09 Dec 2010


See also: IRC log


Jeff Waters and Don McGarry
Jeff Waters


<scribe> Scribe: Jeff Waters

<scribe> ScribeNick: jeffw


jeffw: Welcome to the Dec 9, 2010 meeting of the Decisions incubator!
... You can find the agenda at http://www.w3.org/2005/Incubator/decision/wiki/Decision_Mtg_18_Agenda
... We are pleased to have Dr. Jill Drury and Dr. Gary Klein join us to discuss their decision research.
... There is http://audioconference.mitre.org

<Piotr> what is the valid meeting ID?

jeffw: (various of us got on to the web conferencing system)

The Mitre Corporation: CoAction: Enabling Collaborative Option Awareness for Joint Actions

garyKlein: Collaborative behavior is not often supported by the technology. So I can see you are connected, but I don't know much else which indicates the problem.
... The papers I sent out on option awareness, and I read Eva's position paper, so I thought it might be useful to talk about the connection we made earlier between decision spaces and situation spaces.
... From what I read in the paper, alot of the info communicated about the decision in the paper deals with the situation space. It may convey something about the options, but really that is about the state of the world
... rather than the assessment of the options which we call the decision space. Situation Awareness is defined in 3 levels and its important to not the 1st level is the elements in the space,
... but 2 and 3 are projections, deriving the meaning of the chess pieces and the status of the near future. The reason I bring this up is because most common operational pictures are about situation awareness
... to support decision-making, but in order to do that, it needs to be mapped into the decision space. So when a doctor needs to decide if you have a virtus or a bacterial infection,
... then we look for certain things but that's to map to whether we give you or not an antibiotic. So the comparison of options, to give or not, is the decision space.

jeffw: (summarized his thoughts on that)

jill: we weren't first to come up with this, but we've added some new notions

garyKlein: We've been interacting with a number of the researchers in this work. You've generally got the outline, Jeff, you take dots on a map as level 1, then you consider what courses of action might be taken,
... and then the meaning is brought by a mental model, and then a further model as to what they will do next and that figures into the decision space as to the options I will consider.
... and this distinction was exciting to me. I was trying to understand the differences between decision support systems, some were just databases of things and some were comparing options.
... one is in terms of situation space, i.e. dots on a map, but that is just the beginning of considering the decision space of comparing options. To jointly visualize options and futures is the decision space.
... You can see on our slide the 17 options and we're comparing a sample uncertainty analysis, running them a thousand times with adjusting factors they will impact in environment, and
... we see some low levels of dissidence (which is the cost in this example), whereas other options result in tight set of results, robust options result in good
... The middle line in the box is the median, the box width is the interquartile range, and the end of the line points ... this is the standard representation of a box plot
... The dots are the outliers.

marion: How do you know what the outcomes would be?

garyKlein: there is a model that we used in this case
... So you have options, these go into the models and the endogenous and exogenous variables go into the forecasting models with uncertainty/sensitivity analysis go into plausible futures which drive a scoring model

jill: the work we've done is model agnostic, so you could use your own with your own number of parameters

garyKlein: so whether you use your own cognitive model or some other model, there is always some model applied to bridge to the decision space and choose among the options. A computational model formalizes this
... and allows us to do more future analysis than a person would be able to do.
... these are ideas behind our decision display which allow the box plots to be seen by the users and they can adjust the parameters/weighting on the different aspects of the box plots, max/min/median etc.
... Level 2 is perception and comprehension of relationships between factors underlying the option outcomes
... maybe there is something I can do to mitigate that factor.

marion: is that a drill-down on the edges, the relationship of the variables, that is drill-down on the arrows, to see what is independent of what.

garyKlein: I'm trying to quell the spread of disease and I'm finding 10% antiviral and there are cases turning out badly and people are dying anyway and I drill down and find that is where I didn't hit my 10% target
... so there is a tipping point in the viral situation example, so I'm looking at the drill-down info to get important choice options and level 2 gets you there.
... level 3 might be a quality assurance program to assure I get my anti-virals distributed, but there is a cost to that, so does that reduce the number to offset the cost
... I hope we can collaborate and participate in future meetings to follow-up on this.

jeffw: (summarized the significance of the work on both sides and invited Jill and Gary to participate in future meetings and collaborate off-line and in the wiki)

garyKlein: just to finish up, here is an example of decision-makers who have their own decision-space, but also a mutual decision space where for example sending 4 fire engines due to congestion

marion: do you have a feedback loop?

garyKlein: yes, and all of these are cost evaluations of whatever could occur for each of the possible options. So police and fire working together (4 fire trucks and 2 squad cars independently), but
... we find that if 1 more squad car is sent, then fire only needs to send 2 and you only discover that with considering the joint decision space, which is not obvious, and we just finished role-playing
... experiment which shows this in some detail. So let me give you a taste of what we are doing with collaborative validation, technology can support the collaborative task, but joint work is impacted
... by type of task, type of interdependence, type of coordination, and characteristics of the task environment. So when I'm looking at the protocols, are we communicating the rationalization,
... the justification (the external values we're trying to achieve), and we've looked at the structure of the decision (the network of factors that impact a decision), also looking at the types of coordination,
... are we communicating during the process (like loading a couch onto a truck), are we engaged in plans with schedules, do we have routines (like driving on road has rules so less communication required)
... and there are collaborative behaviors, and I can skip to different class of task processes, and interdependence and it all comes together, the situation awareness, the task process, the coordination
... with what we need to discuss to collaborate correctly. I'd like to talk more about these things.

jeffw: Thanks much, Gary and Jill
... Let's follow up on all of this next time but also online in meantime through email and wiki. I'd like to make sure the papers are accessible to all, so we can link them.

eblomqvi: How can I get the papers?

garyKlein: I can send them to all on the list.

International Semantic Web Conference and Workshop on Ontology Patterns

eblomqvi: My report is I tried to look around and find something related to our work and I didn't find anything very closely related to decision format but the conference in general, the semantic web challenge
... on linked data etc. are interesting. I will post links to the challenge page and you can look at those with demos

<eblomqvi> http://challenge.semanticweb.org/

eblomqvi: it's under previous challenges and you can see this year and last year's, the semantic web challenge is applications of semantic web technology to make things useful and nice that
... use link data and other things as well, one was like wikipedia but dbpedia, using linked data and allowing people to create that, doing it the wiki way, and Jeff you've been talking about the
... application that uses some forms for different things and this may be related and inspiration for looking at that

<eblomqvi> shortipedia

<eblomqvi> http://shortipedia.org/index.php/Main_Page

eblomqvi: for the workshop, I presented our position paper and there was some nice discussion, people seemed to think it was needed and interesting, but no one there who was acutally working with decisions,
... then we also proposed modeling problems for the pattern writing and one was selected, how to represent ordered lists, and so this was one of the things we worked on at the workshop

<eblomqvi> http://ontologydesignpatterns.org/wiki/Submissions:List

eblomqvi: and the conclusion was yes it would be nice, so I've reengineered the ontology into three patterns and I posted them in the ontologydesignpatterns portal and I put the links here
... so this list pattern is there and hopefully we will have some comments and it's there and we can try to use it.
... I mentioned the modeling problems and I got some interesting ideas and also today, so we have a new set of ideas to model with.


jeffw: Here is the link http://www.w3.org/2005/Incubator/decision/wiki/CDEP_Examples_in_RDF

marion: distinction between situation space and decision space seems new, but presenting options has been around in some time, although not necessarily bringing it out to the user.

garyKlein: this frequency approach has taken off and allows people to understand the uncertainty much better than in the past.

jeffw: Thanks all for attending. It has been an interesting hour, we're getting wonderful new ideas and I look forward to exchanging links, papers, slides, etc. and collaborating off-line as we head to the next incubator meeting in 2 weeks.

<Piotr> I've submitted the criterion and criterion setter pattern to the ODP portal

<Piotr> http://ontologydesignpatterns.org/wiki/Submissions:CriterionSetter

jeffw: Thanks Piotr!

<eblomqvi> ah, I didn't notice that! I'll have a look as soon as I can, Piotr

<Piotr> http://ontologydesignpatterns.org/wiki/Submissions:Criterion

Summary of Action Items

[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.135 (CVS log)
$Date: 2010/12/09 16:49:27 $

Scribe.perl diagnostic output

[Delete this section before finalizing the minutes.]
This is scribe.perl Revision: 1.135  of Date: 2009/03/02 03:52:20  
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: RRSAgent_Text_Format (score 1.00)

Found Scribe: Jeff Waters
Found ScribeNick: jeffw

WARNING: No "Present: ... " found!
Possibly Present: Piotr ScribeNick eblomqvi garyKlein jeffw jill marion
You can indicate people for the Present list like this:
        <dbooth> Present: dbooth jonathan mary
        <dbooth> Present+ amy

Agenda: http://www.w3.org/2005/Incubator/decision/wiki/Decision_Mtg_18_Agenda
Got date from IRC log name: 09 Dec 2010
Guessing minutes URL: http://www.w3.org/2010/12/09-decision-xg-minutes.html
People with action items: 

[End of scribe.perl diagnostic output]