INGI Seminar

April 30, 2019



Nyquist Room - Maxell building a.164

How to (not) use an autograder: lessons learned after 5 years using INGInious

by Guillaume Derval, researcher at UCLouvain

Autograding code exercises is a difficult task. Most of us have used autograders such as INGInious to give some parts of their courses, and have faced multiple difficulties.

Some of them have been solved, some other not, but we never took the opportunity to share these experiences. This talk aims at sharing these various lessons learned.

We will discuss how to make an engaging course, what tools can be used to simplify the creation of an exercise, and what are the mistakes to avoid, from a « learning » perspective but also from a more technical point of view.

The talk will be followed by a discussion on how we can further improve our way of creating autograded exercises and how we can improve our tools (i.e. INGInious).

Guillaume is a third-year PhD student here at UCLouvain. His work is focused on maximal-sum submatrix mining, which lies in the intersection between combinatorial optimization and data mining. Prior to the start of his thesis, Guillaume was one of the main developers of INGInious, which he still helps to maintain during its free time.