Background: Know: causality graph Recognize:
Reducing a causality graphEdit
The causality graph may include information that does not contribute to correlation analysis. For example, a cycle represents causal equivalence. A cycle of events may thus be aggregated into a single event.
Similarly, certain symptoms are not directly caused by any problem but only by other symptoms. They do not contribute any information about problems that is not already provided by these other symptoms that cause them. These indirect symptoms may be eliminated without loss of information.
Henceforth, we will assume that a cauality graph has been appropriately pruned.In the figure on left hand side shown below, symptoms 3, 4 and 5 are in a cycle and thus can be aggregated into a single event. Similarly, symptoms 7 and 10 are not directly caused by any problem rather, they are caused by other symptoms, therefore these tqo nodes can also be eliminated. Node 8 in the figure is neither symptom nor problem and it too can be ignored and eliminated. The figure on right hand side is the simplified graph.