KAA-WS

The central part of KAA is the KAA Web Service (KAA-WS), which is a stateless web service implementation through which third party components request and retrieve the knowledge. KAA-WS implements the core learning algorithm, concretely an extended and improved instance-based learning technique. There is an internal ontology model of Grid service performances, resources and used invocation parameters which is called Knowledge Assimilation Schema (KAS). KAA consumes information generated by third party components such as Gemini (a monitoring infrastructure) and the Workflow Composition Tool (WCT) and transform them into KAS representation.

KAA-WXA

New information are added upon completion of any grid workflow, which is detected by a component called KAA-WXA. KAA-WXA also checks the GWES XML database for new workflows, gets the context from the workflow (based on XPath technique) and sends it to the UAA in a note format. If UAA has the same note (same context and input data) the old note is overwritten by the new one so users have always actual information about new results.

KAA-Gemini

Using a retrieved workflow identifier, a component called KAA-Gemini retrieves information about Grid services invocations and transforms them into KAS representation.

OnTal

Using a retrieved workflow identifier, a component called KAA-Gemini retrieves information about Grid services invocations and transforms them into KAS representation. OnTal follows one of general ideas of K-WfGrid, that broadening the knowledge of the system, stored in the form of semantic resources, allows more effective automatic workflow composition (performed by such components as WCT and AAB). Moreover, as the tool is semi-automatic, it is helpful for users who have to update the ontologies during services' registration.