decision-xg -- 16 Sep 2010

<scribe> scribe: jeffw

<scribe> scribenick: jeffw

jeffw: Welcome to the 13th meeting of the Decision Incubator. We are in the midst of modeling efforts to develop the decision format. Eva has done a very nice job applying patterns to develop the "core" components of a decision and we received some nice feedback this week from Piotr Nowara raising some good questions and supportive discussion. One of the main topics for our meeting today is to have Piotr on the line and with Eva, I'd like to suggest

the discussion. I'd like to get educated on the issues and how they impact our thinking.

Update and Insights on the "Core" Decision Model

jeffw: Hi Piotr, will you be dialing in?

<piotr_nowara> hello! sorry for being late. I'll join you in brief second

jeffw: No problem

<piotr_nowara> I got some tewchnical problems

jeffw: any students who will join us at your university Eva?

eblomqvi: perhaps in the spring, there is a masters program, so there may be some students at the appropriate level to assist with us assuming we can match interests

jeffw: I'm kind of excited about our progress, we have some good use cases, we've identified patterns addressing those and the major components of the decision model
... Our paper provides a good summary of our goals, our progress and begins to lay a roadmap for where we are headed.
... So if we list out our patterns, it begins to be a matter of implementation and assigning who wants to work on what and then getting some implementations and integrating
... eva, I've been busy, sorry, have you had any updates to your work

eblomqvi: I did update some of the core component portions of the wiki based on the discussion with Piotr, and I intended to have some more examples, so some new components
... new versions of the existing components that were there and I updated the changes only listing the issues from last meeting and Piotr's comments

jeffw: ok, that's great, since Piotr may be struggling to connect, can you educate me on the one or two major issues and any resolution?

eblomqvi: there is a page listing the issues

<eblomqvi> http://www.w3.org/2005/Incubator/decision/wiki/Use_Case_OLD_Core_Decision_Model#Draft_Solution

eblomqvi: I kept the old files, I just renamed the files "_Old", so you can see those if you want, there is a list of the issues on the page
... description of the options versus realization of the options concept, we decided last time that realization may not be the best way to model this but rather the description
... then this was also one of the things Piotr remarked on when he provided his comments, so he was agreeing with our discussion that there might be a better way to model
... I changed the model and used the "description" pattern, more about abstract descriptions of concepts, now of course this is now up for discussion
... Piotr remarked this may be too much meta-modeling, and I agree that the description pattern it is usually used with "situation" and it appears at the meta level
... so perhaps this can interpreted as too abstract, so perhaps just forgetting this pattern, since the pattern is designed to help, explain something, to provide a higher level model
... if it's more of a confusion, if too abstract, then I agree with Piotr, that understanding with users is important, so perhaps we don't import this pattern, altough we may be following this same type of description
... So we could start to model some examples in meantime and decide if we need the pattern or not, so perhaps Piotr has comments on this as well
... The second issue was about criteria. Again, this is indicated more about choice of modeling, how do we model this, before I had modeled as an abstract concept
... but you can do it as an actual situation, it doesn't have so many abstract components, you need to say these critera are for these options, more on the level of modeling a dialog
... in the ontology the criteria is an abstract class of concept and that is what Pitor was arguing against, the criteria connects some values to the options, so for me it was an abstract concept to connect, subclass of concept
... Piotr was arguing it is a concrete situation where we are applying these values, so he was arguing that the criteria was a kind of situation, but we agreed on the properties that the criteria ...
... let's get Piotr's thoughts on this since he's on the line now

piotr_nowara: regarding the options v. descriptons seems to make sense, but the descripton pattern appears to be better to be more for meta-modeling, so the question is should the options be treated abstractly?
... may be useful for different contexts of similar options, but given the use cases we have now, perhaps it could be more concrete

eblomqvi: Yes, I agree we shouldn't just add the pattern, it should be useful in some way, so perhaps the description pattern isn't appropriate, so I think you have a good point

jeffw: Is there a pattern here?

piotr_nowara: I'm not sure one exists

eblomqvi: I think the situation pattern may fit, we are working sometimes top-down and sometimes bottom-up, so if we don't find something, when we model some real data or decisions we'll see if we need more
... No, not the situation pattern, it doesn't really matter if we don't have a pattern at this moment, let's just leave it for now,

jeffw: so the consensus is let's just model some things, I do feel there is an implicit pattern there but it's maybe the one we're producing

eblomqvi: yes, but the abstract description of the option, is the one which may not need a pattern

piotr_nowara: criteria as a subclass of the concept is the second issue, criteria is a kind of measure and maybe the situation pattern applies
... I started on a criterion pattern and I need a model to determine how criteria are determined by something by like proposals

eblomqvi: piotr and I agreed on this and I wil make some updates on the model and if you have some example or model of what you are doing, I would like to see it and we can discuss it

piotr_nowara: when I succeed in my task of modeling the criteria, then I will share

jeffw: how does the situation pattern match the criteria need?

eblomqvi: you can view it as you apply this term onto this option for this particular decision, so it can be viewed as a situation where you have applied a criteria parameter to some options
... yes, we are talking about applying the criteria as opposed to what is the criteria when we talk about applying the situation pattern

<piotr_nowara> I got disconnected... I'll try to dial in again

jeffw: were there any other issues discussed? (ok, sorry we lost you Piotr, I think we lost Eva as well since here battery was dying)
... We lost Eva on chat as well. Piotr do you want to continue our discussion on chat or if not, we can end our telecon here
... Thanks, Eva and Piotr, for participating and all the great discussion and work on bringing our decision model to life.
... I'll go ahead and summarize thoughts on a couple other topics quickly, just to get them in the notes.

Pattern Implementation Schedule and Future Papers

<piotr_nowara> Thank you! And once again: sorry for being late...

jeffw: As a side note, I mentioned previously that I'd like to use some of our documentation and some slightly rewritten portions of our recently submitted paper to fill in the final report outline that we have on the front page of our wiki.
... So that's a task for me and hopefully I will make some progress on that and we can discuss next time.

<eblomqvi> sorry, I lost the connection... I try to call again, even if just to say bye... I found a power plug :-)

jeffw: Hi, Eva, welcome back. I was just inquiring whether you and Piotr wanted to continue the discussion on chat.

<eblomqvi> sure, we could to that

jeffw: But I didn't see a response from Piotr, so he may be having technical difficulties or had to depart.
... So I went on to summarize a few thoughts on other topics

<eblomqvi> ok, should I call in again?

jeffw: One thing I wanted to note is if there any other major issues that you and Piotr discussed, but why don't we say given we don't have much time left on the call, that they are mentioned on the wiki page you pasted in earlier

<eblomqvi> yes, that's true, they are there

jeffw: Since we only have 5 minutes left, I'd say we can end the telecon here so you don't have to worry about getting things plugged in

<eblomqvi> and there were not any other major issues, just smal ones

jeffw: ok, great, again, thanks for all the excellent work, on the paper, on the models, on the discussion.

<eblomqvi> thank you!

<piotr_nowara> The last issue I wanted to discuss today was if need Concept class at all?

jeffw: We'll pick it up next time and I will bring some things to the table as well

<piotr_nowara> ok

<eblomqvi> yes, and we can continue the discussion via e-mail also! as we did recently

<eblomqvi> and then update the wiki accordingly

<piotr_nowara> that's the good idea!

<piotr_nowara> I'll stay in touch with you

<eblomqvi> do that! I look forward to the continued discussion

jeffw: ok, and Piotr, we can add our combined thoughts and documenting on the wiki is easier for people to track I think after our email discussions
... Thanks to both of you and we'll continue discussions on email and add to the wiki.
... Our time is up. Take care!

<eblomqvi> thanks! bye for now

<piotr_nowara> Bye

jeffw: Here is just a note added to our minutes as an overview of our decision components and the potential patterns implied.
... Regarding the list of patterns, I realize that the patterns are derived from the use cases so these need to go hand-in-hand;
... however, we have some use cases and have suggested some patterns. Once patterns are identified, it is time to reuse previous implementations (from designontologypatterns.org) or to develop the implementation, then specialize as needed to address the use case.

One way to get a handle on the workload is to list out these patterns and begin to assign them to folks for reuse and specialization. We have the beginning list of patterns on our wiki at http://www.w3.org/2005/Incubator/decision/wiki/Final_Report_Decision_Components_And_Patterns.

On this wiki page, we mention the patterns which are implied by the basic components of the decision, which are (1) Question (Statement with Variable pattern); (2) Options (Filter and Aggregation/List patterns);

(3) Criteria and As (Normalizer, Weighting, Aggregation/List patterns); (4) Assessment (OrderedList pattern) which is the application of the Criteria to the Options; and (5) Choosing the Answer (Selection pattern)

jeffw: These are just suggested components and potential patterns, but I like the idea of using this page to document our components and the patterns that we use as an overview and quick reference.
... This also could be a way to help manage design work from implementation work if we have people interested in helping at different levels of experience with the ontology development.
... Also this overview might help us to get a handle on how much work is left and help to schedule out the work given the time remaining. Of course, just documenting the use cases, the required decision components, and the applicable patterns is great progress as well.
... Thanks!

- DRAFT -

decision-xg

16 Sep 2010

Attendees

Contents

Update and Insights on the "Core" Decision Model

Pattern Implementation Schedule and Future Papers

Summary of Action Items

Scribe.perl diagnostic output