Dependency preserving decomposition is a property of database normalization where functional dependencies in a relational schema are preserved after decomposing the relation into smaller sub-relations.
A decomposition of a relation R into sub-relations R1, R2, ..., Rn is dependency preserving if the functional dependencies (FDs) of R can be enforced by enforcing the FDs in each sub-relation without needing to join the decomposed relations.
Consider a relation:
R(A, B, C)
with functional dependencies:
A → B, B → C
Now, suppose we decompose R into two relations:
R1(A, B)R2(B, C)Here:
A → B is present in R1.B → C is present in R2.Since all functional dependencies of R are preserved in R1 and R2, the decomposition is dependency preserving.
If a decomposition loses some functional dependencies, it is not dependency preserving, meaning we might need to join tables to enforce some constraints.
For example, if we decompose R(A, B, C) into:
R1(A, C)R2(B, C)The dependency A → B is lost because neither R1 nor R2 contains both A and B, making it not dependency preserving.
A good decomposition should be: