A general framework for building networked expert systems

Based on factor space theory, a general framework called the expert machine, for building fuzzy expert systems is introduced. This machine helps experts capture intuitive expertise without the assistance of knowledge engineers, automatically builds the knowledge base, simplifies knowledge representation, reduces computer storage, and provides high-speed inference suitable for on-line problem solving, especially for networked diagnosis and data fusion (object detection). As an expert shell, it provides an innovative method for knowledge representation different from most current fuzzy logic-based expert systems that rely on ad hoc membership functions obtained from intuition and experience. A high-speed synthetic reasoning algorithm is provided so that the inference process is computationally efficient. It acts like a configurable blackboard that converts raw input data from experts and compiles them into a source data base, which is downloaded into a TVFI (truth value flow inference) module. This technique presents great potential applications in both military and commercial systems.