ILP Miner

This node implements the ILP Miner to discover a Petri net from an event table. Multiple algorithms based on transition theory and ILP are provided.

Options

Event classifer

The attribute to be used as an event classifier.

Filter Type

The filter type includes the following choices:

• NONE: no filter.
• SEQUENCE_ENCODING: the sequence encoding filter specifies at what level a branch should be cut off.
• SLACK_VAR: the slack variable filter specifies what portion of constraints should be filtered out.

Noise Threshold

Threshold for filtering out noise.

Objective Function

The objective function for ILP. The following options are supported:

• WEIGHTED_ABSOLUTE_PARIKH: weighted parikh values, using absolute frequencies.
• WEIGHTED_RELATIVE_PARIKH: weighted parikh values, using relative frequencies.
• UNWEIGHTED_PARIKH: unweighted parikh values.
• MINIMIZE_ARCS: minimize arcs.

Variable Distribution

Setting LP variable. The following options are supported:

• DUAL: two variables per event.
• HYBRID: one variable per event, two for an event which is potentially in a self loop.
• SINGLE: one variable per event.

Discovery Strategy

The Discovery Strategy. The following options are supported:

• CAUSAL_FLEX_HEUR: mine a place per causal relation (flexible heuristics miner).
• TRANSITION_PAIR: mine a connecting place between each pair of transitions.

Views

Interactive View: Petri Net

Workflows

• No workflows found

Links

• No links available

Developers