Re: UniProt RDF as JSON-LD please test and give feedback,

Hi Markus, Nicholas,  Kingsley,

Thanks for the great feedback.

On 19/03/14 18:17, Markus Lanthaler wrote:
> Hi Jerven,
>
> That's great news.
>
>
>> This is now public, but not yet announced. One can see an example here
>> http://www.uniprot.org/uniprot/P77967.jsonld and a simpler one
>> http://www.uniprot.org/taxonomy/9606.jsonld
>
> Your context is at http://www.uniprot.org/contextjsonld (dot missing before
> "jsonld"?) and is served as text/html. You should serve it as
> application/ld+json.
>
Fixed in master will be live in the announced release.
> In general, I would suggest to avoid compact IRIs (rdfs:seeAlso) and use
> simple terms (seeAlso) instead.
As the JSON-LD is using our RDF writer I am worried this can cause 
collisions in the long run.
> For things like rdfs:seeAlso you should also
> set @type to @id in order to get rid of all the @id-objects.. you would end
> up with a simple array of strings.
There are cases where these are not simple strings and then I would end 
up with something like this.

rdfs:seeAlso": [
     {
       "@id": "http://www.ensembl.org/Homo_sapiens/Info/Index",
       "up:reviewed": true
     },
     "https://www.msu.edu/~heslipst/contents/ANP440/sapiens.htm",
     "http://en.wikipedia.org/wiki/Human_taxonomy",
]

Which makes the JSON-LD more unpredictable which is IMHO more difficult 
for plain JSON developers.



> On the other hand, you wouldn't need the
> type-coercion of booleans if you would use real booleans in the data:
>
>    "up:reviewed": "true" --> "up:reviewed": true
That is simpler, and will be in the released versions.
>
>
>> It is very similar to our other RDF serializations, except for evidence
>> tagging (provenance of an annotation). In the RDF/XML & Turtle this is
>> done with reification and in JSON-LD with graphs.
>
> I would be interested in hearing why you made that decision.
I thought that the graph based approach would give nicer JSON-LD now I 
am not so sure and going back to putting in quads so that the data is 
identical.

I attached an example of the changes on a different entry.
It won't play nice on the playground as the context uri does not exist.
>
>
>> I would be very happy if you all could have a quick look at it and let
>> me know if there are any bugs in it.
>> Then I can put it in the news and have it announced for the next
>> UniProt release as a public format.
>
Regards,
Jerven Bolleman