The point here is for you to make an assessment of the best approach to analyse sets, based on the understanding that the sets you are interested in may be related to each other in a “generative” way and that you may be able to generate those sets in a fashion that leads you to identify the “best” set under given criteria.
This allows you to avoid in some cases the usual performance traps on combinatorial problems.
The sets can represent anything as long as they are mapped to numbers because this is a constructive numerical approach. For instance, I applied this approach in a project to identify several ways by which items in a given list could be adjoined to map to items in another list, both lists being related to financial transactions and their values.
You can download the paper here.
Compliments of the Author 