How to get rule reuse (creation of utility rules)?

Question

This would seem like a common problem. I need rule reuse. I need to create some utility rules. For example:
I want a "utility rule" to verify that a US State is a valid US State.
I can create a simple rule table where the condition is the State's abbreviation or State's full name and the conclusion is the State's validity.
But, how do you reuse that rule across multiple entities that may contain a State abbreviation that need verification?
This is a simple example, there are many similar cases where this is needed. How does the community handle this? It is probably somethng simple I am missing.
Thank you

Answer 1

That is a common challenge. If the data is not normalized on the mainframe side it could still be normalized on it's way to OPA (e.g. when constructing XML if SOAP or in code if integrating through the API). I guess the real decision is whether the logic in OPA is going to represent the business side of the world (i.e. an address is an address regardless of where and how it is represented in the tech stack and all addresses must comply with a single set of validation rules). If that is a goal, then the normalized model in OPA is the way to go. However, if there is a decision against normalization then there really can't be "rule reuse" other than copy, paste, edit because the distinct data model fields each require there own validation rules. (i.e. there is nothing in the model to represent that the two text fields actually have the same logical meaning so they can't possibly share the exact same validation rule).
In some cases, the data can be normalized within the rules as well. Roughly the scenario might be ... all state entities are asserted with their associated names and abbreviations. Each entity has one rule to deduce the relationship to the state entity based on it's state code. The validation rule per entity then simplifies to any entity that does not have a relationship to exactly one state entity must have an invalid state code. Of course, this approach is only practical when there is a relatively small number of mappings from codes to entities that needs to take place (e.g. states would work, but doing the same for all zip codes would be impractical).
Generally speaking, the normalization needs to occur if the rules are going to model the business policies for each logical business entity 1:1. Otherwise, just like with a database design, replication will occur and will impact the number of rules and complexity of rule maintenance.