<brinkwoman> Welcome back Jeremy!!
Jeremy: I am able to again turn my attention to this group and appreciate Linda keeping the continuity for this group
[agenda review]
Ed: privacy and ethics has been a
long discussion, renewed in light of trying to track and trace
for COVID
... 'exposure notification' being the term used instead of
tracking
<eparsons> https://www.edparsons.com/2020/06/the-ethics-of-geospatial-the-four-es/
Ed: big question if we should be
doing this and prompted me to write a blog article
... context to our work beyond the technology... not just the
data but services
... boiled down to four topics, struggled to get them all to
start with E
... Efficacy
... how can you know if you have had the virus or not
... Equability
... not everyone can afford latest smartphones...
... Execution
... how do you create a service minimizing the amount of data
being collected or exposed
... given individuals' location in time and space remains
sensitive
... what can be done to reduce privacy risks
... what do you do with the data after you are no longer using
it? what is the eradication process?
... this is certainly sensible during outbreak, how to ensure
it doesn't become the new normal?
... how to unbreak the glass/put the genie back in the
bottle?
... interested in getting feedback from this audience. there
are other more in-depth works on this topic
... and other geo organizations seeking to address the ethical
questions as well
Linda: is there something about this topic that could feed into our Best Practices document or should it be a separate work?
<MichaelGordon> One such initiative would be https://benchmarkinitiative.com/ from OS and Omiydar network
Rob: would you be able to
identify a best practice in this space at this point?
... certainly can identify aspects of concern
Joseph: I am in agreement with
Rob, there is alot to this
... related to non-authotitative data for decision making
discussion
Peter: ethical use of geospatial
technology is profoundly important
... we have suggestions for presentations on ethical use of
geospatial data for the upcoming maps workshop
Ed: thanks for the feedback. I
agree it is too early for specific practices
... tracking people in realtime risk needs to be made
clear
... differential privacy
Jeremy: I see three possible ways to take this forward
Ted: we are currently struggling with this for Auto group's graph project, making real data available to demo standards and struggling with different ways to expose including stripping PII
RobS: wanted to comment on
execution side. for webvmt we have an image potentially of an
individual with time and location
... as best practices is later on the agenda, should we defer
on what we can include there?
Jeremy: I think there are
potentially three ways forward. a new best practice, note on
ethics for spatial data, lastly simply 'here be dragons'
warning
... heard it is too early for new best practices. option two
for starting note and gap analysis for a best practices
document
RobS: my thoughts are we can do
both. creating the 'here be dragons' makes clear we are working
on this
... it is easy to say it is bad and we need good ethics but
question is how
... I see design, guidance and reassurance
... all are quite important. besides designing systems
correctly, need to guide people on decisions they are making
and communicate assurance on risk/reward for sharing
information
Jeremy: short term: here be
dragons gap and in parallel try to develop that body of
knowledge including anonymization Ted mentioned
... clearly plenty of work taking place in different places. it
would be useful to try to bring that together in one
place
... unsure still what type of document to produce initially
[it can start rough and get polished over time]
RobS: it is possible to share
data/images without location for instance
... or conversely location without identification
Jeremy: we are not starting from a blank page. what we can do is pull together different efforts
Ed: spatial data best practices started as an emerging field of questions
<jtandy_> Ack brinkwoman|2
<jtandy_> Ack brinkwoman|
Ed: we had a vague idea of what we wanted to cover and seems we are at the same point here and we can start a best practices like document that covers what we can recommend
Linda: Note is easy to publish and get wide input and review. there is an OGC group we can line up
Scott: it is a rather large international issue including AI usage, I like the idea of aligning alot
Joe: ethics of AI is a pretty big field in and of itself
Linda: I was thinking the same thing. maybe we want to address first ethics of sharing of location before going into AI uses
Jeremy: clearly hear interest in
this
... Linda, you, Rob, Ed and I can create a scoping statement
and then involve Scott and Ted on drawing in from OGC and W3C
groups
<scribe> ACTION: Jeremy do some work on that :)
Clemens: I reviewed the document
per last call and made a number of comments
... the structure and contents hold up and still valid.
examples can be updated
... we can draw from implementation reports
... OGC API chapter
... noticed a few other things, reading through it again when
it gets to relationships it has a heavy rdf bias
... whereas there is more around in OGC around web linking than
rdf
... over the last five years we are seeing more tiled concepts,
making it easier to rapidly access content
... 3d containers mentioned for instance and maybe should be
elaborated on
Jeremy reviews Clemen's observations
Scott: [on tiles] there is a conceptual effort and API work taking place
Jeremy: too early for best practices type recommendations?
Clemens: several are established
standards and other tile work already being established
... it is mature enough
Scott: I agree
Jeremy: Bill, can you give a general statement and clarifications to Clemen's comments?
Bill: generally agree. worth
highlighting a version 2 of DCAT
... that data exchange working group is looking at future
extensions so worth synching up with them
... would be useful to tie in non-spatial data with spatial
considerations
... there is some mention of this but can be expanded upon
<SimonCox> Did Bill say that DCAT2 'missed a trick'?
Bill: if we think of things that can enhance DCAT2 to make it more relevant to spatial data, there is an opportunity to provide that input
<SimonCox> DCAT is proposed to be an 'evergreen' standard. However, there is still very limited implementation ...
Jeremy: giving more of a
schema.org flavor, ability to link spatial/non-spatial data and
aligned with Clemen's comments. thank you Bill
... is JSON-LD outside our scope?
Clemens: JSON-LD 1.1 makes
GEO-JSON easier with LD
... other related topic is how to mix web linking with rdf?
<jtandy_> (How many folks need to depart in 6-minutes?)
<eparsons> me
<eparsons> another call sorry
<scottsimmons> scottsimmons does
<billroberts> I have to leave on time too, (though happy to postpone the item on stats)
<billroberts> +1 to FAIR
<jtandy_> * thanks @bill
[Rob shares Available Representations of Linked Data]
RobA: most people do poor job at
using JSON-LD, usually due to underlying tools
... thousands of definitions across unrelated domains
<SimonCox> That's because JSON people don't think of it as RDF.
RobA: when we start publishing data models we tend to do so with different formats
<SimonCox> ... they just work with the JSON model often through a JSON-schema, which isn't the same at all
RobA: at OGC we are moving
towards making the data models themselves as linked data
... what is the gap between OGC API and best practice? it comes
down to how JSON-LD is implemented
... here is a data model done with standard model, people tend
to do copy and paste... without clarifying what is different
with a given model
... there may be extra restrictions, etc
<jtandy_> * will try to wrap things up v quickly with some next steps once @RobA is done
<SimonCox> instead of JSON-Schema they would be better using SHACL, but that is pretty much unknown in the JSON world
<eparsons> Sorry guys need to drop off - bye !
<jtandy_> * thanks Clemens ... will continue discussions on the mailing list
<billroberts> sorry also got to go. Bye all and thanks for all the good stuff today
<PeterR> Nice to see everyone today - bye for now
<SimonCox> Also SHACL has now appeared as validation tool
<SimonCox> For stats-on-the-web need to look at work underway in DDI !
<roba> generating SHACL by formalising implied "shapes" is a key function intended here ;-)
<SimonCox> +1 to Clara
<josephabhayaratna> +1
<RobSmith> +1
<roba> +1
<MichaelGordon> For context, Clara is our Data Standards Manager at Ordnance Survey, working with Peter Parslow
<MichaelGordon> Sorry folks, got to dash to another meeting - great to see everyone again!
<jtandy_> * thanks MichaelGordon ... bye
<SimonCox> make sure to save the minutes