Re: Semantics for EAI

Hi. I'll try to better depict what my research is for. I think that
narrowing the scope would help (me) better understand what I'd like to
build.

The idea of the project is to "augment" an ESB for EAI platform and to
enable it allowing it to make "inferences" regarding which routes to use,
"discovering" sources / destinations of an event message(s) which then it
transforms / enriches according destination "semantics" and format(s).

This featuring the exposure of a generic facade which allows to see in an
"homologated" view the applications or services and their data, schema and
behavior (actions) that could be integrated into the tool.

Different integrated applications are enriched with this facade and with
the events that, given the inferred routes and transformations, augments
theirs data, schema and behaviors, invoking activities corresponding to
each destiny semantics.

Regards,
Sebastián.


On Sat, Jun 29, 2019, 6:37 PM Sebastian Samaruga <ssamarug@gmail.com> wrote:

> Yes, sure. And it's not only ERM & CRM but any transactional system /
> contents repository an organization could have deployed inside (BI, BPM /
> Workflow / any source of domains "linkeable" and augmentable knowledge).
>
> My efforts are towards an unified "client" layer, in a given protocol
> (SPARQL / OData for example) for whatever "domain connectors" could be
> deployed / implemented (datasources). Then enable operations over this
> aggregated layer performing augmentations: aggregation (hierarchical
> regression) / alignment (attributes / links clustering) / activation (type
> / ontology matching).
>
> Although abstracting everything into (a layered architecture of organized)
> URIs and reactive and event driven models will help a lot, I lack the
> knowledge for develop / deploy many of the tasks needed for having what I'd
> like to be implemented.
>
> Any type of collaborations / comments / schema / backends or language
> recommendations / sample ontologies are welcome. Regards,
>
> Sebastián
> http://snxama.blogspot.com
>
>
> On Fri, Jun 21, 2019, 8:13 AM Amirouche Boubekki <
> amirouche.boubekki@gmail.com> wrote:
>
>> Getting together a vocabulary to allow ERP and CRM to interop is a very
>> difficult task.
>>
>> Le ven. 14 juin 2019 à 03:12, Sebastian Samaruga <ssamarug@gmail.com> a
>> écrit :
>>
>>> I've been very enthusiastic lately because, around the stuff I've been
>>> reading and researching, I see great chances of a solution being
>>> implemented that, despite looking a lot of work, has the same underlying
>>> backbone in all of it's functionality. Let me state first that I'm
>>> convinced of the vast role of semantics into the EAI niche.
>>>
>>> The basic attempt is to achieve a "semantic overlay" over ERP, CRM and
>>> application domains of organizations allowing for enhancements and
>>> extensions in the field of the extended knowledge that such applications
>>> could "learn" in base of the information contained in their data and
>>> adjacent knowledge integrated from various origins.
>>>
>>> In the beginning I'd use an "abstraction layer" approach, reuse backends
>>> and existing applications with semantic "enhancements" and, also, leverage
>>> integration with existing facilities with frameworks and techniques in use
>>> today (ORM with Hibernate / Spring, for example, implementing custom JDBC
>>> driver). This leveraging extension for new requirements or use cases as
>>> also for existing schema "enrichment" with inferences and learning
>>> "semantic" in the more transparent manner.
>>>
>>> As a begging there exists Tryton (and its derivatives: GNU Health) that
>>> in principle seems like an excellent departure point for an example ERP /
>>> CRP integration example, highly configurable and customizable. It could be
>>> the main "external" component considering the backbone of a proof of
>>> concept for the integration of applications knowledge.
>>>
>>> The components to be developed also will leverage being consumer /
>>> producer of data streams for frameworks such as Apache Metamodel, JBoss
>>> Teiid y JBoss Drools / JBPM. It could also leverage standards and
>>> frameworks such as OData and R2RQ for consumption and production of
>>> "enriched" information and knowledge. This will enhance capabilities of
>>> implementing connectors to consume and provide services and backends for
>>> later integration.
>>>
>>> In respect to an Enterprise Java ecosystem (Spring) an MVC / DCI IoC
>>> pattern implementation is to be facilitated by the existence of a JDBC
>>> driver who procures ORM / OGM (Object Graph Mapper). JCA (Java Connector
>>> Architecture) / Activation JAF (Java Activation Framework) facilitate
>>> descriptive and declarative implementation of process flows and state
>>> exposed as services.
>>>
>>> Features:
>>>
>>> Declarative hypermedia:
>>> REST / HAL / HATEOAS.
>>> SOAP / WSDL.
>>> Services (endpoints) "learnt" from systems and they integrations.
>>>
>>> CMS / Wiki (API / Protocol / DAV). Docs. Forms (Docs Flows).
>>>
>>> Facets: BI Views. Functional, Dimensional, Semiotic. BI APIs.
>>>
>>> Distributed Persistency. Meta Model, Facets / Levels, Encoding:
>>> semiotics, relations, categories, groups, sets, lists. Signatures /
>>> inferences (routes / mappings).
>>>
>>> Semantic Backend: upper ontologies. Model Alignment, inference services
>>> (Model Resource encoded).
>>>
>>> Learning Backend:  Model Augmentation services (Model Resource encoded)..
>>>
>>> Reactive Dataflow Backbone. Nodes: Protocol (dialog) Message flows
>>> (Augmentation) services (Model Resource encoded)
>>>
>>> This repository and attachments are fuzzy "scrapbook" notes:
>>> http://github.com/snxama/scrapbook
>>>
>>> Best,
>>> Sebastián Samaruga.
>>>
>>>