- Approve minutes from previous three meetings.
Note, the minutes are linked from http://www.w3.org/2005/Incubator/urw3/#Minutes
- One item was corrected (Thomas was not the scribe on
September 19th).
- Ken discussed the issue of attendance at our meetings.
In the summer it was relatively low, but as expected (due to vacations).
- Chairs (Ken Laskey, Kathy Laskey) have declared nine
attendees to be a quorum.
- Minutes from previous three meetings approved as
amended.
- Review status of action items from previous meeting
Action
1: Ken to check attendance and send out email.
Ken sent email to group to encourage participation. Action
closed.
Action
2:
[previous mtg action 1] Editing team to go over IRC log and
capture essential topics to include in our report.
Paulo did a first crack at the final report, going through wiki and irc
logs. Action closed.
Action
3:
[previous mtg actions 2,4,5] Coordination with HCLSIG (1) Vipul to
coordinate with Eric Neumann who has SWCG action, (2) Mike Pool to
assist.
This action was not done. Ken to send Eric N's contact info
to Mike (done during meeting).
ACTION: Action 3 from
previous meeting is still open; Mike to
coordinate with Eric N. and Vipul.
Action
4:
Anne will look into POWDER and report back to us.
Note: POWDER Working Group (http://www.w3.org/2007/powder/
) has recently published Working
Drafts.
Anne Cregan sent Kathy L. a summary of POWDER to include in minutes.
Here is the summary:
In brief, the idea of
POWDER is to provide a mechanism, using machine readable meta-data,
where a description of a group of web resources can be associated with
the web resources and retrieved, without retrieving the web resources
themselves.
This allows the author to say things like :
- All resources on the travel.example.com
domain
are suitable for display on mobile devices
- All resources on broadcast.example.com/clips are video
clips that are suitable for all ages.
Other parties can check
that the author actually said this about these resources.
POWDER is based on RDF and a vocabulary working draft was released on
25 September 2007.
Kathy suggested we should continue to monitor the work on POWDER.
Thanks Anne for summary. Action closed.
Action
5:
Peter to send email to group for author to annotate potential
disagreements in uncertainty ontology with who entered them.
Peter sent email but there was no response. Action closed.
ACTION: Kathy will move
this section to discussion page and note that
author is unknown
Action
6:
Kathy to send email to group asking what open issues remain in
uncertainty ontology (in preparation to annotating use
cases)
Action 6 will be superceded by agendum 4 today. Action closed.
Action
7:
Ken and Kathy to meet with editing team, discuss Discovery, report back
to group
Report is agendum 3. Action closed.
- Report from editing committee.
- Paulo gave an update on the actions of the editing team.
- Editing committee has done a first draft of the report.
- Kathy suggested that the report should be posted where
everyone can see it.
- ACTION: Ken
to investigate how we can post our draft
report where group can see it
- Kathy opened discussion of linking use cases to
uncertainty ontology.
- Ken suggested that the use cases should be at a
somewhat abstract
level (don't get into solutions).
- Kathy suggested that this discussion should be deferred
until all
can see the draft report.
- Paulo gave his explanation of the issues of Value
Assessment
(Uncertainty nature: Epistemic; UncertaintyType: Randomness).
"Given a set of properties of interest to the user, there is
uncertainty in trading off the user's target property values
against those of items in the catalogue. Example: User is a 42
years old male, graphical designer, who wants a notebook with
properties hasDisplaySize="15", hasWeight="lessThan5Pounds",
hasBluetooth="true", hasBatteryAutonogy="Between2and3Hours", and
hasColor="Black". Catalog doesn't have an item that fulfills all
five, but has one notebook that meets 4 and two that meet 3 of the
user requirements. Instead of presenting a "no item found", the
system needs to return a prioritized list. Not necessarily the item
with 4 positive requirements will be the best option for the
customer.
- Paulo (continuation): Limitations of deterministic
representation:
A typical catalog system based on deterministic representation and
reasoning might use rules to sort out items that did not completely
meet a request. However, most real life catalog systems will have
to deal with a high number of properties and customer profiles,
making the use of rules a sub-optimal approach (i.e. cases not
predicted by the rules will default to a configuration that might
not be ideal). Possible Approach(es): A probabilistic ontology can
be used to estimate the value of a given property given a user
profile, based on statistical and anecdotal data. In this example,
the result would be a prioritized list built upon the historical
preferences of customers with that specific profile.
- Nico stated that this poses a bootstrap problem.
- Paulo has agreed - we need to discuss that.
- ACTION: Paulo
to insert his more detailed specifics in discussion
section of discovery use case.
- Nico state that they are using distances for ranking
the hits returned from matchmaking.
- Kathy reiterated that distance is a good technology
for matchmaking.
- Ken said that use cases help us to state problem.
- Kathy said that in agendum 5 we will talk about how to
start capturing solution methodologies and how to attach to use cases,
but use cases are intended to be problem statements not solutions.
- Completion of uncertainty ontology
- Kathy said that we need to identify person or persons to
take ownership of maintaining uncertainty ontology and getting into
final form for report.
- Mitch volunteered to fulfill this role.
- Peter stated that we need feedback on the uncertainty
ontology. We try to annotate use cases, and we may find we need to
revise the ontology.
- Anyone with suggestions of how to modify uncertainty
ontology should contact Mitch.
- Ken said he would hope that work on Discovery will
provide example
for others to follow, e.g., example of how to annotate using
uncertainty ontology.
- Mitch noted that the Discovery use case shows numerous
types of uncertainty, in particular both randomness and vagueness. Thus
this use case suggests that there is a real need to combine
different kinds of uncertainty in query answering.
- Ken said that there is a need to combine probablistic and
fuzzy.
- Peter said: First step to solve this problem, combining
probabilistic and fuzzy, would be an example of user query answering.
- Thomas said that an example of a fuzzy-probabilistic
combination can be found here (joint work with Umberto Straccia):
"Description Logic Programs Under Probabilistic Uncertainty and Fuzzy
Vagueness" at ECSQARU-2007. Thomas will put a link to the paper on his
web page. Currently it is in LNCS 4724: http://www.springerlink.com/content/a127481u872607t4/?p=11cb61db3b51443ba481d5498854d1e0&pi=0
- ACTION: Peter
and Mitch - a short discussion on how probabilistic and fuzzy could be
combined.
- Ken suggested that annotating Discovery use case with
uncertainty ontology will also test completeness and consistency.
- Approach to identifying target methodologies
- Ken re-stated that everyone has been wanting to discuss
solutions and he has kept saying that we need to capture the problem.
With some use case work done, now is time to start discussing
solutions.
- Kathy thought that now we are at the point of thinking of
solutions. Need to discuss how to go about doing that.
- Ken reminded that http://www.w3.org/2005/Incubator/urw3/wiki/reference_list reference
list begins collection of possible solution methologies.
Identifying target methdologies will be next step before we
begin looking at information for selected methodologies. What is
process XG should use to proceed.
- Kathy stated that there is expertise in this group;
people to say they could handle a given problem. Should look at what
people are actually doing.
- Trevor asked whether we are looking for an approach or
available tools?
- Kathy said that if no tools are available yet, then we
shouldn't be standardizing yet.
- Ken agreed that methodology needs to be mature enough
that we can identify information to exchange.
- Kathy has summarized the approach to identifying target
methodologies:
- First go through use cases
- Tie use case to ontology. Paulo added: And the
uncertainty ontology. Then we get to the solutions.
- Consider which methodologies would be appropriate to
the use case and how those methodologies relate to the ontology (i.e.,
uncertainty type, uncertainty model).
- Paulo invited everyone to write something in section 4,
once the initial report is complete.
4. Most
Commonly Used Approaches to Uncertainty for the WWW.
4.1 Bayesian Models
4.2 Fuzzy Logic
4.3 and on - expected
contributions from
the group.
- Kathy thought that those (Bayesian and fuzzy) are not the
only two approaches. Also not sure "Bayesian Models" is the right term.
- Paulo explained that this is just a initial take.
- Peter referred to his paper about experiments combining
fuzzy and Bayesian.
- Mitch said he is not sure what is meant by "methodology".
- Kathy said there is technology (e.g., fuzzy logic,
Bayesian networks) and process (e.g., how I go about capturing the
uncertainty, annotating resources with metadata, etc.)
- Mitch said process is more concrete than technology or
methodology.
- Kathy: we are still not completely clear on this issue,
but as Mitch says, "Let's go through the use cases and see what
happens."
- ACTION: Editing
team to appropriately connect Discovery use case to uncertainty
ontology and then start identifying "methodologies" for addressing the
identified problems.
- Face-to-face at SUM.
- Kathy, Ken, Mike live in DC area. Peter and Thomas will
attending SUM.
- Kathy: We will have a mini f2f.
- Nico: none from UniBA, sorry.
- Ken: We will schedule an agenda item for next meeting
about whether we should arrange a f2f for this group.
- Nico: Claudia is coming to Busan.
- Kathy: here is a list of Busan attendees on the wiki:
http://www.w3.org/2005/Incubator/urw3/wiki/People_from_URW3_who_are_going_to_URSW_2007_in_Busan.
There currently are six people listed: Peter, Pavel, Thomas,
Trevor, Fernando, Claudia. Anyone else who is going to Busan should
enter his or her name there.
- Other business
- Ken will be attending W3C Tech Plenary, Advisory Committee, Advisory Board, in Cambridge, MA on November 6-9.
- Next telecon is October 17.
- Meeting was adjourned.