Discovering frequent episodes in sequences extended abstract
Heikki Mannila, Hannu T. T. Toivonen, A. Inkeri Verkamo · 1995
Sequences of events describing the behavior and actions of users or systems can be collected in several domains. In this paper we consider the problem of recognizing frequent episodes in such sequences of events. An episode is defined to be a collection of events that occur within time intervals of a given size in a given partial order. Once such episodes are known, one can produce rules for describing or predicting the behavior of the sequence. We describe an efficient algorithm for the discovery of all frequent episodes from a given class of episodes, and present experimental results. Introduction Most machine learning and data mining techniques are adapted towards the analysis of unordered collections of data. However, there are important application areas where the data to be analyzed has an inherent sequential structure. For instance, in telecommunications network monitoring or empirical user interface studies it is easy to log a lot of information about the behavior and actions ...