基于数据流的频繁模式挖掘应用研究-计算机应用技术专业论文.docxVIP

万方数据 万方数据 摘要 摘 要 数据流是一种海量并潜在无限的、连续快速的、并随着时间快速变化的有序数据 序列。由于数据流本身的特性，使得传统的频繁模式算法难以适用。作为数据挖掘研 究领域的热点之一，频繁模式挖掘已成为数据流挖掘中的基础任务，并得到了广泛的 深入研究和应用。 针对数据流特点，论文概述数据流挖掘的研究背景及主要任务，并对数据流现状、 处理模型、处理技术和关键问题进行了分析研究，并对一些现存的数据流频繁模式挖 掘经典算法进行了介绍和总结。 在此基础上，本文提出了一种基于数据流的实时近似频繁项挖掘算法 ECP-Stream 和基于滑动窗口的数据流闭合频繁项集挖掘算法 WHCFP-Stream。ECP-Stream 算法使 用 ECD ? tree 结构保存数据流中频繁项，随着数据的流入不断更新结构，同时提高删 除过期数据的时间效率，并保证输出数据的精度，满足在线实时处理的分析要求。理 论分析和真实数据及模拟数据实验结果表明该算法具有很好的效率。基于滑动窗口模 型的 WHCFP-Stream 算法采用 WHCFP-tree 结构保存数据流中的闭合频繁模式，随着数 据流的流入，采用深度优先遍历删除过期数据，并简单快速的插入新的闭合项集。理 论分析和真实数据实验表明了该算法的有效性。 关键词：数据流，数据流挖掘，滑动窗口，频繁项，闭合频繁项集 —I— 基于数据流的频繁模式挖掘应用研究 Abstract A data stream is unlimited potential sequence of data which arrive at a high speed and change unceasingly with the time. The features of data stream itself make the traditionary algorithm be hard to mining the frequent items. As one of researching hot area of data mining, frequent pattern mining has been a basic task of data mining and been researched widely and profoundly. According to the feature of data stream, the paper summarizes the research background and the main task of data stream,and researches and analyses the status of data stream, processing models, technologines and key issues, also introduces and summarizes some classic frequent pattern mining algorithms. On this basis, the paper proposes a real time algorithm ECP-Stream for mining approximate frequent item over data stream and a algorithm WHCFP-Stream for the closed frequent items based on the data stream sliding window. The algorithm ECP-Stream uses the ECD-tree structure to save the data stream frequent item, and constantly updated with the data structure, and improves the time to delete outdated data, also ensures data accuracy. Theoretical analysis and experimental results show that the algorithm has a good efficiency. The WHCFP-Stream algorithm Based on sliding window model uses WHCFP-tree structure to save the closed frequent itemsets, with the data stream flows, using depth-first traversal