- From: Adam Sobieski <adamsobieski@hotmail.com>
- Date: Sun, 18 Dec 2022 00:08:40 +0000
- To: Martynas Jusevičius <martynas@atomgraph.com>
- CC: David Booth <david@dbooth.org>, "semantic-web@w3.org" <semantic-web@w3.org>
- Message-ID: <PH8P223MB0675A3459E6BA4957F1155D5C5E79@PH8P223MB0675.NAMP223.PROD.OUTLOOK.COM>
Martynas,
Thank you. The considered syntax, e.g., "ex:template??(ex:alice ex:bob)" and "ex:template??(("a1" ex:alice) ("a2" ex:bob))", maps with URIs using the query string portions. It appears that the LDT project was also thinking about URI patterns with "ldt:pattern".
I had been thinking about syntactic sugar for templating and for reading templates from the Web. I hadn’t yet thought about syntactic sugar for writing data to the Web. Thank you for sharing LDT.
With respect to defining simple templates, we can consider a N3 syntax extension resembling:
@define ex:template1 (?x ?y)
{
?x foaf:knows ?y .
?y foaf:knows ?x .
}
With respect to reading content from the Web using a template, if there were a built-in, language-integrated mechanism for dereferencing a URI (see also: https://github.com/w3c/N3/issues/111), e.g., "@dereference", we could express something like:
@define ex:template2 (?x ?y)
{
@dereference pattern "http://example.org/index.php?x={?x}&y={?y}#template2" .
}
Also possible is that, resembling other modern programming languages, e.g., C#, a general-purpose syntax could be useful for using string templates with RDF literals inside of template definitions:
@define ex:template3 (?x ?y)
{
?x ex:says $"Hello {?y}" .
}
In this case, we could do away with the "pattern" keyword modifying the "@dereference" keyword:
@define ex:template4 (?x ?y)
{
@dereference $"http://example.org/index.php?x={?x}&y={?y}#template4" .
}
So, a language-integrated template definition syntax, as shown above, could enable the expressiveness to optionally bind template parameters to parts of a URI, and beyond query string portions. It could be the case that multiple keywords, or a more intricate syntax, are needed for dereferencing URI. The case shown above encapsulates content retrieval. Developers might desire more customizability and configurability. Also possible is that, inside of template definitions, a syntax could be created for using string templates for arbitrary RDF literals.
There could also be @query and @update keywords for SPARQL, e.g., using SPIN. There are many exciting possibilities!
Best regards,
Adam
________________________________
From: Martynas Jusevičius <martynas@atomgraph.com>
Sent: Saturday, December 17, 2022 8:42 AM
To: Adam Sobieski <adamsobieski@hotmail.com>
Cc: David Booth <david@dbooth.org>; semantic-web@w3.org <semantic-web@w3.org>
Subject: Re: Templates, Generics, and Parametric Polymorphism in Semantic Web Languages
For read-write Linked Data mapping to SPARQL, based on URI patterns and with parameter support, you can take a look at Linked Data Templates.
https://atomgraph.github.io/Linked-Data-Templates/
On Thu, 15 Dec 2022 at 02.47, Adam Sobieski <adamsobieski@hotmail.com<mailto:adamsobieski@hotmail.com>> wrote:
When reading a traditional function (e.g., C/C++/C#/Java/JavaScript), the statements in “{}” brackets are to be interpreted as occurring in a sequence. When reading a graph, the contents of “{}” brackets are to be interpreted as a set. Accordingly, the concept of a function which returns a graph faces an uphill battle in terms of syntactical clarity when using “{}” brackets. I will think about clearer examples.
Here are some sketches of templates, macros, and functions which utilize the “{}” brackets to produce sets of nested contents.
@prefix list: <http://www.w3.org/2000/10/swap/list#> .
@prefix log: <http://www.w3.org/2000/10/swap/log#> .
@prefix ex: <http://example.org/#> .
@prefix graph: <http://www.w3.org/2000/10/swap/graph#> .
@prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#> .
ex:ng??(?x ?y) :graph
{
?x rdf:type foaf:Person .
?y rdf:type foaf:Person .
?x foaf:knows ?y .
?y foaf:knows ?x .
} .
{
ex:ng??(ex:Alice ex:Bob) :graph ?graph .
?graph graph:member ?stmt .
?stmt log:equalTo { ?s ?p ?o } .
}
=>
{
?s ?p ?o .
} .
The intended semantics of that “ex:ng??(?x ?y)” syntax is that, for each valid argument for ?x and ?y, a named graph is provided with a string-based pattern for its URI. That is, wherever the URI-pattern is utilized, a named graph is provided with contents as described, in the above case, four statements.
When I look at that sketch, I want to extend it to include guards for the input parameters, as if it were a function. I want to express that ?x and ?y must be rdf:type of foaf:Person and distinct individuals for the arguments to be valid. One complication is that there is a matter of "in which graph or dataset(s) must these conditions hold"?
Maybe we could add a named graph parameter into the mix? With a URI-based named graph, we could use a syntax like “ex:ng??(?kb ?x ?y)”.
@prefix list: <http://www.w3.org/2000/10/swap/list#> .
@prefix log: <http://www.w3.org/2000/10/swap/log#> .
@prefix ex: <http://example.org/#> .
@prefix graph: <http://www.w3.org/2000/10/swap/graph#> .
@prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#> .
@prefix owl: <http://www.w3.org/2002/07/owl#> .
ex:ng??(?kb ?x ?y) :graph
{
@requires ?kb holds { ?x rdf:type foaf:Person } .
@requires ?kb holds { ?y rdf:type foaf:Person } .
@requires ?kb holds { ?x owl:differentFrom ?y } .
?x foaf:knows ?y .
?y foaf:knows ?x .
} .
{
ex:ng??(:this ex:Alice ex:Bob) :graph ?graph .
?graph graph:member ?stmt .
?stmt log:equalTo { ?s ?p ?o } .
}
=>
{
?s ?p ?o .
} .
Could those three "@requires" be a one-liner:
@prefix list: <http://www.w3.org/2000/10/swap/list#> .
@prefix log: <http://www.w3.org/2000/10/swap/log#> .
@prefix ex: <http://example.org/#> .
@prefix graph: <http://www.w3.org/2000/10/swap/graph#> .
@prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#> .
@prefix owl: <http://www.w3.org/2002/07/owl#> .
ex:ng??(?kb ?x ?y) :graph
{
@requires ?kb holds { ?x rdf:type foaf:Person . ?y rdf:type foaf:Person . ?x owl:differentFrom ?y . } .
?x foaf:knows ?y .
?y foaf:knows ?x .
} .
{
ex:ng??(:this ex:Alice ex:Bob) :graph ?graph .
?graph graph:member ?stmt .
?stmt log:equalTo { ?s ?p ?o } .
}
=>
{
?s ?p ?o .
} .
With something like those sketches, above, I’m hoping to express that the arguments provided must be as indicated, in this case two distinct people, as holding in a provided knowledgebase (a named graph or SPARQL endpoint, ?kb). The described graph-based template/macro/function would then produce a graph containing two statements: “?x foaf:knows ?y” and “?y foaf:knows ?x”.
It’s unclear what would or should happen if a template/macro/function were “invoked” with invalid arguments. Maybe it would or should, then, return an empty graph and/or maybe an error or exception would or should be thrown?
What is the construct in N3 with its variables? I think of it as a template/macro/function. The current approach to invocation involves a URI-based pattern which resembles binding arguments to parameters to invoke the template/macro/function.
What are the values which are bound to the variables? Thus far, I am using URIs (xsd:anyURI) in the sketches. In theory, the graph-based templates/macros/functions should work with built-in datatypes like strings (xsd:string) or numerical types (e.g., xsd:integer). So, the actual values bound to the variables are, thus far, URIs.
Hopefully this content, above, clarifies what I'm hoping to express in the sketches while also introducing some new possibilities.
Also, while I’m excited about the mapping which provides a means of accessing and utilizing query string portions of URIs, and about its potential uses with graph-based templates/macros/functions, one potential limitation of the approach is that there exists an upper limit on URI string length (2048 characters?).
Best regards,
Adam
________________________________
From: David Booth <david@dbooth.org<mailto:david@dbooth.org>>
Sent: Wednesday, December 14, 2022 4:58 AM
To: Adam Sobieski <adamsobieski@hotmail.com<mailto:adamsobieski@hotmail.com>>; semantic-web@w3.org<mailto:semantic-web@w3.org> <semantic-web@w3.org<mailto:semantic-web@w3.org>>
Subject: Re: Templates, Generics, and Parametric Polymorphism in Semantic Web Languages
Hi Adam,
I am not yet understanding how to interpret the examples you have given.
To understand your suggested template syntax, I need an example to
clearly differentiate these 4-5 things:
1. What is template prior to expansion? (Think of this as a
function.)
2. Which parts of that template represent the variables
that will be replaced with actual data during template expansion?
(Think of these as the formal parameters to that function.)
3. For each of those variables, what is the actual value
to be substituted into the template, in place of that variable?
(Think of these as the actual parameters to that function.)
4. What is the final result of template expansion, after values
have been substituted in for the variables? (Think of these
as the function result.)
5. If you are also showing a syntax for invoking template expansion,
then what part of the example is illustrating that? (Think of this
as showing the function invocation.) The syntax needs to:
a. identify the template to be expanded (like identifying the
function to be called); and
b. bind an actual value to each template variable (like binding
an actual parameter to each formal parameter).
I assume that your example included some or all of the above, but it
isn't easy for me to see which bits are which. Could you clarify please?
Thanks,
David Booth
On 12/14/22 00:12, Adam Sobieski wrote:
> Existing query-based solutions, e.g., SPIN, could be of use for
> expressing the preconditions and effects of operators in planning
> domains. Beyond the processing and inferencing functionalities of
> solutions for templates and rules, there is the comparative
> expressiveness of the templates and rules to consider, and also the ease
> of consuming and inspecting the templates and rules to consider for each
> solution. I am thinking about how convenient it can be made for
> developers to utilize a more expressive planning domain definition
> language / ontology and its resultant APIs and object models while
> developing next-generation planning algorithms and solvers.
>
> With respect to the proposed mapping for Semantic Web languages which
> enables the access and use of URIs' query string portions, I recently
> changed the syntax in the sketches and examples from “^^” (which had
> some complications) to “??” (which I hope reminds developers of the URI
> query string).
>
> Understanding that "ex:ng??(ex:Alice ex:Bob)" maps to a URI, here is an
> updated example of what a /named graph template/in N3 could resemble:
>
>
> @prefix list:<http://www.w3.org/2000/10/swap/list#> .
>
> @prefix log:<http://www.w3.org/2000/10/swap/log#> .
>
> @prefix ex:<http://example.org/#> .
>
> @prefix graph: <http://www.w3.org/2000/10/swap/graph#> .
>
> ex:ng??(?x ?y) :graph
>
> {
>
> ?x rdf:type foaf:Person .
>
> ?y rdf:type foaf:Person .
>
> ?x foaf:knows ?y .
>
> ?y foaf:knows ?x .
>
> } .
>
> {
>
> ex:ng??(ex:Alice ex:Bob) :graph ?graph .
>
> ?graph graph:member ?stmt .
>
> ?stmt log:equalTo { ?s ?p ?o } .
>
> }
>
> =>
>
> {
>
> ?s ?p ?o .
>
> } .
>
>
> I am brainstorming about other new uses of and scenarios for the mapping.
>
>
>
> Best regards,
>
> Adam
>
>
> ------------------------------------------------------------------------
> *From:* David Booth <david@dbooth.org<mailto:david@dbooth.org>>
> *Sent:* Tuesday, December 13, 2022 5:16 PM
> *To:* semantic-web@w3.org<mailto:semantic-web@w3.org> <semantic-web@w3.org<mailto:semantic-web@w3.org>>
> *Subject:* Re: Templates, Generics, and Parametric Polymorphism in
> Semantic Web Languages
> Since every rules language (including N3) in essence defines a template
> language for the rules, I think it's worth asking: why are the existing
> template facilities inadequate for my use cases, and how might they be
> extended to handle them? If we can improve them to be more flexible,
> that may be better than adding more bandaids.
>
> Bear in mind that SPARQL is also a commonly used "rules language", using
> either CONSTRUCT or INSERT. ShExMap also acts as a rules language (and
> template language), as does SPIN.
>
> David Booth
>
> On 12/13/22 16:41, Adam Sobieski wrote:
>> Martin,
>> All,
>> Hello. I updated the Automated Planning and Scheduling Community Group's
>> wiki to showcase /Reasonable Ontology Templates (OTTR)/:
>> https://github.com/w3c-cg/planning/wiki/Ontologies
> <https://github.com/w3c-cg/planning/wiki/Ontologies>
>> <https://github.com/w3c-cg/planning/wiki/Ontologies
> <https://github.com/w3c-cg/planning/wiki/Ontologies>> .
>> I am also interested in computation and Semantic Web technologies. I am
>> thinking about inspectable, reflectable, reifiable, semantic templates,
>> macros, and functions (see also: the Function Ontology (FnO)). It could
>> be that the reification of language-integrated semantic templates and
>> functions would involve OTTR and/or FnO models.
>> With respect to automated planning and scheduling, planning domains'
>> operators' preconditions and effects can be represented as inspectable,
>> reflectable, reifiable functions using Semantic Web technologies (in
>> particular using n-ary semantic predicate calculus expressions).
>>
>> Semantic templating can be of use for implementing generic types and
>> functions, e.g,. "Widget<T>" and "function<T>(x, y, z)".
>>
>> I am thinking about how best to represent conditional branching logic in
>> semantic templates. These scenarios interest me in the contexts of
>> planning domains and computer lexicons. Also interesting, with respect
>> to OTTR, are patterns and compression.
>> Utilizing the mapping I shared, we could do new things. While I'm still
>> thinking about the specific syntax (e.g., "^^" or "??"), we could now
>> access and utilize the query string portions of URIs in Semantic Web
>> languages. I'm arguing that this would be generally useful and that it
>> would specifically benefit the technologies under discussion.
>> Best regards,
>> Adam
>>
>> ------------------------------------------------------------------------
>> *From:* Martin Georg Skjżveland <martige@ifi.uio.no<mailto:martige@ifi.uio.no>>
>> *Sent:* Tuesday, December 13, 2022 8:23 AM
>> *To:* william.vanwoensel@gmail.com<mailto:william.vanwoensel@gmail.com> <william.vanwoensel@gmail.com<mailto:william.vanwoensel@gmail.com>>;
>> adamsobieski@hotmail.com<mailto:adamsobieski@hotmail.com> <adamsobieski@hotmail.com<mailto:adamsobieski@hotmail.com>>
>> *Cc:* doerthe.arndt@tu-dresden.de<mailto:doerthe.arndt@tu-dresden.de> <doerthe.arndt@tu-dresden.de<mailto:doerthe.arndt@tu-dresden.de>>;
>> semantic-web@w3.org<mailto:semantic-web@w3.org> <semantic-web@w3.org<mailto:semantic-web@w3.org>>
>> *Subject:* Re: Templates, Generics, and Parametric Polymorphism in
>> Semantic Web Languages
>> Hi,
>>
>> there seems to be enough keyword matches to allow me to mention a
>> project that I have been working on, Reasonable Ontology Templates
>> (OTTR), see http://ottr.xyz <http://ottr.xyz> <http://ottr.xyz <http://ottr.xyz>>.
>>
>> OTTR is a template language for in particular RDF. It does not have all
>> the features you mention, its primary feature is to be able to
>> compactly represent complex statements as template instances that are
>> expanded via a simple recursive template expansion mechanim to plain
>> RDF.
>>
>> Example; (using a modelling pattern from the Pizza ontology), this OTTR
>> template instance
>>
>> [ ottr:of <http://tpl.ottr.xyz/p/pizza/0.2/NamedPizza
>> <http://tpl.ottr.xyz/p/pizza/0.2/NamedPizza
> <http://tpl.ottr.xyz/p/pizza/0.2/NamedPizza>>> ;
>> ottr:values (x:argument1 x:argument2 (x:argument3-1 x:argument3-2))
>> ]
>>
>> expands to:
>>
>> x:argument3-1 a owl:Class .
>>
>> pz:hasTopping a owl:ObjectProperty .
>>
>> x:argument1 rdfs:subClassOf pz:NamedPizza ;
>> rdfs:subClassOf [ a owl:Restriction ;
>> owl:onProperty pz:hasTopping ;
>> owl:someValuesFrom x:argument3-1
>> ] ;
>> rdfs:subClassOf [ a owl:Restriction ;
>> owl:onProperty pz:hasTopping ;
>> owl:someValuesFrom x:argument3-2
>> ] ;
>> rdfs:subClassOf [ a owl:Restriction ;
>> owl:hasValue x:argument2 ;
>> owl:onProperty pz:hasCountryOfOrigin
>> ] ;
>> rdfs:subClassOf [ a owl:Restriction ;
>> owl:allValuesFrom [ a owl:Class
>> ;
>> owl:unionOf (
>> x:argument3-1 x:argument3-2 )
>> ] ;
>> owl:onProperty pz:hasTopping
>> ] .
>>
>> x:argument3-2 a owl:Class .
>>
>> pz:hasCountryOfOrigin
>> a owl:ObjectProperty .
>>
>>
>> This example is found here: https://tpl.ottr.xyz/p/pizza/0.2/NamedPizza
> <https://tpl.ottr.xyz/p/pizza/0.2/NamedPizza>
>> <https://tpl.ottr.xyz/p/pizza/0.2/NamedPizza
> <https://tpl.ottr.xyz/p/pizza/0.2/NamedPizza>>
>> More templates here: https://tpl.ottr.xyz/ <https://tpl.ottr.xyz/> <https://tpl.ottr.xyz/
> <https://tpl.ottr.xyz/>>
>> Playground here: https://weblutra.ottr.xyz/ <https://weblutra.ottr.xyz/>
> <https://weblutra.ottr.xyz/ <https://weblutra.ottr.xyz/>>
>>
>> Cheers,
>> Martin
>>
>> On Tue, 2022-12-13 at 12:59 +0000, Adam Sobieski wrote:
>>> William,
>>> All,
>>>
>>> Thank you. This proposed language feature is generally useful and, in
>>> my opinion, merits consideration across Semantic Web languages (e.g.,
>>> N3, Turtle, TriG). I'm working on more concrete arguments
>>> here: https://github.com/w3c-cg/planning/discussions/27
> <https://github.com/w3c-cg/planning/discussions/27>
>> <https://github.com/w3c-cg/planning/discussions/27
> <https://github.com/w3c-cg/planning/discussions/27>> .
>>>
>>> Some clarifying key points:
>>> 1. There exists a bidirectional mapping between the proposed
>>> syntax and URIs/URLs with their query string portions utilized.
>>> 2. These URIs with their query string portions utilized can be
>>> used wherever a URI node or named graph can, across Semantic Web
>>> languages.
>>>
>>> Expanding on point 1, the following:
>>>
>>> ex:Widget^^(ex:Foo ex:Foo2 ex:Foo3)
>>> maps with:
>>>
>>> http://example.org/?T[]=http%3A%2F%2Fexample.org%2F%23Foo&T[]=http%3A%2F%2Fexample.org%2F%23Foo2&T[]=http%3A%2F%2Fexample.org%2F%23Foo3#Widget <http://example.org/?T[]=http%3A%2F%2Fexample.org%2F%23Foo&T[]=http%3A%2F%2Fexample.org%2F%23Foo2&T[]=http%3A%2F%2Fexample.org%2F%23Foo3#Widget> <http://example.org/?T[]=http%3A%2F%2Fexample.org%2F%23Foo&T[]=http%3A%2F%2Fexample.org%2F%23Foo2&T[]=http%3A%2F%2Fexample.org%2F%23Foo3#Widget <http://example.org/?T[]=http%3A%2F%2Fexample.org%2F%23Foo&T[]=http%3A%2F%2Fexample.org%2F%23Foo2&T[]=http%3A%2F%2Fexample.org%2F%23Foo3#Widget>>
>>> and the following:
>>>
>>> ex:Widget^^(("p1" ex:Foo) ("p2" ex:Foo2) ("p3" ex:Foo3))
>>> maps with:
>>>
>>> http://example.org?p1=http%3A%2F%2Fexample.org%2F%23Foo&p2=http%3A%2F%2Fexample.org%2F%23Foo2&p3=http%3A%2F%2Fexample.org%2F%23Foo3#Widget <http://example.org?p1=http%3A%2F%2Fexample.org%2F%23Foo&p2=http%3A%2F%2Fexample.org%2F%23Foo2&p3=http%3A%2F%2Fexample.org%2F%23Foo3#Widget> <http://example.org?p1=http%3A%2F%2Fexample.org%2F%23Foo&p2=http%3A%2F%2Fexample.org%2F%23Foo2&p3=http%3A%2F%2Fexample.org%2F%23Foo3#Widget <http://example.org?p1=http%3A%2F%2Fexample.org%2F%23Foo&p2=http%3A%2F%2Fexample.org%2F%23Foo2&p3=http%3A%2F%2Fexample.org%2F%23Foo3#Widget>>
>>> .
>>>
>>> This mapping technique is recursive; templates' arguments can be
>>> templates.
>>>
>>> This mapping technique is generally useful across Semantic Web
>>> languages and enables a number of interesting and useful features.
>>> Some of these envisioned features are powered by additional
>>> technologies, e.g., PHP and N3.
>>>
>>> Pertaining to N3 specifically, while one could model templates and
>>> their parameters as you have indicated, the proposed solution enables
>>> N3 "named graph templates", with a rather nice syntactic sugar, as
>>> prototyped
>>> here: https://github.com/w3c-cg/planning/discussions/27#discussioncom
> <https://github.com/w3c-cg/planning/discussions/27#discussioncom>
>> <https://github.com/w3c-cg/planning/discussions/27#discussioncom
> <https://github.com/w3c-cg/planning/discussions/27#discussioncom>>
>>> ment-4388356 .
>>>
>>> While it's a work in progress, I'm excited about this one. Comments,
>>> questions, and feedback are welcomed!
>>>
>>>
>>> Best regards,
>>> Adam
>>>
>>> From: William Van Woensel <william.vanwoensel@gmail.com<mailto:william.vanwoensel@gmail.com>>
>>> Sent: Tuesday, December 13, 2022 6:48 AM
>>> To: Adam Sobieski <adamsobieski@hotmail.com<mailto:adamsobieski@hotmail.com>>
>>> Cc: Doerthe Arndt <doerthe.arndt@tu-dresden.de<mailto:doerthe.arndt@tu-dresden.de>>; semantic-web@w3.org<mailto:semantic-web@w3.org>
>>> <semantic-web@w3.org<mailto:semantic-web@w3.org>>
>>> Subject: Re: Templates, Generics, and Parametric Polymorphism in
>>> Semantic Web Languages [Was Re: Syntactic Options for Probabilistic
>>> Semantics]
>>>
>>> Hi Adam
>>>
>>> Why introduce a new syntax (I believe you had a similar discussion
>>> before with Doerthe before on probabilities :-)
>>>
>>> Unsure whether I fully understand your proposal - so feel free to
>>> criticize - but an option could just easily be:
>>> ex:Widget tpl:parameters ( ex:Foo ex:Foo2 ex:Foo3 ) .
>>> Or
>>> ex:Widget tpl:parameters ( [
>>> tpl:label “param1” ;
>>> tpl:value ex:Foo
>>> ] [
>>> tpl:label “param2” ;
>>> tpl:value ex:Foo2
>>> ] [
>>> tpl:label “param3” ;
>>> tpl:value ex:Foo3
>>> ] ) .
>>>
>>> A set of N3 rules could then directly operate on these parameters (as
>>> they are also written in N3) to instantiate the template.
>>>
>>>
>>> W
>>>
>>> > On Dec 12, 2022, at 10:58 PM, Adam Sobieski
>>> > <adamsobieski@hotmail.com<mailto:adamsobieski@hotmail.com>> wrote:
>>> >
>>> > With respect to syntactic possibilities for templates, generics,
>>> > and parametric polymorphism in Semantic Web languages, it occurs
>>> > that all the brackets are already in use...
>>> > How, then, might we represent templates and generics? One
>>> > possibility is to utilize a list-like syntax for arguments and some
>>> > sort of connector, like:
>>> > <http://www.example.org/#Widget
>> <http://www.example.org/#Widget
> <http://www.example.org/#Widget>>>^^(ex:Foo ex:Foo2 ex:Foo3)
>>> > ex:Widget^^(ex:Foo ex:Foo2 ex:Foo3) Here is how it would look with
>>> > recursive structure:
>>> >
>>> > <http://www.example.org/#Widget
>> <http://www.example.org/#Widget
> <http://www.example.org/#Widget>>>^^(ex:Foo ex:Foo2^^(xsd:integer)
>>> > ex:Foo3) ex:Widget^^(ex:Foo ex:Foo2^^(xsd:integer) ex:Foo3) Do any
>>> > other syntactic possibilities come to mind?
>>> > Best regards,
>>> > Adam
>>> > From: Adam Sobieski <adamsobieski@hotmail.com<mailto:adamsobieski@hotmail.com>>
>>> > Sent: Monday, December 12, 2022 2:37 PM
>>> > To: William Van Woensel <william.vanwoensel@gmail.com<mailto:william.vanwoensel@gmail.com>>
>>> > Cc: Doerthe Arndt <doerthe.arndt@tu-dresden.de<mailto:doerthe.arndt@tu-dresden.de>>;
>>> > semantic-web@w3.org<mailto:semantic-web@w3.org> <semantic-web@w3.org<mailto:semantic-web@w3.org>>
>>> > Subject: Re: Syntactic Options for Probabilistic Semantics
>>> > William,
>>> > All,
>>> >
>>> > Thank you for the useful example. I've been recently thinking about
>>> > "graph reification" which is showcased in your
>>> > example:http://ppr.cs.dal.ca:3002/n3/editor/s/ztVCkf2R .
>>> >
>>> > N3 Syntactic Sugar
>>> > As interesting, here are some more rough-draft syntactic sketches:
>>> >
>>> > @foreach(rdf:Statement ?stmt in example:graph1 insert
>>> > example:graph2)
>>> > {
>>> > ...
>>> > }
>>> > @foreach(ext:Graph ?g in example:dataset1 insert example:graph2)
>>> > {
>>> > ...
>>> > }
>>> > @foreach(ext:Graph ?g in example:dataset1)
>>> > {
>>> > @foreach(rdf:Statement ?stmt in ?g insert example:graph2)
>>> > {
>>> > ...
>>> > }
>>> > }
>>> > N3, Client-side Preprocessing, Templates and Generics
>>> >
>>> > In addition to client-side PHP [2], I am also thinking about N3 as
>>> > a potential "semantic preprocessing" technology. With client-side
>>> > preprocessing [1], runtime arguments – including entire graphs or
>>> > knowledgebases – can be provided to preprocessors which can then
>>> > access the graphs or knowledgebases to utilize program logic to
>>> > vary output contents, e.g., TriG content. As envisioned, graphs or
>>> > knowledgebases would be provided to preprocessing environments as
>>> > implicit arguments alongside possible explicit arguments which
>>> > could be URIs or data types, e.g., boolean values, strings, or
>>> > numerical values.
>>> > Towards using N3 in a semantic preprocessing capacity, one could
>>> > define a reserved input graph name, e.g., “urn:reserved:input”, for
>>> > an input graph or knowledgebase to be provided at runtime. There
>>> > would, then, be a matter of how one might best express other input
>>> > parameters (e.g., T, or T1, T2, T3) which would, similarly, be
>>> > bound to argument values when invoking the preprocessor.
>>> > Scenarios considered include, broadly, templates and generics. For
>>> > a concrete example, “ex:Widget<ex:Dog>” and “ex:Widget<ex:Person>”
>>> > might result in output content which differs based on facts
>>> > asserted in the provided graph or knowledgebase about dogs and
>>> > people, e.g., a dog has four legs and a person two. The
>>> > preprocessing logic which would produce these generic types would
>>> > receive an implicit argument, a graph or knowledgebase, and an
>>> > explicit argument value, in this case a URI, for a template
>>> > parameter T, e.g., "ex:Dog" or "ex:Person".
>>> > As envisioned, output content from semantic preprocessing could be
>>> > semantic graphs. In the running example, the output content might
>>> > represent a type or class. Resultant types or classes might have
>>> > URIs resembling:
>>> > “http://www.example.org/?T=http%3A%2F%2Fwww.example.org%2F%23Dog#Wi
>>> > dget” and
>>> > “http://www.example.org/?T=http%3A%2F%2Fwww.example.org%2F%23Person
>>> > #Widget”.
>>> > While I’m looking at these preprocessing scenarios towards
>>> > advancing expressiveness for planning domain definitions, other
>>> > applications of interest to me include machine lexicons. For
>>> > instance, “verb:Walk<ex:Dog>” and “verb:Walk<ex:Person>” might have
>>> > similar but differing semantic content, e.g., definitions differing
>>> > as dogs have four legs and people have two.
>>> >
>>> >
>>> > Best regards,
>>> > Adam
>>> >
>>> > [1] https://github.com/w3c-cg/planning/discussions/21
> <https://github.com/w3c-cg/planning/discussions/21>
>> <https://github.com/w3c-cg/planning/discussions/21
> <https://github.com/w3c-cg/planning/discussions/21>> [2]
>>> > https://github.com/seanmorris/php-wasm
> <https://github.com/seanmorris/php-wasm>
>> <https://github.com/seanmorris/php-wasm
> <https://github.com/seanmorris/php-wasm>>
>>> > From: William Van Woensel <william.vanwoensel@gmail.com<mailto:william.vanwoensel@gmail.com>>
>>> > Sent: Monday, December 12, 2022 9:35 AM
>>> > To: Adam Sobieski <adamsobieski@hotmail.com<mailto:adamsobieski@hotmail.com>>
>>> > Cc: Doerthe Arndt <doerthe.arndt@tu-dresden.de<mailto:doerthe.arndt@tu-dresden.de>>;
>>> > semantic-web@w3.org<mailto:semantic-web@w3.org> <semantic-web@w3.org<mailto:semantic-web@w3.org>>
>>> > Subject: Re: Syntactic Options for Probabilistic Semantics
>>> > Hi Adam
>>> >
>>> > See here for some examples on how you can infer new statements
>>> > based on the contents of a graph:
>>> > http://ppr.cs.dal.ca:3002/n3/editor/s/ztVCkf2R
> <http://ppr.cs.dal.ca:3002/n3/editor/s/ztVCkf2R>
>> <http://ppr.cs.dal.ca:3002/n3/editor/s/ztVCkf2R
> <http://ppr.cs.dal.ca:3002/n3/editor/s/ztVCkf2R>>
>>> >
>>> > (These are arbitrary examples; if you had something concrete in
>>> > mind, I can try to write a better example.)
>>> >
>>> > I think the syntactic sugar you propose could be intuitive for this
>>> > kind of use case (a more imperative style of programming).
>>> >
>>> >
>>> > W
>>> >
>>> > > On Dec 11, 2022, at 3:37 AM, Adam Sobieski
>>> > > <adamsobieski@hotmail.com<mailto:adamsobieski@hotmail.com>> wrote:
>>> > >
>>> > > Dörthe,
>>> > > All,
>>> > >
>>> > > With respect to syntactic options for probabilistic logic and
>>> > > semantics, I recently created a GitHub discussion
>>> > > thread:https://github.com/w3c-cg/planning/discussions/26#discussi
>>> > > oncomment-4369123 . The most recent comment, there, broaches some
>>> > > syntactic options inc
Received on Sunday, 18 December 2022 00:08:59 UTC