MOOC and Mechanized Grading

As many others, we too are developping a Massive Online Open Course or MOOC. This MOOC will teach recursive programming to beginners and will heavily use an already existing infrastructure for mechanical grading (Queinnec, 2010). This position paper discusses how these two components are combined in order to increase students’ involvement. Developping a MOOC is now a common activity in the Academia and so are we doing. This paper is a position paper that presents the main characteristics of our MOOC: it makes an heavy use of an infrastructure to mechanically grade students’ programs. How we intend to combine our MOOC with that infrastructure and how we want to create incentives for the students in order to increase their involvement is addressed in this position paper. Programming exercises are proposed in between the videos of the course. These programming exercises are mechanically graded and grading reports are sent back to the students. Reading grading reports allows students to evolve their programs but additional incentives may better help students. They may program in pair: video-chatting around a shared program, that both students can edit, recreates conviviality. Epsilon-better peeping propose to peep slightly better programs from other students’ and learn from them. Finally, recommending programs that were helpful, benefit to all. Section 1 presents the main lines of our MOOC and Section 2 presents the grading infrastructure while its new features appear in Section 3. Proposed incentives are described in Section 4 and Section 5 concludes this position paper. 1 PREPARATION OF A MOOC We are currently developping a MOOC on recursive programming. The e-learning part is based on a course created in 2000 (Brygoo et al., 2002) and since then delivered every year at UPMC to hundreds of young scientific students as an introduction to Computer Science. The course material was, from 2000 to 2003, provided as a physical CDrom then, from 2004 to 2006, as a CDrom image (300MB) downloadable from UPMC web servers. An innovative characteristics of these CDroms was that they contained a programming environment (based on DrScheme (Felleisen et al., 1998)) with a local mechanized grader, see (Brygoo et al., 2002) for details. The students could then read the course documents, write programs, run them and be graded without requiring an Internet connection. This new MOOC, adequately named “Programmation récursive”, is an endeavour to extend this course to a broader French-speaking audience, to experiment with the aspects related to social networks and, finally, to collect and study students’ answers to the proposed exercises in order to build appropriate error taxonomy and thus better future editions of the MOOC. Of course, the context has dramatically evolved from 2000 to now. The mechanical grader of 2000 which was grading Scheme programs (Scheme is the programming language used by that course) has now evolved into a multi-language grading infrastructure running in the cloud (Queinnec, 2010). The MOOC uses CourseBuilder (Google, 2013), is hosted on Google App Engine for elasticity, uses YouTube for video streaming and a Google group for forum. The programming environment, named MrScheme is now provided as a Scheme interpreter, written in Javascript by Frédéric Peschanski and his colleagues (Peschanki, 2013). This programming environment runs locally in students’ browsers. With help of MrScheme, students can http://programmation-recursive-1.appspot.com