- From: Sebastian Samaruga <ssamarug@gmail.com>
- Date: Mon, 17 Jul 2017 12:45:48 -0300
- To: public-rww <public-rww@w3.org>, DBpedia <Dbpedia-discussion@lists.sourceforge.net>, pragmaticweb@lists.spline.inf.fu-berlin.de, W3C Semantic Web IG <semantic-web@w3.org>
- Message-ID: <CAOLUXBse-7ja_XxqTEEPNiXQ=z3pxUp0C2gEwW++v7HrssHsOA@mail.gmail.com>
I'll try to narrow the scope of the description of what I'm trying to build and that I've published previously: Besides some basic initial type inference I'm not focusing (only) on ontology engineering or modeling but more in providing an engine (some like DOM / JavaScript is for HTML) for whatever RDF I can provide to it from whatever sources, not only DBpedia but seamless integrating RDBMSs / REST / LDP or any other backend / datasource / protocols. Maybe I'll try to 'distill' sources of statements into layers (data, information and knowledge) and this could be helpful for whatever ontologies classes and instances modeling due to ontology alignment. A goal for me will be to allow to not need to have a previous knowledge of sources / schemas for being able to work with them. Regards, Sebastian. On Jul 16, 2017 2:01 PM, "Sebastian Samaruga" <ssamarug@gmail.com> wrote: > Hi. I know I've been lately posting some stuff which may have seem very > fuzzy or even non sense (including for me when I read it again, I admit) > but, for some reason, and this I think I share with almost all of SW > community, since I've realized I've some point regarding semantics I feel > in the enormous need of communicate and let the world know of the big > amount of benefits they are missing... > > So, despite of my lack of academic politeness, I'm trying (again) to put > this in a few lines. At this point, I admit, it will be most worthwhile > what I could learn from others, as also what I'm learning while I'm doing > this work, than the benefits of only trying to propose 'my' version of a > way of doing things in semantics. So, whatever suggestions of joining in an > open source project or participate in SW developments shall be welcome. > > Here is what I've came up on thinking of what could be some 'patterns' I > should follow when building my next applications. > > Again, despite being non sense, and my english is not so good, the most > I'd like to get is what I'll learn of what I'm doing wrong. Here I don't > know any playground in where to learn how to do things in SW so this is > only almost by myself: > > The statements below are a 'list' of what I 'should' do in a 'framework' > with a very brief description of each 'feature'. I think they are the ones > I'll need when trying to build a business application. > > --- > > Achieve plain RDF serialization to-from any service / backend through Node > Bindings (below). > > Model statements into layers, each one having its own 'abstractions' for > the roles each statement parts play: > > Data layer: Infer type information, Classes / Kinds roles. Properties > (domain / range role Kinds). From facts layer statements (data) infer which > property kinds holds for which domain / range role kinds: > > Who is a Person. > Who is a Business. > Which relations (properties) holds between them. > > Information (schema) layer: Events which entail Flows given certain > statements (relations) roles. > > aPerson (hired), worksFor, someCompany (hires); > > Knowledge (behavior) layer: Given certain Rules some Messages are > ‘activated’. > > betterJobOffering, employeeBehavior, jobSwitching; > > Knowledge, information and data behavior (CRUD) entails ‘propagation’ into > upper / lower levels of statements updating them accordingly via dataflow > graphs (below). > > Data example: aProduct, price, 10; > > Information example: aProductPrice, percentVariation, +10; > > Knowledge example: aProductPriceVariation, tendencyLastMonth, rise; > > Features should be provided by the framework / datastore (alignment): > > Align identity: merge equivalent entities (with different URIs). Align > types (schema / promotion: infer type due to role in relation). > > Align attributes / links: augment knowledge (properties and values) about > entities (because of type / role alignment). > > Align ordering: sort entities regarding some context / axis (temporal, > causal, composition and other relations). > > Implement functional query / transformation language: for a given entity / > concept (Monad) being able to browse / apply a function which entails some > other result entity (Monads). > > Resource endpoints. RESTFul interfaces for metamodels interaction. > Provides / consumes (feeds / streams) events / messages declared via > dataflow engine. The datastore should be a HATEOAS web application. > > Implement XSL Driven declarative dataflow engine: Streams ‘pipes’ > declaratively stated in XSL for reactive behavior definitions. Functional > query / transform semantics. Datastore application publish / subscribe to > this (request / response 'filters'). > > Implement a 'Node' abstraction: Integration of diverse (wrapped into > metamodels) datasources / backends / services / protocols via the > implementation of service contracts and exposing a plain RDF IO (dialog > protocol) interface. A Node Binding to other Nodes shall allow to integrate > via dataflow semantics diverse datasources / datastores. > > --- > > Best Regards, > Sebastian Samaruga. >
Received on Monday, 17 July 2017 15:47:12 UTC