Towards Personalization of Peer Review in Learning Programming

Abstract

Peer review is one of the effective processes for sharing knowledge and improving overall learning performance. This became more popular by the use of ICT. However, it is challenging to implement peer review in learning programming languages due to the complexity of the subject matter. A group of peer reviewers may have different overall performance but similar weaknesses on a given aspect of the programming tasks. Hence, they may not be able to help each other to address individual needs. In this paper, we present a personalized approach to peer review with consideration to criteria based assessment and individual performance on specific programming tasks. This is achieved using a novel peer-matching algorithm to create reviewer groups. The algorithm assigns peer-reviewers in such a way that each student gets reviews from at least three peers with different levels of competence (low, medium and high). Peer matching is tailored to individual student needs with respect to specific aspects of learning programming. This work implemented a web based peer review system, and carried out user-based evaluations with computer science students. There are indications that personalized peer matching, based on relevant assessment criteria, can improve individual learning achievement in programming courses.