Quality assessing algorithm of solving programming problems

Abstract

The possibility of automating the qualitative assessment of solutions to programming problems in teaching secondary school students is considered. The currently existing methods of evaluating the solution of programming problems are considered, their advantages and disadvantages are highlighted. Based on the considered methods, the possibility of using the representation of the source code of the problem solution in the form of a semantic network for its further qualitative evaluation by machine methods is substantiated. The concept of the similarity function of two semantic web based on the program code is introduced. This function is defined as the ratio of the sum of the sizes of subgraphs composed of common elements of semantic webs to the sum of the sizes of thise semantic webs. The properties of the resulting function based on its definition are given, and an algorithm for calculating the similarity function created on the basis of the ISD algorithm. Various modifications of the specified algorithm and influence of each of them on the final result are considered. To test the algorithm's operability, an experiment was set up comparing various solutions for the problem of finding the maximum element, as well as comparing solutions for various tasks of the basic programming course with each other. The results of the experiment are presented in a summary table and allow us to justify the chosen modification of the algorithm for calculating the similarity function. The results also demonstrate the prospects of using this method to determine the competence of students in the programming language. As a side effect, the algorithm of qualitative evaluation of the program code based on its semantic web allows us to automaticly search for plagiarism of the program solution of the problem.