See also: IRC log
<trackbot> Date: 14 April 2010
i can scribe
<MacTed> something's weird with Zakim again...
<hhalpin> scribe: melvster
tho my IP will reset in about 55 minutes
<hhalpin> http://www.w3.org/2010/04/07-swxg-minutes.html
<hhalpin> approve?
<hhalpin> APPROVED: Minutes from 4-07 approved.
<MacTed> for attendance ... IRC (currently) shows : yoshiaki pchampin hhalpin oshani MacTed melvster AlexPassant tinkster
hhalpin: looking at policy languages
<hhalpin> XACML
hhalpin: next week larger overview, outside of sem web, inc. XACML
<hhalpin> APPROVED: meet next week 4-22
+1
<rreck> wonder how zakim knew me
<rreck> they must have gotten the email
scribe: everyone that has an open
action isnt on the phone
... will send people personal reminders
<hhalpin> ACTION: hhalpin to send personal reminders that final report actions should be done before next meeting [recorded in http://www.w3.org/2010/04/14-swxg-minutes.html#action01]
<trackbot> Created ACTION-137 - Send personal reminders that final report actions should be done before next meeting [on Harry Halpin - due 2010-04-21].
scribe: let's go over to
oshani
... from MIT DIG
<oshani> Slides of my talk: http://people.csail.mit.edu/oshani/talks/swxg.pdf
scribe: use case where someones
picture was taken from flickr
... advertising campagn
... but it was legal acc. to the rules, to use it
... pen friend
<hhalpin> http://www.parentdish.com/tag/dump%20your%20pen%20friend/
oshani: slides on irc
... im a phd student at DIG MIT
... focuses mainly on policies for the web
... how it relates to sem web
... first a small example, of why it's important to make people
policy aware
... discuss a few tools we've developed in our group
... talk about 'respect my privcay'
... lots of content on the web 3.6bn on flickr, 1 trillion web
pages
... people put their content on the web, and expect people to
properly attribute
... cc licences is the de facto licence
... meant to be a rights expression, rather enforcement,
expressed in human/machine readable format
<hhalpin> Creative Commons I think was one of the main use cases for the creation of RDFa (i.e. RDF markup inside HTML)
oshani: human readable, icons,
text for lawyers
... machine readable too
... 100m flikr images CC licenced
<pchampin> slide 6
oshani: cc licences support
different uses
... Share Alike most common
... non commercial use
... no derivatives, means you must use as is
... people can mix content
<pchampin> slide 7
oshani: most search engines support CC, yahoo google
<pchampin> slide 8
oshani: most content sites
support CC
... you can generate your own CC licences
... from creativecommons.org
... you can generate from a wizard
cc: AttributionName
... cc:AttributionURL
<pchampin> slide 10
cc: cc:morePermissions
... using RDF to describe licences
... licence has a URI
<pchampin> slide 11
<rreck> you could color code the edges
cc: human readable format is easy
to read
... also uses RDFa
... will show you the permissions
... eg free to share / remix and the permissions beyond are
availailble at <URI> attribution, and can properly
attribute the person
<pchampin> slide 13
cc: wanted to see if people are
actually aware of these things
... 78% - 94% are not aware of CC
... we used flickr images
... CC did a survey saying images are the most commonly used
content
... flickr has the largest collection of CC images
... nice URI
... to extract photo information
... used technorati to generate samples
... randomized selection of flikr image URIs
... authority rank, varies over time, and we collected 3
samples
<pchampin> slide 17
cc: used CC attribution
... one of the well defined CC licences out there
... use username
... use URL
... close to wherever that content is used
... flickr username or full name of user, URL was flickr user
url
... either /or
... questions?
hhalpin: was my intution, but
this is the first statistics
... is there any related work?
oshani: not that i know of
<pchampin> slide 18
oshani: raw violations 78% - 94%
<hhalpin> oh, I'd also be interested in stats of use of ccRel in microformats vs rdfa in the wild
<hhalpin> or just in HTML as a picture.
oshani: if it's your image you own the copyright
<hhalpin> My guess is image in HTML will predominate...
oshani: if you're not using the
licence, then you're setting a precident
... precision analysis
... low precision
... needed to remove
... non self attributed images
... found a higher rate
... not bad people
... lazy / dont know terms / dont know how to properly
attribute people
<pchampin> slide 21
http://dig.csail.mit.edu/2008/WSRI-Exchange/results
hhalpin: did you look at only flickr? does it use microformats or rdfa?
oshani: they now use rdfa, but not during the experiment
<hhalpin> great work helping them use RDFa.
hhalpin: interesting to see stats on how much machine readable format is out there
<pchampin> slide 22
oshani: have several tools to
enable policy awareness
... one is flickr cc validator
... idea is you can give any uri with flickr images
... can validate violations
... will show images that violate any licence, the owner, and
how to correct
... limitations : can only validate images that have the flickr
image uri
... it wont work with an alternate URI
... wont detect getty images etc. if they had been uploaded to
flickr
MacTed: images that are hotlinked ... do you have anything that looks inside the image file, eg in the headers?
oshani: can look at meta data, but CC data isnt there unfortunately, just a key value pair that people can overwrite
MacTed: if someone is casually violating, do you detect that
oshani: there are tools that
allow you to embed any key value pairs
... better alternative, XMP where you can put RDF, I was
proposing that to flickr
<hhalpin> +1
oshani: not sure whether they'll change
<hhalpin> for embedding it themselves, make it as easy as possible for enduser
oshani: want them to support XMP
http://dig.csail.mit.edu/FlickrCC/validator.cgi
scribe: you can try it out
... other type of tool is more general
... any type of content
<pchampin> slide 27
scribe: so long as they have a CC
licence in right expression lang
... right now only impl. for images
... FF extension
... if you're browsing images, it will show you if you can copy
an image
... if you want to find images that are used for commercial
purposes
... you can use the tool to highlight those images
<pchampin> slide 28
scribe: right click on an
image
... context menu will allow you to copy, copy with licence,
will make you automatically licence compliant
... limitations we would want rdfa, most content doesnt have
rdfa
... hopefully supported in subsequent versions
<pchampin> slide 30
http://dig.csail.mit.edu/2007/tab/release/unofficial/tabulator-latest.xpi
scribe: try it out
... install tabulator
... then you can use it
MacTed: any chance it will be divorced from tabulator?
oshani: tabulator is in active
development
... several unofficial uses
... it uses the tabulator rdf parser library
... so the code wont be replicated
MacTed: sandbox security settings
are needed for tabulator
... if you dont do that tabulator doesnt work
oshani: i can do that
MacTed: i would encourage you to do so
oshani: policies on content
reuse
... how to give proper credit
... people worry about things like privacy on soc nets
... two of my colleagues created a project called 'repsect my
privacy'
<pchampin> slide 32
oshani: specify
restrictions
... facebook has access control settings, but what happens if
the data somehow gets out?
<pchampin> slide 33
oshani: private information can
be copied and reposted
... inferences can be made ... 'gaydar'
<hhalpin> this work is actually rather scary!
oshani: analysed MIT network and
selected 1500 males, and based on inference principles, they
can infer sexual orientation
... policy changes can also occur
... no clear cut solution
... information accountability
... specify under what conditions data can be reused
... similar to CC
<pchampin> slide 34
<pchampin> slide 35
oshani: make people aware of
usage restrictions
... supported across facebook, opensocial, tabulator
... ask the user to specify privacy preferences
... displays data and highlights restrictions
<pchampin> slide 36
oshani: RMP restrictions
... no commercial
... no depiction ... please dont use my picture
... no employment
... no financial
... no medical dont use this information for any medical
decisions
... eg insurance cos. / hospitals
... motivated by several complaints
<pchampin> slide 37
http://apps.facebook.com/respectmyprivacy/
scribe: went further
... tried to use the semantic web
... facebook applications are limited
... just a label
... used FOAF converter by Matt Rowe
... uses RMP decisions
... not allowed to keep exported data off facebook for > 24
hours
... makes it restrictive
<hhalpin> hmm how do they enforce that time restriction?
scribe: also used on
tabulator
... uses application panes
... supprots edit in place (SPARUL)
... data pane
... social pane
... get yourself a webid
... using webDAV
... browse FOAF profiles
<pchampin> slide 42
scribe: extended foaf pane
... includes RMP
... include privacy restrictions
... if your FOAF has SPARUL it will persist
... can add licences to FOAF
... example slide 43
... can intermix CC and RMP
... that's about it for RMP ... questions?
<hhalpin> like the mixing, thats RDF
hhalpin: how does facebook legally enforce the 24h export data rule
oshani: it's the privacy
policy
... dont know how they enforce it
<hhalpin> any other questions?
i can download that extension from the link above, yes?
pchampin: what does the fb app. do? does it change privacy setting to be confomant with rmp?
oshani: no, we cant touch those,
we can display a badge
... before you install the app, it gives the philosophy behind
RMP, more a notice, encouraging people to be honest
<hhalpin> questions?
hhalpin: we want distributed policy in access control languages, only a few uses, what are your thougths on CC and the other work being done, can we make it more general
<hhalpin> but it could be part of an access control langauge
oshani: CC is data uses rather than access control
hhalpin: some use cases eg dump your penfriend, are content reuse, some issues solved by tagging, can people be notified
<hhalpin> yes that's correct
oshani: that used CC attribution licences, and vigin mobile properly attributed the person that uploaded that image, so they were licence compliant
<hhalpin> Virgin was license compliant!
oshani: settled out of court
<hhalpin> weird
<hhalpin> nothing to Flickr?
( my connection will die in about 2 minutes )
scribe: if someone comes and
reuses your image, they can do certain things
... cc is not that explicit, people need to be more policy and
licence aware
... there should be tools that let you do that
<rreck> thanks for presenting
<hhalpin> standardization?
<hhalpin> ccRel well-known
<hhalpin> seems to being used correctly?
<hhalpin> RMP introduced some new terms that might not have consensus
<hhalpin> so maybe W3C could look in that direction.
<hhalpin> Meeting Adjourned
<pchampin> bye bye
<rreck> bubye
<melvster1> hi sorry about that
<hhalpin> no problem melvster I got the final minutes
<melvster1> great! :)
<hhalpin> trackbot, end meeting
This is scribe.perl Revision: 1.135 of Date: 2009/03/02 03:52:20 Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/ Guessing input format: RRSAgent_Text_Format (score 1.00) Succeeded: s/APPORVED/APPROVED/ Succeeded: s/UIR/URI/ Found Scribe: melvster Inferring ScribeNick: melvster Default Present: pchampin, MacTed, melvster, yoshiaki, +34.91.639.aaaa, oshani, rreck, +049172247aabb Present: pchampin MacTed melvster yoshiaki +34.91.639.aaaa oshani rreck +049172247aabb Found Date: 14 Apr 2010 Guessing minutes URL: http://www.w3.org/2010/04/14-swxg-minutes.html People with action items: hhalpin[End of scribe.perl diagnostic output]