CAJ | 학술논문

针对传统的Apriori算法在执行过程中存在着需要扫描多次数据库，执行效率低和需要大量的内存来存储候选项集的缺点，有人已经提出了引入索引结构的Apriori改进算法，但是该算法还是传统的Apriori算法都不适合应用到云计算平台。因此，在本文中我们提出了将引入索引结构的Apriori算法的设计思想应用到Apriori算法MapReduce并行化改进的过程中。该算法的核心思想：将数据进行MapReduce并行化分块，并在每个分块中采用索引的执行模式。该算法结合了索引结构的优点和算法并行化的优点，不仅大大提高了算法的执行速度，而且实现了在云环境下的应用。
침대전통적Apriori산법재집행과정중존재착수요소묘다차수거고，집행효솔저화수요대량적내존래존저후선항집적결점，유인이경제출료인입색인결구적Apriori개진산법，단시해산법환시전통적Apriori산법도불괄합응용도운계산평태。인차，재본문중아문제출료장인입색인결구적Apriori산법적설계사상응용도Apriori산법MapReduce병행화개진적과정중。해산법적핵심사상：장수거진행MapReduce병행화분괴，병재매개분괴중채용색인적집행모식。해산법결합료색인결구적우점화산법병행화적우점，불부대대제고료산법적집행속도，이차실현료재운배경하적응용。
Aiming at the shortcomings of traditional Apriori algorithm existed in the implementation process, such as needing to scan databases many times, low efficiency, and needing a lot of memory to store the candidate set, some people has proposed the improved Apriori algorithm introducing index structures, however, both this algorithm and traditional Apriori algorithm is not suitable for application to a cloud computing platform. Therefore, in this paper we propose to apply the design idea of the Appriori algorithm introducing index structure into the Apriori algorithm to improve MapReduce to parallelize the process. The core idea of the algorithm: The data are MapReduce parallelized blocking, and the index execution mode is used in each block. The algorithm combines the advantages of index structures and algorithms parallelization, not only greatly improving the execution speed of the algorithm, but also realizing the application in the cloud environment.