Research And System Realization Of Automatic Judgment Of Program With Dynamic And Static Method

In the online examination system for subjective programming usually adopts dynamic judging: compile and run the program with using the test case set in advance.However,dynamic judging means to be compiled in case of program can give the corresponding scores;it is not objective for program evaluation.Therefore,through the research of the automatic judging technology program and the analysis of teachers' manual review process,we put forward a dynamic program judging method which combines static and automatic marking method.The method use program's value(dynamic)and contents(static)two aspects to evaluate the program,it makes the judging result of the program more objective,accurate and comprehensive.We proposes the static method because it needn't compile the program.However we can't provide all standard programs,so we judge the program by the combination of dynamic and static method.The method will be more accurate and comprehensive based on the output and the code of the program.The work of this paper is as follows:Firstly,we get statements which are free of lexical and syntax error.We use flex and bison to finish this task.Secondly,we compute the similarity between the student program and the standard program.The computation process has four steps: pretreatment,standardization,intermediate representation and similarity algorithm.During the pretreatment stage,we remove the part which have no influence of the structure of program,for example,the annotations,successive lines,consecutive spaces.In order to reduce the diversity of the program,we put forward some standardization rules.Choose control flow graph as the intermediate representation and get the deep firstly traverse of control flow graph.Then we calculate the LCS(Longest Common Sub-sequence)based the traverse of the student program and the standard program.Further more,we provide a similarity algorithm according to the traverse of student program,the traverse of standard program and LCS.Thirdly,we develop an online exam system.The system contains three roles:administrator,teacher and student.It can judge choice questions,true-false questions and program questions automatically.The system uses the dynamic and static method assess program questions.Through the analysis of scores in the manual method,dynamic method and the combination of dynamic and static method,we proves that the combination of dynamic and static method is closer to manual method.The combination of dynamic and static method makes program assessment more objective and reasonable,it reduces the workload of teachers,and has more important use certain reference for the research of intelligent assessment technology of program.