Mark: various site changes in collaboratory completed by Megan
Ken: I am changing logical data
model to be city data model and physical to specification
... we will have separate pages in collaboratory for different
properties of a class
Mark: correct
Mark: I can start with my list and ammend with feedback
[projects Concerns.pdf]
https://github.com/w3c/tocc/blob/master/management/InformationViewpointConcerns.md
Mark: we don't address analytics directly
'What dynamic metadata exists regarding the employed analytic methods used to produce the aggregated/fused data?'
Mark: may be a side effect
... I have not included any of the data rights
... to me this is outside of scope of what we are doing
Ken: doesn't everyone have to deal with it in order to share information?
Mark: absolutely but not as part of definition for city data model
Ken: then I cannot share any data...
Megan: models and standards information can still be shared
Ken: without agreement on data rights, no exchanges can take place
Clemens: we are not building any system but model/ontology
Mark: agree but this is not our focus
Ted: I see similar concerns in
several other groups who also have deemed it out of scope, huge
issue for data the world over.
...there are also numerous efforts, a few I'm partially familiar
with but haven't found any that convince me solve the problem and
will see wide adoption
...ones I like tend to capture metadata that accompanies the data
representing who that is authorative gave permission to use what
data with whom for what purposes
Megan: some of these are considerations that should be included with the data
Mark: one approach of dealing with data ownership, usage rights is to introduce a whole other ontology on it
Ken: if we actually want to connect more than two systems, it is much more difficult to leave it up to individual deployments
Mark: I agree with you, one is
embed in ontology or a wrapper as you allude to wrt
correspondence rules
... that is outside of the city data model ontology
Ken: I am willing to accept that,
however we need some way to indicate how that can be recorded
somewhere
... are we expecting different definitions of how that would be
done or standardized and leveraged by this project
Mark: we have quite a bit of work
around city data classes without getting into data rights,
usage etc
... it is important but outside our scope
Ted: University of Passau along with
some researchers from an institute in Lyon are working on Layered
Policy Language and choosing to represent it using XACML (Oasis
spec) which is a perhaps the most widely used policy language
...they have presented to W3C Auto folks a few times in
the past and want to focus on
automotive use
cases currently
...I'm interested in hearing of other efforts
Ken: I can accept that is not the
focus of this group but would like us to agree to a single or
small number of solutions we find acceptable
... a placeholder is worthwhile
... that or we defer to every country or individual
deployment
... we need to acknowledge this as an issue
Mark: agree, where should we put that... in the collaboratory?
Ken: yes
Mark: it could be an intramodel
on data usage although hesitate to use term metadata
... category 2 is external to data model, system enforces
restrictions
Ted: not sure it is _the_ solution or one even possible given complexities noted with myriad of laws internationally
Mark: does anyone want to speak towards Data Quality
Ted: yes, at least sampling methodology but also precision and fault tolerance
Ken: knowing what the data is, an average or precise measurement etc
Clemens: on speed example (multiple in vehicle gps, speedometer, drive train etc) isn't that the definition?
Ken: sampling method etc can likely be done within the data model
Mark: ISO 21972 includes
measurement ontology, allows for representation on how things
are measured and may provide us with most of what we
want/need
... WG11 SmartCities see that as part of our foundation layer
along with time, location, units of measure
... do we need such foundation layers
Clemens: there are lots of community specifications for these sorts of things. if we try to choose one as basis we may get lost in long debates instead of focusing on higher level
Mark: for instance if we want to use a unit of measurement in a property we can, they tend to be equivalent
Ken: may hinder interchanges
Clemens: how far do we want to go?
Ken: we need precision for inambiguous transformations or we cannot interop
Clemens: @@
Ken: we need more than a logical data model, your higher level focus
Clemens: we need to agree on our
scope and we have some differences in expectations
... I want to bind concepts to given standards to understand
how they relate to each other but wasn't expecting us to solve
dynamic interop
... corresponding to concrete mapping
Ted: we could again defer and leave that to those to build individual bridges for that automatic interop
Ken: again, ok with decision but want to be clear on what our goals are
Mark: both are important but for first phase we want to clarify shareable classes and properties before we dig into particulars and mapping/interoperability
JTC1 WG11
Ted: I was hopefully we can identify core ontologies, ideally public at eg schema.org for core concepts
Mark: 21972 leverages OWL Time, PROVO and QUDT (NASA or independent, some thought W3C)
Megan: affiliated with NASA but used in W3C specs
Mark: NIST has guide for it
Ted: when we don't have clear winners and find ourselves arguing in circles, we can revert to identifying who uses which and let people do their own mapping/translations
Mark: any other obvious normative choices?
Clemens: SOSA and SSN
Mark: back to the higher level question are we trying to address data quality?
Ken: if we keep focus on higher level, no
Mark: similar with integrity
concerns, all important points but out of our initial
scope
... data storage, persistence, destruction - all very useful
but not for initial phase
... data conformance and evolution... there are many
interpretations of 'metadata' are we referring to the ontology
itself or meta to it
Ken: I was pulling from other documents but instead use static (from model itself) and dynamic (eg how sampled)
Mark: next meeting in 2 weeks, 10 Eastern and continuing on this scoping