I have a table that satisfies the following relations:
R(A, B, C, D, E, F, G, H, I, J, K, L, M, N)
{A, B, C} is the primary key.
{D, E} forms a candidate key.
The following functional dependencies exist among the attributes of the relation:
{A, B} -> {H,I}
{D} -> {F,G}
{J} -> {K,L}
{E} -> {M,N}
I need to decompose this to the 2nd normal form, so I separate {AB->HI}, {D->FG}, {E->MN}
into independent tables. But how about {J->KL}
? How should I normalize this one?
A table is in 2nf if and only if all nonprime attributes are functionally dependent on all the attributes of every candidate key, not just on a subset of attributes of any candidate key.
According to your comment, you're given two keys, {ABC} and {DE}.
In the case of {AB}->{HI}, the attributes H and I are dependent on part of a key. ({AB} is part of the key {ABC}.) So you're correct in projecting {HI} from R. Ditto for {D}->{FG} and {E}->{MN}.
When it comes to the attributes J, K, and L, you have to ask the same question.
What do you think about J, K, and L?