基于用户反馈的关联规则挖掘在试卷分析中的应用

传统试卷分析系统一般只是对考试成绩做简单的整体统计,如平均分、等级、不及格率等,用户缺乏对试题知识点掌握情况关联程度的准确了解。针对该问题提出一种基于用户反馈的关联规则挖掘算法。首先对原始数据进行预处理,得到试卷知识点评分权重表和二进制的学生得分率表。然后建立一个根据用户选择层数输出关联规则,以及查询与选定知识点相关的规则的新方案。最后,提出了一个考虑用户反馈、支持度和置信度阈值的关联规则挖掘算法,以过滤无用规则,提高挖掘效率。对VB试卷数据应用该算法,发现了基于用户反馈的有趣关联规则。实验结果表明基于用户反馈的关联规则挖掘优于其他关联规则挖掘算法,更易获得有趣的关联规则。

User Feedback Based Association Rule Miningwith Application to Examination Paper Analysis

Traditional examination paper analysis system was generally only to do a simple test scores overall statistics, such as the average score, grade, failure rate, etc.. Users lacked accurate understanding on the mastered situation and degree of relevance of the examination knowledge point. A user feedback based association rule mining algorithm was proposed to solve this problem. First, original data were preprocessed to a knowledge point scoring weight table for the examination paper, and a binary scoring rate table for students. Then, this paper designed a new scheme which outputted the association rules according to the number of layers users selecting, and queried the relevant rules according to the knowledge point of user"s choice. Finally, an association rule mining algorithm based on user feedback, support degree and confidence threshold was proposed to filter the useless rules and improve the efficiency of mining. It applies the algorithm to the VB test paper, and finds the user feedback based interesting association rules. The experimental results show that the association rules mining based on user feedback is better than other association rules mining algorithm and easier to get interesting association rules.