one of these things to focus on, then we start to think about the near term deliverables to help 16:50:38 shape the plenary agenda 16:51:37 are we expecting david? 16:52:53 em - David told Mick he was with his family - he would try but may have trouble phoning in (no coverage) 16:53:59 hmm.... s/tasks/components for first instance 16:54:13 thanks AndyS 16:55:53 em: the real benefit is if we get out of IMS, we say this is harder than just writing PERL scripts, but then when we do VRAcore it seems its still 16:56:09 not saving us any time, but its when its the third it becomes more cost effective 16:56:51 mickBass: final comment, for action, is this notion of consolidation and focussing on specific instances is a good idea 16:57:16 the one Mark has proposed is the primary reasons we undertook SIMILE, this has got to be in scope, but I'm not convinced this 16:57:25 q+ 16:57:28 is the only one, or that we want to focus just on this use case 16:57:59 MacKenzie: we agree, its just a question of priorities 16:58:16 q- 16:58:32 I think authority control and annotation have things in common, so we can boil this done to 2 or 3 and then order them 17:01:13 mickBass: CNRI debrief - MacKenzie, I'm not sure how well it will work to do that on this call, can we schedule it for another time? 17:01:30 q+ to suggest Kapor defrief as well 17:02:05 MacKenzie: Yes I think we need to do that. I think we don't just want a short term relationship on DSpace, if we want one we need to consider the longer term also 17:02:23 mickBass: So can I action John and MacKenzie to set that up? 17:03:15 ACTION: JSE and Kenzie to schedule SIMILE debrief regarding CNRI 17:04:13 JSE: The other thing is DOI not DOI, the other thing is handle not handle 17:05:56 em: for a future agenda item, I would like a Mitch Kapor Chandler debrief 17:06:25 they are engaging in a university setting, MIT may be one of the universities, how do you like to action that 17:07:00 MacKenzie: Yes lets do that next week 17:07:31 is someone archiving the minutes? 17:07:39 how does RRS Agent work? 17:07:56 AndyS has been able in the past to refer to a URL that has just this topic. 17:08:03 em set the topic at the beginning of the call 17:08:14 eric do you know how to find the appropriate URL? 17:12:58 I'll find the find the URL - it has teh date in it which needs tweaking ... 17:13:11 http://www.w3.org/2003/06/20-simile-irc : Eric - it is 403 again. 17:21:54 Thanks Andy, I will post. 17:27:47 mickBass has left #simile 19:01:48 Zakim has left #simile