Sequential patterns graph and its construction algorithm

Sequential pattern mining is an important data mining task with broad applications that include the analysis of customer behaviors, web access patterns, process analysis of scientific experiments, prediction of natural disasters, treatments, drug testing and DNA analysis, etc. Agrawal and Srikant first introduced the sequential pattern mining problem. Over the last few years considerable attention has been focused on the achievement of better performance in sequential patterns mining but there is still the need to do further work in order to improve results achieved so far. Questions that are usually asked with respect to a better performance in sequential pattern mining are: What is the inherent relation among sequential patterns? Is there a general representation of sequential patterns? We propose a novel framework for sequential patterns called Sequential Patterns Graph as a model that can be used to represent relations among sequential patterns. This model has features that: (i) it can be used to represent all the sequential patterns mined in a mining task; (ii) it is the foundation of structural knowledge from which other new patterns can be obtained. Based on this model, a novel concept, Post Sequential Patterns, is proposed that involves graph patterns composed of sequential patterns, branch patterns and iterative patterns. The properties and construction algorithm of SPG are presented also.