Peer Review Data Warehouse: Insights From Different Systems

Peer assessment is widely used at all levels of education. Students give and receive feedback from their classmates, and thereby produce a wealth of information that can potentially be used to improve the assessment process. But thus far, each online peerassessment system has been an entity unto itself. There has been no attempt to compare the approaches taken by such systems, for example, the rubrics or the structuring of the assessment process. Our PeerLogic project is an attempt to change that. We are constructing a data warehouse of millions of peer reviews, from at least half-a-dozen systems, that can be mined to determine how differences in the assessment processes translate into differences in peer assessments. This paper reports on some of the issues that arise in the construction of the warehouse, and how we have resolved them in a way that will work for all constituent systems. We also presented an example of comparing data coming from two systems that are based on rating and ranking.