See also: IRC log
Frederick: Sets up the meeting,
how sensors have come up in places like sysApps
... is chair of the DAP WG
<scribe> scribe: phila
<scribe> scribeNick: phila
Frederick: Is there enough interest in a new IG/WG in this area
ddahl: I'm chair of local
interaction WG
... we've looked at things like glucose meters
... sensors are burgeoning, how do we get it standardised
ericp: I'm HCLS team contact. I'm interested in how we can use this to fill gaps in medical data
Klaus: I have an interest from a
Web research, WoT, POV. I'm also diabetic so glucose sensors
are important to me
... This is not just restriced to medical, but also things like
different diets
dsr: I'll be talking about Web of
Things later
... is this about sensors talking to browsers?
Frederick: I don't think it's that well defined
??: W try and make use of proprietary sensors (Tomo Digtal??)
Gentleman from Tencent... looking at connection to physical devices
... using IM
Another gentleman from Tencent - how can healthcare data be socialised (? hope I got that right, forgive me)
... sharing data within a community that has relevant privacy and security controls
SteveB: GS1 has a huge healthcare community. Every device has its won ID etc
... unique ID is clearly important
... we talk about authentication etc. so ID is critical
Francis from Shanghai - I'm interested in implementing a healthcare system in Shanghai. The medical staff can use their mobile device to track the patient anad find out the patient's condition
... need a central repository for this data
... we make use of HTML5
Tim from Google - we're trying to make sense of sensor data
Gentleman from Fraunhofer
scribe: we have expertise on
accessibility. We don't make APIs for sensors - we mostly do
R&D projects funed by EU
... we can help with requirements that we have from the
sensors
<Steve_Bratt> GS1 healtchcare standards work: http://www.gs1.org/healthcare/standards
scribe: looking at how we can have sensors and then make sense of that data for users (scribe note I hope I got that right)
Pater from Panasonic
Panasonic makes sensors - we want to be able to integrate them more into the WoT
s/pater/Peter/
fjh: When you move from standalone to integrated, what's the biggest challenge?
Peter: Right now everything's standalone
fjh: So issues like privacy and security are issues but also underlying protocol layers?
Peter: yes
... Not really anything Web-specific yet
scribe note - sorry missed this gentleman's name and affiliation
Hyunjin from Korea
<paulliu> Paul Liu from Huawei Technologies
My company makes a healthcare device...
wonsuk from Samsung
As you know we have a lot of different devices...
scribe: we're trying to make a
Web platform like a Tizen
... we need to expose Web API for sensors like pedometers
... how can we handle this kind of API on the Web.
Jungkee from Samsung
scribe: talking about the same things as Wonsuk
phila: Apologises sincerely for my inability to catch so many names properly. I hope no offence is taken by those present
Ken: At MITRE I support ?? which
is trying to get a lot of health data from a lot of
people
... so I need to be able to advise them on what's coming down
the pipe wrt standards
... My other issue is how different is this from a vehicle API?
Privacy may be different
Steve_Bratt: If you're having a medicine delivery system, the couterpart might be a sensor to measure effect
phila: Community group on APIs
http://www.w3.org/community/hydra/
... Automotive Business group http://www.w3.org/community/autowebplatform/
ericP: Social Apps are different but how do we share a bunch of different devices send different sources
fjh: I was thinking Annotations might be relevant to noting properties
ericP: I also noted security and
general interrogation of devices
... in terms of possible attacks - you need to protect things
like pace makers, subdermal insulin systems etc. These have all
been attacked by promximity to other systems
dsr: Live data is not the same as periodic
ericP: You're going to need to turn strings of 1s and 0s into strings that can be processed
ken: Everything has policies
related to it (P&S etc.)
... devices will have policies attached and we need to
understand that
fjh: Do people read policies?
usability is an issue
... data overload/ sensor overload is an issue
ericP: It's up to us to understand the policies...
Claes: How much should we look at
other orgs' work...
... I'm not aware of the current status of other work but maybe
we need to be?
... What can we reuse from their work?
fjh: I'm not informed
??: Things like HL7 are used to exchange medical data
ddahl: I wanted to say something about... there's a lot around privacy, security, authorisation - how can control the robot etc.
dsr: Talked about mobile phones as sensors - can't program that with JavaScript
ddahl: Robots are even more diverse than sensors
<Zakim> ddahl, you wanted to mention that some of these concerns also came out in the earlier robots session
ericP: I see 2 pieces ... coding requires attaching a long tail of capabilities and observation codes
fjh: The DAP has had a long conversation about the balance between generic and specific
... a lot of stuff is event driven
... I'm trying to figure whether the Web model being event driven fits
... and what Eric was saying about the long tail suggetss we need a more generic approach
... in DAP we moved away from that because we wanted to work in a short time frame
... generic opens the door to fingerprinting
... to avoid fingerpringinting, we reduce the amount of data available
<ddahl_> robots session had a lot of the same concerns -- authorization, privacy, security, discovery of capabilities, timing (e.g. realtime vs scheduling) what messaging API's are best? and are there concerns related to low-level protocols?
Fingerprinting is the idea that I can identify you without an ID for you as I look at what I can see and your behaviour. I can identify you without an ID
... I can probably gather enough info to identify you. I can get your browser etc.
So fingerprinting can become an issue
<ddahl_> also in the robots session we talked about graceful degradation
... a generic sensor API may be open to fingerprinting
ericP: So that assumes that the attacker has access to that data
... the other way is to give the user to have accecss control
fjh: So in DAP we looked at a security model but it didn't work for the Web
... we decided in the end just to use explicit consent dialogues
Evangelos: (Franhofer) I think it would be useful to identify 3 situations
... immediate need, longer term for doctors etc.
... and proactive healthcare
... this is a hot topic at the moment. They are t rying to prevent people from having a health problem
... and are we covering elderly people, children?
... target groups need identifying
fjh: I'm trying to forumlate an understand of how that relates to W3C. I think here we do more horizontal
... if you do that then they should apply across the board including automotive, healthcare...
Ken: the idea from getting from generic to specific. On specific you'll forever be coming up with new API descriptions
Ken: The idea would be that the standard can handle the horizontal and then CGs can write the verticals
<ddahl_> we also talked about capabilities vocabularies for different kinds of robots that could be plugged into a generic API
Ken: So the horizontal would be the basic and then CGs would extend and feed back
phila_: That matches the schema.org system
dsr: We're launching a WoT Interest Group to look at overall picture, and this may then lead to specific WGs
fjh: That makes sense but that has the poss problems faced by SysApps and DAP
ericP: For a large class of these devices, we care more about protection than things like fingerprinting
... you said reduce the capabilities to prevent fingerprinting
... we don't mind if a device has bluetooth or NFC connection - that can be public. But the fact that it's a glucose sensor is more private
yusuke: Dave, you said you were looking for a common standpoint for missions of device
<fjh> we did look at discovery etc in dap with webintents
yusuke: As you know, that's hard
yusuke: A big company might take the market...
yusuke: So I think we need a strategy to handle marketing and standardising
fjh: I'm not sure we should talk about market segments
dsr: The Interest Group can look at what is needed and that will cover open and closed markets
<Zakim> fjh, you wanted to ask about layering and discovery
fjh: I think discovery is a big area, layering and discovering... not knowing whether it's an insulin pump or not is an issue
ericP: You need to be authorised to find that out
dsr: For healthcare - your doctor needs to know who you are
ericP: when you have a sub dermal defribulator will have an ID
fjh: We have a number of general requirements gathered today
we have the notion that a generic approach has value for scalability
we have that an approach similar to schema.org with a horizontal base with CGs can be spun off to handle verticls
ericP: The security model is an issue to address
... authentication required for interrogation
fjh: With authentication, more data is appropriate
fjh: The next step is to express interest in the Interest Group
dsr: Explains W3C process for establishing and joining new groups
yusuke: This IG will be about Web of Things, not just healthcare?
Dsr: yes
... we want to see what the areas havea in common etc
fjh: Did we miss anything?
... feel free to add to the IRC if you have projects etc. to look at