基于难度系数的加权关联规则在试卷评估中的应用  

WEIGHTED ASSOCIATION RULES BASED ON THE COEFFICIENT OF DIFFICULTY IN THE ASSESSMENT OF PAPERS

作　　者：陈世保[1] 徐峰[1] 吴国凤[2] 

机构地区：[1]安徽财贸职业学院,安徽合肥230601 [2]合肥工业大学,安徽合肥230009

出　　处：《井冈山大学学报（自然科学版）》2013年第1期70-74,共5页Journal of Jinggangshan University (Natural Science)

基　　金：安徽省高等学校重点教学研究项目(20101766)

摘　　要：针对传统的关联规则在试卷评估中应用出现的问题:由于试题的难易程度不同,被答对的概率也不一样,即数据集中数据项发生的概率不一样,数据项具有倾斜支持度分布的特征,选择合适的支持度阈值挖掘这样的数据集相当棘手。文章提出了基于试题难度系数加权的关联规则挖掘算法,从而解决因试题难度不同而导致数据项出现的概率不均的问题,发现更多有趣的关联规则,并且理论上证明了基于难度系数的加权关联规则算法保持频繁项集向下封闭的重要特性。With the wide range of data mining applications,the association rule mining algorithm is applied to the paper assessment in the literature.Traditional association rule data mining problems in the papers assessment, such as the degree of difficulty of questions is different,the probability of being correct answers are not the same, that is to say,the data set is not the same as the probability of data entry,data entry with a sloping support the distribution of the characteristics of mining such data sets is very difficult to select the appropriate support threshold.We present the association rules mining algorithm based on item difficulty coefficient weighted to solve the problem of uneven frequency of data items appear different item difficulty and find more interesting association rules.Furthermore,we prove theoretically that the weighted association rules based on the coefficient of difficulty to maintain the important features of the frequent item sets is downward closed.

关 键 词：APRIORI算法 试卷评估 加权关联规则 数据挖掘 难度系数 

分 类 号：TP274[自动化与计算机技术—检测技术与自动化装置]