<McCool> https://github.com/w3c/wot/issues/939
Linda welcomes attendees, introduces topic
MichaelM: geolocation is
certainly metadata we may want to keep track of as part of
WoT
... we are working on a discovery service and in the process of
rechartering
... use case is you might want to find sensors in a given area.
we want to try to align. we use geolocation API within
browser
... it is not consistent with other geolocation done in the
semantic web
... what I hope to do in this meeting is figure out how we are
going to collaborate, establish goals and provide overview of
our issues
Linda: sounds good
MichaelM: we use github and have
been labeling issues related to geospatial
... discovery needs geospatial queries while retaining privacy.
we need to decide how to encode metadata
... we want a defined vocabulary and consistent way to access
location data and know a device's location
... we have a few different use cases documents and looking for
where geolocation comes up. in smartcities, shipping...
... we are trying to track and determine requirements. we have
been doing some experimental prototypes and how things might
look like
... we want your input on that
... we are considering a best practices for spatial data in
WoT
... in some cases the devices will not know their location and
provide that data but we would want to be able to add as an
annotation
... we will want to align with things like SSN accuracy
... we ideally would select instead of define a vocabulary. we
lack geolocation experts and trying to recruit such
expertise
... we are aiming for May for our final drafts before CR during
the current charter
... we are expecting to perhaps issue an updated CR with
privacy considerations
... one of the constraints is one shouldn't broadcast location
metadata
... connecting to a directory service and authenticating prior
to getting location data is one possibility
... we are looking at proposed extensions to DNS to include
geospatial queries or we could possibly handle in our directory
service
... we will need a standardized way to encode the data
... there has been some discussion in IETF groups and
encouraging them not to provide location data directly in
DNS
... instead provide a web service
<brinkwoman> https://w3c.github.io/wot-usecases/
Christine: IoT is mostly about
connected sensors as I understand it
... I want to know what are the use cases for needing location
data
... in particular I am interested in AR discovery, in your
document it doesn't seem related to a virtual guide
... I want to be able to expose where there are AR experiences
near a user. currently a challenge to query them
MichaelM: our discovery mechanism allows queries across various semantic information using SPARQL, XPATH...
Kaz: I have been working with Rob Smith on this use case. I want to include 'virtual' location
Christine: I know Rob and will follow up with him directly outside this meeting
RobS: happy to have a chat and let's take this offline
MichaelM: we are in discussion
with the Linked Building Data Community Group
... we have various issues around tracking deliveries, traffic
management
... this document needs some reorganization, it has had a
number of individual contributors
RobA: some of the thinking in
SOSA/SSN in describing observations does not include who is
providing metadata
... that is more imporant in IoT, who is reporting
... a temperature sensor location makes a big difference but
the sensor itself might not know that
... do you have a reference architecture?
MichaelM: SSN is on our minds. Thing Descriptions include a link description
<brinkwoman> https://www.w3.org/TR/wot-architecture/
MichaelM: there are indeed many ways you might get access to a thing. discovery is meant to help define how you get that metadata
<brinkwoman> https://w3c.github.io/wot-discovery/
MichaelM: it is a two phase
process, meant to not leak any metadata but provide a URL
... URL as such needs to be opaque
... you might have P2P self describing device or arrive at it
through a directory service
... you can't download the entire directory but query it. we
have three query mechanisms, SPARQL the most powerful
RobA: how does someone know where the directory is?
MichaelM: we want to be able to
include location as part of directory query
... what is the introduction process to find the
directories?
... we do not yet have a way to find directories based on
location which is why the DNS approach is being discussed
... as mentioned, metadata will be withheld until
authenticated
... there may still be tracking concerns since you will be
provided a fixed URL for a given location
... we are discussing hierarchical directories of directories
on the web as a way to scale across a large area
... it is more a campus than country scale thing, for latter we
feel DNS solution needed
RobA: I can see leveraging DNS as a starting point
MichaelM: we are designing an
architecture to support future discovery mechanisms
... we have periodic plugfests where we try out various
things
... we have played around with how to include geolocation in a
thing description
<brinkwoman> https://github.com/w3c/wot-testing for plugfest examples
MichaelM: there is a simple geolocation ontology defined a long time ago on schema.org that defines terms we can use
<kaz> specifically Thing Description resources for the Plugfest during TPAC 2020
MichaelM: this is only good for
static metadata, not properly handling moving devices
... we do not want to constantly update a Thing Description
RobA: that model of long/lat is
flawed, not handling things like continental drift or delivery
drones will be smacking into sides of buildings
... you will need a more extensive model
MichaelM: correct. we have looked
at other variants, using a different structure including
accuracy information
... when I look at web api for html, it includes heading and
speed but lacks unit and accuracy information not even
size/shape of the thing
... for dynamic location, if you have a geolocator service
coming from GPS on device we are thinking of providing link to
that property
... this handles dynamic, static and even if another device is
providing location information
<McCool> https://github.com/w3c/wot-testing/tree/master/events/2020.09.Online/TDs/Intel
Matthew Purrs: DGGS gives you an implicit level of precision
scribe: better than long/lat schemas. we are exploring if this is a good fit
MichaelM: if there is something
we could leverage that would be great
... we have different requirements for static, dynamic and
discovery
... I want to discuss process on how we can work together in
our remaining time more than specific solutions
<ghobona> Discrete Global Grid Systems (DGGS)
MichaelM: personally I want to
see a joint task force and possibly work on a best practices
document to address these concerns
... separately I am concerned about how ontologies and web apis
often do not align
Linda: we should take advantage of the participants on the call and see if anyone wants to speak out, expressing their interest in chat or vocalize
<roba> I am working on projects looking to leverage SSN/SOSA with FIWARE/NGSI-LD - and hence on approaches to generate modular JSON-LD contexts to support this - so I would be interested in some aspects of this.
Steve: I am interested and have seen some odd use cases which can be helpful
<McCool> https://github.com/w3c/wot/issues/939
MichaelM: please comment on this issue
<RobSmith> I'm interested, as there is a significant overlap with WebVMT
MichaelM: a task force call could be monthly
Linda: we could use the SDWIG for some of this as a number of interested parties are there
<cperey> I am also interested, Linda. please include me on follow up
Linda: we would be happy to collaborate on best practices
George Percival expressed interest in chat. Ted Guild also interested
MichaelM: we need to collect
ideas and references initially and figure out what to produce.
I was thinking a best practices document
... it would be nice to align with our next round of spec
publication toward the end of the year
... I'll setup a doodle poll
Kaz: please send me an email if you're interested