摘要
对大数据的频繁项集挖掘是关联规则挖掘的关键步骤,通过有效的频繁项挖掘提高大数据量数据库的访问效率。传统方法中对大数据的频繁项集挖掘采用FP-Growth的粗糙集挖掘算法,扩展性和容错性不好。提出一种基于贝叶斯粗糙集的大数据频繁项挖掘技术,引入后缀项表的概念,通过后缀项表的构建,保留频繁项集的完整信息。构建FP-Tree,生成闭频繁项集,计算样本的密度,并抽取高密度区域的点集作为聚类中心集合,进行后缀项表的构造,按支持度分成若干集合,对各约简集内的属性集合进行融合,用变精度粗糙集的贝叶斯粗糙进行数据挖掘算法改进,仿真结果表明,算法不受可变参数的影响,鲁棒性较高,数据挖掘的准确度较高,运行时间较短。算法将在人工智能和数据挖掘领域具有更广的应用前景。
The frequent itemsets on data mining is a key step of association rule mining, through frequent item mined effec?tively, it can improve the access efficiency of large quantities of data database. The rough set algorithm for mining frequent item sets in the traditional method on data mining using FP-Growth, scalability and fault tolerance is not good. Put forward a kind of data mining technology based on large Bayesian Rough set of frequent items, introducing the concept of suffix ta?ble, by constructing a suffix table, complete information remain frequent item sets. Construction of FP-Tree, the generation of closed frequent itemsets, calculate the sample density, and extract the regions of high density point set as the clustering center, constructed suffix table, according to the degree of support is divided into a plurality of sets, attribute of each reduc?tion set within the set of fusion, using Bayesian variable precision rough sets rough data improved data mining algorithm, simulation results show that the algorithm is not affected by the impact of variable parameters, high robustness, data mining is of high accuracy and short running time. The algorithm will have more wide prospect of application in the field of artifi?cial intelligence and data mining.
出处
《科技通报》
北大核心
2015年第6期211-213,共3页
Bulletin of Science and Technology
基金
四川省教育厅自然科学基金No.13ZA0136
关键词
贝叶斯粗糙集
频繁项挖掘
大数据
Bayesian Rough Set
frequent item mining
large data