Day 1 (Jan 15)
Agenda - Workshop home page
Sam Critchely reported on the distributed architecture breakout.
<link> Peter Mika's slides
Peter Mika summarized the discussions of the breakout on data mining / semantic Web.
John Kemp reported the discussions around businees models for social networks.
<link> Karl Dubost's slides
Karl Dubost summarized the discussions of the breakout regarding privacy and trust.
<link> Topic introduction
<link> Julian's slides
<link> Simon's slides
<link> Dan's slides
Privacy aspects of all the solutions are considered to be important or are they nice to have. Are there any business drivers to influence a higher priority on the location based solution agenda?
Sam: Yes, it is important and location is a category that warrants privacy. Younger and Older users, aren't so worried about it. However, families are really worried about it.
Privacy is partly about regulations and about control of users' data and location with operators.
Are they not really concerned or are they not aware? I think they aren't aware of this.
Simon: I don't know. I think users are aware of it being stored.
Imito (JP), FB - explain a lack of privacy of teenage use.
Sam: NY magazine has teen interviews where teens don't really care about their implications until they reach 30s.
John Kemp: Attitudes of people will change.
Sam: As GypSii, we wouldn't do this, however, there are US college kids who are uploading porn videos regularly based on cultural norms.
2 reasons why privacy may not be very concerning yet. 1 - there hasn't been any serious repercussions based on privacy violations (max. job loss, no large crisis yet such as 9/11, financial loss, etc.), 2 - geo location issues haven't really surfaced yet.
It is good to build for it in the technology stack, and not trivialize it based on teen feedback and current cultural norms. What is within scope within standardization - what is that people want that is not there within the current APIs?
Is privacy the only big hole?
Simon: Privacy is not the biggest technical hurdle yet.
Sam: Cambridge you are using much more than location, has everyone signed up to it?
Simon: Not a good case as it is an office environment, but couple of people didn't. Use cases were useful - desktop porting, meet when they are around, etc. Many benefits to efficient interaction.
When you ask to track someone, does it tell you are being tracked?
Do uncertain about what they know about you?
Is there a standard for time based location (APIs) - where was someone 2 days ago?
Dom: W3C has the API based on location and time, so you can track and no notion
Kaushik: User research talk about 'Relationship issues' and 'need for control'. Proactive vs. reactive modes is preferred.
Socialight - Loopt is one, that has problems - synchronized application that uses asynchronized data (few users, and very stale information), big issues
GypSii - not yet a problem, but if we give proximity alerts, then the problem of deadlocks, relationship issues will emerge (Doppler, TripIt - via LinkedIn, people know about it)
Voda - we haven't done this
Henry: When everyone has this problem, then it becomes natural and disappears.
Sam: Do you think it is something
If I were able to check a system to check to find out where my friends and what they did.
"Posts in the air ". Are they APIs that help us get to find what you did and tracking what your friends did? Don't want to mash on demand but as a path-to-track-what-your friends did. Socialight has APIs but proprietary. Erin Kopec at Flicr is also doing some stuff - tag photos of say meal and you get a geoRSS feed out of that, and string those together. This is essentially "auto archiving of lifestreaming based on location". Example: Tagging of objects (transforming real world stuff into URLs).
Sam: Can you get processing power to solve these contexts? Voda?
Voda: benefits for the user is huge, and getting to an efficient context is vital.
Simon: Temporal aspects of location is very important. Where shall I go in Barcelona? What is more interesting is Where did my friends spend more time in Barcelona? This should be done by short-term state things of query on aggregated data.
Sam: Do you think it is easy to solve the computational problems of a 3D representation - timelines, individual elements, media/data?
Socialight - CitySense, yes it is possible. They do heavy computational work. They are using data to hedge fund managers as to heat mapping users and which retailers they are going to.
Telecom Italia: Only lat/long is not very important. Semantics in addition to location and context of what user activity is vital. "Semantic-based location systems"
Simon: Isnt this a large-scale project.
Henry: Mereology (we need a logic or ontology for this type of inference engine to arrive at semantic-based location systems).
Karl: What do you need to solve this?
- research, mereology, person (4D - time, space, role.., ..)
... We already have haptic systems and collaborate with them on this? (Wii, Apple)
Simon: It is a challenge to deal with concentric rings of contexts.
Sam: Is time a factor we are prepared to share, given that we may be hesitant to share our current information?
Nokia: We had issues with location data, we found people can say favs (at specific times at specific places, they can say things - predictability of historical data is a privacy concerns).
Sam: History is ok, future and current is not ok. Home is not ok, a foreign trip is ok.
But it depends on who is asking? - my boss, friend, family
Simon: Should there be any plans for best practices with user interface design?
Should enable with APIs
Karl: Who? When? What? Where? - knowing all this before sharing is important.
May be more important to have a data set standard to satisfy the UI requirements. Semantic has to be somewhere as people use different identities on various SN to express themselves. Min pieces of data is one way to express the UI.
Dunbar theory: There are studies that talk about people having not more than 150 friends and max of 3 different social circles.
Henry (Uof Edin): Its not entirely a myth. This was an anthropological study (man has brain 4x than chimps, so 4x friends), so a bit weird. Some people have lot of contacts, some have less. What is interesting is cognitive load may limit the number. Media intervention can variate communication.
To me, location - who sees it, when, how long before it expires, and the number of people it is limited to.
Simon: Do you see this as given in a society? Accept it as we go on?
Henry: This is the way how people used to be in the village/
Simon: Benefits of ambient awareness (village analogy - know what 12 others were doing, where they were going, supportive effect for an isolated urban scenarios, ties with U of Cambridge )
Having control helps people share more, esp. by location.
Simon: In villages, haven't we been subject to the peer influence?
It is important that we do get subject to the peer influence. Esp. more if it is global.
Karl: We should be careful with the village analogy - our behavior became because of that. We shouldn't think that our behavior now will change because of a village analogy.
Sam: Can we encapsulate our recommendations into a standard? Should this be a bigger context?
: Geolocation standard that matches our needs for social purposes.
: Context is too large to capture into a W3C standard
Simon: You can quite a lot with location based systems based on sensors. We should limit context to location alone.
Sam: Smoke alarms in phones, body temperature sensing, accelerometer on all phones
Nokia: Shouldn't we standardize accelerometers?
Sam: Should we include other things (not just GPS)?
: Yes, we should get lat/lon/accelerometer cords
Sam: Marriage of SN in marriage with Context management systems is vital
: Needs a definition of terminology - list of context, features, interfaces to be defined etc. before approaching as a standard. Ex: You are in the office, mall, - some applications just need to know this abstraction and exact lat/lon. Defnition of location is more important than the actual location!
Is an umbrella effort required?
1.Should define APIs based on application, for each API - define different function
2.W3C has lots of standard around context, and doesn't look at location. CCPP, User Agent profiles, buying habits etc. are there. We should've started umbrella effort years ago. Modularized approach is missing and sure we should get into a standard.
3.Geolocation APIs should be extended to include: semantic context/definition of location (home, work, etc.) in addition to lat/long
4.We have a ubiquitous web API (applications working group - ontology, context information) - not going too fast
5.There are multiple working groups, but we need a best practices working group for SN (which points to the specific APIs and standards - how to use them, use cases, list individual APIs)
6.There is Palm with its OS, iPhone OS open, Google Gears is very cool. Geo-location working group is working with Gears and Google is authoring this.
8.Produce a paper of some kind to reflect - what's in and what's out
9.To rebut not having a working group for everything, APIs aren't a good model for asking questions, a paper on query APIs is the way to go through the TAG (tech architecture group)
10.Simon: There is broad consensus that a lot of work is going on in APIs, but we should publish to highlight APIs, use cases, esp. with context within SN, and recommendations on what is available on how to use them. Suggest: "W3C Context in SN" and chapters: Overview of the context imperative, Specific Use Cases, What's available in each APIs (based on each context), What's needs standard APIs, Recommendations, Communicating context through SN for users and for partners wanting to use this information.
<link> Presentation from Henry
<link> Dom's slides
* Towards creating an XG based on Harry's proposal Social Web XG
* Identify gaps in the protocol landscape
* Build the business case for decentralized architectures
* Outside W3C would be best? Big Social Networks would probably not discuss their business models with us.
Candidate for combination with Harry's Social Web XG
* W3C is not the right place to develop business cases or business models
* Business models should be an input on the W3C, not an output?
* Business cases and models should be transversal to any W3C activity; makes no sense to start a dedicated "Business Models" XG or IG.
* Maybe we have enough suspicions to start a dedicated IG/XG on micropayments? --- A more technical subject.
* Half the room agrees that micropayments are necessary to ensure Social Networks' future
* Do we need micropayment protocols or just the concept?
* We just need to get the ball rolling, get the discussion started, figure out if we do indeed need protocols and a common approach or if that has been done already, in which case we'd cut it off after 3 months.
* Very mild support in the room to actively participate in a micropayment XG. People like the idea of micropayments but feel that big players (operators, banks) are missing. Or they support the idea but feel they can't bring much to it.
<scribe> ACTION: TimAnglade to explore opportunity to start an XG around micropayments
* Educational role for W3C on that front.
* Relation to [P3P] (http://www.w3.org/P3P/)?
* Probably keep that line of discussion as part of [PLING] (http://www.w3.org/Policy/pling/wiki/Main_Page).
* P3P is the best technical opportunity right now? Maybe through a dedicated profile?
* What about the toolkit and the guidelines? => heavily linked to the educational mission.
* 3 initiatives
1. P3P Profile dedicated to Social Networks: interest expressed by some in the room.
<scribe> ACTION: Sören to send feedback to PLING based on discussions at workshop
2. Ontology for Access Control: reasonable amount of interest. W3C generally doesn't accept ontologies as recommendations. There will be a follow-up with the FOAF community, beginning of a talk through a mailing list, TBD.
<scribe> ACTION: AlexPassant to check on opportunity/need for standardizing an access control ontology
3. Toolkit and best practices: creation of an XG? Draft charter to be written by Christine and ... Henry Story.
<scribe> ACTION: Christine to propose a draft charter around the creation of a Web Privacy Toolbox, with Henry Story
* All feedback should go to existing W3C groups (WAI Education & Outreach, Mobile Web Initiative)?
* Consumer research and pooling is necessary. A place to collect. Probably not something that would be inside the W3C.
<scribe> ACTION: HennySwan to send feedback on outreach needed to WAI E&O
<scribe> ACTION: Dom to send feedback on outreach needed to MWI
* Lots of elements already covered by existing W3C groups, including Geolocation WG, Ubiquitous Web Applications WG, Web Applications WG.
<scribe> * New elements include:
* API/Ontology to annotate semantically user context: space / time / roles / social context
* Good amount of interest in furthering this work.
* The most appropriate shape would be: inside Henry's Social Web XG? A separate XG?
* Creation of an explaratory XG?
* Looking into what the values are, the interests are and then spinning a few separate XGs from that first XG.
* Would that be the same as Henry's Social Web XG? Henry seems to disagree this should be one and the same. Proposes to keep his very low-level practical and do another, high-level one.
* Ori disagrees, thinks those two cannot be separated. Others argue in the same direction, saying one group is better, as tangential subjects can be discussed collaboratively.
<scribe> ACTION: Claudio to create a draft charter on annotating user contexts
* Requirements / Use cases for sensor APIs
ACTION SamCritchley to draft a draft charter around requirements/use cases for context sensitivty APIs, with help from Marcel Medina and comments by Dan.
* Notion that something is part of something else. Too early?
<scribe> ACTION: HenryStory to send feedback to existing OWL WG
Use the **firstname.lastname@example.org** mailing list. You can subscribe. It's a public W3C mailing list, archived.
Great place to continue our discussions, also with existing list members that may have not been here. But also discuss charters and the XGs.
The slides, draft minutes and report will be sent by email to all registered participants.
Send pointers sto blog posts, photos, notes to dom by email.
* Time horizon for a next workshop? 18 months? At TPAC (Novembre '09, in California)
* The target of that next conference should be network operators
* Web conference in Madrid? => Academics
* Spot conference => Semantic Web => Academics
* The next conference *needs* to be in the Valley.