Research On Peer Assessment Mechanism In Massive Open Online Courses

With the rapid development of Massive Open Online Courses(MOOC)in the world,peer assessment became not only an effective means for large-scale evaluation of subjective assignment,but also an inevitable requirement for massive learning and assessment.At present,optimizing the existing peer assessment mechanism and assignment assessment strategies to improve its accuracy are the big challenges and characteristics of research facing MOOC,which are both of theoretical research and practical application significance.The thesis analyzed the existing peer assessment mechanism and its effect,and some research work being carried out consist in the following three aspects: first,optimizing the peer assessment mechanism by bringing in the preparatory training before peer assessment and mutual supervision,and also by adopting the experimental methods of A/B test to analyze the optimized results;second,having designed a calculation method for peer assignment assessment based on distribution of hybrid parameter and then adjusted the scores deviation of peer assessment and compared it with the existing calculation method;based on the “MengKe” platform,using two approaches mentioned above comprehensively to optimize the results from peer assessment of Cyberspace Security of 2016 and a contrastive analysis was done between the results of 2016 and the results from 2013 to 2015.The analysis shows that the assessment accuracy of scores from trained students was improved by 6.02% compared with that direct from untrained peer assessment,and after algorithm adjustment,the consistency of scores from peer assessment and scores from teachers is improved by 12.5%.Therefore,the accuracy of peer assignment assessment is increased.In the future research,other factors,such as the assignment allocation strategy and the engaged time on peer assessment will be taken into consideration comprehensively to further optimize the peer assessment mechanism,and improvement will be done to prior distribution and parameter of deviation and reliability to optimize the evaluating strategies in the pursuit of an improved accuracy in peer assessment.