大型数据库中利用强化学习改进treap的关联规则挖掘算法  

Association rule mining algorithm using improving treap with interpolation algorithm in large database

作　　者：辛春花 郭艳光 鲁晓波[1] Xin Chunhua;Guo Yanguang;Lu Xiaobo(Dept.of Computer Technology&Information Management,Inner Mongolia Agricultural University,Baotou Inner Mongolia 014109,China)

机构地区：[1]内蒙古农业大学计算机技术与信息管理系,内蒙古包头014109

出　　处：《计算机应用研究》2021年第1期88-92,共5页Application Research of Computers

基　　金：国家自然科学基金资助项目(31660602,31660701,31960361);内蒙古自然科学基金资助项目(2017BS403);内蒙古自治区高等学校科学研究项目(NJZY20055)。

摘　　要：信息的爆炸式增长使数据挖掘分析过程更加困难,针对普通关联规则挖掘算法很难在短运行时间和低关联度的前提下完成大型数据库中变量关系的评估和发现的问题,提出利用强化学习算法改进treap的大型数据库关联规则挖掘算法。提出的算法首先计算数据库中每个变量的优先级;然后,在优先级模型中利用强化学习算法改进的build-treap程序构建treap数据结构;最后,通过遍历程序和generateRule程序完成数据库中所需的关系查找。在对提出的算法进行稳定性分析后进行了仿真验证实验,实验结果表明,提出的算法在其最次和最佳案例分析中分别能够完成O(n log n)次和O(n 2)次挖掘,能够在较短时间内完成低关联度的大型数据库中变量关系挖掘任务,相对于改进型Apriori算法和改进型FP生长算法有较大提升。The explosive growth of information makes the process of data mining and analysis more difficult.It is very difficult for the common association rules mining algorithm to evaluate and discover the relationship between variables in large database under the premise of short running time and low correlation degree.This paper presented an algorithm for mining association rules in large databases based on improved treap.Firstly,the algorithm calculated the priority of each variable in the database.Then,it constructed the treap data structure by the interpolation algorithm to improve build-treap program in the priority model.Finally,it found the relationship of the database by traversing the program and generateRule program.After the stability analysis of the proposed algorithm,the simulation results show that the proposed algorithm can mine the O(n log n)times and O(n 2)times in the worst-case analysis and the best-case analysis,respectively.The algorithm can complete the task of variable relational mining in a large database with low correlation degree in a short time,which is much better than the traditional Apriori algorithm and FP growth algorithm.

关 键 词：改进型treap算法 强化学习算法 大型数据库 优先模型 关联规则 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]