Scribe: Eric Monk
Action #4 - Vipul and Mitch and anyone else who cares to contribute to uncertainty ontology
Group
discussion on Action #4. Discussion of whether 'Sentence' is the
correct concept or maybe it should be changed to Propositions? Do
distinctions between Sentence and Propositions need to be made at this
point? Kathy mentions discussion on ontolog forum and IKL effort.
Discussion of Aleatory vs Epistemic. Discussion of uncertainty types
and other definitions such as proposition. Mike says Peter Suber's
glossary of FOL is a good reference for definitions. Link to Peter
Suber's definition of proposition.
Was decided group needs more discussion on this topic. See Action 2.
Action #5 - Vipul to work w/ Ken/Kathy on coordinating with HCLSIG
Discussion on Action #5. HCLS has been having interesting discussions (in biomedical research and clinical practice)
surrounding evidence and interpretation. Should URW3 be better coordinating with HLCS? Vipul provided link
to work going on in HCLS regarding uncertainty http://esw.w3.org/topic/HCLS/UncertaintyUseCases.
Ken does a walk through of the Discovery use case. Discussion regarding whether Vagueness and/or Ambiguity apply. Vagueness and ambiguity are disjoint but could still both apply in given solutions. Possibilities could include a simple sentence having one or the other, but a compound sentence having both. Kathy provides other situations where they may both apply. Kathy mentions some are working on probabilities on fuzzy membership functions and this might be an area to explore for this use case. Someone asks Ken whether machine processible contents of user vocabulary or catalog vocabulary has to be an ontology. Answer is no, it can be just XML tags. Peter points out solution to use case indicates an ontology matching / mediation problem. Mediation is the overarching term - and an ontology matching solution could be used when ontologies are involved. Francis raises question regarding "Does mediation include data translation resulting from the matching"? Discussion leads into short discussions of other use cases - but discussion is limited due to time. One point regarding User preference modeling, is that it differs from discovery use case slightly in that it is connected to combination functions of bayesian networks.
Group proceeds to discuss how to continue discussion on
these use cases and how it relates to final report. What is structure
of final report? Needed to figure out which discussion points /
information regarding use case discussion should be included. Mitch
comments that discussion on Reasoning should be part of final report.
More discussion regarding reasoning and next steps for this use case.
This leads to Action 3 and Action 4.
