Suggestions
by Sebastian Samaruga
Hi, sorry for bothering. I'm just sharing some thoughts regarding what I
think 'semantic' applications could look like (I'm mostly looking forward
existing implementations, not shure if I'm able to build such a framework,
any suggestion regarding projects or learning is welcome):
Could it be possible that existing applications (backends, datasources /
services) as they are deployed today could benefit from semantic tools
alignments, augmentation and learning (inferences) enabled by a framework
leveraging a 'virtualization overlay' which covers not only 'intelligence'
in a given application domain but between applications. This domain
'intelligence' being 'encoded' as domain behavior knowledge allowing to
describe 'alignments' ('translations') between such applications.
The point is if it is possible maybe, to some degree, 'automate' this
domain (application: backends, datasources and services) virtualization,
integration and translations via the use of some form of heuristics,
inference and learning enabling them to be augmented themselves and in
respect to other applications domains with learning and knowledge
capabilities in the most transparent manner.
Those applications should be 'plugged' in a streaming bus (Nodes). Adapters
(Backend streams / IO) allowing streaming adapter IO synchronization
(domains 'gestures' translation) performing corresponding domain's
'effects' given context's actions.
Features (enabling previously mentioned capabilities): aggregation
(alignment: identity merge, augmentation: attribute / rels discovery,
regression: entity 'role' in context discovery) by means of an uniform
messaging layer and declarative 'assets' (components) described through an
uniform Metamodel layer (Semantic Resources Metamodel REST APIs).
Client Nodes: ad-hoc application extension assets (entities, schema, flows
declarative descriptions) as means to augment bus applications with new
functionalities. Custom declarative endpoints that expose APIs through
protocols.
Dashboard: virtualized domains visualization and assets management (domains
use cases flow management).
Example: data / schema / behavior flows in one application / domain
generate 'triggered' transactions between applications / domains (CRUD +
rules / flows application kinds). Infer backend 'contexts' (DCI /
Metamodel).
Example: classification (document oriented application kinds). Flows
(trays: state / form action templates) classify images / documents (folders
/ labels) by features. Automatic tagging (labels).
Example: query custom endpoints (protocols: inference / predictions). Apply
'views' transforms over aggregated bus domains. Expose knowledge in custom
protocols (REST, SOAP, SPARQL, etc.). Complete missing information.
Best Regards,
Sebastian.
6 years
- 1
- 0
- ← Newer
- 1
- Older →