对比模式挖掘研究进展

对比模式挖掘是数据挖掘的一个重要和集中的子领域,主要涉及数据集的模式挖掘和对比处理。它的目的是寻找有趣的对比模式,描述满足各种不同条件的显著差异的数据集。对比的条件可以在类、时间、位置、或其他“维”中定义,当然也可以在他们的组合中定义。对比模式可以代表类之间的不同差异,随时间推移的有趣的变化或者空间趋势变化等等,通过分析两类或多类样本中的对比信息能够得到新的未知信息。对比模式挖掘发展至今,已有了众多的相关技术和算法,在许多领域得到有效应用。本文对现有的对比模式挖掘技术进行了全面的解读,其中包括了它的背景介绍、基本概念、技术算法、相关应用、研究展望等内容,能够为对该方向感兴趣的研究者提供详尽参考。

Research Progress in Contrast Pattern Mining

Contrast pattern mining is an important and focused subarea of data mining, it mainly deals with pattern mining and comparison processing of data sets. Its aim is to find interesting contrast patterns that describe significant differences between datasets satisfying various contrasting conditions. The contrasting conditions can be defined on class, time, location, other “dimensions” of interest, or their combinations. The contrast patterns can represent nontrivial differences between classes, interesting changes over time, interesting trends in space and so on,new and unknown information can be obtained by analyzing the contrast information in two or more types of samples. Contrast pattern mining has been developed so far that there are many related techniques and algorithms, which have been applied effectively in many fields. This paper makes a comprehensive interpretation of the existing contrast pattern mining techniques, including its background introduction, basic concepts, technical algorithms, related applications, research trends and other content, which can provide a detailed reference for researchers interested in this direction.