My problem is that in the system I design with each question person can assess one of objects of type A, B, or C with for example number or some value from fixed (for this particular question) answer set. In some complex situations, even objects of type A, B and C can be taken as an assessment value.
To demonstrate, there are simple questions, such as:
Please rate A12 on the scale from 1 to 100
and complex ones, such as
Please pick best object of type A for object B42
with answers
- A31,
- A6,
- A42.
Each of the object types - A, B and C cannot be generalised (at least, directly). They surely don't have any common properties and in some scenarios can be related to each other.
Answers given by people are the most crucial data in the system.
I should also mention that all other data fits perfectly into a relational database.
Normally, I prefer control I get from relational databases, but now I consider noSql solutions, as they may fit better.
So, I am asking you to help me answer these questions:
- Which database should I use - relational, noSql or some mixed solution (for example, postgresql with array/json columns)?
- If relational database is used
- do I use sparse relations (tables) for describing questions and answers
- or should I normalise schema, creating dozens of tables?
Aucun commentaire:
Enregistrer un commentaire