Question

Question: how possible could it be to accomplish development and
implementation of a real business application with a semantics-enabled
yet-to-be developed framework?

Some of the features needed are similar of those of Drools rules, JBPM
flows and events / CEP processing. Inference and reasoning should provide
those as well as declarative means for backend / frontend use case
declaration and implementation.

Application instances should have a 'cognitive' BI (Business Intelligence)
interface layer over instances of their KB DB models.

Basically the proof of concept I have in mind is that of a 'social network'
in which individuals, profiles, resources, purposes and goals are managed
with their connections. There will be a 'working tray' use case with inbox
and outbox, indicators, tasks, issues, resource management, availability,
etc. Requirements for example for task completion, profile selection and
appropriate / necessary resource administration are handled by the concept
of 'Purposes'.

Given the framework architecture its KB could be feed and feed multiple
current EIS (ERP / CRM) with data, alongside having its own interfaces,
services and clients (Agents).

Another example of a framework application instance will be that of a
'smart' organizations projects management with dynamic roles and resources
assignment with the goal of pursuit a 'global' Purpose arranged according
its sub-purposes or tasks.

Applicaton instance exposes its 'generated' services according KB knowledge
schema and client 'Agents' expose necessary interface for clients use case
interactions (ie.: GUI) executions. Models data and inferred schema allows
for operations 'inference' exposed through an uniform protocol / interface
which can be ported to specific architectures such as SOAP or REST.

Purpose examples would be: Trip planning, Weight loss, Health attention,
Language / profession learning, Win football tournament, Margins /
processes optimization (via sub-purposes: increase sales, reduce costs) for
an organization, etc.

An advantage that a semantics-enabled approach could have is that being the
'schema' so de-normalized (in my approach 'tables' are 'data') then a
'model by example' paradigm could be implemented as the one used in Neo4j
graph database and other NoSQL DBs. Thanks to this approach schema and
'code' or interfaces of the application domain services are product of an
initial ETL load and aggregation.

This kind of 'declarative' application modelling is expected to be
accomplished by the ongoing development of this framework based upon
elemental RDF triple structures aggregated through logical and set theory
arrangement operations.

I'm currently finishing analysis phase with a few deliverables with some of
this guidelines detailed. I would like to join someone to review this
before going into implementation. Don't know if this is the adequate list,
but looking for help. Reviews will be highly appreciated.

Thanks,
Sebastián.