Knowledge Engineering for Practical Applications

A method is necessary to develop any piece of complex software. There are many methodologies which span the spectrum of software development. These range from specialised methods which accommodate real-time issues, to methods which embrace object-oriented systems. In addition, for Knowledge Based System (KBS) development, the pre-eminent methodology is KADS1.The argument advanced in this paper is that, when building KBS for practical applications, the method used for analysis, design and development may not necessarily be a specialised KBS method. There are reasons for this decision. Firstly, the new generation of KBS form only a part of an IT solution. That is, the KBS may not be the dominant IT component. Therefore any method must encompass the whole IT solution. Secondly, as will be shown by the case studies outlined in this paper, the term “KBS” now encompasses many disparate AI technologies. Specialised KBS methodologies may be restrictive for the development of systems utilising Case Based Reasoning (CBR), Constraint Logic Programming (CLP), Fuzzy Systems, Adaptive Neural Networks, Object-Oriented Systems (OOS), or their combinations.