17-03-2014, 09:22 PM
ABSTRACT
Solving a non-tivial problem can rarely be reduced to
performing a single, simple task. Within any complex
knowledge-based system, there exists a number of tasks to be
performed to solve the problem. Making these tasks explicit
has been a recurrent concem over the past few years. This
has led to functional architecture for knowledge-based
systems.
The purpose of this paper is to assess the use of functional
architecture for knowledge-based systems. The discussion
will be based on the experience gained while designing
DIVA, an expert system for vibration-based monitoring of
large rotating machinery. We first describe the system and
show how its task decomposition results from the
characteristics of the problem. We then describe how this
approach enables to build more sophisticated explanations of
reasoning. We present how knowledge acquisition has been
achieved in this framework. And lastly, we discuss the
maintainability of such systems.
1. INTRODUCTION
Solving a non-trivial problem can rarely be reduced to
performing a single, simple task. Within any complex
knowledge-based system, there exists a number of tasks to be
performed to solve the problem. Making these tasks explicit
has been a recurrent concem over the past few years. This
has led to functional architecture for knowledge-based
systems.
The purpose of this paper is to assess the use of functional
architecture for knowledge-based systems. The discussion
will be based on the experience gained while designing
DIVA, an expert system for vibration-based monitoring of
large rotating machinery.
The paper is structured as follows: Section 2 reviews the
motivations for functional architectures; Section 3 gives an
overview of DIVA; in Section 4, we discuss how
explanation, knowledge acquisition and maintenance may
benefit from designing DIVA at the task level.