The following log was recorded during the actual workshop by workshop volunteer participants. Thanks a lot!!!
Subject: IRC log from SWIW |
[08:59] <bengee> is this channel logged? [09:00] <bwm> I was wondering about that [09:00] <bwm> I'm logging the chat session from my irc client but it would be great to get a proper logger in here if you know how to do that [09:01] <bengee> I think they are directed by edd or dajobe [09:01] <bwm> its a bit early for dajobe :) [09:01] <bengee> true [09:02] <bengee> shall we use #swig instead? I don't think folks would complain about the noise [09:02] <bengee> wegot the chump there as well.. [09:03] <bwm> Could do - the organizers are primed to announce this channel [09:04] <bwm> I was thinking other workshops might use that [09:04] <bengee> ah, I see [09:05] <bwm> suggest we chump stuff in #swig but use here for the workshop notes - if there is a workshop report it will be produced from the notes here [09:07] <bwm> we are starting [09:07] <bwm> avi introduces wkshop chairs [09:08] <bwm> avi: have some talks in the morning [09:08] <bwm> ... have breakout groups in the afternoon [09:08] <bwm> ... based on themes from the papers and other ideas that come up [09:08] <bwm> duane: talk about issues and processes for involving users in building applications [09:09] <bwm> ... got two colours of postits [09:09] <bwm> ... green ones are for [lost] [09:09] <bwm> ... yellow ones for general ideas and comments [09:09] <bengee> green: UI/usabiliy-related notes/questions/ideas [09:10] <bengee> ;) [09:10] <bwm> avi: 6 talks maybe with demos this morning [09:10] <bwm> ... in the pm duane will organize the breakout sessions [09:11] <bwm> ... we will try to organize dinner for this evening [09:11] <bwm> irc channel is #swis [09:11] <bwm> s/swis/swiw/ [09:12] <bwm_scribe> ... presenters for first paper not here yet [09:12] <bwm_scribe> ... so beginning with second paper [09:13] <bwm_scribe> ... end user evaluations ... [09:14] <bwm_scribe> [general milling around looking to connect mac to the projector] [09:15] <bwm_scribe> talk: Search on Tap - Rob McCool and Andrew Cowel [09:15] <bwm_scribe> ... structure context, Search on Tap, motiviation, overview of test ... [09:16] <bwm_scribe> ... ARDA funded - disruptive technologies office [09:16] <bwm_scribe> ... part of NIMD program [09:16] <bwm_scribe> ... Knowledge Associates for Novel Intelligence [09:17] <bwm_scribe> ... stanford KR and reasoning [09:17] <bwm_scribe> ... IBM info extraction [09:17] <bwm_scribe> ...??? user interface [09:17] <bwm_scribe> ... start with unstructured information [09:17] <bwm_scribe> ... used a bunch of data found on the web [09:17] <bengee> ... paper at http://www.ifi.unizh.ch/ddis/fileadmin/events/iswc2005ws/CameraReady/Cow ell_EnduserEvaluations_016.pdf [09:18] <CaptSolo> is the workshop programme online? [09:18] <bwm_scribe> ... motivation [09:18] <bwm_scribe> ... Search On Tap is a structured search tool [09:18] <bwm_scribe> ... search over data scrapped from web sites [09:18] <bwm_scribe> ... augments keyword search results [09:19] <bwm_scribe> ... extra modes that traditional search cannot do [09:19] <bwm_scribe> ... can do "what is" analysis [09:19] <bwm_scribe> ... assists in the info gathering phase [09:19] <bwm_scribe> ...helps disambiguating terms [09:19] <bwm_scribe> ... does group and comparison queries [09:20] <bwm_scribe> ... does specific fact search - tell me all you know about a particular thing [09:20] <bwm_scribe> ... a list of different kinds of queries [09:21] <bwm_scribe> ... example skyscrapers constructed during the nixon era [09:21] <bwm_scribe> ...fields covered, music movies, etc [09:21] <bwm_scribe> ... stat 31 sites, 12 topics, ... [09:21] <bwm_scribe> ... motivation for the user evaluation [09:22] <pshab> examples url: http://sp11.stanford.edu/examples.html [09:22] <bwm_scribe> ... wanted an informal user evaluation [09:22] <bwm_scribe> ... have an unusal group of users [09:22] <bwm_scribe> ... don't read manuals or go on courses [09:22] <bwm_scribe> ... expect tools to fit existing workflow [09:22] <bwm_scribe> ... don't get to choose their own tools [09:23] <bwm_scribe> ... discount usability framework [09:23] <bwm_scribe> ... 10 dimensions of heuristic evaluation [09:23] <bwm_scribe> ... 2 usability experts [09:24] <bwm_scribe> ... 17 issues identified [09:24] <bwm_scribe> ... e.g. speak the users language [09:24] <bwm_scribe> ... too long to display system status [09:24] <bwm_scribe> ... shows before and after screenshots [09:24] <bwm_scribe> ... result is more google like [09:25] <bwm_scribe> ... participants were 8 working battelle analysts [09:25] <bwm_scribe> ... age 30-40 - described as old :( [09:26] <bwm_scribe> ... 3 individuals are enough to identify 70% of usabilty issues in an interface [09:26] <bwm_scribe> ... analysts encourage to speak aloud [09:26] <bwm_scribe> ... example findings [09:27] <bwm_scribe> ... wanted to have drop downs for property names and values [09:27] <bwm_scribe> ... UI used space to right of google search results [09:28] <bwm_scribe> ... but that space is traditionally used for advertising and often ignored [09:28] <bwm_scribe> ... so moved to the left [09:28] <bwm_scribe> ... http://tap.stanford.edu/ [09:28] <bwm_scribe> conclusion: user evaluation [09:28] <bwm_scribe> ... got debrief from NIST [09:29] <bwm_scribe> ... unofficially they found it pleasant and easy to use without any formal training [09:29] <bwm_scribe> q: you made the point about swapping the panes left and right [09:30] <bwm_scribe> ... should one sometimes break the mold [09:30] <bwm_scribe> ... go for non-web interface [09:30] <bwm_scribe> a: users were used to a web interface [09:30] <bwm_scribe> ... and we only had 4 months to do the whole thing - not long enough to change the paradigm [09:31] <bwm_scribe> q: is 40 old? [09:31] <bwm_scribe> a: it is for analysts [09:31] <bwm_scribe> q: any difference between different age groups [09:31] <bwm_scribe> a: [lost] [09:31] <bwm_scribe> q: how does this help disambiguate terms [09:32] <bwm_scribe> a: thats part of the tool [09:32] <bwm_scribe> ... if you type china, it will show you a list of different kinds of china [09:32] <bwm_scribe> ... you can select kind you meant [09:32] <bwm_scribe> ... try the examples [09:33] <bwm_scribe> q: is there a demo of this interface anywhere [09:33] <bwm_scribe> a: http://sp11.stanford.edu [09:34] <bwm_scribe> avi: we are planning to put papers on a public domain site for workshop papers [09:34] <bwm_scribe> ... we need permission to publish work there [09:35] <bwm_scribe> ... you don't give up copyright, only give permission to publish [09:35] <bwm_scribe> ... so please sign this form [09:35] <bwm_scribe> Karun Bakshi, works with David Karger [09:35] <bwm_scribe> talk: personalized semantics web applications [09:36] <bwm_scribe> Karun: why applications? [09:36] <bwm_scribe> ... designed to support particular tasks [09:36] <pshab> session papers at http://www.ifi.unizh.ch/ddis/313.0.html [09:36] <bwm_scribe> ... but problems [09:36] <bwm_scribe> ... information can be buried [09:37] <bwm_scribe> ... lack of integration between applications [09:37] <bwm_scribe> ... too many windows on the screen [09:37] <bwm_scribe> ... application you want may not exist [09:37] <bwm_scribe> ... sources of these problems [09:37] <bwm_scribe> ... information needs are fluid but applications are static [09:38] <bwm_scribe> ... there are not enough developers to build all the apps we need [09:38] <bwm_scribe> ... so let the users build their own applications [09:38] <bwm_scribe> ... give them application fragments they can aggregate on the fly [09:38] <bwm_scribe> ... we believe that semantic web will enable this [09:39] <bwm_scribe> ... by providing a common data model [09:39] <bwm_scribe> ... and support for fine grained metadata [09:39] <bwm_scribe> ... information tasks match the structure of the semantic web [09:39] <bwm_scribe> ... sweb supports personalized ontologies - decentralization [09:40] <bwm_scribe> ... sweb benefits also - users will like building their own apps [09:40] <bwm_scribe> ... approach - build task workspaces [09:40] <bwm_scribe> ... specify content - queries, views and operations [09:40] <pshab> Haystack homepage http://haystack.lcs.mit.edu/ [09:40] <bwm_scribe> ... haystack provides a lot of these fragments [09:40] <bwm_scribe> ... we build tools to provide to the users [09:41] <bwm_scribe> ... haystack has a triple store [09:41] <bwm_scribe> ... has a ui framework for recursive rendering of views [09:41] <bwm_scribe> ... view in haystack can be described in rdf [09:41] <bwm_scribe> ... so they are just data [09:42] <bwm_scribe> ... can collect arguments - lets users specify some arguments - i.e. currying [09:42] <bwm_scribe> ... channel manager for persistent queries [09:43] <bwm_scribe> ... view designer to select properties to be displayed [09:43] <bwm_scribe> ... workspace builder - like a portal building tool [09:43] <bwm_scribe> ... look at a demo [09:43] <bwm_scribe> ... user is a neurologist writing a conference paper [09:43] <bwm_scribe> ... screen is layed out in rectangular sections [09:44] <bwm_scribe> ... user creates a new workspace and edits it [09:44] <bwm_scribe> ... specify attributes [09:44] <bwm_scribe> ... split the canvas up into sections using a control [09:45] <bwm_scribe> ... define a persistent query - volumes of brain structures [09:46] <bwm_scribe> ... can map an attribute over a collection of items (?) [09:46] <bwm_scribe> ... select previous query as a parameter of this query [09:47] <bwm_scribe> ... this is sample groups of patients - [the data source is some structured info in the triple store?] [09:47] <bwm_scribe> ... open presentation pane [09:47] <bwm_scribe> ... drag the [something] over to it [09:48] <bwm_scribe> ... select from the available list of views [09:48] <bwm_scribe> ... each item will be displayed using that view [09:48] <bwm_scribe> ... here is the final usage view that you get [09:49] <pshab> [something] was previously defined query? [09:49] <bwm_scribe> ... operations dynamically bind to the current selections [09:49] <bwm_scribe> ... specify the top pane as the source [09:49] <bwm_scribe> ... can specify todo items [09:50] <bwm_scribe> ... can specify just the type of the item they want a query is automatically created for it [09:51] <bwm_scribe> ... workspace is live [09:51] <bwm_scribe> ... so agents can update things and the changes will propagate through other panes [09:52] <bwm_scribe> ... that tools that we use to build workspaces is itself a workspace [09:53] <bwm_scribe> ... evaluation [09:53] <bwm_scribe> ... did some work with neurolgists in boston hospital [09:54] <bwm_scribe> ... hoping to plan a more involved study with them [09:54] <bwm_scribe> ... its a complex task to figure out what information sources you need to perform a task [09:54] <bwm_scribe> ... one might object that this is a lot of work to build an application [09:54] <bwm_scribe> ... but its worth it [09:55] <bwm_scribe> ... one person can build an application and then share it [09:55] <bwm_scribe> ... others can tweak it [09:55] <bwm_scribe> ... advantages [09:55] <bwm_scribe> ... quick switching between tasks [09:55] <bwm_scribe> ... extensibility [09:56] <bwm_scribe> ... a common interaction model for different applications [09:56] <bwm_scribe> ... sweb enables a new flexible interaction mechanism [09:56] <bwm_scribe> ... speaker is from Oracle [09:57] <bwm_scribe> q: do you show all ? to the user [09:57] <bwm_scribe> a: yes we show them all [09:57] <bwm_scribe> q: you said something about using uri's [09:57] <bwm_scribe> a: that's about operations - they have names the user sees but there is an underlying uri [09:58] <bwm_scribe> q: how did you do the evaluation [09:58] <bwm_scribe> a: we have done a demo so far , the proper evaluation is in the works right now [09:58] <bwm_scribe> q: were you able to draw on any of the usability research on portals and why portals fail e.g. not mainted [09:58] <bwm_scribe> a: mostly we tried to get design principles [09:59] <bwm_scribe> ... its biased [09:59] <bwm_scribe> ... portals aren't normally a personal tool [09:59] <bwm_scribe> ... people will invest more time in this because its personal to them and affects their productivity [09:59] <bwm_scribe> q: building an interface is an involved task, how did users react [10:00] <bwm_scribe> a: users said we didn't know you could do this, but it seemed like a lot of work [10:00] <bwm_scribe> bijan: it looked like a lot of work to me [10:00] <bwm_scribe> q: was it their data [10:00] <bwm_scribe> a: yes - we are giving them a better way to work with it [10:01] <bwm_scribe> ... we are trying to figure out if they can conceive of tasks in this abstrac tway [10:01] <bwm_scribe> q: why didn't you start with a formal evaluation and then work back to design the system [10:01] <bwm_scribe> a: we wanted to try to build a tool to see if folks could use the semantic web info [10:02] <bwm_scribe> q: that's a common practice [10:02] <bwm_scribe> ... but it might be more efficient to figure out the process and then design the tool to support the process [10:02] <bwm_scribe> a: then you build a specific tool for those people, and we wanted to build a generic tool [10:02] <bwm_scribe> q: some would say some usability is better than very little usability [10:03] <bwm_scribe> done [10:03] <bwm_scribe> [bijan approaches the front] [10:07] <bwm_scribe> duane: who here is working on things for public use - for general users [10:07] <bwm_scribe> ... about one thirds [10:07] <bwm_scribe> ... what about knowledge workers, office environments, etc [10:07] <bwm_scribe> ... even more [10:08] <bwm_scribe> ... who is working on tools [10:08] <bwm_scribe> [load music] [10:08] <bwm_scribe> bijan introduces himself as Jennifer Golbeck [10:08] <CaptSolo> :) [10:08] <bwm_scribe> [shows movie with more load music] [10:09] <bwm_scribe> ... clip from Titantic [10:09] <bwm_scribe> ... forget it boy you'll never get next to the likes of her [10:09] <bwm_scribe> WHAHAHAHAHOOO! [10:10] <pshab> url: http://trust.mindswap.org/FilmTrust/ [10:10] <bwm_scribe> ... FilmTrust is a movie recommender system [10:10] <bwm_scribe> ... social networking website [10:10] <bwm_scribe> ... users list friends [10:10] <bwm_scribe> ... friendship is not necessarily symmetric relationship [10:11] <bwm_scribe> ... trust ratings given to friends - trust in their recommendations that is [10:11] <bwm_scribe> ... expect to use semweb data to drive the system [10:11] <bwm_scribe> ... social network used in individual movie info pages [10:11] <bwm_scribe> ... rating section of the page show 266 users rated it [10:12] <bwm_scribe> ... the recommended rating however is much lower and uses the social network [10:12] <bwm_scribe> ... weighted average based on trust rating [10:12] <bwm_scribe> ... trust rating may be given directly or computed through the network [10:13] <bwm_scribe> ... conducted study [10:13] <bwm_scribe> ... compare our results with average and ? [10:14] <bwm_scribe> s/?/collaborative filtering system/ [10:14] <bwm_scribe> ... shows graph comparing [10:14] <bwm_scribe> ... errors [10:15] <bwm_scribe> ... if I have common tastes the other algs work just as well [10:15] <bwm_scribe> ... but if I have unusual tastes then the social network algorithm works better [10:17] <bwm_scribe> http://trust.mindswap.org/ [10:17] <bwm_scribe> bijan: shows his film trust page [10:18] <bwm_scribe> ... he likes a clockwork orange [10:18] <bwm_scribe> ... that's it [10:18] <bwm_scribe> questions? [10:18] <bwm_scribe> q: what does acf stand for [10:19] <bwm_scribe> a: automatic collaborative filtering [10:19] <bwm_scribe> q: doesn't that take into account divergent opinions [10:19] <bwm_scribe> a: acf fails if I have broadly common opinions, but have some outliers [10:20] <bwm_scribe> q: user defines trust rating, there is little computation [10:20] <bwm_scribe> ... the effort is in the building trust relationships [10:21] <bwm_scribe> a: well you only rank your immediate circle, the rest comes from the network [10:22] <bwm_scribe> q: saw that the personal trust network outperformed the general algorithms [10:22] <bwm_scribe> ... did people trust the right people? [10:23] <bwm_scribe> a: give me an analysis of whether people like the same movies as me [10:23] <bwm_scribe> ... if you produce profiles on people, there was a study [10:24] <bwm_scribe> ... to look at the quality of peoples personal trust judgements v looking at analysus [10:24] <bwm_scribe> ... results so far are inconclusive [10:24] <bwm_scribe> ... this system is based on actual judgements of people folks know [10:26] <bwm_scribe> q: how much of weighting comes from immediate neighbours [10:26] <bwm_scribe> a: often neighbours haven't rated a film you are interested in [10:27] <bwm_scribe> ... we don't have the numbers at present - ask Jenifer [10:27] <bwm_scribe> ... feel free to join [10:27] <bwm_scribe> ... http;//trust.mindswap.org/ [10:27] <bwm_scribe> q: is it a long lived thing? [10:28] <bwm_scribe> a: we expect to leave it up indefintely [10:28] <bwm_scribe> q: how is scaling? [10:28] <bwm_scribe> a: we hooked to imdb [10:28] <bwm_scribe> ... you can run into scaling problems [10:28] <bwm_scribe> ... we haven't optimised the implementation of the algorithms [10:28] <bwm_scribe> ... we think it'll be ok [10:29] <bwm_scribe> ... but we don't mind getting swamped with users so we have to fix that [10:29] <Laurian> If you dont have enough friends, you may add random users :) http://trust.mindswap.org/FilmTrust/tour/big/3.jpg [10:30] <EtnaRosso> moin Laurian [10:30] <bengee> break [10:31] <EtnaRosso> but the core of this speech was acf algorithm? [11:07] <pshab> Session starting ... [11:07] <bwm> duane: we have one green post it [11:08] <pshab> ... want discussions to be very active this afternoon... [11:09] <pshab> ... having spent time in sem web and user centered design... need to understand what the questiosn are... [11:09] <pshab> looking for questions/issues... raw material for future discussions... [11:10] <pshab> next talk [11:10] <pshab> rdf is machine friendly not user friendly [11:11] <pshab> Emmanuel Pietriga(?) Fresnel - A Browser-independent Presentation Vocabulary for RDF [11:11] <pshab> cant' expect users to work direclty with rdf [11:12] <pshab> ... examples... visualistions e.g. rdf author, IsaViz... based on displaying the graph natur eof the data [11:12] <pshab> ... another tools, web based e.g. Longwell, Noadster, Brownsauce, Piggy bank [11:13] <pshab> ... rich clients e.g. haystack [11:14] <pshab> ... presentation knowledge, need to enable programmers/users to choose content to dispaly and how to display and format it... [11:14] <pshab> ... currently diff tools use different ad hoc mechnanisms and vocabs [11:15] <pshab> ... so designed Fresnel... browser independant, with a set of core vocabs (simple as pos.) and allows extension vocabs [11:16] <pshab> ... based on the concept of lenses, formats and external style sheets [11:16] <bengee> (paper at http://www.ifi.unizh.ch/ddis/fileadmin/events/iswc2005ws/CameraReady/Pie triga_Fresnel_009.pdf) [11:17] <pshab> ... core lens vocabs, content selection and ordering e.g. domain, show/hide properties, merge properties, ... [11:17] <pshab> can have different lenses apply to the same resources [11:17] <pshab> ... applications might allow usrs to choose what lens to use [11:18] <pshab> ... varying levels of detail etc. [11:19] <farh0rizon> fresnel looks interesting - i have something similar in principle for presenting rdf as swing data models [11:19] <pshab> ... core formatting vocab - propertFormatDomain, classFormatDomain... the values afect how somethign is displayed... [11:20] <pshab> ... example of PersonLens [11:21] <pshab> ... can do recursive instructions e.g. foaf:knows and display usign a different lens... [11:22] <pshab> ... selector languages - used to define the lens and format domains [11:22] <pshab> ... basic selectors - single URI (tyep text / URI test)... [11:23] <pshab> ... can use FSL - XPath like graph traversal expressions... [11:23] <pshab> ... can use SPARQL selectors... [11:23] <pshab> ... Demo of an implimentation [11:25] <pshab> ... demo uses one lens to display a list of names (just names) clicking on a name brings up a more detailed view and other lenses can be applied via a dropdown(? I think I missed that) [11:25] <pshab> ... implimentations - horus (the previous demo), Simile Longwell, IsaViz [11:26] <avi> mailing list at www.w3c.org/2005/04/fresnel-info/ [11:27] <pshab> ... the work is an open community project ... many people from simile, Haystack, and many others... [11:27] <avi> question: What are your thoughts on a default interface [11:28] <avi> ... Should there be a default style-sheet? [11:28] <avi> Answer: No, we haven't thought about that ... but you should be able to implement such a stye-sheet in fresnel! [11:28] <pshab> ... yes... could write soem default stylesheets, but might not particularly helpful [11:29] <pshab> ... basic problem is finding rdf data that you dont' know anything about, you need a way to look at it... [11:30] <avi> Question: Is there a standard x-from language for links? [11:30] <pshab> thanks avi [11:30] <avi> :-) [11:30] <pshab> question: why didnt' you jsut go ahead and use XPath...? [11:31] <pshab> ... this is more focused on hte rdf nature of hte data [11:32] <pshab> ... again missed the question, sorry [11:32] <avi> question: What other datasets did you try? [11:32] <avi> Answer: None yet, will try some more as soon as we have implemented more! [11:33] <bengee> CropCircles coming up? [11:33] <bwm> looks like it [11:33] <bwm> I will scribe if no one else will [11:33] <bengee> Next: Bijan Parsia: CropCircles [11:34] <avi> Paper: Visualizing Web Ontologies with CropCircles [11:34] <avi> Bijan Parsia, Taowei Wang, Jennifer Golbeck - University of Maryland, College Park MD, USA [11:34] <bengee> paper at http://www.ifi.unizh.ch/ddis/fileadmin/events/iswc2005ws/CameraReady/Par sia_VisualizingWebOntologies_018.pdf [11:34] <avi> Paper at: http://www.ifi.unizh.ch/ddis/fileadmin/events/iswc2005ws/CameraReady/Par sia_VisualizingWebOntologies_018.pdf [11:34] <bengee> ...motivation [11:34] <bengee> moreand more onts [11:34] <bengee> .. quite large, too [11:34] <bengee> ... overlaps [11:35] <bengee> ...needs: [11:35] <avi> but how can users understand these large ontologies [11:35] <bengee> ...the "right" level of understanding [11:36] <bengee> ... basic structure etc [11:36] <bengee> ...straight graph visualization more or less a non-starter [11:36] <avi> Typical size of onts they look at: 50-2700 classes [11:37] <bengee> .. shows a sample collection of onts [11:38] <bengee> ... approaches [11:38] <bengee> ... tree widgets [11:38] <avi> .. tree editors are problematic given nesting, can't really see more than one branch, ... ... [11:38] <bengee> .. tree maps [11:39] <bengee> .. not much space for nodesat the end of the branches [11:39] <bengee> .. cropcircles was designed targetted at putting as much structure in as possible [11:39] <bengee> ... key features / basics [11:40] <bengee> .. onts: colored circles [11:40] <bengee> ... classes: circles [11:40] <bengee> ... shows koala ontology as an example in swoop [11:41] <bengee> .. switches to CropCircles view [11:41] <bengee> ... nested and nicely arranged circles [11:42] <bengee> .. nested circles represent subclass hierarchies [11:42] <bengee> ... layout heuristics [11:43] <bengee> ... checks for a "dominant child", "all children have the same size", ... [11:43] <bengee> ... dblclick on circle does a zoom [11:44] <bengee> ... additional list of concepts in a swoop sidebar [11:44] <bengee> .. and tooltips in order not to clutter the cropcircles graphx [11:45] <bengee> q: wouldn't it makesense to show labels within the circles [11:45] <bengee> a: up to a certaindegree, perhaps [11:46] <bwm> q: what are we supposed to see other than something aesthetically pleasing [11:46] <bwm> a: I'll get to that [11:46] <avi> demo available at http://www.mindswap.org/2005/cropcircles/ [11:48] <bengee> I'll upload my photos later, though [11:50] <bengee> ... colors are used for different highlights [11:50] <bengee> ... e.g. multiple inheritance [11:50] <bwm> ... can right click on a cirle and it will show related classes [11:51] <bengee> ... name origin [11:51] <bengee> .. we al saw "signs" [11:51] <bengee> s/al/all/ [11:51] <avi> q: Did you you think about changing the layout to adopt for multiple inheritance? [11:52] <bwm> q: monica might ask what sort of usability analysis we did [11:52] <bwm> a: none [11:52] <bwm> that's called a preemptive strike, I think [11:52] <avi> a: yes, but didn't do anything about it yet? [11:53] <bengee> ... some use cases [11:53] <bengee> .. analyzing OWL-S models (which entities are unconnected etc) [11:54] <bengee> ... other example which nicely shows isolated circles [11:54] <bengee> ... benefits [11:55] <bengee> .. beter viz of structure [11:55] <bengee> ... partitions and overlaps [11:57] <bengee> (a fly in front of the beamer pretends to be a mouse pointer) [11:58] <bengee> .. links [11:58] <bengee> q&a [11:59] <EtnaRosso> q&a? [11:59] <bengee> q: different layout rules, what's the motivaton? [11:59] <avi> q: ... isn't that confusing? [12:00] <bengee> a:they alow tailoring the different vizs you may be interested in [12:00] <bengee> allow even [12:00] <bengee> a: patterns vs. dominant nodes etc [12:00] <avi> q: expected users are programmers or end users? [12:01] <avi> a: the former! ... someone who can use eclipse! [12:01] <avi> a: ... not for end users! [12:02] <avi> q: The square layout of shrimp allows peoeple to edit in the graphical layout. Aren't circles limited as you can't do that? [12:03] <avi> a: Mmmm yes/no it is hooked up to an editor, hence you can edit in the editor and use the layout to navige, understand, etc.? [12:03] <avi> q: is this related to grokker ( ... grokster???)? [12:04] <avi> a: No, absolutely not! [12:04] <pshab> http://www.grokker.com/ [12:05] <pshab> uses similar visualisation approach... [12:05] <avi> q: (didn't get the q...) [12:07] <ajcowell> grokker searches over collections on the web (e.g., amazon) as well as Google, files on the users drive, etc - makes pretty pictures, not convinced on how useful it is (have it on my laptop if anyone wants to see it) [12:07] <avi> q: related to icons and labels... are the labels needed.... [12:08] <bengee> the qbefore: something related to lots of paradigms, the presented solution is not going to satisfy everyone, or so [12:08] <avi> a: Yes, but icons try to embedd the meaning in an image ... and we aren't as we have circles! [12:08] <avi> a: put in labels will ruin the layout - possible view? [12:08] <bengee> (the critics are speaking up ;) [12:09] <avi> interruption by chairs - this seems to be one of the discussion issues forthe afternoon. [12:09] <bengee> people want more info in the circles, bijan defends the clean design [12:10] <bwm> comparison with grokker is interesting - it has more info [12:10] <bwm> 3 presenters for the next talk [12:10] <bengee> ---- /presentation ---- [12:11] <bwm> avi helpfully points out that they are between us an lunch :) [12:11] <avi> paper: mSpace Mobile: A Mobile Application for the Semantic Web [12:11] <avi> Max Wilson, Alistair Russell, Daniel A. Smith, Alisdair Owens, m.c. schraefel - University of Southampton, UK [12:11] <avi> @ http://www.ifi.unizh.ch/ddis/fileadmin/events/iswc2005ws/CameraReady/Wil son_mSpaceMobile_015.pdf [12:12] <pshab> http://mspace.fm/mobile/ [12:12] <bengee> ...a location-aware mobile app [12:12] <avi> mspace is about location aware use of the SW [12:12] <bengee> ... key UI areas [12:12] <avi> Introduces the main setup of the UI, which has 5 elements [12:12] <bengee> ... coloumn browser (slicing and dicing) [12:12] <avi> ... 1) Column browser to slice/dice lists [12:13] <Blackat> salve [12:13] <bengee> ... OLAP-y operations [12:14] <bengee> .. choose/fix one col, populate the others [12:14] <avi> ... drill down akin to apple finder browser in NeXT mode [12:14] <bengee> (a well-prepared/-designed presentation, btw) [12:15] <avi> .. shows demo in film which clearly indicates that sub-panes zoom in ... [12:15] <avi> ... includes context dependent map-browser in one of the panes [12:15] <avi> ... Selector: [12:16] <avi> .. inlcudes addictional domains (eg., IMDB, music, etc) [12:16] <avi> .. favorites pane, (is what you thought it is) [12:16] <bengee> ..demo is too fast, but impressive [12:17] <avi> ... Zedpanes - zooming in an out mentioned before ... [12:17] <avi> ... akind to a fisheye view [12:17] <Blackat> good morning....I'm a student in Computer Science from University of Catania(Italy, Sicily) [12:18] <avi> ... shows rating system with comments? [12:18] <avi> (combine it with the trust work by Jen) [12:18] <ajcowell> I think the video is available at http://mspace.fm/mspaceVids/mspace-mobile.mov [12:19] <avi> ... intorudce the three tiered architecture of mspace [12:19] <avi> ... to minimize computation on the client handheld device [12:19] <bengee> ... app - query server - knowledge server [12:20] <avi> ... (the whole thing reminds me of vindigo) [12:21] <pshab> vindigo: http://www.vindigo.com/ [12:21] <bengee> ... uses open guide data [12:22] <pshab> http://london.openguides.org/ [12:22] <avi> ... export RDF [12:22] <bengee> ... next step is to re-export the data as RDF [12:22] <avi> FUTURE WORK [12:22] <avi> ... - Domain Switching [12:23] <pshab> other openguide link http://openguides.org/ [12:23] <avi> example: assume that you have Cinemas and Movie from open guide and more movie info from IMDB ... how can you hook them up? [12:24] <avi> ... - Column locking [12:24] <avi> same thing, but hooking up two parts of the same data source [12:24] <avi> ... - trust [12:25] <avi> Intend to use FilmTrust ( :-) ) [12:25] <avi> .... - Ghost data [12:25] <avi> essentially collect the data of where you have been in order to do recaollection [12:26] <avi> q: wouldn't ghost data just be another dimension [12:26] <avi> a: (explain the idea ... ) [12:27] <avi> ... - fresnel [12:27] <avi> want to use fresnel and specify some lenses [12:28] <avi> Q & A ---------- [12:28] <avi> q: how do you think that the interface would have to be different if there would be people freom different age groups using it? [12:29] <avi> a: The interface does have a learning curve and can be confusing... we didn't do such an valuation [12:29] <avi> a2: there is also the issue of accesibility [12:30] <avi> which we haven't entered in [12:30] <avi> q: Have you looked into folksonomies? [12:31] <avi> a: Yes, that would be hooked up using ghost datat [12:32] <avi> q: I saw you had itinirary ... any planning/searching/....? [12:33] <avi> a: not yet, but we are planing to that... [12:34] <avi> q: mSpace is cool ... but is there someting new by putting it on a small screen or does it also apply to a desktop? [12:35] <avi> a: the project was about that you have it with you and not focused on small screens [12:36] <avi> - chair [12:36] <avi> 1) Please contribute ideas [12:36] <avi> 2) Will try to heat room (brrrrrr) [12:37] <bwm> adjourned [12:37] <bwm> recommencing at 2pm sharp Eire time [14:06] <bwm> resuming for the afternoon session [14:06] <bwm> duane: we are having breakouts this pm [14:07] <bwm> ... one session before coffee, one after with collective sessions to share what we have learned [14:07] <bwm> ... we'd like to begin with a discussion with what is different about the user experience for the semantic [14:07] <bwm> web [14:08] <bwm> ... from that what do we think is going to be the areas we concentrate on from that [14:08] <bwm> ... if we can have 15 minutes to tackle that subject [14:09] <bwm> ... lightweight generation of metadata - what are the usability issues [14:09] <bwm> ... break that in two things [14:09] <bwm> ... need to generate metadata [14:09] <bwm> ... how to make that easy [14:09] <bwm> ... people don't fill in the fields in content management systems [14:10] <avi> ... capture all kinds of information (not just ontology data but also instances) [14:11] <bwm> ... capture data for machine use [14:12] <bwm> ... having metadata allows you to describe how data should be presented - so you don't need specialised applications [14:12] <bwm> ... so don't have to purpose build applications for new types of data [14:13] <bwm> ... information is both classified and presented in domains [14:14] <bwm> ... a question of whether the user needs to understand this [14:14] <bwm> ... is learning required [14:14] <bwm> ... the question of contribution to categorizing [14:15] <bwm> ... if usability has to be traded against automatic generation of information [14:16] <bwm> ... there may be a competition between design interface and the generic presentation of information [14:16] <bwm> ... there is the whole issue of choices and personalization in this [14:17] <bwm> ... its really about how to choose or personalize both interactions and views [14:18] <avi> ... Mathematitions/logicians are designing some of the SW, how do we bridge the precise/logical approaches with users [14:18] <avi> ... comptetion btw formal logic and the imprecision of users [14:18] <bwm> ... competition between the precision of formal logic and the imprecision of users [14:19] <bwm> ... can bring in new datasets on an adhoc basis and users want to see how that relates to what they already have [14:20] <bwm> ... there is a break in the metaphor - we now have documents on a desktop ... [14:20] <bwm> ... is it still a desktop really or are we killing the last bit of that metaphore [14:20] <bwm> ... an integrated environment rather than 'types' of information and presentation [14:21] <bwm> ... we had things of a certain type and apps that operated on them - now we have a web of information [14:21] <avi> New question: What frustrates you now that should change with the semantic web [14:21] <bwm> ... what disatisfies you now that you think will get better [14:22] <bwm> ... how can end users benefit from the advantages of the semweb without getting lost in the technicalities [14:23] <bwm> ... because we can control and manipulate the data at the user level [14:23] <bwm> ... we don't know what will happen if we can do federated query [14:24] <bwm> ... what will it be like to jump from one interface to another [14:24] <bwm> ... how do we explore across web scale spaces that are richly interlinked and associated [14:24] <bwm> ... we don't have a good paradigm for this [14:24] <bwm> ... people can disoriented pretty fast when searching and browsing [14:25] <bwm> ... this will be much worse iwth a deeper linked dataset [14:25] <bwm> ... one of the frustrations is that info is buried deep in a web page [14:25] <bwm> ... I can just bookmark the page but we should be able to do better [14:26] <bwm> ... bbc are interested in how data chunking is happening - can produce snippets [14:26] <bwm> ... whilst this allows higher degree of selection, can also lose the context which might be important [14:27] <bwm> ... producing knowledge in little chunks and using the ontology to create context is different [14:27] <bwm> ... it is being able to generate documents... [14:28] <bwm> ... or to see information in the way you want to see it [14:28] <bwm> ... you can control how those facts are assembled into a document [14:28] <bwm> ... there is both the ability to assemble chunks into your own view [14:28] <bwm> ... or collection [14:29] <bwm> ... but then there is the potential need to structure the chunks into views [14:30] <bwm> ... I'm getting worried that so much info is contained in context - once we remove that are we losing the essence [14:30] <bwm> monica: we have models of people's blogs with rss feeds [14:30] <bwm> ... whilst they can be experienced in an RSS viewer the link does go back to the source [14:31] <bwm> ... this is one of the exciting things about the semantic web - the link does go back - you are not just cutting and pasting the text [14:31] <bwm> avi: ok you can do it, but will people do it. [14:31] <bwm> ?: the semantic web does have a context [14:31] <bwm> ... a printed document has several contexts [14:32] <bwm> ... on the semweb the context is so complex that you need filters to pick out some of the context [14:32] <bwm> ... this is all about filtering and reshaping and lensing in the fresnel sense [14:32] <bwm> monica: this is a opportunity for interface designers [14:32] <bwm> duane: if we are filtering and applying context [14:33] <bwm> ... how do we capture the context [14:33] <bwm> ... I write something and six months later don't know why I wrote it [14:33] <bwm> davidK: sounds more like an application and a document [14:33] <bwm> ... the distinction might evaporate [14:34] <bwm> ... a document may be a whole application for viewing the information [14:34] <bwm> ... documents could become incredibly accurate [14:34] <bwm> bijan: haven't all active document systems failed - they are hard to maintain, they don't print well, ... [14:34] <bwm> davidK: then we'd better solve those problems [14:35] <bwm> bijan: this has been tried a couple times - OpenDoc, Ole, even applets [14:35] <bwm> ... pretty much they all suck [14:35] <bwm> ... why will it work this time [14:35] <bwm> davidK: those systems embedded complex applications, we are embedding ontologies [14:36] <bwm> bijan: spreadsheets are documents that have subapplications [14:36] <bwm> ... but that is really just one application [14:36] <bwm> ?: a db backed website is an application [14:37] <bwm> bijan: take amazon - I don't think of amazon as an application but as a bunch of documents [14:37] <bwm> davidk: i think amazon does express some of the things we are looking for from the semweb [14:37] <bwm> ... faceted navigation [14:37] <bwm> ... review mechanism [14:37] <bwm> ... annotations [14:37] <bwm> bijan: but are those active documents [14:39] <bwm> duane: could be a breakout session on the line between the application and data [14:39] <bwm> ... what about the idea of the language and the ontologies themselves [14:39] <bwm> ... is that changing anything - opportunity or challenge? [14:40] <bwm> ?: folksonomy [14:40] <bwm> duane: the capturing language and the idea of surfacing common or maybe useful language [14:40] <bwm> ?: or common sense [14:41] <bwm> duane: ok [14:41] <bwm> ... we will post these up [14:41] <bwm> ... may use in breakouts [14:41] <bwm> ... based on the notes that we got from the morning [14:41] <bwm> ... and from some of the papers [14:41] <bwm> ... break group into 5 groups [14:41] <bwm> ... we have 40-45 people [14:42] <bwm> ... discussion groups of 8-9 people [14:42] <bwm> ... possible topics: [14:42] <bwm> ... how do we model and understand semantic data sets [14:42] <bwm> ... creating ontologies [14:42] <bwm> ... top down, bottom up [14:42] <bwm> ... formal and/or informal [14:43] <bwm> ... 3: query and search construction [14:43] <bwm> 4: learning and use: are semweb applications har dto learn or only useful to "knowledge people" [14:43] <bwm> 5: information integration at the user interface level [14:51] <bwm> ... one breakout group around 1&2 [14:51] <bwm> ... another around 3&4 [14:51] <bwm> ... two around 5 [14:53] <bwm> I'm not sure what the 5th group is though [14:53] <bwm> - there are only 4 [15:36] <bwm> gathering again [15:37] <bwm> breakout group 3: on integration [15:37] <bwm> ... integration at the ui level [15:38] <bwm> ... in the wysiwyg sense [15:38] <bwm> ... spoke about adding [15:38] <bwm> ... what types of information sits in the computer as a whole [15:38] <bwm> ... acts as a interface [15:38] <bwm> ... the goal is to let the users produce more information to share [15:39] <bwm> ... how do we get enterprise to adopt it [15:39] <bwm> ... enablers - it should be easy for the users to do [15:39] <bwm> ... what is in it for me [15:39] <bwm> ... challenges [15:40] <bwm> ... credit and trust [15:40] <bwm> ... it is natural that when you are looking at something you can change it [15:40] <bwm> ... how do you credit for your contribution [15:40] <bwm> ... people want to know the provenance of information [15:41] <bwm> ... perceived cost/effort benefit [15:41] <bwm> ... big problem for semweb [15:41] <bwm> ... cost of implementation is high [15:41] <bwm> ... much discussion at the methodological level [15:42] <bwm> duane: we'll take a break now and come back for the rest of the debriefs [15:42] <bwm> ... lets be back by a little past 4 [15:42] <bwm> adjourned [15:42] <mc-iswc> pro bono vs what's in it for me: schraefel's paradigm for usable design - must increase immediate benefit to achieve long time large scale good FIRST rather than hold out the promise of personal benefit after "for the good" of the community is achieved [16:11] <bwm> resuming after coffee [16:11] <bwm> group 4: user integration [16:12] <bwm> ... goals syndicated views [16:12] <bwm> ... exploiting relationships [16:12] <bwm> ... understnding what relationships [16:12] <bwm> ... freeing information from being locked into applications [16:12] <bwm> ... need shared vocabularies and vocabulary translation [16:13] <bwm> ... need transparency of behvaiour - user needs a model of what is happening behind the scenes [16:13] <bwm> ... so they can understand when things go wrong [16:13] <bwm> ... applications lock in data - not in the interest of the vendors to free it up [16:14] <bwm> duane: questions or comments [16:14] <bwm> davidK: I'm of the opinion that we should come up with common ontologies [16:14] <bwm> ... thoughts on how practically taht might be done [16:15] <bwm> duane: who is next [16:15] <bwm> [points to Bijan] [16:16] <bwm> bijan: I'm not doing it [16:16] <bwm> ... we realised that the main problem is there is a big gap between knowing what folks want to do and what technology can do [16:16] <bwm> ... e.g. if you are enriching an ontology you would do something different [16:17] <bwm> ... you'd figure out what would be useful for that task [16:17] <bwm> ... that is where most of our discussion is based [16:17] <bwm> ... querying the content of an ontology you don't care about its structure [16:17] <bwm> ... when its small you care less [16:17] <bwm> ... when it gets bigger you need help to figure out where to put new concepts [16:18] <bwm> ... all the tools have different benefits for different activities [16:18] <bwm> duane: what is number 3 [16:18] <bwm> ?: that is we can't remember what number 3 is [16:18] <bwm> davidK: I'm nervous about what people want to do with ontologies [16:19] <bwm> ... as opposed to ontology developers [16:19] <bwm> ... who can be given powerful tools [16:19] <bwm> ... but what do the end users need [16:19] <bwm> ?: that's what we need to do now - figure that out [16:19] <bwm> ... that's the future [16:20] <bwm> duane: there is an intersection point with the usability community [16:20] <bwm> ... who have picked up the idea of design pattersn [16:20] <bwm> ... as a way of looking at tasks, objectifying tasks [16:20] <bwm> ... seems like a good intersection point with this [16:20] <bwm> duane: who is talking for the 4th group [16:20] <bwm> ?: [16:21] <bwm> ?: I can't read the flip chart [16:21] <bwm> ... we were talking about query and search construction [16:21] <bwm> ... what would have to happen to make semweb usable for the end users [16:21] <bwm> ... we split into two ... [16:21] <bwm> ... there is no strict differentiation between user and developer [16:21] <bwm> ... user can rearrange data [16:22] <bwm> ... in a certain way the user becomes a developer [16:22] <bwm> ... so usability must consider both [16:22] <bwm> ... but we focussed more on the end user [16:22] <bwm> ... concerning the challenges, what we saw ... [16:22] <bwm> ... the majority of us wanted to start from a simple interface like a user is used to [16:23] <bwm> ... but still include the posibilities of RDF [16:23] <bwm> to be able to pass a query or an input to the semweb system and translate results back into user terms [16:24] <bwm> ... we didn't come up with a solution [16:25] <bwm> ... that is it [16:26] <bwm> duane: that is good [16:27] <bwm> ... what are the things that we can do to involve the users [16:27] <bwm> ... four subject areas to consider [16:27] <bwm> ... 1: how to make collection of data or metadata easy [16:28] <pshab> ... user analysis, user profiles, interaction design, development, usibility testing and eval, data contribution/maintencance [16:29] <bwm> what sort of thing would we want to involve the users [16:29] <bwm> ... identify context [16:29] <bwm> making sure the user realizes they have a lot of metadata already [16:29] <bwm> ... its just in the wrong place [16:30] <bwm> ... a lot of metadata exists and just gets thrown away [16:30] <bwm> duane: explore the user's recognition or valuing of existing metadata [16:32] <bwm> davidK: I have doubts about users creating metadata as a separate activity [16:32] <bwm> ... metadata will be generated automatically and then fixed by users [16:32] <bwm> duane: things that would allow users to ... [16:32] <bwm> davidK: inline correction of metadata [16:33] <bwm> duane: can I suggest that some activities around interaction design, ... [16:34] <bwm> paulS: because of an error in metadata you never get to see the resource because it will be hidden [16:34] <bwm> duane: how would you design an activity to figure that out [16:34] <bwm> avi: doesn't this lead to a software requirement that you must deal with incorrect metadata [16:35] <bwm> ?: use trust mechanisms [16:35] <bwm> ... we can label metadata as machine generated [16:37] <avi> bwm: Shouldn't a bot be able to determine what we do (here at the workshop) given the content of this IRC channel and the URLs that we type [16:37] <avi> --> hence what cues should we be looking for (or ask the users what are reliable cues)? [16:38] <pshab> ... Assemble "chunks" into "document"? [16:38] <bwm> ... what sort of things do we want to do there? [16:39] <pshab> ... e.g. user analysis, user provlies, interactive design, development, usability testing and eval, data contribution/maintenance [16:39] <bwm> avi: there is a whole chunk of research ... you have to understand that users will look for genre cues to understand what a document is about [16:40] <bwm> ?: one can work top down or bottom up [16:40] <bwm> duane: what experiment do you want to do? [16:40] <bwm> ?: [missed] [16:40] <bwm> ... top down is a particular goal - higher quality but less freedome [16:41] <avi> ( bottom up is more serendipitoues ... (probably misspelled)) [16:41] <bwm> duane: I suggest one activity might be to test those assumptions [16:41] <bwm> ... are their other organizing principles [16:42] <bwm> ... are there things that are identified by uri [16:42] <bwm> s/.../?:/ [16:43] <bwm> duane: there is a challenge there, but in terms of an acitivty how can you facilitate their creation of a uri where one is needed [16:44] <avi> bwm [missed it] [16:44] <bwm> bwm: identifying chunks that users use [16:44] <bwm> davidK: are we pushing programmer needs onto a pool of users [16:45] <bwm> ... is there some evidence that we can offer a benefit [16:46] <bwm> duane: is about assembly or synthesis [16:46] <bwm> ... much discussion about reuse [16:46] <bwm> paulS: one area where there is a lot work on this is in e-learning [16:47] <bwm> ... taking composite learning objects, taking them apart and reconstiting it [16:47] <bwm> davidK: are they showing its valuable or doing it [16:47] <bwm> paulS: they assume it would be useful [16:49] <avi> bwm: Maybe to diferent situations: [16:49] <avi> 1) One publisher that rechuncks information [16:49] <avi> seems economical [16:49] <avi> 2) Users exploring the web that pool and repool data [16:49] <bwm> avi: I know that at least two companies are doing this [16:50] <bwm> duane: there are many motivations for doing these things [16:50] <bwm> ... we seem to be moving on [16:50] <bwm> ... we have heard conversations today about the general and a specific situation or task [16:51] <bwm> ... in ontology maintenance we want to investigate the kind of maintenance tasks there are for ontologies [16:51] <bwm> ... are there others? [16:51] <bwm> ?: what are the parameters that get changed [16:52] <bwm> ... what is the space in which personalization occurs [16:52] <bwm> duane: uses the "C" word [16:52] <bwm> ... that context [16:53] <bwm> ... how do we help someone communicate their context to semantic applications [16:53] <bwm> ... what kind of things would we want to explore to find that out [16:53] <bwm> ?: I don't think we want to tell the computer about our context [16:53] <bwm> ... I want it to figure it out [16:54] <bwm> duane: how would we involve users in creating a better bob [16:54] <bwm> ... bob=paper clip [16:54] <bwm> ?: we need more sensors [16:55] <bwm> ?: we have got a lot of information and we can explore any of them [16:55] <bwm> ... we need the environment to work out what we are interested in [16:55] <bwm> ... see how users react to big lists of possible connections from where they are [16:56] <bwm> duane: any other tasks we might want to do to look at identifying user's context [16:56] <bwm> ... is there a modeling task to look at how existing applications can expose a little more about the context they know about [16:57] <bwm> davidK: clippie is objectionable because the system jumps at you [16:57] <farh0rizon> there has been lots of research in automatically capturing user context from environmental cues [16:57] <farh0rizon> summary: it's *really* hard [16:57] <farh0rizon> and not very effective [16:58] <bwm> duane: there are interesting experiments in cognitive interaction space [16:58] <bwm> ... monitors response over time and adapts more or less to the degree of receptivity of the user [16:58] <farh0rizon> btw, bob != clippy [16:58] <avi> I disagree n that it is not really effective. Studies by Fogarty et al and others show quite some promise... [16:58] <bwm> s/interaction space/ adaptive interface space/ [16:59] <bwm> oops - bob != clippy [16:59] <bwm> duane: moving on to the fourth one [16:59] <bwm> ... knowing where the data comes from [16:59] <bwm> ... a couple of differnt groups talked about info load [16:59] <farh0rizon> avi: do you have refs? [16:59] <farh0rizon> or at least a title? [16:59] <bwm> ... what would we want to do to explore with users [17:00] <bwm> ?: some users are open and trusting and don't care [17:00] <bwm> ... only users who are interested in provenance will have the problem [17:01] <bwm> duane: looking at user profiles or personas and sensitivity to overload, ability and desire to investigate and/or trust what they are seeing [17:01] <bwm> davidK: i hear losts of people talking about trust [17:01] <bwm> ... but a lot of evidence from the web is that people trust everything [17:01] <bwm> ... one response is don't worry about it [17:02] <bwm> ... a more pessimistic attitude is that we have a disaster in the making [17:02] <bwm> ... current trust on the web is based on a web site, and as data smears out this could no longer be viable [17:02] <bwm> duane: they trust the source because they beleive the source [17:03] <bwm> bijan: distinguish between trust as a security issue and the way we used it as a similarity metric [17:03] <avi> forh0rizon: look at Fogarty, J., Hudson, S.E., and Lai, J. "Examining the Robustness of Sensor-Based Statistical Models of Human Interruptibility," Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (CHI 2004), 2004 for a starter [17:03] <bwm> ... in order to do anything on the web you need to have trust - we don't have records of when people didn't trust it - didn't enter credit card number [17:04] <bwm> duane: let me ask you a follow on question [17:04] <bwm> ... how would you investigate which things they would rate, trust etc [17:04] <bwm> ... is there a way to learn that [17:04] <bwm> bijan: I don't u nderstand the questions [17:05] <bwm> duane: are there types of things or amounts of things or frequency of asking that predispose people to trust [17:05] <bwm> bijan: I don't have anything on that [17:05] <bwm> davidK: asking people to assign trustworthyness is the what's in it for me problem [17:05] <bwm> ... having used info why should invest effort in asserting trust [17:06] <bwm> ... so just watch what they do with it [17:06] <bwm> ?: people rate on ebay after they have used them [17:07] <avi> duane: what other areas should we explore with users? [17:07] <avi> bwm: We may want to look at what kinds of information people are more or less sensitive about the provenance [17:08] <bwm> ?: currently on the web there is remote scripting and interactivty in the browser [17:08] <bwm> ... it would be interesting to see if these new interfaces are being accepted by users because they can be a huge advantage [17:09] <bwm> ... the question is whether these new interfaces are already mass market accepted [17:09] <bwm> duane: are users becoming comfortable iwth newer in page dynamic interactions [17:09] <bwm> ... looking at some of the things that are out there now [17:09] <bwm> ... over time users become more familiar and then more comfortable [17:10] <bwm> ... we have a little time left [17:10] <bwm> davidK: how do we involve ourselves [17:10] <bwm> ... can we do stuff communally [17:10] <bwm> ... that would advance things more rapidly [17:10] <farh0rizon> avi: thanks for the reference [17:10] <bwm> duane: any thoughts anyone [17:11] <farh0rizon> interesting, but I'm not sure it directly addresses the context issue [17:11] <bwm> monica: [can't here - someone closer type] [17:11] <bwm> s/here/hear/ [17:11] <farh0rizon> maybe I'm not interpreting the discussion correctly (quite possible) [17:11] <avi> mc: Things that david do are already a good start [17:11] <avi> ... make ourselves aware about the our groups [17:12] <bwm> ?: there may be unusual opportunties for concerns in the semantic interface that don't come up in other areas [17:12] <farh0rizon> but I took it to be a discussion essentially about inferring /semantic/ context and goal detection [17:13] <bwm> ... hearing how other people have done completely different things even with the same background and goals [17:13] <bwm> davidK: we have tools and collections of RDF [17:13] <bwm> ... if you have some you could add to our repository [17:14] <bwm> ... google for the semantic bank [17:14] <avi> bwm: Is there a large enough community here for a mailing list? [17:16] <bwm> avi: seems to be [17:16] <bwm> davidk and avi volunteer to investigate [17:17] <bwm> duane: I talk a lot to people in the usability and hci community [17:17] <bwm> ... are there questions/ideas [17:17] <bwm> ... what do we want [17:18] <bwm> ... there is a tremendous hci community [17:18] <bwm> ... here we have a bunch of semweb folks - but we are not the hci community [17:18] <bwm> ... we each have half the skills needed to solve the problem [17:19] <bwm> ... is there a way to get the hci community to treat this as a major problem [17:19] <bwm> monica: there is a feeling that a lot of that is done - interacting with geeks with semantics isn't interesting [17:19] <bwm> ... need to post novel problems [17:20] <bwm> ...need to formulate the questions of interest in a new way to that community [17:20] <bwm> duane: i would second that [17:21] <bwm> ... its not getting into the interface [17:21] <bwm> davidK: so explaining how what we want is different is what is wanted [17:21] <bwm> ... there is lots of work on large datasets with a schema [17:21] <bwm> ... what about when there is no schema [17:21] <bwm> bijan: what tasks [17:22] <bwm> davidK: sense making, finding, all the traditional tasks [17:22] <bwm> duane: any other points [17:22] <bwm> avi: before we are done let me say a few things [17:22] <bwm> ... please give me your email addresses so we can inform you when we set up the email list [17:23] <bwm> ... we tried to do something different from a conventional workshop format [17:23] <bwm> ... can we get some feedback in form of show of hands [17:23] <bwm> ... who think this alternative format which didn't allow so much papers [17:23] <bwm> [general support] [17:24] <bengee> bwm++ for scribing [17:24] <bwm> adjourned [17:24] <farh0rizon> thanks bwm from the remote audience too