| Summary: | 碩士 === 國立成功大學 === 工程科學系 === 82 === With the improvement of software techniques and thedemands of
system applications, more and more evidences show that the
demand of integration of expert system and database system.
This thesis describes the design and implementation of an
intelligent database system. It consists of knowledge data and
database system. Rules from inductive learning are used to
assist the database management system for data transaction .
Knowledge acquisition is regarded as the bottleneck of an
expert system. To solve this problem, this thesis proposes an
inductive learning algorithm to achieve the goal. Information
is stored in a structured and organized manner in adatabase；
this provides attractive features for machine learning. Beside
that, in this thesis also uses a fuzzy classificatory
membership function to solve the problem of numeric type data
categories. The concept tree is used to deal with the
characters and string type data. During the generalization
process, the records that can not form a rule are removed.
After applying the inductive learning algorithm, we will get
rules and add them into the knowledge base. Rules are based on
the data of database system, the correctness will increase.
Because the development environment of database and expert
system is same, this system is distinguish from the traditional
rule_based database system. In that way, We do not have to
design another interface to integrate these two systems. This
feature increases the extendibility and usability of our system.
|
|---|
