Web-based system for automatic evaluation of java algorithms

Automatic evaluation of source code can be an important mechanism for learning programming languages. Several tools and systems perform automatic evaluation of source code through the use of input/output tests, but very few make use of software engineering metrics. Considering the limitations of current work, we present in this paper a 4-stage approach for the automatic evaluation of source code: i) the source code is compiled and an error-check is performed; ii) the compiled code is tested against a set of JUnit (see IV-A) tests provided by the teaching staff; iii) a set of software engineering metrics are used to compare the student´s solution with the teacher´s solution; iv) and finally, based on the results of the previous stages, automatic feedback is provided to the student in order for them to identify the areas in which they need further study. At the end of the paper, we present the results of the system´s evaluation that show that our approach can facilitate the learning and knowledge acquisition processes of the student while learning a programming language.