Information Technology Reference
In-Depth Information
(2) knowledge
Knowledge update is necessary due to the
incompleteness of the initial system knowledge. The newly-generated knowledge
may be conflict with what systems already know, thus a process of conflict
elimination is necessary to maintain the consistency of the knowledge base.
Knowledge maintenance in knowledge-bases systems is hard and expensive,
while CBR systems learn by acquiring new knowledge as cases thus making
maintenance easier.
(3) Improvement in problem solving efficiency: Problem solving in CBR can
reuse the previous solutions, and is not from the scratch as that in classic
inferences. Particularly, the previous records can help in advance the solutions
doomed to failure.
(4) Improvement in problem solving quality
maintenance
The records of previous failure help
in the current attempt to avoid failure in advance.
(5) Improve user acceptance degree: Users are ready to accept a conclusion only
when they are shown how it is reached. CBR systems draw conclusions on the
basis of historical facts. Facts speak louder than words; conclusions in CBR
systems are more convincing.
The Key Laboratory of Institute of Intelligent Information Processing,
Institute of Computing Technology, Chinese Academy of Sciences has carried
out a series of research in CBR. Z. Shi et al. propose a memory network
model and case-based search algorithm (Shi, 1992) in 1991. H. Zhou develops
EOFDS, a CBL-based design system for internal-combustion engine oil
product (Zhou, 1993). In 1994, H. Xu develops a CBR-based weather
forecast system, while two years later, J. Wang develops FOREZ, a
CBR-based predication and deployment system for Wang's Dam in Huanghe
River. S. Ye designs and implements a CBR-based system for fishing ground
predication. (Ye, 2001)
5.2 Basic Notations
Analogical problem solving is the process of finding the solution to a new
problem by applying an analogy between the recalled similar old problems
and the new problem, and by the analogy, new knowledge is generated
through reasoning to help problem solving. The solution to the new problem
can also be generated on the analysis of, and adjust to the previous problem
solutions. So in addition to the ability to record the solutions to similar old
