← 🧩 Normal Forms & Decomposition
Subtopic 5 of 6

5.5BCNF decomposition and 3NF synthesis

The idea

BCNF decomposition splits a violating relation along the offending FD. 3NF synthesis builds one relation per FD of a canonical cover.

Think of it like this: BCNF: keep cutting until no rule is broken, even if a rule gets split across pieces. 3NF: build pieces from the rules first, so none is lost.

Why it exists

You are asked what the output of each algorithm looks like and what it guarantees.

How it works

  1. BCNF: pick a violating X → Y, replace R by (X ∪ Y) and (R − Y); repeat.
  2. Always lossless; may lose dependencies.
  3. 3NF synthesis: canonical cover, relation X ∪ Y per FD, add a key relation if none contains a candidate key.
  4. Always lossless and dependency preserving.

Worked example: R(S, C, I), SC → I, I → C.

BCNF splits on I → C into IC and SI, which loses SC → I. A 3NF design keeps SCI as one relation.

Now play with it

  • →Run both algorithms on the "classic" button and read the Lossless and Preserving flags.
  • →Try the chain A → B, B → C.
  • ABCD violates BCNF through A → BC: split into ABC and AD
  • ABC violates BCNF through B → C: split into BC and AB
  • BC is in BCNF
  • AB is in BCNF
  • AD is in BCNF
RelationsAB, AD, BC
Lossless join?yes
Dependency preserving?yes

BCNF decomposition is always lossless but may lose dependencies. 3NF synthesis is always lossless and dependency preserving.

Check yourself

Which is guaranteed for every relation?

Remember: BCNF: lossless, maybe loses FDs. 3NF synthesis: lossless and preserving.
Exam notes (the dense version)

BCNF decomposition: while some Ri has a violating X → Y, replace Ri by (X ∪ Y) and (Ri − Y). It is always lossless but may lose dependencies. 3NF synthesis: take a canonical cover, make one relation per FD (X ∪ Y), add a relation with a candidate key if none contains one. It is always lossless and dependency preserving. Classic case: R(S, C, I) with SC → I and I → C has no BCNF decomposition that keeps SC → I.

Previous year questions: Normal Forms & Decomposition (11)

PYQs are grouped by topic, so this is the full set for Normal Forms & Decomposition; try the ones that match this subtopic. Open all 11 on the topic page.