Re: description of known limitations when deserializing JSON-LD to RDF

> On Jun 17, 2021, at 2:40 PM, Peter F. Patel-Schneider <pfpschneider@gmail.com> wrote:
> 
> The main reason that JSON-LD goes beyond RDF is that predicates can be blank nodes.  See the first bullet in
> 
> https://www.w3.org/TR/json-ld11/#relationship-to-rdf
> 
> JSON-LD does not support serialization of all generalized RDF triples as it does not allow literals as subjects or predicates, at least as far as I can tell.

Yes, Peter is correct; Blank Nodes can be used as predicates, although this is disarranged if not deprecated. Otherwise, it encodes RDF.

Additionally, the JSON native types for boolean and particular number have analogs in XSD datatypes, but JSON numbers, in particular, do not neatly divide into integer, decimal and float. The algorithms describe how they are transformed back and forth, and it is symmetric for all practical purposes.

There are other parts of JSON-LD which are lost when transforming to RDF, such as index maps.

JSON-LD also has explicit, if experimental, support for directional text using, e.g., the i18n namespace for literal datatypes. However, thsese literals can also be described in all other RDF serializations, although the RDF specs haven’t caught up with the behavior. Similarly, there is an rdf:JSON datatype added, for JSON literals, which can also be represented in other RDF serializations.

Gregg

> peter
> 
> 
> 
> On 6/17/21 5:03 PM, David Booth wrote:
>> On 6/17/21 4:51 PM, Nicolas Chauvat wrote:
>>> . . .
>>> Would someone on the list have good pointers on this [JSON-LD] topic, with maybe
>>> examples of JSON-LD documents that can not be mapped to RDF or a list
>>> of JSON-LD constructs that one should not use if a bijection is
>>> required between the JSON-LD document and say a Turtle document?
>> 
>> Try asking on the JSON-LD list:
>> https://lists.w3.org/Archives/Public/public-json-ld/
>> 
>> David Booth
>> 
>> 
>

Received on Friday, 18 June 2021 03:49:24 UTC