Research On Online Judge Method Of Program Based On Web

The development of computer technology has made it more and more important in teaching.It has become an indispensable teaching method and brings great convenience to teachers and students.Among them,the research of program intelligence evaluation method has been a hot research direction.The research of program similarity detection method has far-reaching significance in application fields such as program plagiarism,software property protection,code reuse,and software system analysis.In this paper,the online intelligent evaluation of the program is realized through the study of the intelligent detection method of program similarity,and it is extended to the similarity detection method of large-scale code.The study in this paper is divided into two parts:Aiming at the problems such as poor flexibility,poor adaptability,and lack of intelligence in current detection methods of similar programs,this paper proposes a program similarity detection method based on semantic analysis,which provides new ideas and solutions for intelligent evaluation of programs.From the perspective of program semantic analysis,this paper establishes a program intelligence scoring model based on the human evaluation process.The whole intelligent scoring process is divided into three steps: First,the program is preprocessed by standardized rules;then,we improve the representation of program dependency graphs and the student program and the template answer program are converted into the intermediate representation of the program dependency graph;finally,from three aspects of the program structure matching,the program statement matching and the program knowledge point matching to analyze the students' program semantics,we get the result of the program similarity detection,and then give the students a reasonable and accurate evaluation result.The experimental results show that the method proposed in this paper can quantitatively evaluate the correctness of the student program through semantic analysis,and only has a small difference from the manual evaluation result which achieves better results.In view of the complexity and high cost of the large-scale code similarity detection algorithm,this paper extends the research on the intelligent detection method of program similarity,and proposes a method of combining program dependency graph and token string to realize similarity detection of large-scale code.First,the program dependency graph is parsed and the large-scale code is divided into several code blocks.Each code block is composed of token strings.Then,token sort filtering algorithm and token position filtering algorithm are established to reduce the size of the candidate sub-block index and reduce the number of comparisons between the candidate sub-blocks.Finally,similarity detection algorithm is used to obtain large-scale code similarity detection results.Through experiments show this method has a good effect on the similarity detection of large-scale code from the aspects of execution time,recall rate and accuracy.