Which DBMS are suitable for record selections that depend on each other?

Question

Database

If, for example, we take a relational database, then in the table it is necessary to find records that depend on each other in a certain way. Multi-entry combinations may depend on other multi-entry combinations.
There may be millions of entries.
In a relational DBMS, this can be done using a certain number of joins. But since there can be millions of records, after the first join, the DBMS will deal with a trillion records, and so on. And the request in this case can be processed indefinitely (already checked).
Perhaps there are subds designed just for such tasks?
While I'm looking at graph DBMS (neo4j), I haven't figured out yet whether it will work or not.
Advise in which direction to at least move.

Answer 1

give an example from your difficult case

Multi-entry combinations may depend on other multi-entry combinations.

.
In relational DBMS, it is necessary first of all to build a data schema - for different data sets that describe different entities, build different tables, describe the relationships between them.
Read about data normalization, enough about the first 3 normal forms.

Answer 2

Are you going to solve problems on graphs?!
Then you just need to choose one of several graph representations
(You can also store RDBMS)
But still, in the limit, you may end up with an NP-complete problem.
Oops, nothing will help. :-)