Dataset of Program Source Codes Solving Unique Programming Exercises Generated by Digital Teaching Assistant

This paper presents a dataset containing automatically collected source codes solving unique programming exercises of different types. The were generated by the Digital Teaching Assistant (DTA) system that automates massive Python course at MIREA—Russian Technological University (RTU MIREA). Source small programs grouped type solved task can be used for benchmarking code classification and clustering algorithms. Moreover, data training intelligent program synthesizers or mutation testing frameworks, more applications are yet to discovered. We describe architecture DTA system, aiming provide detailed insight regarding how why was collected. In addition, we algorithms responsible analysis in system. These use vector representations based on Markov chains, compute pairwise Jensen–Shannon divergences programs, apply hierarchical order discover high-level concepts students while tasks. proposed approach incorporated into courses when there is need identify approaches implemented students.