Blog

Update

Selecting the Best Concept

07. February, 2022 | Simon Heni


The Authorization Robot generates different solutions. It takes into account the preferences of the user. So before the calculation is started, the user can assign preferences.  Here, we set for example, that the number of composite roles should be kept as low as possible (5 stars), also the positive deviations should not become too high (4 stars). Positive deviations are authorizations that a user gets but does not really need. They arise from the fact that we usually want to assign a role to several users in order to keep the reusability and transparency of the authorization concept at a high level. However, this automatically results in positive deviations. Two users almost never need exactly the same authorizations.

Once the computation is done, we can consider different solutions. Generally, each solution is optimized with respect to the user preferences. However, there is a best solution for each criterion. You can currently recognize this by its name.

In a dashboard we can now compare several solutions. We now select two different solutions and see very clearly in a spider diagram how they differ in the various criteria:

In the legend you can see the selected solutions: Minimum amount of single roles (Lsg. 1) and minimum amount of positive deviations (Lsg. 2) The differences are quickly recognizable: Lsg. 1 has managed to generate only 9 single roles and also only 8 composite roles. However, the positive deviations here are at 260. Lsg. 2, on the other hand, has significantly more single roles: 71 and slightly more composite roles: 12. The positive deviations, on the other hand, are 0.

Numerous box plots are also available for detailed analysis of a specific criterion. Here is an example of three solutions on the criterion of positive deviations ( number of positive deviations):

In the box plot, you can quickly see how this criterion is distributed among the selected three solutions: 0, 260 and 359. Quantiles, median and outliers are of course also displayed, but are only interesting when comparing a higher number of solutions. The upper quantile is shown here as an orange rectangle.