In The Object-oriented Software Design Mode Of Intelligent Extraction Technology Research

High quality software systems typically apply expert design experience which has been called design patterns. Though they have been extensively applied in the software, the information of them always is lost in the large number of software system. It causes that many good designs can not be played importance role in the design of software system. At the same time, it cannot be reused and enhanced the maintaining in system. Mining the instances of design pattern from the source code of object-oriented software systems is to get the information of design pattern which is lost in the source code or in the design charts in the system. It can assist in the understanding of the system. More importantly, it also helps to trace back to the original design decision.This master thesis presents that the use of intelligent algorithm of the decision tree is extracted design pattern instances from object-oriented software system and implements a tool for the method.The tool training source system of JUnit3.8.2and generates a rule of design pattern. Then the use of the design pattern rule is to extracte design pattern instance from source system of JhotDraw、JRefacy、Java.awt. The experiment result shows that decision tree can extract not only the standard design pattern instances, but also variants in design pattern instance.Even it can extract a new design pattern.Meanwhile, the master thesis proposes a data preprocess system. One of the main obstacles of applying decision tree for design pattern extraction is the difficulty of collecting training examples.Because a design pattern instance typically includes a group of classed with certain relationships. Thus, the possible combinations of the group of classes can be enormous which results in huge training sets making the application of decision tree impracticable. In the master thesis, we propose a data preprocess method using matrix transformations to cluster the training examples.Also, the preprocess system optimize samples.Finally, the master thesis proposes a method estimates the result of design pattern extracted from source system. The method is called design pattern of benchmark.