Re: Relationship and difference between N3 and RDF*

On Jun 9, 2021, at 10:18 AM, William Van Woensel <william.vanwoensel@gmail.com> wrote:
> 
> 
> Hi Joy,
> 
>> But in practice, I still don't know which one to choose.
>> 
>> In most cases, can I use N-quads or named graph, Like this:
>> 
>>         ex:bob a ex:Person .
>> 
>>         ex:bob  ex:haschild  ex:meli  ex:ng1 .
>> 
>>         ex:ng1  ex:saidBy  ex:lucy .
>> 
>> Does this have the same effect as RDF* ?
>> 
> Pierre-Antoine's explanation summarizes it quite nicely - it depends on your needs really. Notation3 allows you to insert a quoted graph (a.k.a. cited formula) into an statement and make assertions about it (i.e., describe the graph), whereas RDF* focuses on describing individual statements. 
> 
> Using named graphs:
> 
>     :lucySays { ex:bob ex:haschild ex:meli } . :lucySays ex:saidBy ex:lucy ; ex:saidAt "2021-09-06"^^xsd:date .
> 
> Using N3:
> 
>     { ex:bob ex:haschild ex:meli } ex:saidBy ex:lucy ; ex:saidAt "2021-09-06"^^xsd:date .
> 
Formally, this is somewhat more complicated. While it is common usage for a graph name to denote the graph it names (particularly in JSON-LD usage), formally RDF 1.1 Semantics to not provide a chain of reasoning for an IRI subject or object to necessarily be correlated with a graph it may name. See RDF Datasets <https://www.w3.org/TR/rdf11-mt/#rdf-datasets>. N3, in contrast, directly supports this, and also has some distinction with shared blank nodes.

Gregg

> In addition, about “N3 adds an If-Then style of decision making in the form of logical implications and variables”, I would like to know the current progress in this respect, what kind of expression will be used?   Does it have anything to do with ruleML or DMN?
> 
>> 
> Yes, Notation3 supports logical implications, as do the various RuleML formalisms. N3, in contrast to RuleML, builds on RDF (in particular, the Turtle notation) and extends it with quoted graphs and collections (or lists) - this may be more convenient for people coming from the Semantic Web world. N3 currently supports a rich set of builtins for operating on graphs, lists, strings, time, etc.. See here for a complete list of builtins.
> 
> Personally, what I believe sets N3 apart from other rule languages are its ties into the Web of Data: it includes builtins to access remote data (e.g., log:semantics builtin), inferring any conclusions from it, and/or checking whether it supports, or does not support, a given set of facts (e.g., log:includes, log:notIncludes builtin).
> 
> You can find some N3 examples here: https://github.com/w3c/N3/tree/master/examples . (For a comprehensive set of tests, see here).
> 
> You can try out N3 examples here: http://ppr.cs.dal.ca:3002/n3/editor/ (I'd recommend using Eye as the reasoner)
> 
> 
> 
> Regards,
> 
> William
> 
>>  
>> 发件人: William Van Woensel
>> 发送时间: 2021年6月7日 21:23
>> 收件人: Joy lix; public-rdf-star@w3.org
>> 抄送: Dörthe Arndt
>> 主题: Re: Relationship and difference between N3 and RDF*
>> 
>>  
>> 
>> Hi Joy,
>> 
>> Thanks for looking into Notation3 (N3). Re the W3C specification - it builds on top of prior "complete" W3C submissions (e.g., see here and here) - hence the new incomplete spec may be giving a bit of a misleading view on the current state of N3: there are already mature systems that implement N3 (Eye and cwm). Our effort at https://w3c.github.io/N3/spec/ is meant to further standardize and flesh out N3, since the prior documents are a bit vague on certain aspects.
>> 
>> Re the difference with TriG: N3 allows quoted graphs (a.k.a. cited formulae) to be utilized within triple statements, e.g., ":william :said { :william :wrote :moby_dick }", whereas TriG supports graph statements such as ":authorship_lies { :william :wrote :moby_dick }" that allow specifying named graphs. Both serve similar purposes - attaching context, provenance, and general metadata to statements - but have different semantics. Well, AFAIK named graphs do not really have an agreed-upon semantics - see here for a list of possible interpretations of named graphs - while cited formulae have more of a fixed semantics as referentially-opaque RDF graphs (this corresponds to a particular interpretation of named graphs).
>> 
>> You may notice that this is where there is an overlap with RDF*, since it also means to allow attaching metadata to statements - in their case, by simplifying reification within RDF.
>> 
>> Hope that answers your question.
>> 
>>  
>> 
>> Regards,
>> 
>> William
>> 
>>  
>> 
>> On 2021-06-05 7:25 p.m., Joy lix wrote:
>> 
>> I'm a beginner in semantic technology, and I'm learning about RDF*.
>> 
>> I notice that another W3C specification is also being drafted : https://w3c.github.io/N3/spec/ . Also it mentioned: “N3 adds an If-Then style of decision making in the form of logical implications and variables”.
>> 
>> But I feel like it's not quite up to speed with RDF*, because there are a lot of sections that still say "bla...".
>> 
>> I don't know how it is related to or different from RDF*, there is also the W3C's TRIG, would some expert explain this briefly?   Thanks.
>> 
>>  
>> 
>> Regards,
>> 
>> Joylix
>> 
>>  
>>