See also: IRC log
<trackbot> Date: 11 February 2015
<fjh> Agenda: https://lists.w3.org/Archives/Public/public-annotation/2015Feb/0062.html
<fjh> ScribeNick: bjdmeest
RESOLUTION: 4th February 2015 minutes are approved: http://www.w3.org/2015/02/04-annotation-minutes.html
bigbluehat: not that much activity on list, we should wait for Paolo next week
<fjh> Continued Discussion of http://w3c.github.io/web-annotation/protocol/wd/
fjh: draft, there has been discussion, anything wants to say anything specific?
dwhly: could we discuss getting annotator on the other draft?
fjh: that is underway
shepazu: we can do that
... we need the infrastructure set up
... on webplatform.org
... I will get the annotation system put in place there
<fjh> https://lists.w3.org/Archives/Public/public-annotation/2015Feb/0040.html
fjh: [about protocol] Nick had some issues
... about a lot of annotations
<fjh> https://lists.w3.org/Archives/Public/public-annotation/2015Feb/0067.html
fjh: LDP's minimal container could solve this issue
... also another discussion about containers and their types
... matching the rdf types and so on
... [about search] is that something we have to figure out?
azaroth: I am not aware of a standard search pattern
... that is less complex than SPARQL
... any ideas about other ways?
http://linkeddatafragments.org/
<fjh> bjdmeest: we have linked data fragments, basically SPARQL interface
<fjh> bjdmeest: allows specific data dumps
<fjh> bjdmeest: can index specific parts to ease querying
<fjh> bjdmeest: now have lightweight server, with more work on server, but could change this
<azaroth> azaroth previously: Important to understand the differences between bulk / web clients to see how the protocol needs to support both
azaroth: [about LDP containers] 3 types:
... ??? contain: endpoint
... direct container: basic container with one piece of additional functionality
... adding a resource creates link from resource to the created object
... [example]
... third type: indirect container
... also looks at RDF that gets posted
... new created results: picks a predefined relation, uses that to create a new link
... eg
... if you create a new body
... body owl:sameAs url:X
... system creates annotation hasBody url:X
... currently we use basic containers for the top-level ones
... so direct containers creates more triples
... triples aren't extra, system creates them for you
... use case is: if you want a non-rdf resource as part of the annotation
... (and image has no URI)
... then you would have to create a bodyless annotation, as the annotation has no URI
... then publish the image, and update the annotation
... direct container automates the last step
... we are not going to look at containers, mostly
... more a way of interacting
... paging will help
... not something a client needs to do
... we need to easily filter annotations
... [will help scaling]
<fjh> azaroth: those that have a given target etc
azaroth: we need to see how current clients work with current servers
shepazu: how tightly bound are our deliverables?
... e.g. twitter using the datamodel, but not change the infrastructure to LDP
... those tweets should still be annotations
... people might be less inclined to adopt this if the protocol is very strict
azaroth: I don't see them being tightly bound (the deliverables)
... If you already have some API, exposing annotations in the correct model will improve interoperability
... publishing in the correct format would be a big win already
... if you implement the protocol
... then we need to be more specific about the interactions
... so avoid twitter-specific APIs (as we have now)
fjh: so: protocol can be useful, but you don't need it
... doing the protocol without the datamodel would not make a lot of sense
shepazu: would there be something like: you need to implement this protocol to publish this datamodel?
... that would be a bad idea
... the minimum viable protocol: does it need to be LDP?
... is there something more abstract
... and still be interoperable?
... [LDP] is a very specific solution for a more general problem
... especially as a transitional thing
azaroth: agree that current implementations may not change to the protocol we specify
... but it may be good to suggest a current specific system
... we could suggest and abstract API
... Community group Hydra is about describing hypermedia API using JSON-LD
... Hydra is kinda like WSDL to SOAP
<fjh> Rob notes that me might consider Hydra, in some ways like WSDL, if we want to be more abstract
<bigbluehat> Hydra Community Group URL: http://www.hydra-cg.com/
azaroth: to move forward: it would be good to have a thickened protocol
... having an alternative would be good
... to see what would be acceptable, and not too specific
fjh: concern is real, but should start something completely different because of that?
... at some point, you need details
azaroth: so: we should write out the spec and see what the reactions are
... next steps:
1: prototype implementation, and report pain points and complex interactions
... e.g. hypothesis or annotator group to implement?
2: look at container types
... are these necessary? or could we provide discovery mechanisms of where these containers could live?
... e.g.: creating a binary resource as the body might be out of scope of the protocol?
... e.g. restrict the fact that you cannot create annotations of offline resources
azaroth: I will try to build a very quick implementation, someone else would also be good
... so, don't wait for me
fjh: decision: 26 april
<fjh> F2F 22 April, in conjunction 23, 24 April, SF
fjh: in conjunction with iAnnotate (23, 24 april)
dwhly: I have made some preliminary reservations
... about 11 rooms
RESOLUTION: F2F is 22 April in SF, in conjunction with iAnnotate 23, 24 April
... if you want a place to stay: please let me know
... let us know if you want help with the hotel
<dwhly> Fort Mason Center
<dwhly> Golden Gate Room
RayD: logistical info? location F2F and conference?
<dwhly> For both
<fjh> ACTION: fjh to post F2F message [recorded in http://www.w3.org/2015/02/11-annotation-minutes.html#action01]
<trackbot> Created ACTION-4 - Post f2f message [on Frederick Hirsch - due 2015-02-18].
dwhly: they are all in the same place
dwhly: we are almost at capacity for iAnnotate
... so iAnnotate.org and register
<shepazu> http://iannotate.org/2015/
azaroth: other WGs their discussion, and connection them with our work
... [LDP] all in favor of having an event
... they were very keen on having feedback on LDP
... but no event planned yet
... there is talk of having an LDP 1.1
... so comments are welcome
... so our discussion about protocol could be good feedback for LDP
... [Social Web]
... discussions about extend of requirement of JSON-LD
... or output being compatible with JSON-LD
... some interactions have been about RDF subclassing
... should that be part of Social or not?
... we have same concerns
... if we need reasoning, we loose 99% of our adopters
... is there something compatible, but does not need RDF savvies?
... good discussions happening there, we should follow those, and have the same discussions
... [DPUB IG] discussion about footnotes
... and marginalia
<shepazu> Footnotes: http://davidmacd.com/blog/html51-footnotes.html
azaroth: and make that more convenient for website creators
... e.g., using a specific HTML element
... request: is this something to be taken up by their a11y workgroup, or by us?
... doug and I believe this falls under HTML serialization of annotations
shepazu: blogpost: perspective of making HTML-docs more accessible
... Word is very accessible using footnotes