Table of Contents

Frequency Abstraction Miner

The Frequency Abstraction Miner is a plugin for ProM which enables to mine processes in a successive and user-definable fashion, zooming to the desired level of precision and detail. The importance of events and transition is evaluated by frequency, i.e. more frequently observed artifacts are considered more important.

Note: This plugin is retained for historic and research purposes. For actual practical applications, I strongly recommend you consider using the Fuzzy Miner, which excels the Frequency Abstraction Miner in most respects, and can be considered its mature successor.

Input

The Frequency Abstraction Miner takes any log file as input.

Interface

The result interface of the Frequency Abstraction miner, which appears directly after its being called, looks as follows:

On the left, the process graph is shown. Regular (i.e., unclustered) activities are shown as rectangular nodes, while clusters of less-important activities are shown as green octagonal nodes.

It is possible to click on clusters, which will open a dialog revealing their inner composition.

On the right, a slider allows the user to zoom into the desired level of precision and detail. After moving the slider, the graph will be adjusted automatically.

The option attenuate edges will render edges in a grey tone, which corresponds to their respective frequency.

The option merge sibling clusters will trigger similar clusters (with the same incoming and outgoing neighbors) to be merged into one, thus further simplifying the displayed model.