关联规则挖掘算法及其在Web挖掘上应用的研究-计算机软件与理论专业论文.docxVIP

哈尔滨工程火学硕士学位论文摘 哈尔滨工程火学硕士学位论文 摘 要 由于应用性的数据的爆炸式增长(如商业分析)，及计算机的软硬件技术 的发展完备，数据挖掘已经成为近几年来应用最广泛的分析数据的工具。挖掘 重要数据仍然需要配合许多其他领域的技术才能得到完善有效的结果，其中 包括机器学习，人工智能，统计学原理，数据库系统，数据可视化等。然而目前 大多数挖掘关联规则的算法往往必需多次扫描事务数据库才能达到要求的目 标，这样重复性的数据库存取动作将会导致过多的执行时间浪费在I／O动作 上另外在频繁集生成规则时现有的算法没有考虑到规则的大量冗余，为了解 决这个问题本文提出了一个不需要生成候选集同时有效的去除生成规则时产 生大量冗余的算法QAIS，来提高关联规则生成的速度，并且在此算法基础上 提出了新颖的关联规则增量更新算法AIU，通过应用合成数掘验证了 QAIS／AIU算法的有效性，由试验结果来看这个方法确实能更有效且准确地获 得事务数据库的关联规则，尤其适合挖掘超大数据库中高支持度下长模式的 挖掘。然后论文介绍了关联规则的一种扩展一时间序列模式挖掘方面的算法， 给出了Web同志文件挖掘的高效、新颖的改进算法，并用实验验证了此种方法 的有效性。论文的最后讨论了关联规则挖掘领域面临的几个问题。 关键词：数据挖掘：知识发现；关联规则；增量更新；时间序列模式；Web 挖掘 哈尔滨工群人学硕士学位论文AB 哈尔滨工群人学硕士学位论文 AB STRACT Data mining has been becoming more and more popular in the past few years due tO the growing demands of database application and the advances in computer technology．Data Mining merges many important research fields including machine learning，artificial intelligent，statistics，knowledge—base systems and data visualization，etc．However，current algorithms proposed for datA mining of association rules require several passes over the analyzed database The I／0 overhead in scanning the large database can be extremely high。 An ef蠡cient algorithm QAIS iS proposed that uses the ef五cient method to reduce database access activity,and present a novel algorithm AIU based on this algorithm．it iS fit for mining association rules and incremental updating．It iS especially effective in VLDB，mining long patterns，and high support．The Performance of QAIS／AIU iS verified on the basis of synthetic dataj experiments show that the proposed algorithm can mine association rules more efficiently by not generating candidate itemsets and reducing the redundancy ol、fiequent itemsets while generating association rules． And then，a extent of association rules—Time Series Pattern iS discussed， and an efficient improved algorithm of web mining is presented．The Performance of algorithms iS verifled on the basis of synthetic data too． At last，the problems faced in the mining association rule