多维概念格与多维序列模式的增量挖掘

多维序列模式挖掘旨在将一个或多个背景维度信息中发现的关联模式与有序事务序列中发现的序列模式有机结合,从而为用户提供信息内容更加丰富、更具有直接应用价值的多维序列模式.目前虽有一些挖掘多维序列模式的工作,但其关联模式与序列模式的发现过程是基于不同的数据结构分开进行的.提出一种新的概念格结构——多维概念格,它是对概念格的延伸与泛化,其内涵更加丰富,不仅具有多个有序的任务内涵,而且具有多个无序的背景内涵.设计实现了基于该结构的增量式多维序列模式挖掘算法,该算法使用统一的数据模型实现关联模式与序列模式的高效同步挖掘.在合成数据集上的实验结果验证了算法的有效性.同时,算法在实际的银行数据集上的应用效果也说明了算法的实用性.

Multi-Dimensional Concept Lattice and Incremental Discovery of Multi-Dimensional Sequential Patterns

Multi-dimensional sequential pattern mining is the process of mining association rules from one or more dimensions of background information in which the order of the dimension values is not relevant, alongside mining sequential patterns from one or more dimensions of information in which the order is important. Multi-dimensional sequential patterns are much more informative frequent patterns which are suitable for immediate use. Although some work has been conducted for mining multi-dimensional sequential patterns, association patterns and sequential patterns are mined separately based on different data structures. In this paper, a novel data model called multi-dimensional concept lattice is proposed, which is the extension or generalization toward concept lattice. The intension of multi-dimensional concept lattice is more informative, which is constituted of one or more ordered task-relevant dimensions and one or more unordered background dimensions. Moreover, an incremental multi-dimensional sequential pattern mining algorithm is developed. The proposed algorithm integrates sequential pattern mining and association pattern mining with a uniform data structure and makes the mining process more efficient. The performance study on synthetic datasets shows the scalability and effectiveness of the proposed algorithm. At the same time, the application on the real-life financial datasets demonstrates the practicability of the approach.