- From: Martynas Jusevičius <martynas@atomgraph.com>
- Date: Thu, 9 Nov 2017 23:28:54 +0200
- To: John Flynn <jflynn12@verizon.net>
- Cc: semantic-web <semantic-web@w3.org>
- Message-ID: <CAE35VmzoOWhZx90hGpdGHLh202HFyksU3zHWJ+jOcdgt4pZeCQ@mail.gmail.com>
There is no need for new languages -- we have SPARQL already. No procedural language can accommodate RDF better than it does. What is usually not done yet, is expressing not only application's domain model, but also its operations and its every state as RDF. With this, we can have an ontology-driven architecture and a single well-defined generic protocol (or API, if you will) for all SPARQL-backed Linked Data applications. After 6+ years of Linked Data application development, we observed this architecture emerge, which we eventually contributed as the Linked Data Templates specification: https://atomgraph.github.io/Linked-Data-Templates/ Martynas atomgraph.com On Thu, Nov 9, 2017 at 11:01 PM, John Flynn <jflynn12@verizon.net> wrote: > There have been previous threads on this basic idea. I sent the following > out in 2012 in response to one such discussion. > > "I am somewhat surprised that someone hasn't created a dedicated Semantic > Web applications development language by now. I tried, unsuccessfully, to > get government funding for this idea shortly after the DAML to OWL > transition. > It seems to me one of the handicaps to the widespread development of > Semantic Web based applications is the lack of a specialized computer > language that uses OWL ontologies and instance data as native data > structures. Such a language would include constructs designed to manipulate > and access specified ontologies and data. It shouldn't be particularly > difficult to design a language along these lines and to implement it in a > compiler. It would probably make for a fairly decent computer science > master's thesis." > > So, what I was talking about back in 2012 was developing a procedural > language that would use the declarative capabilities of RDF/OWL as it > primary native data structure. The language would include all the standard > procedural constructs to allow the user to manipulate data in terms of > algorithms, heuristics and other data processing capabilities. What it > would add is the built-in (native) use of RDF/OWL ontologies as data > structures and it would also include new data processing constructs > specifically designed to manipulate ontology graphs and associated instance > data. This is different from other approaches where an attempt is made to > bring the ontology data into object oriented data structures which are > already native to the procedural language. The new language would include > facilities to handle new (non-declared) instance data generated via > ontological reasoning. The core of the procedural language could > potentially be any of the current/past procedural languages such as Java, > Python, C++, C #, etc., although consideration should be given to which > language would provide the best constructs to accomodate the semantic > nature of the RDF/OWL data structure (graph). Building off an existing, > well known procedural language would facility the learning curve for users. > However, it might also turn out that an entirely new procedural language > would provide a better fit for manipulating semantic data. I still think > this would make a good computer science master's thesis. > > John Flynn > http://semanticsimulations.com > > > > > >
Received on Thursday, 9 November 2017 21:29:20 UTC