"The Knowledge Acquisition Problem in Monotone
Boolean Systems"
Encyclopedia of Computer Science & Technology,
(A. Kent and J.G. Williams, Eds.), Marcel Dekker, Inc., New York, NY,
Vol. 39, pp. 89-106, (1998).
by E. Triantaphyllou and J. Lu
Abstract:
Learning is the main property that characterizes an intelligent system. As similar
situations will usually appear again and again in a given environment, the goal of learning
is to gain experience that can be used to improve the performance of the system when similar
situations occur in the future. Gaining experience often requires the system to build up an
internal knowledge base that can efficiently fetch the information when needed.
The learning process can be divided into two phases: the knowledge acquisition phase and
the rule generation phase. In the knowledge acquisition phase, it is relatively easy to gather a
large amount of knowledge or facts. However, critical facts are usually difficult to collect or
easy to neglect. Therefore, the time, manpower, and other resources spent on knowledge acquisition can
be very large if this process is not managed properly.
This article examines the knowledge acquisition problem for learning in a monotone Boolean system.
In such systems, it is assumed that all examples are represented by binary vectors in space En
and each bit of a vector represents an attribute of the example. The attributes are assumed to be
binary (i.e., to be either "True" or "False" ("0" or "1", respectively)). All examples are decided into two
classes and are thus regarded as positive and negative examples. The relation among the examples can be
expressed in the form of a monotone Boolean function, in which an example is regarded as a positive example
when the function value for the example is 1 and as a negative example when the function value is 0. The
goal of the knowledge acquisition phase in a monotone Boolean system is to infer the function and thus be able
to determine the class membership for all examples in the problem space.
(size = 1,070 KB)
