关键词: 教育考试, 大数据, 分层过滤, 二次校正, 评价分析

Abstract: To increase the efficiency and accuracy of test data analysis and evaluation, a second revision based analysis and evaluation method for education and test data was proposed. The positive and negative attributes of the test were defined, and a gradient evaluation rules was constructed. Based on the defined attributes and evaluation rules, a novel spectral clustering algorithm was presented to specify the data into different clusters. Then, the second revision based analysis and evaluation method was implemented on the different clusters to revise the scores of the testers. This second revision method could find the special groups whose scores were low while their real ability were excellent. A real analysis and evaluation platform was developed based on cloud technology. The practical applications from more than 100 cities in 13 provinces showed that this method can increase efficiently of the test data obtaining, storage, analysis, and evaluation for the education management department. This platform has owned more 40% market share of the education and test data services in china.

Key words: Education and Test, Big Data, Layered Filter, Second-Time-Revision, Analysis and Evaluation