【作者】 李少年；

【导师】 孟志青；

【作者基本信息】 湘潭大学 ， 计算机应用技术， 2003， 硕士

【摘要】 随着商务活动日趋频繁和企业信息化程度的提高，有越来越多的信息积累，其中大部分均以时态数据形式存在。这样，时态数据挖掘作为数据挖掘的一个新的课题出现，引起了人们极大的兴趣。国内外时态数据挖掘的研究起步不久，面向时态数据的挖掘算法目前效果还不是非常理想，还存在许多问题亟待解决。本文对此进行了研究。 首先，我们介绍了数据挖掘有关的概念、技术和研究现状，论述了时态数据挖掘的研究背景和进展，给出了本文研究的内容。 然后，介绍了时态型、时态因子和时态粒度的基本概念和性质，给出了一种时态关联规则的数学模型，并对给出了几个具有实际意义的时态关联规则。 其次，讨论了单事件相同时态、周期时态内关联规则挖掘的两个算法，并给出了相应的试验结果。同时针对双事件时态关联规则挖掘提出一个增量算法，并给出了实验结果。另外还给出一个基于兴趣度的关联规则挖掘算法和试验结果。 最后，我们论述了Markov Chain在时态数据挖掘中的应用。 本文所获得的主要成果为：1、给出了一种时态关联规则的数学模型，2、提出了单事件相同时态、周期时态内时态关联规则挖掘的两算法，以及双事件时态关联规则挖掘的一个增量算法和一个基于兴趣度的时态关联规则挖掘算法。更多还原

【Abstract】 With the frequent interaction of business and the elevation of corporation’s information degree, more and more information are being accumulated, most of which exist as a form of temporal data. Therefore, temporal data mining (TDM) becomes a very interesting field of data mining. The study of temporal data mining is at its early stage, till now the effect of the arithmetic for temporal data mining is not idea. There exist many problems that need to be studied, therefore we discuss TDM in this paper.First, we discuss the background and development of TDM, and describe the researching content in this paper.Second, we introduce the basic concepts about temporal type, temporal factor and temporal granularity, propose a kind of mathematical model about temporal association rules, and describe several temporal association rules with practice significance.Third, we discuss the temporal association rules mining algorithm of single event during the period of same temporal factor, cycle temporal factor and analyze the experiment results. To mine temporal association rules in two events, we propose a new increment algorithm. Moreover, their experiment results are given. We propose a new association rules mining algorithm based interest measurement, and give the experiment results.Finally, we discuss applications of Markov Chain in TDM.In this paper, we obtain these results: 1 a kind of mathematics model of temporal association rules; 2 propose an algorithm of single event during the period of same temporal factor, an algorithm of cycle temporal factor, an increment algorithm about temporal association rules mining of two events and an algorithm based interest measurement.更多还原