“Classical” Expert Systems
- 614improject
- Mar 17, 2016
- 1 min read
Classical Expert Systems
Mycin
Category - Diagnosis
Domain - Bacterial blood infections
Characteristics - Used certainty factors to handle uncertainty.
Used backward chaining as inferencing method.
Prospector
Category - Advice
Domain - Mineral Geology
Characteristics - Used fuzzy logic to handle uncertain information. Produced results matching those of a human very closely.
Oncocin
Category - Planning
Domain - Medical - cancer treatment
Characteristics - Automated process of knowledge acquisition by consulting human domain experts directly using an interview session at the computer terminal.
Reduced time and cost in maintaining knowledge base and eliminated neccesity for knowledge engineer and programmer to enter new domain knowledge.
Dendral
Category - Classification
Domain - Molecular chemistry
Characteristics - Used forward chaining.
The earliest expert system taking over 20 person-years to develop.
Internist
Category - Diagnosis
Domain - Medical diagnosis of internal diseases
Characteristics - Combined forward chaining and backward chaining to provide results.
Knowledge base had over 100,000 rules.
Strips
Category - Not strictly an expert system - used in Planning systems
Domain - Varies. Initially used in Robotic developments.
Characteristics - Provided instructions to allow a robot to navigate a room.
Major influence in development of planning expert systems.
R1/Xcon
Category - Planning
Domain - Computer configuration
Characteristics - Domain was such that no certainty handling was required.
Mainly used the specificity conflict resolution strategy.
OPS5
Category - Not an actual expert system
Domain - Production systems
Characteristics - Introduced the use of the RETE algorithm. Implemented using LISP.
PDF File : Classical Expert System ( Chapter 10)
Bình luận