摘要：Learning analytics is focused on the educational challenge of optimizing opportunities for meaningful learning. Assessment deeply influences learning, but at the same time data about assessment are rarely considered and utilized by learning analytics. Current approaches to analysis and reasoning about peer-assessment lack rigor and appropriate measures of reliability assessment. Our paper addresses these issues with a geometrical model based on the taxicab geometry and the use of the scoring rubrics. We propose and justify measures for calculation of the final grade in peer-assessment and related inter-rater and intra-rater reliability measures. We present and discuss a geometrical model for two important peer-assessment scenarios.
