Copy the page URI to the clipboard
Harrison, Alan
(1995).
DOI: https://doi.org/10.21954/ou.ro.0000fb85
Abstract
There is increasing evidence that designers of future real-time embedded systems are turning to knowledge-based techniques in order to solve complex problems where algorithmic techniques have failed to produce a solution. In addition, many applications have been mandated to use the Ada programming language for all implementation software, including the knowledge-based components.
This thesis identifies three essential requirements needed to support the construction of these systems: first, the need to provide a library of Ada knowledge-based components that supports a variety of knowledge representation paradigms to model the diverse expert domains being encountered in complex applications; second, the need to provide the user with the means of creating and controlling multiple independent instances of the knowledge-based components to cope with the complexity and scale of the implementations; and third, the need to provide an integrating architecture in which the knowledge-based components may be embedded directly into an application environment.
These requirements have been satisfied by using ideas derived from the concept of abstract data types to construct a library of knowledge-based components; the components have been called abstract knowledge types. Subsequently, multiple instances of the abstract knowledge types have been integrated in modules called knowledge sources, which model specific problem knowledge domains. The knowledge sources have been used to construct a blackboard architecture.
The abstract knowledge types have been used to build a prototype university timetabling system in order to demonstrate their use. The research has shown that the abstract knowledge type integration approach results in a uniform implementation strategy for both conventional and knowledge-based components.
CORE (COnnecting REpositories)
CORE (COnnecting REpositories)
