AUTOMATION OF PERSONALIZED FEEDBACK IN THE PROGRAMMING STUDIES COURSES
The use of automatic feedback can significantly increase the success of beginners in programming, especially for those who have to study inside a large group, and the teacher’s time is limited. The article proposes an approach to the creation of automatic feedback based on previous solutions. The approach is to form a solution space of programs – weighted graph. The nodes in the graph are the program code; the edge weight is the number of changes and actions that need to be performed in order to go from one state to another. To reduce the number of unique solutions, the source code is normalized using a number of transformations and the construction of an abstract syntax tree. Feedback is a hint of the next step, which can be generated after a new solution is added to an existing graph and the path leading to a more correct state is identified. Thus, with the help of feedback you can reach the right decision. Using the solution space also allows you to find out which solutions are most common, which errors occur and which ways they can be corrected students prefer. Since this approach is based solely on data, the teacher does not need significant interaction with, which makes it scalable and adaptable.
Keywords: intelligent tutoring system, programming courses, automatic feedback, educational data mining, learning analytics.