MapReduce框架下近似概念格的并行构造算法.pdfVIP

第2期 微 处 理 机 No.2 2期 . . 2017年4月 MICROPROCESSORS Apr. ，2017 ·微机软件 · MapReduce框架下近似概念格的并行构造算法* 谭富林，姜 麟 （昆明理工大学理学院，昆明650500） 摘 要：具有缺值的形式背景称为不完备形式背景，相应的概念格扩展模型称为近似概念格。近 似概念格构造中，在数据规模大的情况下采用串行算法效率低，完备形式背景下概念格并行构造算 法不适用于不完备形式背景。针对这些问题，对近似概念格的特征进行深入分析，提出了在MapRe- duce框架下的两种分布式构造算法，包括一种并行合并算法和一种增量式并行算法。实验结果表明， 相比串行算法，两种并行构造算法可以提高近似概念格的建格效率。 关键词：不完备形式背景; 近似概念格；概念格构造；MapReduce框架；并行构造算法 DOI：10.3969/j.issn.1002-2279.2017.02.011 中图分类号：TP301.6 文献标识码：A 文章编号：1002-2279-（2017）02-0045-07 ParallelConstructingAlgorithm ofApproximation Concept Lattice Basedon MapReduceFramework Tan Fulin,Jiang Lin (FacultyofScience,Kunming UniversityofScience and Technology,Kunming650500, China) Abstract: Formal context with missing values is called incomplete context, and the concept lattice expansion model in incomplete context is called approximation concept lattice. In the approximation concept lattice construction, serial algorithm is low efficiency in the case of large data and parallel constructing algorithm under complete context is not appropriatefor incomplete context. In order to solve these problems, through deep analysis of the characteristics of approximation concept lattice, the paper introduces two parallel constructing algorithms of approximation concept lattice based on MapReduce framework, including a parallel union algorithm and a parallel constructing algorithm based on an incremental constructing algorithm. The experime