Abstract

Sequences of events arise in many applications, such as web browsing, e-commerce, and monitoring of processes. An important problem in mining sets of sequences of events is to get an overview of the ordering relationships in the data. We present a method for finding partial orders that describe the ordering relationships between the events in a collection of sequences. The method is based on viewing a partial order as a generative model for a set of sequences, and applying mixture modeling techniques to obtain a descriptive set of partial orders. Runtimes for our algorithm scale linearly in the number of sequences and polynomially in the number of different event types. Thus, the methods scales to handle large data sets and can be used for reasonable numbers of different types of events. We illustrate our technique by applying it to student enrollment data and web browsing data.