Web Annotation Working Group Teleconference

HTML Serialization

TimCole: goal is to have something in week to 10 days that WG can vote on

<bigbluehat> discussing http://w3c.github.io/web-annotation/serialization-html-note/index-respec.html#annotations-embedded-in-html-as-json-ld

<bigbluehat> s/http://w3c.github.io/web-annotation/serialization-html-note/index-respec.html#annotations-embedded-in-html-as-json-ld/http://w3c.github.io/web-annotation/serialization-html-note/index-respec.html

csarven: need to cover the text of the document and also the examples
... main options we're using are json-ld and rdfa, maybe mention Turtle also
... we don't need in a Note to cover everything

<bigbluehat> http://w3c.github.io/web-annotation/selector-note/#frags

bigbluehat: writing a section about fragment selectors in context of HTML, e.g., with citation and reference within same doc

<ivan> +1 to ben

bigbluehat: one thing about Turtle, probably not worth including examples, but need to mention that you can use other RDF options
... but caution that other serialization may not be as friendly to HTML (i.e., less tooling)>

ivan: agree with what's been said. Additional - about whether it's worth making clear that if you use JSON-LD and RDFa in same file there are some details
... about whether this create minor issues. So maybe a cautioin that it's better not to use both.
... agree Benjamin

<bigbluehat> +1 to calling out the "unclear usage" for JSON-lD *and* RDFa in a single doc

<bigbluehat> it's pretty much unexplored/defined afaik

TimCole: first, the identifier given to the annotation(s)

<csarven> 'given to...?'

TimCole: are we assuming that the annotation id is typically an HTML fragment id

azaroth: from API standpoint you typically get more than just annotation
... if a representation exists independently, then that id is better

ivan: From RDFa lens, people make a differnece between HTML id attribute and id of the RDFa (e.g., using the about attribute)
... not sure if id attribute is used in RDFa
... so maybe a warning aobut HTML id as annotation id

<bigbluehat> ...that would mean in Tim's examples the fragment identifiers are referencing the <script> element in the DOM...not the contents of the <script> tag...correct? http://w3c.github.io/web-annotation/serialization-html-note/index-respec.html#annotations-embedded-in-html-as-json-ld

bigbluehat: so the json-ld id and id of script are being conflicted

ivan: yes, so not only RDFa but json-ld

<bigbluehat> DOM-wise in most modern browsers the DOM id's become JS variables fwiw

<csarven> I haven't seen (or recall) @id getting picked up by an RDFa parser.

<bigbluehat> so if you were to console.log(window['anno-588a322026bbcc00203fd0fb']) you'd get the JSON-LD (afaik)

<ivan> Yep, csarven, I think at some point we decided to separate these

TimCole: in practice we give wholly different ids

<bigbluehat> csarven: yeah...I'm not worried in this case about RDFa exactly, just that all references point to the same thing

ivan: mint a different id

<csarven> +1 to removing @id from script. Confusing (IMO)

ivan: remove id attribute from script tag (and maybe also class) since not relevant to illustration in Note
... editorially id should be shorter so easier to read

bigbluehat: For javascript id on script tags become global variable. Is id on script tag referring to contents of the script?

ivan: really it is representing a point in the DOM tree, not the content

csarven: +1 to focus on what needs to be shown in the illustration
... so my example also needs to be trimmed down
... re whether should use HTML frag id, one case is that the html is the complete annotation (RDFa example)
... so what would be the case when you have multiple annotations in same HTML doc

<csarven> http://csarven.ca/web-science-from-404-to-200#2055991574

csarven: the marginalia is added into the document
... maybe useful to differentiate between annotations that do exist externally and those only inthe document
... when annotation only exists in the document, and you have multiple, you need to use frag

ivan: for me the natural thing is annotation is disjoint
... describing both cases would be good

<bigbluehat> +1 to RDFa Lite support

ivan: agree RDFa would be better if re-formatted. Could the example be expressed as RDFa Lite
... there are tools that understand RDFa, but not necessarily understand full RDF

csarven: Will see if I can use RDFa Lite for the example in the document; could also talk about when full RDF is required

<bigbluehat> https://www.w3.org/TR/rdfa-lite/#respecContents

ivan: about is not in RDFa, resource instead sometimes
... the number of prefixes maybe could be reduced?

<bigbluehat> https://www.w3.org/2011/rdfa-context/rdfa-1.1

<bigbluehat> https://www.w3.org/TR/rdfa-core/#s_initialcontexts

csarven: remind reader how the context has to be handled differently in RDFa than json-ld?
... motivation for including is to support potential copy-paste

ivan: happy to go through document as editorial check

TimCole: Would it be useful for RDFa to show how it distills and can be framed into json-ld that looks like the model

ivan: semantically the json-ld example has a problem in trying to use the script identity as the identity of the annotation.

azaroth: no requirement that annotation id be de-referenceable in model or vocab

csarven: whether any of the annotation URIs should be derefernceable ... so should be okay for the Note

<azaroth> From the model, the only requirement is that it MUST have an IRI as an id

<azaroth> It could be urn:uuid:1234-...

<azaroth> Dereferencability comes from https://www.w3.org/TR/annotation-protocol/#annotation-retrieval

csarven: id on script happens to be the same, not clear what that means
... so not really a requirement to be de-referenceable
... one other use case, adding json-ld as supplemental metadata, versus duplicative text , e.g. author name appears in both

ivan: isolating annotation in a div element may be doable, but may be too much for this Note
... so we should record Takeshi's example as a postponed issued on the Note for now

takeshi: will do

ivan: to recognize that the whole doc except the annotation div is the target would require more than current tools do.

TimCole: fix examples - Tim and Sarven; add example - Benjamin?
... write down the issues and cautions - put them somewhere and then reogranize
... text has to be written around the examples, once the examples are more readable
... add in more references as needed

ivan: timeline - PR ends 14 Feb.
... if no issues time to publish Rec minimal
... expect 21 Feb could be target for publishing Rec and Selector Note
... This Note should also be published
... Both Notes will need votes
... can we get this note ready in time
... need complete rough draft by 3 Feb, cleanup week of 6 Feb, final by 10 Feb
... don't have to have real implementations.