Testing-Based Automatic Grading: A Proposal from Bloom´s Taxonomy

The test-first approach has been repeatedly advocated to learn both programming and software engineering. Besides, automatic graders are broadly used, both in programming contests and in everyday teaching. The sort of problems stated in both situations is always placed at the application and higher levels of Bloompsilas taxonomy, and it demands some degree of mastery on students. In this paper we address the combined use of automatic grading and the test-driven approach from a pedagogical view. In particular, we address assessment at the levels of comprehension, application, analysis and synthesis in Bloompsilas taxonomy. We make a proposal, based on sorts of exercises and suggested examples. We also gather a set of features that exercises should include as metadata to facilitate its use from pedagogical criteria.