Minutes for 17 October 2007 URW3 Meeting

Attendees:

Agenda:

  1. Approve minutes from 03 October 2007 meeting.  Note, the minutes are linked from http://www.w3.org/2005/Incubator/urw3/#Minutes.
  2. Review status of action items from previous meeting.
  3. Report from mini-F2F at SUM conference in Maryland and recommendations stemming from the meeting for use cases (Kathy).
  4. Presentation of HCLS use cases (led by Vipul).
  5. F2F at Busan and plans for full F2F.
  6. Other business

Scribe: Paulo Costa

Discussion summary:

  1. Minutes from previous meeting were approved with no changes.

  2. Review status of action items from previous meeting

    Action 1: Mike to coordinate with Eric and Vipul
    Status: Done! Mike, Eric, and Vipul have exchanged emails. Vipul will go through a use case today and will invite URW3 members to annotate HCLS wiki.

    Action 2: Kathy to move the possible disagreements section to discussion page and note that author is unknown.
    Status: Done! Fixed wiki page to move possible disagreements section to discussion and add attribution.

    Action 3: Ken to investigate how we can post our draft report where group can see it.
    Status: Ken sent out email pointing to link in wiki discussing attachment feature and questioned the editing team on whether this is sufficient for what we want to do. Kathy suggested that the draft report should be on the member only web site. The editing team is going to submit periodically a draft report to Ken, who will take continuing action so he will publish the draft in the members only part of the urw3 site. [ACTION 1:] Ken to continue action 3 to come up with member-only way to view evolving draft. **UPDATE on 2007.10.26**: Group to define how the draft final report will be edited [recorded in http://www.w3.org/2007/10/17-urw3-minutes.html#action02].

    Action 4: Paulo to insert his more detailed specifics in discussion section of discovery use case.
    Status: Did it partially. [ACTION 2:] Paulo to take continuing action on Action 4 [recorded in http://www.w3.org/2007/10/17-urw3-minutes.html#action03].

    Action 5: Peter and Mitch - a short discussion on how probabilistic and fuzzy could be combined.
    Status: They will continue the discussion and [ACTION 3:] and provide email and/or lead to later discussion on how probabilistic and fuzzy could be combined [recorded in http://www.w3.org/2007/10/17-urw3-minutes.html#action04].

    Action 6: Editing team to appropriately connect Discovery use case to uncertainty ontology and then start identifying "methodologies" for addressing.
    Status: Editing team to take continuing action on Action 6, [ACTION 4:] appropriately connecting Discovery use case to uncertainty ontology[recorded in http://www.w3.org/2007/10/17-urw3-minutes.html#action01], and then [ACTION 5:] start identifying "methodologies" for addressing the issue of annotating the use cases. [recorded in http://www.w3.org/2007/10/17-urw3-minutes.html#action05].

  3. Report from mini-F2F at SUM conference in Maryland and recommendations stemming from the meeting for use cases (Kathy).

    1. Major conclusion: we should be looking at UIF (Uncertainty Interchange Format), and should not be designing by example a new representation language (whether Fuzzy, Probabilistic, or any other approach) for the Web.
    2. Kathy mentioned that we are an XG, thus our report will say what a working group will standardize. Giorgios asked whether we should ignore previous formats and W3C standards for uncertainty. Kathy, Peter, and Anne agreed that we should define a format for exchanging the kind of info needed for uncertainty reasoning, rather than a particular uncertainty language, but the result should be consistent with previous W3C standards.
    3. Kathy and Peter questioned whether there are any previous formats for uncertainty, and Mike added the issue on whether the existing formats (if there are any) are sufficient for representing uncertainty.
    4. Anne mentioned that interoperability with OWL or at least RDF should be in consideration. Mike pointed out that this is a distinct issue, since consistency with OWL, RDF, and others should not be a design constraint.
    5. Kathy asked whether extending OWL wouldn't be an option, and Ken replied that it might be a conclusion of the XG, but in general we will be discussing what needs to be in an interchange format.
    6. It is up to the XG to decide whether interoperability with OWL or RDF should be a consideration, which Kathy pointed out it is different from a design constraint. Vipul says we have a design constraint and Giorgos added that one can always see OWL as a knowledge interchange format of the SHOIN DL (Giorgos later corrected, explaining he meant to refer to OWL DL, not SHOIN DL, and added that OWL DL is SHOIN(D+) minus annotation properties - http://www.cs.vu.nl/~frankh/abstracts/JWS03.html).
    7. Thomas says there are some papers from Ian Horrocks on this relationship. Kathy suggests: Ian Horrocks, Peter F. Patel-Schneider: Reducing OWL entailment to description logic satisfiability. J. Web Sem. 1(4): 345-357 (2004).
    8. Kathy mentioned that RDF/OWL is more than just a data format and we should not contradict the underlying data/object model.
    9. Peter thinks this is a task for an WG, not for an XG. Ken replied that if the group realizes that we need to move to an WG then this would be a conclusion of the XG.
    10. Trevor concluded that what we need is an agreement on what needs to be exchanged and what it means. The group has agreed on: 1) We will charge the WG that we evolve to, not do it ourselves (= developing the UIF is not within the scope of the XG). 2) XG's goal is to identify information that needs to be exchanged for Web applications to appropriately handle uncertainty. 3) To identify this, we will use our use cases, and annotate them with what information needs to be exchanged for them to work. 4) We are not directly concerned with the internals of how a given web application does its uncertainty processing, if one has to be developed, then we should move to an WG. We are concerned rather with what needs to be exchanged between applications
    11. Anne summarized further, stating that the XG will identify the data that would need to be exchanged by an UIF, which would then be developed by the WG based on the annotated use cases and conclusions of the XG.
    12. Ken pointed out that the XG will indicate focus and rationale for future work, but the XG itself may not lead into a WG or the XG conclusions may be taken up by another WG, such as RIF. Instead, our job is to put together compelling conclusions that others would agree should be pursued.
    13. Referring to the annotations, we need to organize the information in accordance with the uncertainty ontology. Mike asked whether we want to use the ontology as is, extend it, or not use it, while reinforcing the need to standardize the data we are producing (Mike understands that a glossary or something similar would do this standardization job). Anne added that the uncertainty ontology is a good starting point, and a bottom up approach would be the way to go.

  4. Presentation of HCLS use cases (led by Vipul).

    1. Vipul discussed HCLS use cases linked from wiki: http://esw.w3.org/topic/HCLS/UncertaintyUseCases
    2. His presentation included issues of hypothesis uncertainty, interpretation/classification uncertainty, prediction oriented uncertainty, and others. He noted the distinction between prediction and hypothesis, where the first is more oriented towards the future while the latter is usually based on present time. In other words, prediction looks towards hypothesis in the future.
    3. Beliefs are linked to prediction.
    4. Kathy brought up the distinction between class and instance. As an example, 70% of patients with BCRA1 get breast cancer, which is a class uncertainty. Mary Smith having 80% chances of having breast cancer is uncertainty about an instance of a class.
    5. Vipul thinks there should be separate category of reputation based uncertainty.
    6. Peter says that part of UIF is the way of data acquisition, data quality, and method of data cleaning. Kathy added that UIF should have reputation and trust of the person asserting the belief, and Peter mentioned that HCLS exchange can point to information necessary in UIF (e.g. what is the uncertainty information relevant in HCLS exchange formats?).
    7. Health care is a good domain for applying uncertainty.
    8. [ACTION 6:] Vipul to expand one of the HCLS use cases with types of uncertainty exhibited (using uncertainty ontology as guide) [recorded in http://www.w3.org/2007/10/17-urw3-minutes.html#action06]
    9. Peter offered to work on part of the extraction use case. [ACTION 7:] Peter to work on the extraction use case.

  5. F2F at Busan and plans for full F2F.

    1. Ken noted that ACTION 6 summarizes what we are after from agenda item 3.
    2. People that are going to Busan (so far: Peter, Thomas, Trevor, Claudia, and Fernando) are encouraged to organize a meeting and submit a report. Thomas volunteered.
    3. Focal point to Busan meeting: Thomas Lukasiewicz.
    4. Paulo and Thomas to post a comment on the URSW program regarding the URW3 meeting. It might include plans, tasks, time, agenda, etc.
    5. Full F2F is not required for an XG. The major issue preventing a URW3 full F2F would be the feasibility of bringing everybody to a unique place.
    6. Ken added that a F2F in Busan would be great, but we still need to consider a full F2F.

  6. Other business.

    1. None discussed.

Action items

  1. ACTION (1): Ken to continue action 3 to come up with member-only way to view evolving draft. **UPDATE on 2007.10.26**: Group to define how the draft final report will be edited.
  2. ACTION (2): Paulo to take continuing action on Action 4.
  3. ACTION (3): Peter and Mitch to provide email and/or lead to later discussion on how probabilistic and fuzzy could be combined.
  4. ACTION (4): Paulo, Trevor and Mitch to work on appropriately connecting Discovery use case to uncertainty ontology.
  5. ACTION (5): Paulo, Trevor and Mitch to start identifying "methodologies" for addressing the issue of annotating the use cases.
  6. ACTION (6): Vipul to expand one of the HCLS use cases with types of uncertainty exhibited (using uncertainty ontology as guide)
  7. ACTION (7): Peter to work on the extraction use case.

Last update on 27 Oct 2007.