In the last years the datasets available have grown tremendously, and the development of efficient and scalable data mining
algorithms has become a major research challenge. However, since the data is more dynamic than static there is also a strong
need to update previously discovered rules and patterns. Recently, a couple of studies have emerged dealing with the topic
of incremental update of discovered knowledge. These studies mostly concentrate on the question whether new rules emerge or
old ones become extinct.
We present a framework that enables the analyst to monitor the changes a rule may undergo when the dataset the rules were
discovered from is updated, and to observe emerging trends as data change. We propose a generic rule model that distinguishes
between different types of pattern changes, and provide formal definitions for these. We present our approach in a case study
on the evolution of web usage patterns. These patterns have been stored in a database and are used to observe the mining sessions
as snapshots across the time series of a patterns lifetime.