See also: IRC log
<scribe> Scribe: Jeff Waters
<scribe> ScribeNick: jeffw
Bruce and Jeff are on the phone line
Welcome to our 8th meeting
and I'm learning alot in the process about ontologies, our tools, the extreme design approach and thinking about the decision issues.
Piotr was going to join us today but if he isn't able to, there are a couple ontologies that he provided for our review.
Eva: He appears to have a nice start and it will be interesting to see his goals and vision.
<esela> is the conference code 2244?
Why don't we put those into our minutes.
<eblomqvi> This is the basic ontology that he is working on, more or less an upper ontology: http://code.google.com/p/basic-ontology/
<eblomqvi> This is the actual decision ontology: http://code.google.com/p/decision-ontology/
Piotr has not been able to join us yet, but we have links to his ontologies that we will review.
Eva: Open Linked Data, Jeff has
done a good bit of work, but we need a first model of decision
and Jeff started more from the dataset end, so
... we've done a nice collaborative work from different ends.
And I set up a small page of the modeling of decisions that I
was thinking about
... And started to really think about what a decision is. In
our use cases, we have different views of what is a decision.
For example, in the earthquakes with
... Jeff's sparql queries, the focus is on relevant pieces of
data for a report. While in other cases, it's more what are we
going to do with the data, like where do I want to live.
... I then use the data for this higher level decision. So we
are looking at this in terms of subdecisions and more overall
complex decisions. We will probably start with
... the data selection kind of decisions but we may want to
represent the high level decisions as well. More like a general
question that we can't evaluate with a
... simple query. We need to consider both kinds. In one case,
if we have what data is most relevant then the outcome is a
subset of the data while in the other case, it is more
... of a high level statement, like I will move to Los Angeles
or wherever. So somehow we need to incorporate all of this in
the model. I had some thoughts back and forth.
... A decision is something about my next action or what I am
going to do and we need to incorporate both in the model. I
don't know if you have any thoughts on this?
(Jeff summarized thoughts on multi-metrics and gray scale)
Eva: It also depends on who is
making the decision, cause I may get some help from the
datasets, but in the end I may decide some other thing than
what is indicated by the datasets and metrics
... So I'm thinking we may want to separate the decsiion and
...
Don: I'm observing and taking this all in
Eva: I added one subpage to the Open Linked Data use case for Core Decisioin Modeling where I was thinking about what we need just to record that decision.
Jeff: Anything Don on Situational Awareness?
Don: Not yet, I'm thinking of Golden Phoenix and how this might support that exercise or scenario, that would probably be good. We'd have something tangible to dig deep into.
(JeffW mentioned interest in infrastructure components, backend triple stores, Virtuoso, anybody have recommendations?)
Eva: I've heard Virtuoso is a good framework but I don't have personal knowledge.
(Jeff discussed SPARQL v. inferencing and coupling the two)
Eva: Yes, we need the inferencing support as well.
My use case for Metrics and Assessment is http://www.w3.org/2005/Incubator/decision/wiki/Use_Case_Metrics_for_Auto_Assessment
My progress on that is here http://www.w3.org/2005/Incubator/decision/wiki/Use_Case_Metrics_Progress
<esela> I'm afraid I have to leave now, thanks for having me.
(Jeff summarized his metrics work based on the links, any thoughts?)
Eva: I can put some information
on the e-mail and the SPARQL construct and you select some of
the data that correspond to the triple patterns and the
CONSTRUCT creates
... a new RDF graph. For example, you can make the earthquake
instances of a new graph, and you can include it in your
ontology as a new addition
One of the issues is lists of things, in xsd you can put tags around lists of things, how is it best handled in rdf?
Eva: To create the extra class of "Metrics" v. "Metric" doesn't really help, so the interesting thing is to have the collection of individuals in some way
(Jeff went on to raise issues: How do you best represent a question in RDF? Also how do you model the question? Reificiation?)
<dmcgarry_> Jeff- I have to step away early for another meeting
Eva: This could be a linguistic ontology and we could look in that direction, but I'm not entirely sure we want to proceed in that direction
<dmcgarry_> Thanks again for another great call & discussion!
Eva: To model the meaning of the
question and reason on it, then we really are doing advanced
stuff. That may be the last thing to do.
... Regarding reification, I edited the agenda and added a
couple of links to help explain that and added the links to
patterns for n-ary relations in OWL
... So we could take this to another level and not talk about
the individual statements, and we are not in control of the
data sets, we could take it to the level of modeling, there is
some
... of this in the open linked data sets. Something in line of
how to model data entry in the data sets and the n-ary
relations are good ways to do that.
... I have reached out to Jim's group and the Open University
and see if I can have one or more participate in our
discussions, perhaps brief us on their work.
That would be great, because what we're doing builds so much upon the excellent work of both of those organizations, that we would really benefit from having their advice and updates on their work
<piotr_nowara> Hi! Has the meeting begun? Should I call or will we communicate via irc?
and being able to discuss with them our concepts and vision for how we understand and are leveraging their work.
Piotr, Hello! So nice to hear from you.
I apologize that there appears to have been a mixup.
We just completed our meeting.
We started at 10am EST, which was 1 hour ago.
Our meetings are just 1 hour long, so everyone has disconnected from the call.
I apologize if I misstated the time.
We put in our irc chat the links to your ontologies that you sent and we took it as an action item to further review your work and to get in touch with you to follow-up.
<piotr_nowara> I think it's my fault
I'm sorry I should have tried to contact you. I thought you might join us part way through.
Here is a suggestion, if you want to consider this.
<piotr_nowara> No problem for me.
<piotr_nowara> Sorry for destroying your meeting plan...
If you have a paper, or perhaps just want to write up a paragraph or two, describing your goal, vision, status, schedule, use cases, or whatever,
then I could create a page on the wiki and post that information as part of our overall effort.
<piotr_nowara> I've already prepared a brief outline of my decision ontology for this meeting. It summarizes the current state of my project. It can be downloaded from http://code.google.com/p/decision-ontology/downloads/list
<piotr_nowara> I'va also emailed it to you and Eva some hours ago
Ok, great, I will make sure we have a page on our wiki that links to that and includes some of it as appropriate.
<piotr_nowara> Thank you
Can you remind me, do you belong to an organization that would be interested in joining the W3C?
<piotr_nowara> Please let me know if that will do or I should write more
(I don't think they allow individual memberships)
Are you affiliated with a university or non-profit organization (since I think the membership fees are more reasonable for those groups)?
<piotr_nowara> I work for KAMSOFT S.A. its relatively small company (less than 1000 employees). We work almost exclusively in health care domain,. They are interesting in joining W3C.
We're always looking for interested folks like yourself to join and help with this kind of effort.
I know it's hard for any organization to find funds and time to participate in these standardization efforts; however, I think it can be very worthwhile include for the business development and bottom line.
<piotr_nowara> Thank you for your invitation! I'd like to contribute to your project
There are alot of great companies, universities and non-profits who are members of the W3C, including health care organizations.
<piotr_nowara> I think our income allows us paying the lowest applicable fee
Do you know who to contact at W3C to pursue membership inquiries?
<piotr_nowara> Yes, as far I know KAMSOFT wants to fine business opportunities outside of Poland.
<piotr_nowara> No, I don't know who we have to contact
<piotr_nowara> could you provide me some information on that?
A great contact for membership is Karen Myers, karen@w3.org, 617-253-5509
Karen is wonderful and she'll be happy to answer any of your questions and get you going.
Once your organization joins, any of your employees can join any of the working groups, special interest groups or incubator activities.
You can find a list of all of the W3C groups at http://www.w3.org/Consortium/activities.html
The W3C has a technical staff of I believe about 50 people and this is a real asset and makes the W3C unique in this regard.
<piotr_nowara> Thank you for the information. We'll send an email to Karen tommorow.
Anyway, enough proselytizing on my part. But if your organization can join, we'd love to have you participate in our incubator.
Don McGarry (who co-chairs the incubator with me) and I both participate in a variety of emergency management activities, including participation in the OASIS Emergency Management Technical Committee.
<piotr_nowara> Thank you. I hope you can benefit something for my participation.
The OASIS EM TC fosters standards like EDXL HAVE (Hospital Availability) which is a standard allowing hospitals to report their status in support of effective routing of patients in an emergency.
Anyway, my point is that there are important use cases for our decision work in the medical field and application of the semantic standards of the W3C. So we can do some good collaboration in this regard.
<piotr_nowara> that's interesting. We are always looking for some new technologies
<piotr_nowara> I hope so!
<piotr_nowara> Maybe it's not all that wrong that I miss the today's meeting. at least anyone interested could read my paper.
So I will post up a link to your work and send you an e-mail when I do. Also, whether or not your organization can join, perhaps you can consider attending our next meeting and briefing us on your work and progress as you were planning to do today.
<piotr_nowara> Yes. that's a good idea.
I will be online adding a few more notes to the minutes, if you have any more comments or questions. Otherwise feel free to e-mail.
I will include our discussion here in the meeting minutes, if that's ok with you.
<piotr_nowara> Of course
(The following notes were entered after the meeting and are short summaries of Jeff's comments which were not adequately scribed during the meeting)
Under the topic of Best Practices for referring to statements, the reason this is an issue is because when we specify the Decision question, for example "What computer should I buy?",
this is merely a text string which is not understandable to the computer. It would be better to state this as a triple. In other words, the subject of the decision should be documented and represented as effectively as the rest of the decsion.
<piotr_nowara> I'll be going now... thank you again for your time. I'll talk to my manager and we'll contact W3C tommorow. Please email me if you have any question or comments about my paper. Bye...
So one question is how best to represent a "question" in rdf. In this case, the "computer" could be represented ...
See you, Piotr, sorry again that we missed you this meeting but hopefully we will see you next time...
scribe: as a resource, the concept of "purchase" could be represented as a resource, but then is the concept of "What" to be represented as the third component of a triple to represent the question?
The use case is that an organization or user might want to query for all the decisions involving a computer purchase, or all the decisions involving computers, or all the decisions involving purchases in the last week.
In these cases, if the subject of the decisions were well represented, then these queries could be performed effectively without relying merely on string search.
Also the subject of the decision could be a link that could be followed for drill-down information or additional context.
Eva's recommendation is a good one for looking at how this type of referencing is done for DataEntires in the open linked datasets, as opposed to doing meta-modeling at the linguistic level which is more challenging.
Regarding Best Practices for Handling Lists, Jeff was making the point that in XMLSchema, an element can be repeated many times if so specified without any outer enclosing tag; however,
it may be better practice to have the enclosing tag because of the challenges faced when nulls, empty values and empty lists when doing things like serializing and deserializing the information.
So what is the best practice in RDF? RDF has collections but I was under the impression that they are not well-used or preferred. Perhaps I am wrong about this. What is the best way to handle multiple items in RDF/OWL?
The use case is that a decision will probably have several "metrics", so if we have multiple instances of a Metric class, then how should we aggregate these or should we aggregate them at all?
Participants agreed we need more study on this.
Regarding Eva's observation about the need to separate or consider high-level decisions (which go beyond an simple set of metrics, such as what city should I live in) versus low-level decisions (which might be answered from a simple set of metrics from one dataset),
Jeff agreed that both need to be supported. Even the situation where someone simply wants to note that a decision was made, what the question was, perhaps a few of the options, and what the answer was,
simply recording that with the time and the decision-maker is important information and should be easily supported by our ontology representation. At the same time, there may be a grey scale,
as opposed to a hard dividing line, between a decision with a simple set of metrics using one dataset and another decision with a complex set of metrics using multiple datasets. In either case,
the user may choose an option which is not dictated by the specified metrics or where there are considerations beyond any easily defined set of metrics, so these decisions also must be
capable of being adequately represented in our model.
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: Don Eva Jeff ScribeNick dmcgarry dmcgarry_ eblomqvi esela inserted piotr_nowara 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_8_Agenda Got date from IRC log name: 08 Jul 2010 Guessing minutes URL: http://www.w3.org/2010/07/08-decision-xg-minutes.html People with action items: WARNING: Input appears to use implicit continuation lines. You may need the "-implicitContinuations" option.[End of scribe.perl diagnostic output]