Chapter 11

Relational Database Design

Algorithms and Further
Dependencies

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Decompositions
Definition
The decomposition of a schema R=A1…An is its replacement by a collection
DR = {R1, R2, …, Rm} of subsets of R such that R = R1  R2  …  Rm
Note: schemas Ri‟s do not have to be disjoint!

Example 1
Assume the schema R=ABCD. The following are possible decompositions
of R.

D1 = {AB, CD}
D2 = {AB, ACD}
D3 = {A, BCD}
D4 = {AB, BC, CD, AD}

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 2

1
 Dependency Preservation Property of a Decomposition

Observations
• It would be useful if each functional dependency X→Y specified in F either
appeared directly in one of the relation schemas Ri in the decomposition  or
could be inferred from the dependencies that appear in some Ri.

• We want to preserve the dependencies because each rule in F represents a

constraint on the database.

• If one of the dependencies is not represented in some individual relation Ri of

the decomposition, we cannot enforce this constraint by dealing with an
individual relation. We may have to join multiple relations so as to include all
attributes involved in that dependency.

• It is not necessary that the exact dependencies specified in F appear themselves

in individual relations of the decomposition D. It is sufficient that the union of
the dependencies that hold on the individual relations in D be equivalent to F.

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 3

Dependency Preservation Property of a Decomposition

Definition
Given a set of dependencies F on R, the projection of F on Ri, denoted by
Ri(F) where Ri is a subset of R, is the set of dependencies X → Y in F+ such
that the attributes in X  Y are all contained in Ri.

Hence, the projection of F on each relation schema Ri in the

decomposition D is the set of functional dependencies in F+
such that all their left- and right-hand-side attributes are in Ri.

We say that a decomposition D = {R1, R2, . . . , Rm} of R is

dependency-preserving with respect to F if the union of the projections of F on
each Ri in D is equivalent to F; that is,

( R1(F)  R2(F)  . . .  Rm(F) )+= F+

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 4

2
 Dependency Preservation Property of a Decomposition

Claim
It is always possible to find a dependency-preserving decomposition D
with respect to F such that each relation is in 3NF (to be discussed later).

Example 2a
Assume R= ABCD, and F= {A→B, C → D}. The decomposition
D1 = {AB, CD} is clearly F-preserving. Observe the first rule is kept in R1,
while the second is preserved in R2. Also notice that schemas AB and CD
are in 3NF format.
A B C D F= {A→B, C → D}

A B C D
F1= {A→B} F2= {C → D}

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 5

Dependency Preservation Property of a Decomposition

Example 2b
Assume R= ABCD, and F= {A→B, B →C, C → D}. Evaluate the
decomposition D = {ABC, CD}.

D is clearly F-preserving. Observe that

F1 = ABC(F)= {A→B, B →C} and

F2 = CD(F)= {C→D}

and F = F1  F2

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 6

3
 Dependency Preservation Property of a Decomposition

Algorithm: Testing Preservation of Functional Dependencies

Input: A set F of FDs on schema R, a partition D = (R1 . . .Rm) of R, and a

dependency X→Y with R  XY.
Output: true whenever the dependency X→Y is retained by the
decomposition D, i.e. (Ri F) X→Y, and false otherwise

Method: begin
Z = X;
while changes to Z occur do
for i =1 to m do
Z = Z  [ ( Z ∩ Ri )+F ∩ Ri ];
if ( Z  Y )
then return (true)
else return (false);
end;

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 7

Dependency Preservation Property of a Decomposition

Example 2c. source Ullman, J. "Database Systems", Computer Sc. Press

Consider schema R=ABCD, D = {R1= AB, R2= BC, R3= CD }

subjected to F = { A→B, B→C, C→D, D→A }.

Question: Is D→A lost in the projections?

Answer: Even though there is no segment containing attributes DA the rule
D→A is not lost ( ZD= ABCD  A ).
Why? Let‟s use the F-Preserving Test Algorithm. If D preserves F it must
occur that R1(F) R2(F)  R3(F)  D→A. Observe that the other three
rules in F are naturally retained in R1, R2, and R3.
Let‟s begin with Z = D ( this the LHS of D→A )
1. Use R1= AB. Here Z= D  [( D ∩ AB)+ ∩ AB] = D Skipping some steps
2. Use R2= BC. Here Z= D  [( D ∩ BC)+ ∩ BC] = D Of the iteration
3. Use R3= CD. Here Z= D  [( D ∩ CD) ∩ CD] = DC (a change! ) …
+

4. Use R2= BC. Here Z= D  [( DC ∩ BC)+ ∩ BC] = DCB (a change! ) …

5. Use R1= AB. Here Z= DCB  [( DCB ∩ AB)+ ∩ AB] = DCBA  A. Stop! Rule is kept in D

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 8

4
 Dependency Preservation Property of a Decomposition

Example 2c (cont). source Ullman, J. "Database Systems", Computer Sc. Press

Consider schema R=ABCD, D = {R1= AB, R2= BC, R3= CD }

subjected to F = { A→B, B→C, C→D, D→A }.

Observation:
The rule D→A is preserved in the decomposition (R1, R2, R3)
Although not obvious it is clear that the following FDs are in F+
F + ⊇ { A→B, B→C, C→D, D→A, B →A, C →D, D →C }
Therefore
F1 = { A→B, B →A } on R1=(AB)
F2 = { B→C, C →B } on R2=(BC)
F3 = { C→D, D →C } on R3=(CD)

Finally
(F1  F2  F3)+ derives the FD D →A and consequently it is not lost.

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 9

Nonadditive (Lossless) Join Property of a Decomposition

Observation
Another property that a decomposition D should possess is the
nonadditive join property, which ensures that no spurious (phantom)
tuples are generated when a NATURAL JOIN operation is applied to
the relations in the decomposition

Definition.
Formally, a decomposition D = {R1, R2, . . . , Rm} of R has the lossless
(nonadditive) join property with respect to the set of dependencies F on
R, if for every relation state r of R that satisfies F, the following holds,
where * is the NATURAL JOIN of all the relations in D:
*( R1(r) ,. . , Rm(r)) = r.

The word loss in lossless refers to loss of information, not to loss of tuples. If a decomposition
does not have the lossless join property, we may get additional spurious tuples

The nonadditive join property ensures that no spurious tuples result after the application of
PROJECT and JOIN operations.

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 10

5
 Nonadditive Lossless-Join Decomposition
Example 3
Schema < R=ABC , F = { A→B }> and partition D2 = { AB, BC }

Consider the decompositions

D1 = { AB, AC } and
D2 = { AB, BC }
Observe that D1 is a „good‟
decomposition (lossless) while
D2 D2 is not.

Phantom /Spurious rows created

when using decomposition D2

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 11

Nonadditive Lossless-Join Decomposition

Example 3 cont.
Schema < R=ABC , F = { A→B }> and partition D1 = { AB, AC }

Consider the decompositions

D1 = { AB, AC } and
D2 = { AB, BC }
Observe that D1 is a „good‟
decomposition (lossless)
D1 while D2 is not.

Notice that r’ = AB(r) * AC(r)

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 12

6
 Nonadditive Lossless-Join Decomposition
Example 3 cont.
Schemas < R1= ABC, F1 = { A→BC, C →B } > and
< R2= BCD, F2 = { C →B, B →C, D →B }

Observe that r1  r1‟

D1 r1 tuple <2,2,6> is lost in

the JOIN and does not
appear in r1‟

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 13

Testing Lossless-Join (or Non-Additive) Decomposition

Definition (good only on binary partition)
If D={ R1, R2 } is a decomposition of R and F is a set of FDs on R, then D has
a lossless-join with respect to F if

F ( R1  R2) → (R1 - R2) or F ( R1  R2) → (R2 – R1)

Example 4
Consider the previous problem where R=ABC and F = { A→B }.
Let‟s assess the partition D1 = {AB, AC}. Here R1=AB and R2= AC
therefore R1  R2 = A
R1 – R2 = B
R2 – R1 = C

The question F ( R1  R2) → (R1 - R2) is equivalent to F A → B and we

know this is true because F contains exactly this dependency.

We must conclude the decomposition D1 is lossless with respect to F.

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 14

7
 Testing Lossless-Join (or Non-Additive) Decomposition
Example 4 (continuation)
Consider the previous problem where R=ABC and F = { A→B }.
Let‟s now evaluate the partition D2 = {AB, BC}.
Here R1=AB and R2= BC

therefore R1  R2 = B
R1 – R2 = A
R2 – R1 = C

The question F ( R1  R2) → (R1 - R2) is equivalent to F B → A (or

F B → C). Both dependencies are NOT derivable from F (they are not
in F+).

We conclude the decomposition D2 is NOT lossless with respect to F (we

will call it a lossy decomposition).

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 15

Testing Lossless-Join (or Non-Additive) Decomposition

Example 5
Consider the schema R=ABCD and F = { A→B, C→D }.
Let‟s now evaluate the following binary partitions

R1  R2 R1 – R2 R2 – R 1
1 = {AB, CD}  AB CD No

2 = {AC, BCD} C A BD No (neither C →A

nor C →BD in F+)
3 = {ABC, CD} C AB D Yes C→D in F+

2 = {BD, ACD} D B AC No

We conclude only the decomposition 3 is lossless with respect to F.

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 16

8
 Successive (Non-Additive) Lossless-Join Decompositions
Claim.
If a decomposition D = {R1, R2, . . . , Rm} of R has the nonadditive
(lossless) join property with respect to a set of functional dependencies F
on R, and if a decomposition Di = {Q1, Q2, . . . , Qk} of Ri has the
nonadditive join property with respect to the projection of F on Ri, then the
decomposition

Di = { R1, R2, . . . ,Ri-1, Q1, Q2, . . . , Qk, Ri+1, . . . , Rm }

of R has the nonadditive join property with respect to F.

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 17

Algorithm 11.1. Testing for Nonadditive Join Property

Input: A universal relation R, a decomposition D = {R1, R2, . . . , Rm} of R, and a set F of FDs

Note: Explanatory comments are given at the end of some of the steps. They follow the format: (* comment *)

1. Create an initial matrix S with one row i for each relation Ri in D, and
one column j for each attribute Aj in R.
2. Set S(i, j):= bij for all matrix entries.
(* each bij is a distinct symbol associated with indices (i, j) *)
3. For each row i representing relation schema Ri
{for each column j representing attribute Aj
{if (relation R, includes attribute Aj) then set S(i, j):= aj;};};
(* each aj is a distinct symbol associated with index (j)* )

4. Repeat the following loop until a complete loop execution results in no changes to S
{for each functional dependency X → Y in F
{for all rows in S that have the same symbols in the columns corresponding
to attributes in X
{make the symbols in each column that correspond to an attribute in
Y be the same in all these rows as follows:
If any of the rows has an a symbol for the column, set the other rows to that
same a symbol in the column.
If no a symbol exists for the attribute in any of the rows,
choose one of the b symbols that appears in one of the rows for the
attribute and set the other rows to that same b symbol in the column
;} ;} ; } ;

5. If a row is made up entirely of a symbols,

then the decomposition has the non-additive join property;
otherwise, it does not.

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 18

9
 Algorithm 11.1. Testing for Nonadditive Join Property
Example 6
Consider the schema R=ABCD, subjected to FDs F= { A → B, B → C }, and the
Non-binary partition D1 = {ACD, AB, BC}.
Question Is D1 a Lossless decomposition?

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 19

Algorithm 11.1. Testing for Nonadditive Join Property

Example 7
Consider the schema R=ABCD, subjected to FDs F= { A→B, B→C }, and the
Non-binary partition D2 = {AB, BC, CD}.
Question Is D2 a Lossless decomposition?

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 20

10
 Algorithm 11.1. Testing for Nonadditive Join Property
Example 8
Consider the schema R=ABCD, subjected to FDs F= {A→B, B → C }, and the
Non-binary partition D2 = {ABC, AD}.
Question Is D3 a Lossless decomposition? (Binary test indicates: Yes!)

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 21

Algorithm 11.1. Testing for Nonadditive Join Property

Example 8b – Your turn…
Consider the schema R=ABCD, subjected to FDs F= {A→B, C → D}, and the
Non-binary partitions D4 = {AB, AC, AD} and D5 = {AB, AC, CD}.

Question. Are partitions D4 and D5 Lossless decompositions?

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 22

11
 Algorithm 11.1. Testing for Nonadditive Join Property
Example 9
Consider the schema R=ABCDE, subjected to FDs F= {A→C, B → C, C → D,
DE →C, CE →A}, and the Non-binary partition D4 = {AD, AB, BE, CDE, AE}.
Question Is D4 a Lossless decomposition?

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 23

Algorithm 11.1. Testing for Nonadditive Join Property

Example 9 continuation F= {A→C, B → C, C → D, DE →C, CE →A}

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 24

12
 Algorithm 11.1. Testing for Nonadditive Join Property
Example 9 continuation F= {A→C, B → C, C → D, DE →C, CE →A}

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 25

Algorithm 11.1.
Testing for Nonadditive
(Lossless) Join Property

Example 10

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 26

13
 Prime, Non-Prime and Key Attributes

Key Attributes
A set X of attributes in the schema R is a key for R under the
dependencies F, if X→R and no proper subset Y of X (X Y) has
the same property.

Prime Attributes
An attribute A in a relation schema R is prime when it is part of any
candidate key of the relation.

If A is not included in any candidate key of R, A is called non-

prime.

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 27

Prime, Non-Prime and Key Attributes

Example 11
Consider the relation Schema R=ABCD and its set F of functional dependencies
F={A→BC, C→D, D→B)

Brute-force discovery of all the keys in R under F is a simple but exponential problem.

Steps
1. Find all rules derived from F having only ONE attribute on the left-hand side
A+ = ABCD = R (therefore A is a key)
B+ = B (not a key)
C+ = CDB (not a key)
D+= DB (not a key)

2. Find all rules derived from F having only TWO attribute on the left-hand side
AB+ = ABCD (a superkey)
AC+ = ABCD (a superkey)
AD+ = ABCD (a superkey)
BC+ = BCD (not a key)
BD+ = BD (not a key)
CD+ = CDB (not a key)
Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley
Slide 11- 28

14
 Prime, Non-Prime and Key Attributes

Example 11 continuation

3. Find all rules derived from F having only THREE attribute on the left-hand side
ABC+ = ABCD (a superkey)
ABD+ = ABCD (a superkey)
BCD+ = ABCD (a superkey)

4. Find all rules derived from F having only FOUR attribute on the left-hand side
ABCD+ = ABCD (a superkey)

Summary
Key: A
Prime: A
Non-Prime: BCD

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 29

Prime, Non-Prime and Key Attributes

Example 12 (your turn…)

Consider the relation schema R= A B C D E G H I J K L and the set of dependencies F

AB→ CDEGH
BD → AC
DG → L
EG → L
IJ → K
JK → I
H → IJK

Find all the keys and identify the non-prime attributes

Candidate keys =
Non-prime =

PROBLEM
Write an program to find all the candidate keys of a database schema (R,F).

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 30

15
 Algorithms for Relational Database Design

We will discuss the following three algorithms

11.2.2 Non-Additive Decomposition into 3NF Schemas

Non-Additive Decomposition into BCNF Schemas

11.2.1 Dependency Preserving Decompositions into 3NF Schemas

a) Decomposition Method
b) Synthesis Method

11.2.3 Dependency-Preserving and Nonadditive (Lossless)

Decompositions into 3NF Schemas

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 31

11.2.2 Non-Additive Decomposition into 3NF Schemas

Algorithm 11.2.A. 3NF Decomposition Method

Input: Relation Schema R and its set F of functional dependencies

Output: A decomposition D= (R1, R2, ..., Rm) of R, such that each Ri
is in 3NF and the decomposition is lossless.
Method
1. Find a key K and a transitive dependency, such that
A is a non prime attribute (i.e. A is not part of a key)
K→Y→A
Not (Y→K)
A KY

2. Make Rl= (Y,A) and R2= (R-A)

3. Repeat process on Rl and R2 until they become 3NF

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 32

16
 11.2.2 Nonadditive Join Decomposition into 3NF Schemas

Example 13.A.
Consider the TEACH relation below.

Observe that Key = { Text, Teacher }

The dependency Teacher → Course is such that its L.H.S is not a key.
Therefore we use this rule to partition the schema TEACH into TEACH1 and TEACH2
where:
The attributes in TEACH1 are those in
{Teacher, Course, Text} – {Course} and

TEACH2 contains {Teacher}  {Course}

Finally the 3NF schema is: TEACH1(Text, Teacher)

TEACH2(Teacher, Course)

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 33

11.2.2 Non-Additive Decomposition into 3NF Schemas

A. 3NF Decomposition Method – Example 11

Consider the relation schema
R= Flight#, From, To, Departs, Arrives, Duration, PlaneType,
FirstClass, Coach, TotalSeats, #Meals

Non-Trivial Functional dependencies:

PlaneType → FirstClass Coach TotalSeats
Departs Duration → #Meals
Arrives Duration → #Meals
FirstClass Coach → TotalSeats
FirstClass TotalSeats → Coach
Coach TotalSeats → FirstClass

Candidate keys = { Flight#, (From To Departs), (From To Arrives) }

Non-Prime Attributes = { Duration, PlaneType, FirstClass, Coach,
TotalSeats, #Meals }
Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley
Slide 11- 34

17
 11.2.2 Non-Additive Decomposition into 3NF Schemas
A. 3NF Decomposition Method – Example 11 continuation

1. We find an inappropriate transitive dependency

Flight# → Departs Duration → #Meals

We decompose into two fragments R1 and R2

R l = Flight# From To Departs Arrives Duration PlaneType

Firstclass Coach TotalSeats
With candidate keys:
K l = { Flight#, (From To Departs), (From To Arrives) }
Observe that R l is not in 3NF. Note the transitive dependency
Flight# → PlaneType → FirstClass Coach TotalSeats

R2 = Departs Duration #Meals

With designated key K2= (Departs Duration)
Schema R2 is already in 3NF

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 35

11.2.2 Non-Additive Decomposition into 3NF Schemas

A. 3NF Decomposition Method – Example 11 continuation
2. We decompose the offending R l into two partitions R11 and R12

R11 = Flight# From To Departs Arrives Duration PlaneType

With keys K l l = { Flight#, (From To Departs), (From To Arrives) }
Which already is in 3NF.

R12 = PlaneType FirstClass Coach TotalSeats

With key K12= { PlaneType }

3. We decompose R12 which has the following transitive dependency

PlaneType → FirstClass Coach → TotalSeats
R121 = PlaneType FirstClass Coach now in 3NF format
With key K121 = PlaneType

R122= FirstClass Coach TotalSeats now in 3NF format

With Key K122= FirstClass Coach

Stop, database is already in 3NF format.

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley
Slide 11- 36

18
 11.2.2 Non-Additive Decomposition into 3NF Schemas
A. 3NF Decomposition Method – Example 11 continuation

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 37

11.2.2 Non-Additive Decomposition into 3NF Schemas

A. 3NF Decomposition Method – Example 12
Consider the relation scheme R= {ABCDEG} subject to FDs
F= { A→B, BC → D, D → EG }
STEP 1
R={ABCDEG} Key=AC Violation: AC → D, D → EG
Break R into: Rl= DEG and R2=ABCD
STEP 2
Rl = {DEG} Key= D. No violations - already in 3NF
however R2 is not in 3NF
STEP 3
R2= {ABCD} Key =A Violation: AC→BC and BC→D
Break R2 into: R21= BCD and R22= ABC
STEP 4
R21 = BCD Key= BC No violations - already in 3NF
However R22 is not in 3NF
STEP 5
R22= ABC Key = AC There is a (partial dependency) violation: A → B
Break R22 into R221=AB and R222=AC both in 3NF
SOLUTION: D = { DEG, BCD, AB, AC }
Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley
Slide 11- 38

19
 11.2.2 Non-Additive Decomposition into 3NF Schemas
A. Shortcomings of 3NF Decomposition
1. Time consuming - testing if an attribute is prime is an NP operation.
2. It may produce too many tables (more than we need for 3NF).

Example 13
Assume the relation scheme R=(ABCDE) obeys the following set of FDs
F= { AB→CDE, AC→BDE, B → C, C → B, C → D, B → E }

STEP 1 R={ABCDE} Key= { AB, AC } Violation: AB → C, C → D

Break R into: Rl=CD and R2=ABCE
STEP 2 Rl= {CD} Key= C No violations - already in 3NF, but R2 is not
STEP 3 R2= {ABCE} Key={AB, AC) Violation: AB → B, B → E
Break R2 into R21={BE} and R22={ABC}
STEP 4 R21 = {BE} Key= {B} No violations - in 3NF
R22= {ABC} Key={AC, AB) No violations - already in 3NF

Solution D1= { CD, BE, ABC}

Observation: D2 = { ABC, BDE } is 3NF and includes less fragments than D1

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley
Slide 11- 39

11.2.2 Non-Additive Decomposition into 3NF Schemas

A. Shortcomings of 3NF Decomposition
A serious problem with the 3NF-Decomposition methods is that dependencies
in F may not be enforced on the decomposition.

Example 14
Consider the relation scheme R= {ABCDE} subject to the following functional
dependencies F = { A→BCDE, CD → E, CE → B }
STEP 1
R={ABCDE} Key={ A } Violation: A → CD, CD → E
Break R into the following 3NF tables

D= Rl = {ABCD} where A → BCD,

R2 = {CDE} where CD → E

The decomposition D is lossless but the rule CE → B is not retained in the two
fragments R1 and R2.
Notice that CE → B  { (R1(F)  R2(F) ) + }

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 40

20
 11.2.2 Non-Additive Decomposition into BCNF Schemas

Algorithm 11.2.A. BCNF Decomposition Method

Input: Relation Schema R and its set F of functional dependencies

Output: A decomposition D = (R1, R2, ..., Rm) of R, such that each Ri
is in BCNF and the decomposition is lossless.
Method
1. Set D = {R}
2. While there is a relation schema Q in D that is not in BCNF;
{
a. choose a relation schema Q in D that is not in BCNF;
b. find a functional dependency X→Y in Q such that X
is not a superkey in Q;
c. replace Q in D by two relations schemas (Q – Y) and (X  Y)
}

Claim: We can use decomposition to find a lossless BCNF database

scheme for an initial relation scheme that is not in BCNF.
Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley
Slide 11- 41

11.2.2 Non-Additive Decomposition into BCNF Schemas

Example
Let R= { ABCDE } and let F = { A →BC, BC →A, BCD →E, E →C}
Convert R into a BCNF scheme (if needed!).

Key = {AD, BCD}. Observe BC → A is such that BC is not a superkey.

Therefore R is not in BCNF and should be decomposed:
R1 = (ABC) {A →BC, BC →A} in BCNF (superkey = {A, BC})
R2 = (BCDE) {BCD →E, E →C} not in BCNF (E →C and E is not
a superkey)
We continue decomposing R2 into
R21= (BDE) no FDs. R21 is in BCNF
R22= (EC) {E →C} E is a superkey R22 is in BCNF.

Final solution:
Scheme D= { ABC, BDE, EC } is in BCNF (and it is lossless, prove it!)
Is the rule BCD →E preserved in D?

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 42

21
 11.2.1 Dependency Preserving Decompositions into 3NF Schemas
Algorithm. 11.2.B. 3NF Synthesis with Dependency Preservation

Input: A universal relation R and a set of functional dependencies F on R.

1. Find a minimal cover G for F

2. For each left-hand-side X of a functional dependency that appears in G,

create a relation schema in D with attributes {X  {A1}  {A2} . . . 
{Ak }, where X →A1, X →A2, . . . , X →Ak are the only dependencies
in G with X as the left-hand-side (X is the key of this relation);

3. Place any remaining attributes (that have not been placed in any
relation) in a single relation schema to ensure the attribute preservation
property.

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 43

11.2.1 Dependency Preserving Decompositions into 3NF Schemas

B. 3NF Synthesis with Dependency Preservation

Claim 3.
Every relation schema created by Algorithm 11.2.B is in 3NF.

Observation
• The 3NF-Synthesis algorithm creates a dependency-preserving
decomposition D = {R1, R2, . . . ,Rm} of a universal relation R based on
a set of functional dependencies F, such that each Ri in D is in 3NF.
• It guarantees only the dependency-preserving property; it does not
guarantee the nonadditive join property.

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 44

22
 11.2.1 Dependency Preserving Decompositions into 3NF Schemas
B. 3NF Synthesis with Dependency Preservation

Example 14. Consider the following universal relation:

U(Emp-ssn, Pno, Esal, Ephone, Dno, Pname, Plocation)

Emp-ssn, Esal, Ephone refer to the Social Security Number, salary and phone number
of the employee. Pno, Pname, and Plocation refer to the number, name, and location
of the project. Dno is department number.

The following dependencies are present:

FD1: Emp-ssn → Esal, Ephone, Dno

FD2: Pno → Pname, Plocation
FD3: Emp-ssn, Pno → Esal, Ephone, Dno, Pname, Plocation

The Key is {Emp-ssn, Pno}. A minimum Cover G is given by:

Emp-ssn → Esal, Ephone, Dno
Pno → Pname, Plocation
Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley
Slide 11- 45

11.2.1 Dependency Preserving Decompositions into 3NF Schemas

B. 3NF Synthesis with Dependency Preservation

Example 14 continuation

From rule Emp-ssn → Esal, Ephone, Dno we create the partition

R1 (Emp-ssn , Esal, Ephone, Dno)

From rule Pno → Pname, Plocation we obtain

R2 (Pno, Pname, Plocation)

Comments
Observe the original key {Emp-Ssn, Pno} is broken by the partition of R.

The partition D={R1, R2} enforces the rules originally in F; however the
decomposition is not lossless [it is not true that R1∩R2 → (R1-R2) or (R2-R1)]

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 46

23
 11.2.1 Dependency Preserving Decompositions into 3NF Schemas
B. 3NF Synthesis with Dependency Preservation
Example 15.
Consider r(ABCDEF) subject to the following FDs
F = { A→BCE, B → DE, ABD → CF. AC → DE )
Find a 3NF database for r under F.

Solution: Find a canonical cover for (r, F).

1. After Left Reduction:
Rule ABD → CF becomes A → CF
Rule AC → DE becomes A → DE
F= { A → BCE, B → DE, A → CF, A → DE)
2. After Right Reduction (not really necessary! – use projectivity instead)
Rule A → BCE becomes A → B
Rule A → DE becomes A → E
F= { A → B, B → DE, A → CF, A → E )
3. After Removing Redundant Rule(s):
Rule A → E is redundant (Trans. on A → B, B → E )
F= { A → B, B → DE, A → CF } after compacting similar left-hand-side rules
F= { A → BCF, B → DE )
Final Partition D = ( ABCF, BDE ) (clearly in 3NF format)

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 47

11.2.3 Dependency-Preserving and Nonadditive (Lossless) Join

Decomposition into 3NF Schemas
The following method –which includes a minor modification to Algorithm 11.2-
produces a decomposition D of R that does the following:
• Preserves dependencies
• Has the nonadditive join property

Algorithm 11.4. Relational Synthesis into 3NF with Dependency

Preservation and Nonadditive Join Property
Input: A universal relation R and a set of functional dependencies F on the
attributes of R.
1. Find a minimal cover G for F.
2. For each left-hand-side X of a functional dependency that appears in G create a
relation schema in D with attributes {X {A1}  {A2} . . .  {Ak} }, where
X → A1, X → A2, . . . , X → Ak are the only dependencies in G with X as
left-hand-side (X is the key of this relation).
3. If none of the relation schemas in D contains a key of R, then create one more
relation schema in D that contains attributes that form a key of R.
4. Eliminate redundant relations from the resulting set of relations in the relational
database schema. A relation R is considered redundant if R is a projection of
another relation S in the schema; alternately, R is subsumed by S.
Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley
Slide 11- 48

24
 Algorithm 11.4.(a) Finding a Key K for R Given a set F
of Functional Dependencies
Input:
A universal relation R and a set of functional dependencies F on attributes
of R.

Method:

1. Set K:= R.

2. For each attribute A in K

{compute (K - A)+ with respect to F;
if (K - A)+ contains all the attributes in R, then set K := K - {A}
};

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 49

11.2.3 Dependency-Preserving and Nonadditive (Lossless) Join

Decomposition into 3NF Schemas

Example 17. Algorithm 11.4.

Let us review example 14.
U(Emp-ssn, Pno, Esal, Ephone, Dno, Pname, Plocation)

The Key is {Emp-ssn, Pno}. A minimum Cover G is given by:

Emp-ssn → Esal, Ephone, Dno

Pno → Pname, Plocation
Instead of only two partitions (each representing a rule in G) we will add one more
fragment to include the key (Emp ssn, Pno), the resulting design contains:

R1 ( Emp-ssn, Esal, Ephone, Dno )

R2 ( Pno, Pname, Plocation )
R3 (Emp ssn, Pno)

This design achieves both the desirable properties of dependency preservation and
nonadditive join.

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 50

25
 11.2.3 Dependency-Preserving and Nonadditive (Lossless) Join
Decomposition into 3NF Schemas

Example 18. Algorithm 11.4.

Consider the relation LOTS1 and its
functional dependencies F.
F = { P→LCA, LC →AP, A →C }

A minimum cover G is
{P → LC, LC → AP, A → C }.

In step 2 of Algorithm 11.4 we produce design X (before removing redundant relations)

as
Design X: R1 (P, L, C), R2 (L, C, A, P), and R3 (A, C).

In step 4 of the algorithm, we find that R3 and R1 are subsumed by R2. Hence both of
those relations are redundant. Thus the 3NF schema that achieves both of the desirable
properties
is (after removing redundant relations)

Design X: R2 (L, C, A, P).

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 51

11.2. Summary of Relational Database Design Algorithms

Slide 11- 52
Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 52

26
 11.3 Multivalued Dependencies and Fourth Normal Form (4NF)

Introduction.
Functional Dependencies are not capable of representing all type of
associations between attributes.

In general a FD X→Y represents a many:1 relationship.

Sometimes we want to emphasize a many:many association or multiple 1:many

relationships in the same table.

In this case multivalued dependencies might be more appropriated.

Book# Author Price
Book
Book# Author Price B1 A1 100

B1 { A1, A2, A3 } 100 B1 A2 100

B2 { A4 } 90 B1 A3 100

B3 { A1, A5 } 120 B2 A4 90
Equivalent 1NF
B3 A1 120 relation
NFNF 1:many Book : Author
B3 A5 120
Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley
Slide
Slide 11-
11- 53
53

11.3 Multivalued Dependencies and Fourth Normal Form (4NF)

Example 19.
• Multiple independent 1:many relationships in the same table.
• Observe that ProjectName and DependentName are independent.

Ename Pname DependentName

Smith Proj-X Johnny

Smith Proj-Y Anna
Smith Proj-X Anna
Smith Proj-Y Johnny

Ename Pname Ename DependentName

Smith Proj-X Smith Johnny

Smith Proj-Y Smith Anna

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide
Slide 11-
11- 54
54

27
 11.3 Multivalued Dependencies and Fourth Normal Form (4NF)

Definition
A multivalued dependency XY specified on relation schema R, where X and Y are
both subsets of R, specifies the following constraint on any relation state r of R:

If two tuples t1 and t2 exist in r such that t1[X] = t2[X], then two tuples t3 and t4
should also exist in r with the following properties

t3[X] = t4[X] = t1[X] = t2[X].

t3[Y] = t1[Y] and t4[Y] = t2[Y].
t3[Z] = t2[Z] and t4[Z] = t1[Z] [we use Z to denote (R - (X  Y)) ].

Whenever X  Y holds, we say that X multi-determines Y.

Because of the symmetry in the definition, whenever XY holds in R, so does

X Z. Hence, XY implies XZ, and therefore it is sometimes written as X  Y |
Z.

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide
Slide 11-
11- 55
55

11.3 Multivalued Dependencies and Fourth Normal Form (4NF)

Example 20.a
Consider the tuples t1, t2, t3, and t4 shown below. Observe that
t3[X] = t4[X] = t1[X] = t2[X].
t3[Y] = t1[Y] and t4[Y] = t2[Y].
t3[Z] = t2[Z] and t4[Z] = t1[Z]

Therefore XY and XZ i.e. Ename  Pname , Ename  DependentName

X Y Z

Ename Pname DependentName

t3[X] = t4[X] = t1[X] = t2[X] = “Smith”
t1 Smith Proj-X Johnny
t3[Y] = t1[Y] = „Proj-X‟
t2 Smith Proj-Y Anna
t4[Y] = t2[Y] = „Proj-Y‟
t3 Smith Proj-X Anna
t3[Z] = t2[Z] = „Anna‟
t4 Smith Proj-Y Johnny t4[Z] = t1[Z] = „Johnny‟

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide
Slide 11-
11- 56
56

28
 11.3 Multivalued Dependencies and Fourth Normal Form (4NF)

Observation
A multivalued dependency (or MVD) XY specifies that given a particular value
of X, the set of values of Y determined by this value of X is completely
determined by X alone and does not depend on the values of the remaining
attributes Z of R.

Definition
An MVD X  Y in R is called a trivial MVD if
(a) Y is a subset of X, or
(b) X  Y = R.

Observation
If we have a nontrivial MVD in a relation, we may have to repeat values
redundantly in the tuples (which is unwanted!).

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide
Slide 11-
11- 57
57

11.3 Multivalued Dependencies and Fourth Normal Form (4NF)

Example 20.b
The Employee-Family-Project table shown before in Example 20.a holds the MVDs
Ename Pname DependentName
Ename  Pname
Smith Proj-X Johnny
Ename  DependentName
Smith Proj-Y Anna
Smith Proj-X Anna
Smith Proj-Y Johnny

Ename Pname Ename DependentName

Smith Proj-X Smith Johnny

Smith Proj-Y Smith Anna

Ename Pname DependentName Equivalent representations of the

original relation
Smith { Proj-X, Proj-Y } { Johnny, Anna }

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide
Slide 11-
11- 58
58

29
 11.3 Multivalued Dependencies and Fourth Normal Form (4NF)

Inference rules for Multivalued Dependencies

The following inference rules IR1 through IR8 form a sound and complete set for
inferring functional and multivalued dependencies from a given set of dependencies.

Assume that all attributes are included in a universal relation schema

R = {A1, A2, . . . ,Am} and that X, Y, Z, and W are subsets of R.

IR1 (reflexive rule for FDs): If X  Y, then X→Y.

IR2 (augmentation rule for FDs): {X → Y}  XZ → YZ.
IR3 (transitive rule for FDs): {X → Y, Y → Z}  X → Z.
IR4 (complementation rule for MVDs): {X  Y)  {X  (R - ( X  Y))}:
IR5 (augmentation rule for MVDs): If X  Y and W  Z, then WX  YZ.
IR6 (transitive rule for MVDs): {X  Y, Y  Z}  X  (Z - Y).
IR7 (replication rule for FD to MVD): {X → Y) X  Y.
IR8 (coalescence rule for FDs and MVDs): If XY and there exists W with the
properties that (a) W ∩ Y is empty, (b) W → Z, and (c) Y  Z, then X → Z.
IR9 (Additivity) If X Y and X Z then X  YZ
IR10(Projectivity) If X Y and X Z, then X  Y∩Z, X Y - Z, and X Z – Y

Note: IR9 and IR10 are derived from IR1-IR8. See D. Maier Chp 7 pp 129
Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley
Slide
Slide 11-
11- 59
59

11.3 Multivalued Dependencies and Fourth Normal Form (4NF)

r A B C D E
How to know if a given m.v.d. a1 b1 c1 d1 e1
holds in a relation r ? a2 b3 c3 d3 e3
a2 b3 c3 d4 e4
a2 b3 c3 d3 e4
Is A BC valid in r ? a2 b3 c3 d4 e3
a3 b4 c4 d5 e5
Method a3 b4 c4 d5 e6
a3 b4 c4 d6 e5
1. Decompose r into r1(ABC) and r2(ADE) a3 b4 c4 d6 e6
2. Join r1 and r2 (call it r12) a3 b5 c5 d5 e5
3. Compare r with r12 a3 b5 c5 d5 e6
if both are equal the depenency holds. a3 b5 c5 d6 e5
a3 b5 c5 d6 e6

A D E
r1 A B C r2 a1 d1 e1
a1 b1 c1 a2 d3 e3
a2 b3 c3 a2 d3 e4
a2 d4 e3
a3 b4 c4
a2 d4 e4
a3 b5 c5 a3 d5 e5 1. In this example A BC
a3 d5 e6 is valid.
a3 d6 e5
a3 d6 e6 2. Your turn
try BC CD
Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley
Slide
Slide 11-
11- 60
60

30
 11.3 Multivalued Dependencies and Fourth Normal Form (4NF)
r A B C D E
How to know if a given m.v.d. a1 b1 c1 d1 e1
holds in a relation r ? a2 b3 c3 d3 e3
a2 b3 c3 d4 e4
a2 b3 c3 d3 e4
Testing validity of BC CD a2 b3 c3 d4 e3
a3 b4 c4 d5 e5
Method a3 b4 c4 d5 e6
a3 b4 c4 d6 e5
1. Decompose r into r1(BCD) and r2(ABDE) a3 b4 c4 d6 e6
2. Join r1 and r2 (call it r12) a3 b5 c5 d5 e5
3. Compare r with r12 a3 b5 c5 d5 e6
if both are equal the dependency holds. a3 b5 c5 d6 e5
a3 b5 c5 d6 e6
B C D A B D E
r1 a1 b1 d1 e1
b1 c1 d1 r2 a2 b3 d3 e3
b3 c3 d3 a2 b3 d3 e4
b3 c3 d4 a2 b3 d4 e3
b4 c4 d5 a2 b3 d4 e4
a3 b4 d5 e5
b4 c4 d6
a3 b4 d5 e6
b5 c5 d5 a3 b4 d6 e5 In this example BC BD
b5 c5 d6 a3 b4 d6 e6 is valid.
a3 b5 d5 e5
a3 b5 d5 e6
a3 b5 d6 e5
a3 b5 d6 e6
Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley
Slide
Slide 11-
11- 61
61

11.3 Multivalued Dependencies and Fourth Normal Form (4NF)

Inference rules for Multivalued Dependencies

MVD Transitivity

Rule IR6 (transitive rule for MVDs): {X  Y, Y  Z}  X  (Z – Y)

A B C D E
Example
a1 b1 c1 d1 e1
a2 b3 c3 d3 e3
A BC
a2 b3 c3 d4 e4
BC CD
a2 b3 c3 d3 e4
A  D (i.e. CD – BC)
a2 b3 c3 d4 e3
a3 b4 c4 d5 e5
Observe that
a3 b4 c4 d5 e6
a1 {b1c1}  {d1} a3 b4 c4 d6 e5
a2 {b3c3}  {d3,d4} a3 b4 c4 d6 e6
a3 {b4c4, b5c5} {d5, d6} a3 b5 c5 d5 e5
a3 b5 c5 d5 e6
a3 b5 c5 d6 e5
a3 b5 c5 d6 e6
Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley
Slide
Slide 11-
11- 62
62

31
 11.3 Multivalued Dependencies and Fourth Normal Form (4NF)

Inference rules for Multivalued Dependencies

MVD Coalescense
IR8 (coalescence rule for FDs and MVDs):
If XY and there exists W with
the properties that
(a) W ∩ Y is empty,
(b) W → Z, and
(c) Y  Z, Show Day Hour Channel Tvstation Sponsor Actor
then X → Z. s1 Mo 9p 1 ABC Nike Angeline

s1 Mo 9p 1 ABC Nike Sandra

Example s1 Mo 9p 1 ABC Pepsi Angeline

S  Day Hour Channel s1 Mo 9p 1 ABC Pepsi Sandra

S  Sponsor s2 We 8p 3 CBS ATT Fox

S  Actor s2 We 8p 3 CBS ATT Scully

TVstation → Channel s2 We 8p 3 CBS IBM Fox

s2 We 8p 3 CBS IBM Scully

S → Channel
s3 Tu 7p 1 ABC Toyota Columbus

s3 Fr 7p 1 ABC Toyota Columbus

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide
Slide 11-
11- 63
63

11.3 Multivalued Dependencies and Fourth Normal Form (4NF)

Example 21A. Inference rules for Multivalued Dependencies

Consider the dependencies F = { A  BC, DE  C} over the schema

R = {ABCDE}.

Question
Does AD  BE ?

Proof. Using the Inference Axioms IR1 to IR8 we have

1. A BC given
2. A DE by complementation respect to R
3. DEC given
4. AC transitivity on 2 and 3
5. AD C augmentation of 4 by D
6. AD BE complementation of 5 respect to R

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide
Slide 11-
11- 64
64

32
 11.3 Multivalued Dependencies and Fourth Normal Form (4NF)

Example 21B. Inference rules for Multivalued Dependencies

Consider the dependencies F = { AEI, C  AB } on the schema

R = { ABCDEI }.

Question
Does AC  BEI ?

Proof. Using the Inference Axioms IR1 to IR8 we have

1. C AB given
2. C DEI complementation of 1 respect to R
3. CEIDEI augmentation of 2 by EI
4. A EI given
5. AC CEI augmentation of 4 by C
6. AC D transitivty on 5 and 3
7. AC BEI complementation of 5 respect to R

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide
Slide 11-
11- 65
65

11.3 Multivalued Dependencies and Fourth Normal Form (4NF)

Example 21C. Inference rules for Multivalued Dependencies

Consider the dependencies F = { AEI, C  AB } on the schema

R = { ABCDEI }.

Question
Does AC  B ?

Proof. Using the Inference Axioms IR1 to IR8 we have

1. C AB given
2. C DEI complementation of 1 respect to R
3. AC ADEI augmentation of 3 by A
4. ACB complementation of 3 respect to R

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide
Slide 11-
11- 66
66

33
 11.3 Multivalued Dependencies and Fourth Normal Form (4NF)

Minimal Disjoint Set Basis

Definition:
Given a collection of sets S = {S1, S2, … SP}, where the universe U = S1  S2
… SP the minimal disjoint set basis of S (mdsb(S)) is the partition T= {T1,
T2, …Tq} of U such that
1. Every Si is a union of some of the Tj‟s
2. No partition of U with fewer cells has the first property

Example
Assume S = {ABC, BCD, AD } then mdsb(S) = { A, BC, D }
Observe that ABC = A + BC
BCD = BC + D
AD = A + D

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide
Slide 11-
11- 67
67

11.3 Multivalued Dependencies and Fourth Normal Form (4NF)

Algorithm. Computing Dependency Basis

Input:
A set of multivalued dependencies M over a set of attributes R, and a set X U
Output: The dependency basis for X with respect to M.

Method:
1. Let T be the set of sets Z  X such that for some WY in M, we have WX,
and Z is either (Y – X) or ( R – X – Y )
2. Until T consists of a disjoint collection of sets, find a pair of sets Z1 and Z2 in T
that are not disjoint and replace them by sets (Z1 - Z2), (Z2 – Z1) and (Z1  Z2).
Let S be the final collection of sets.
3. Until no more changes can be made to S, look for dependencies VW in M and
a set Y in S such that Y intersects W but not V. Replace Y by the sets ( Y  W)
and ( Y – W ) in S.
4. The final collection of sets in S is the dependency basis for X.

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide
Slide 11-
11- 68
68

34
 11.3 Multivalued Dependencies and Fourth Normal Form (4NF)

Algorithm. Computing Dependency Basis

Intuition
Assume we are computing DEP(AC) for schema R under M(Functional and
Multivalued ). STEP1 begins with the application of reflexivity AC AC and
complementation AC R-AC. Therefore the initial set TAC includes {A, R-A}.
Exploring rules in M of the form AC  X1, A X2 and C X3 brings to TAC
other attributes directly implied by AC. Observe that by augmentation if A  X2
then AC X2, similarly if C X3 by augmentation AC X3. Therefore fragments
X1, X2, X3 (and their complements) must also be added to TAC.
If no more elements can be included to TAB a fine fragmentation of the existing TAB
components should follow to produce its mdsb(TAC) (called SAC in the algorithm).
In the final step, we try one more refinement of the elements Y in SAC = msdb(TAC).
The Coalescence axiom (IR8) states that if there is a dependency VW in M and a
set Y in SAC such that Y intersects W but not V then AC  W. Therefore the set Y
could be replaced by the finer sets ( Y  W) and ( Y – W ) in SAC.
The final collection of sets in SAC is the dependency basis DEP(AC).
Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley
Slide
Slide 11-
11- 69
69

11.3 Multivalued Dependencies and Fourth Normal Form (4NF)

Claim
To test whether a MVD X Y holds in F, it suffices to determine DEP(X)
and see whether (Y - X) is the union of some sets in DEP(X).

Example 22A
Let F={A  BC, DE  C} be a set of MVDs over ABCDE.
Compute DEP(A)
To compute it we begin with TA = {A, BCDE} (A  A, A  BCDE).
Step 1 allows us to use A  BC (given) and introduce BC in TA.
Observe that A  DE (by use of complementation), therefore DE is also
added to TA = {A, BCDE, BC, DE}.
Refining this set produces SA = {A, BC, DE}.
Using Step3 of the method we find rule DE  C and element BC intersects
the RHS; however BC does not share common attributes with DE (the LHS).
Consequently we replace BC with (BC C) and (BC – C).
Hence DEP(A)={A, B, C, DE}

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide
Slide 11-
11- 70
70

35
 11.3 Multivalued Dependencies and Fourth Normal Form (4NF)

Example 22A continuation

Let F={A  BC, DE  C} be a set of MVDs over ABCDE.
Hence DEP(A)={A, B, C, DE}

Significance
Attribute A multi-determines any union of elements in DEP(A), for
example
AA A BC
AB ADE
AC ACDE
ABCDE
ABDE

Similarly we may say it is NOT true that A  D [notice D is not an element

of DEP(A) ]

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide
Slide 11-
11- 71
71

11.3 Multivalued Dependencies and Fourth Normal Form (4NF)

Example 22A continuation

Let F={A  BC, DE  C} be a set of MVDs over ABCDE.

In a similar way DEP(AD) = { A, B, C, D, E }

DEP(BC) = { B, C, ADE } and so on…

Why?
Let us compute the dependency basis of AD. We begin with a set TAD consisting
of {AD, BCE}. Then we look for MVDs whose LHS is in AD, such as
A  B, A  C, A  A (see first part of this problem)
Therefore TAD = { AD, BCE, A, B, C, BCDE, CDE, BDE, DE, BC }
Refining of this set produces SAD = {A, B, C, D, E } = DEP(AD)

Similarly, computing DEP( BC ) begins with TBC = {BC, ADE} = SBC

Nothing else can be added to TBC. Now consider Y=BC and the rule DE  C.
The RHS of this rule and Y intersect but its LHS has nothing in common with
BC. Step3 of the DEP _BASIS algorithm tells us to replace BC with (BC∩C) and
(BC - C). That is, C and B. Therefore DEP( BC) = {B, C, ADE}.

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide
Slide 11-
11- 72
72

36
 11.3 Multivalued Dependencies and Fourth Normal Form (4NF)

Example 22B
Let F = { AEI, C  AB } be a set of MVDs over ABCDEI.
Find DEP(AC). See problems 21B and 21C

DEP(AC) = { A, B, C, D, EI }
Why?
Let us compute the dependency basis of AC.
We begin with a set TAC consisting of {AC, BDEI}. Then we look for MVDs
whose LHS is in AD, such as A  EI, A  BCD, CAB, C DEI.
Therefore TAC = { AC, BDEI, EI, BCD, AB, DEI }
Refining of this set produces SAC = {A, B, C, D, EI }.

In problem 21B we were asked to show that AC  BEI. Observe that B and EI
are elements in DEP(AC) therefore ACBEI, similarly AC  B (see problem
21C) .

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide
Slide 11-
11- 73
73

11.3 Multivalued Dependencies and Fourth Normal Form (4NF)

Example 22C. (from J. Ullman – Database Systems)

Consider the following relation
R= (Course, Teacher, Hour, Room, Student, Grade)

By observation we find the following MVDs

C  HR C  SG
C →T HT →R
HR →C CS →G
HS →R

Notice that DEP(C) = { T, SG, HR }.

Therefore C  T, C  HR, C  THR, C  TSG, C  HRSG, C  THRSG,

Also observe that the MVD C  R is invalid (Room and Hour are depend on each other)

Other example is DEP(HR) = {HR, TSG} from here we conclude it is not true that HR  SG

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide
Slide 11-
11- 74
74

37
 11.3 Multivalued Dependencies and Fourth Normal Form (4NF)

Definition. Nonadditive Join Decomposition into 4NF Relations

Whenever we decompose a relation schema R into R1 = (X  Y) and R2 = (R - Y)

based on an MVD X  Y that holds in R, the decomposition has the nonadditive
join property.

Property NJB. The relation schemas R1 and R2 form a non-additive join

decomposition of R with respect to a set F of functional and multivalued
dependencies if and only if

R1 ∩ R2  (R1 – R2) or
R1 ∩ R2  (R2 – R1)

Example:
The relation Book(Book#, Author, Price) with multivalued dependencies
F= { Book#  Author, Book# → Price } has a lossless decomposition in
R1 = (Book#, Author)
R2 = (Book#, Price)

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide
Slide 11-
11- 75
75

11.3 Multivalued Dependencies and Fourth Normal Form (4NF)

Definition Fourth Normal Form (4NF)

Fourth normal form (4NF) schema is violated when a relation has
undesirable non-trivial multivalued dependencies (it should be decomposed)

Definition.
A relation schema R is in 4NF with respect to a set of dependencies F (that
includes functional dependencies and multivalued dependencies) if, for
every nontrivial multivalued dependency X Y in F+, X is a superkey for R.

Observation.
In the previous definition the determination of superkey values is solely
based on functional dependencies in F.

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide
Slide 11-
11- 76
76

38
 11.3 Multivalued Dependencies and Fourth Normal Form (4NF)

Algorithm 11.5. Relational Decomposition into 4NF

Relations with Nonadditive Join Property
Input:
A universal relation R and a set of functional and multivalued dependencies
F.

Method
1. Set D:= { R }
2. While there is a relation schema Q in D that is not in 4NF, do
{ choose a relation schema Q in D that is not in 4NF;
find a nontrivial MVD X  Y in Q that violates 4NF;
replace Q in D by two relation schemas (Q - Y) and (XY);
}

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide
Slide 11-
11- 77
77

11.3 Multivalued Dependencies and Fourth Normal Form (4NF)

Example 23.
Consider the previous example where R= {CTHRSG} and the
dependencies are F = { C  HR, C  SG, C → T }.

1. Observe that MVDs in F are not trivial and C is NOT a superkey in R.

2. Therefore we could use the rule C  HR to decompose the table into

fragments: {R1=CHR, R2=CSGT}. The mvd C  HR in R1 becomes trivial.

3. Again, C is not a superkey in R2 and the MVD C  SG is not trivial

4. Therefore we could partition R2 into R21=CSG, R22=CT where R21 holds a

trivial MVD, and R22 has only one rule C → T in which C is a superkey.

5. The final 4NF Lossless database schema is D = { CHR, CSG, CT }

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide
Slide 11-
11- 78
78

39
 11.4 Join Dependencies and Fifth Normal Form (5NF)

Definition. Join Dependencies

A join dependency (JD), denoted by JD(R1, R2, . . . , Rn), on relation schema R,
specifies a constraint on the states r of R. The constraint requests that every legal
state r of R should have a non-additive join decomposition into R1, R2, . . . , Rn

That is, for every such r we have

* ( R1(r), R2(r), …, Rn(r) ) = r

Notice that an MVD is a special case of a JD where n = 2. That is, a JD denoted

as JD(R1, R2) implies an MVD (R1 ∩ R2)  (R1 – R2) (or (R2 – R1) ).

Definition
A join dependency JD(R1, R2, . . . , Rn), specified on relation schema R, is
a trivial JD if one of the relation schemas Ri in JD(R1, R2, . . . , Rn) is equal to R.

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide
Slide 11-
11- 79
79

11.4 Join Dependencies and Fifth Normal Form (5NF)

Definition. Fifth Normal Form (5NF) or Projection-Join NF (PJNF)

A relation schema R is in fifth normal form (5NF) with respect to a set F of

functional, multivalued, and join dependencies if, for every nontrivial join
dependency JD(R1, R2, . . . , Rn), in F+ every Ri is a superkey of R.

Observation
Fifth normal form deals with cases where information can be reconstructed from
smaller pieces of information that can be maintained with less redundancy. The
big difficulty of reaching 5NF is that all possible decompositions of R must
support the lossless join property.

Note: Discovering JDs in practical databases with hundreds of attributes is next to impossible.
Therefore, the current practice of database design pays little attention to them.

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide
Slide 11-
11- 80
80

40
 11.4 Join Dependencies and Fifth Normal Form (5NF)
Example 24. Fifth Normal Form (5NF) or Projection-Join NF (PJNF)

5NF demands that all possible projections of r(R) on a given decomposition DR ,

losslessly regenerate the original table. Consider a decomposition of the example
schema R= (Agent, Company, Product) into the following two sub-schemas:
R1(Agent, Company) and R2(Company, Product)

Using the given data is easy to proof that

R1(r) * R2(r)  r
(phantom tuples such as <Brown, GM, truck>
will be produced) . Hence R is not in 5NF.

However r(R) has a 5NF representation using

the three trivial schemas: R1, R2, R3 given below.

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide
Slide 11-
11- 81
81

11. 5 Inclusion Dependencies

Observation
Inclusion dependencies were defined in order to formalize two types of
inter- relational constraints that cannot be represented with FD/MVD/JDs:
• Foreign key (or referential integrity) relates attributes across relations.
• Inheritance represent a class/subclass relationship.

Definition.
An inclusion dependency R.X < S.Y between two sets of attributes-X of
relation schema R, and Y of relation schema S-specifies the constraint that,
at any specific time when r is a relation state of R and s a relation state of S,
we must have
 X(s(S))   Y(r(R))

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide
Slide 11-
11- 82
82

41
 11. 5 Inclusion Dependencies

Example 25
we can specify the following (referential
Integrity type) inclusion dependencies
on the relational schema in Figure 10.1

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide
Slide 11-
11- 83
83

11. 5 Inclusion Dependencies

Example 26
we can specify the following (inheritance type) inclusion dependencies on the
relational schema in Figure 7.1

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide
Slide 11-
11- 84
84

42
 11. 5 Inclusion Dependencies

Definition
As with other types of dependencies, there are inclusion dependency inference rules
(IDIRs). The following are three examples:

IDIR1 (reflexivity): R.X < R.X.

IDIR2 (attribute correspondence): If R.X < S.Y, where X = {A1, A2, . . . ,An}
and Y = {B1, B2, . . . , Bn} and Ai Corresponds to Bi
then R.Ai < S.Bi for 1i n.

IDIR3 (transitivity): If R.X < S.Y and S.Y < T.Z, then R.X < T.Z.

The preceding inference rules were shown to be sound and complete for inclusion
dependencies. So far, no normal forms have been developed based on inclusion
dependencies.

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide
Slide 11-
11- 85
85

11. 6 Template Dependencies

Rationale
Template dependencies provide a technique for representing constraints in relations
that typically have no easy and formal definitions.

The idea behind template dependencies is to specify a template-or example-that

defines each constraint or dependency.

There are two types of templates: tuple-generating templates and constraint

generating templates.

• A template consists of a number of hypothesis tuples that are meant to show an

example of the tuples that may appear in one or more relations.

• The other part of the template is the template-conclusion. the conclusion is a set
of tuples that must also exist in the relations if the hypothesis tuples are there.

• For constraint-generating templates, the template conclusion is a condition that

must hold on the hypothesis tuples.

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide
Slide 11-
11- 86
86

43
 11. 6 Template Dependencies

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide
Slide 11-
11- 87
87

11. 6 Template Dependencies

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide
Slide 11-
11- 88
88

44
 11. 6 Template Dependencies

Functional Dependencies Based on Arithmetic

Functions and Procedures

Sometimes the attributes in a relation may be related via some arithmetic function
or a more complicated functional relationship. As long as a unique value of Y is
associated with every X, we can still consider that the FD X →Y exists. For
example, in the relation
ORDER-LINE (Order#, Item#, Quantity, Unit-price, Extended-price, Discounted-price)

In this relation,
(Quantity, Unit-price ) → Extended-price

by the formula
Extended-price = Unit-price * Quantity.

Although the above kinds of FDs are technically present in most relations, they are
not given particular attention during normalization.

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide
Slide 11-
11- 89
89

References

David Maier, "The Theory of Relational Databases", Comp. Sc. Press, 1983.

Desai, Bipin, An Introduction to Database Systems, West Publishing, 1990

Ullman, J. Principles of Database Systems. Computer Science Press, 1982.

H. Mannila, K. Raiha, “The Design of Relational Databases”. Addison-Wesley, 1992.

Elmasri, Navathe, “Fundamentals of Database Systems”, 5th Ed. Addisson-Wesley, 2007.

Silberschatz, Korth, Surdarshan, “Database System Concepts”, McGraw-Hill, 2011

William Kent, "A Simple Guide to Five Normal Forms in Relational Database Theory",
Communications of the ACM 26(2), Feb. 1983

Questions ?

Copyright © 2007 Pearson Education, Inc. Publishing as Pearson Addison-Wesley

Slide 11- 90

45