The similarity index defines how rare/common a data combination is. The similarity is measured by counting how many records occurred with a given combination in the model data table. The highest combination gets a similarity value of 1. One of the conditions for triggering a rare combination detector is when monitored data is below similarity.
The topology index defines how good/bad a data combination is. Topology is measured by comparing variables values vis rang set by the engineer. Each variable may have a good and bad range. In some cases, the lower the value, the better the situation is, and for some, the opposite.