KReS

From IKS Project

KReS is the Knowledge Representation and ReaSoning layer of IKS.

Contents 1 Overview 2 Source Code 3 Pattern based approach: ontology scopes and recipes 4 A stack of RESTful services to exploit ontologies and rules 5 References

Overview

The goal of KReS is to exploit semantic web standards and technologies for extending CMS systems with semantic capabilities.

When administrators setup and configure a KReS instance, they can load a custom set of ontologies to be used by that specific instance, and define a set of rules to be executed. Client applications can then access, reuse, and customize the environment for executing their specific tasks.

KReS has been implemented with the intention to deliver higher level services over this basic environment. The alpha version mainly provides basic services, most of the higher level ones are planned for next versions.

KReS is composed by three subsystems: the Ontology Network Manager (ONM), and the Rule manager and inference engine (R&I) - which implement the core part - and Semion, which provides reengineering and refactoring capabilities. Details about KReS architecture can be found here.

Source Code

KReS is part of the Apache Stanbol project and the source code is hosted under

 http://svn.apache.org/repos/asf/incubator/stanbol/trunk/kres/

Pattern based approach: ontology scopes and recipes

KReS implements a pattern-based approach for handling both ontologies and rules:

• Access to ontologies is controlled and centralized:
  1. KReS handles ontologies as nodes of an ontology network
  2. Ontologies are loaded in scopes
  3. Each scope is a runtime ontology network associated with a specific purpose (typically a task) e.g. integrity check of remote data fusion
  4. The same scope can be shared by several clients;
  5. KReS allows to create so called sessions associated with a specific scope, they support unshared computations;
  6. This approach natively supports the handling of web ontologies e.g. FOAF, Dublin Core, SKOS, Ontology Design Patterns (ODP) as well as local ones e.g. defined for the specifics of a CMS content repository.
• Rules are packed in recipes, and can be run by the application as single task oriented macro.
• Recipes are reasoning, alignment and refactoring patterns

A stack of RESTful services to exploit ontologies and rules

KReS functionalities can be accessed by client applications through a set of RESTful services.

1. Do refactor of RDF applying recipes;
2. Run inferences over RDF against ontologies and rules;
3. Reengineer non-RDF resources (XML and DBMS) to RDF
4. Manage the ontology network to be used:
   1. Creating and deleting scopes;
   2. Adding/removing ontologies from the scopes;
5. Exploit the ontologies for unshared operations
   1. Creating sessions over the scopes;
6. Manage rules
   1. Create rules with the KReS syntax (more user-friendly than SWRL, executable as SWRL or SPARQL);
   2. Pack rules in recipes;

Detailed description of the services with curl usage examples.

References

1. Deliverable for the alpha version of KReS. Note: some content can be outdated.
2. Reference standards are: RDF, OWL2, SWRL, SPARQL, OWLLink
3. Read more about ontology patterns at http://ontologydesignpatterns.org .