RE: how does a term definition include all of : type, id, and alias

Hi James,

On 19 Jul 2014 at 21:56, james anderson wrote:
> a term definition in a context can concern (at least) three things:
> - the uri respective the key
> - the type of the respective member values
> - the processing keyword for which the term is aliased.
> the definition of an extended term definition does not appears to provide
> a means to specify all combinations.

No, you can't map a term to a IRI and a keyword at the same time. Similarly,
you can't modify keywords. So a keyword alias is always just a simple
term->keyword mapping.

> that is, it appears as if a context can specify either that a term is an
> or that it has a definition (possibly extended), but not both.

Right. Why would you like to combine the two?

> is there a means to combine the two aspects? or does the lack of a means
> to express both an alias and an identifier in the extended definition mean
> one resorts to encoding the identifier in the key? as e.g..
> {
>    "@context": {
>      "": "@type"
>    }
> }

That's certainly valid but why would you need that?

> i find in the mailing list archive some indications, that one should be
> able to declare, for example,
> {
>   "@context": {
>     "myIdAlias": {"@id": "@id"},
>     "myTypeAlias": {"@id": "@type"},
>     "myValueAlias": "@value"
>   },
>   "myIdAlias": "",
>   "myTypeAlias": "",
>   "": {"myValueAlias": "bar"}
> }

Yep, that works. See

> but i found no description of the of the facility in the recommendation.
> that is, i did not understand either the discussion of aliasing
> (, or the various passages
> (extended) term definitions - for example the description of their valid
> constituents (, to
> how to interpret such a combination.

Simply speaking, terms such as myIdAlias are first replaced with the keyword
they alias (@id) and then processed as usual. The processing is defined in
detail in the JSON-LD Processing Algorithms & API specification:

> i note, also, that as the expression conflates either the processing role
> the universal identifier, or the member term and the universal identifier,
> lacks the degrees of freedom required to specify all combinations.

Could you elaborate?

> is there yet another alternative?

I can tell you if you explain me what you are trying to achieve :-)

> best regards from berlin,
> ---
> james anderson | |

Does Dydra support JSON-LD now?

Markus Lanthaler

Markus Lanthaler

Received on Sunday, 20 July 2014 11:01:42 UTC