Temporal Data Mining: Similarity-Profiled Association Pattern

Abstract

Temporal data are of increasing importance in a variety of fields, such as financial data forecasting, Internet site usage monitoring, biomedicine, geographical data processing and scientific observation. Temporal data mining deals with the discovery of useful information from a large amount of temporal data. Over the last decade many interesting techniques of temporal data mining were proposed and shown to be useful in many applications. In this article, we present a temporal association mining problem based on a similarity constraint. Given a temporal transaction database and a user-defined reference sequence of interest over time, similarity-profiled temporal association mining is to discover all associated itemsets whose prevalence variations over time are similar to the reference sequence. The temporal association patterns can reveal interesting association relationships of data items which co-occur with a particular event over time. Most works in temporal association mining have focused on capturing special temporal regulation patterns such as a cyclic pattern and a calendar scheme-based pattern. However, the similarity-based temporal model is flexible in representing interesting temporal association patterns using a user-defined reference sequence. This article presents the problem formulation of similarity-profiled temporal association mining, the design concept of the mining algorithm, and the experimental result.