27 Jan 2011


See also: IRC log


Jeff Waters and Don McGarry
Jeff Waters


<scribe> Scribe: Jeff Waters

<scribe> ScribeNick: jeffw

<piotr> Hi, Jeff! I haven't had time to work on my decision example since the last Sunday, so maybe I'll just focus on the issues I wrote about in my last email...

Hi, Aaron, Eva, Piotr!

jeffw: Welcome! We are at a good place in our incubator activity! We are finishing up our decision ontology patterns and beginning to combine them together into a complete (or at least an initally integrated) decision format.
... Our agenda for today is to review Piotr's updated Question Pattern ontology: http://www.w3.org/2005/Incubator/decision/wiki/Question_Pattern
... Also we wanted to in reviewing people's interest in continuing work as a Working Group.
... Also final report.
... Piotr can you give us an overview?

piotr: This is under development, not something complete, and I haven't had time to prepare additional documentation. I could prepare for next meeting or upcoming weekend.
... You should start exploring by using the reasoner to see the inferences and look at the instances. My decision scenario, it's quite simple but can happen in real life, stage 1 some patient is diagnosed with infection, doctor needs to prescribe right drug
... Doctor needs to consider potential allergies and info related to dose, such as age, weight, etc. So patient is asked whether or not he has allergy to penicillin, say no, then doctor decides.
... After a couple days, patient shows signs of allergy reaction, which suggests allergy to penicillin, so doctor reconsiders and prescribes some other antibiotic. So a simple scenario, but this gives us enough to consider interesting issues related to decision modeling.
... First question that came up, continuity or identity of decision-making process, it might be hard to tell when one decision process begins and ends
... At end of first stage of process, it might be assumed all ok, so decison-making is complete. But after a few days, it appears the decision-process had to be restarted. But stage 2 suggests the decision-making wasn't finished.
... This issue may be complicated, but for now we should leave this open.
... Our starting point is a Question that initiates a decision-making and an output is an Answer, so we have a Question and an Answer, pretty clear, a decision is a subclass of an answer and an answer is a subclass of a situation.
... I haven't had enough time explore the consequences of this, but we can try that approach.

eblomqvi: The situation Piotr refers to is from the situation pattern, an abstract state of affairs, it provides a state of affairs to connect things and it could work like that.

jeffw: I think of it as "Context". Is that correct?

eblomqvi: Yes

Piotr: My example clearly notes that a decision is different from the decision-making. The outcome product is different from the decision process. From beginning of decision modeling, I was interested that the decision ontology should describe this difference. The decision-making and its outcome. Decision-making is a process, could have multiple outputs.
... Decision is a simple fact, the selection of an option. So I think we need to describe this distinction.

jeffw: Yes that makes sense, you have Answer for a Decision, what did you use to represent the decision process?

piotr: I don't know if there is a pattern for process?

eblomqvi: We tried to model the process of decision-making over the summer, we made a draft and put it in the wiki, when we were discussing the information flow states, and we were moving to a new state of decision-making

jeffw: I see two answers for representing the decision process, one is the aggregation and one is the transition pattern for representing process. So we can bring that back in.

piotr: The pattern must enable us to represent the stages, cause a process consists of stages and we need to represent that. I haven't studied the transition pattern, but that sounds good.
... Another issue is to record a decision stage,even if no outcome. The patient may need to go thru some allergy tests that last a couple days and other facts may be gathered and all together may make the decider want to begin with a new decision-making stage.

eblomqvi: like branching, you go down one path and then you change the idea?

piotr: Yes, that's correct. May need some sophisticated modeling, but some little consideration could be given now.

jeffw: Yes, that's an interesting pattern, to show candidate options that are being considered, but then new ones appear better and we could model that as a general flow of thinking that occurs in other settings.

piotr: In my example, I'm using instances to represent the same person in different stages of the decision process. In a real life application, a property would unambiguously represent one person, like an id or insurance number, but I don't have that yet in my example.
... Using different instance to represent provides a fine-grained view of the id. So the throat infection diagnosed at time A is represented by a different individual, but the same person diagnosed later at time B is a different individual representation.

eblomqvi: With respect to last point, I think it is useful to have snapshots of state of affairs at different times, but you can also model the type of the thing to distinguish between the type of infection and the concrete instance, the realization of the disease in a particular patient, this is a common distinction between the abstract concept and the realization.

jeffw: Can we aggregate these things with some of our other work, Eva, and have one common format that we can all contribute to, understand, etc.?

eblomqvi: Yes, we should be able to aggregate components, we should dig up the things we did in the summer, and begin to put it together, etc.

jeffw: (summarized invited expert roles)

eblomqvi: First step, let's collect what we have now, to create links to latest versions of Piotr's ontology, the decision process model from the summer, and we put them together in the section of the final report and we good, Ontological Patterns and Solutions is the place to aggregate.

jeffw: Eva, maybe we can visit offline, what about continuing this work as a working group? Any interest? I have

eblomqvi: Yes, definitely, and we need an organizational structure to help motivate us despite other demans. I'm all for continuing this in some formal sense. I have no idea about what we need to do for a Working Group, so I will go with your recommendation.

Aaron: definitely,

Piotr: Yes, of course.

jeffw: Thanks everyone for joining today. I will take an action item to follow Marie-Claire's advice to draft a working group charter. Hopefully, I will have that to share by our next meeting. Also I will explore more with W3C staff the formal or informal requirements for forming a working group.

Summary of Action Items

[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.135 (CVS log)
$Date: 2011/01/27 16:07:14 $

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 "Topic:" lines found.

WARNING: No "Present: ... " found!
Possibly Present: Aaron ScribeNick eblomqvi jeffw piotr
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_21_Agenda
Got date from IRC log name: 27 Jan 2011
Guessing minutes URL: http://www.w3.org/2011/01/27-decision-xg-minutes.html
People with action items: 

WARNING: Input appears to use implicit continuation lines.
You may need the "-implicitContinuations" option.

WARNING: No "Topic: ..." lines found!  
Resulting HTML may have an empty (invalid) <ol>...</ol>.

Explanation: "Topic: ..." lines are used to indicate the start of 
new discussion topics or agenda items, such as:
<dbooth> Topic: Review of Amy's report

[End of scribe.perl diagnostic output]