Lecture 2 CP 224: Relational Database Design

❖ Features of Good Relational Design

❖ Atomic Domains and First Normal Form
❖ Decomposition Using Functional Dependencies
❖ Functional Dependency Theory
❖ Algorithms for Functional Dependencies
❖ Decomposition Using Multivalued Dependencies
❖ More Normal Form
❖ Database-Design Process
❖ Modeling Temporal Data

Combine Schemas?
❖ Suppose we combine instructor and department into inst_dept

❖ Result is possible repetition of information

A Combined Schema Without Repetition

❖ Consider combining relations

▪ sec_class(sec_id, building, room_number) and
▪ section(course_id, sec_id, semester, year)
into one relation
▪ section(course_id, sec_id,
semester, year, building,
room_number)
❖ No repetition in this case
Disadvantage of Large/Combine Schemas

Update anomalies:
Modifying the budget in one tuple but not all tuples leads
to inconsistency.

Cannot insert a new department until the first instructor is

hired
 ID is part of PK.

Deleting the last instructor will lose the department

information
 Without Kim, there is no department information (Elec.
Eng.).

 Without Mozart, there is no department information

(Music).
What About Smaller Schemas?

❖ Suppose we had started with inst_dept. How would we know to split up

(decompose) it into instructor and department?
❖ Write a rule ―ifthere were a schema (dept_name, building, budget), then
dept_name would be a candidate key‖
❖ Denote as a functional dependency:
dept_name → building, budget
❖ In inst_dept, because dept_name is not a candidate key, the building and
budget of a department may have to be repeated.
l This indicates the need to decompose inst_dept
❖ Not all decompositions are good. Suppose we decompose
employee(ID, name, street, city, salary) into
employee1 (ID, name)
employee2 (name, street, city, salary)
❖ Next we shows how we lose information -- we cannot reconstruct the
original employee relation -- and so, this is a lossy decomposition.

A Lossy Decomposition
 Example of Lossless-Join Decomposition

❖ Lossless join decomposition

❖ Decomposition of R = (A, B, C)
R = (A, B) R = (B, C)
1 2

First Normal Form

❖ Domain is atomic if its elements are considered to be indivisible units

▪ Examples of non-atomic domains:
✓ Set of names, composite attributes
✓ Identification numbers like CS101 that can be broken up
into parts
❖ A relational schema R is in first normal form if the domains of all
attributes of R are atomic
❖ Non-atomic values complicate storage and encourage redundant
(repeated) storage of data
▪ Example: Set of accounts stored with each customer, and set of
owners stored with each account
▪ We assume all relations are in first normal form.

❖ Atomicity is actually a property of how the elements of the domain are

used.
• Example: Strings would normally be considered indivisible
• Suppose that students are given roll numbers which are strings of
the form CS0012 or EE1127
• If the first two characters are extracted to find the department, the
domain of roll numbers is not atomic.
• Doing so is a bad idea: leads to encoding of information in
application program rather than in the database.

❖ Non-1NF should be normalized.

ID Name Phone_number ID Name ID Phone_number

1 John {6601, 6602} 1 John 1 6601
1 6602
Goal for Normalization — Devise a Theory for the Following

❖ Decide whether a particular relation R is in ―good‖ form.

❖ In the case that a relation R is not in ―good‖ form, decompose it into a set
of relations {R 1, R 2, ..., R n} such that
• Each relation is in good form
• The decomposition is a lossless-join decomposition
❖ Our theory is based on:
• Functional dependencies
• Multivalued dependencies

Functional Dependencies

❖ Constraints on the set of legal relations.

❖ Require that the value for a certain set of attributes determines uniquely
the value for another set of attributes.
❖ A functional dependency is a generalization of the notion of a key.
 Practice

inst_dept (ID, name, salary, dept_name, building, budget ).

For each constraint, write the corresponding functional
dependency
Every instructor (ID) has only one name and one salary.
Ans:
Every department (dept_name) is located in only one
building and has only one budget.
Ans:
Every instructor belongs to only one department.
Ans:
Explain why each of the following functional dependency
doesn’t make sense.
dept_name → ID
Ans:
building → dept_name
Ans:
Use of Functional Dependencies
❖ We use functional dependencies to:
▪ Test relations to see if they are legal under a given set of
functional dependencies.
✓ If a relation r is legal under a set F of
functional dependencies, we say that r satisfies
F.
▪ Specify constraints on the set of legal relations
✓ We say that F holds on R if all legal relations on R
satisfy the set of functional dependencies F.
❖ Note: A specific instance of a relation schema may satisfy a
functional
dependency even if the functional dependency does not hold on
all legal instances.

▪ For example, a specific instance of instructor may, by

chance, satisfy
name → ID.

❖ A functional dependency is trivial if it is satisfied by all

instances of a relation
• Example:
✓ ID, name → ID
✓ name → name
• In general, α → β is trivial if β ⊆α

Closure of a Set of Functional Dependencies

❖ Given a set F of functional dependencies, there are certain
other functional dependencies that are logically implied by
F.
▪ For example: If A → B and B → C, then we can infer that
A →C
❖ The set of all functional dependencies logically implied by F is
the closure of F.
+
❖ We denote the closure of F by F .
+
❖ F is a superset of F.
Boyce-Codd Normal Form
A relation schema R is in BCNF with respect to a set F of functional
+
dependencies if for all functional dependencies in F of the form
α→β
where α ⊆R and β ⊆R, at least one of the following holds:

❖ α → β is trivial (i.e., β ⊆α)

❖ α is a superkey for R

Example schema not in BCNF:

instr_dept (ID, name, salary, dept_name, building, budget )
because dept_name→ building, budget
holds on instr_dept, but dept_name is not a superkey

Decomposing a Schema into BCNF

❖ Suppose we have a schema R and a non-trivial dependency α →β causes a

violation of BCNF.
We decompose R into:
• (α U β )
• (R-(β-α))
❖ In our example,
• α = dept_name
• β = building, budget
and inst_dept is replaced by
• (α U β ) = ( dept_name, building, budget )
• ( R - ( β - α ) ) = ( ID, name, salary, dept_name )

BCNF and Dependency Preservation

❖ Constraints, including functional dependencies, are costly to check in

practice unless they pertain to only one relation
❖ If it is sufficient to test only those dependencies on each individual
relation of a decomposition in order to ensure that all functional
dependencies hold, then that decomposition is dependency preserving.
❖ Because it is not always possible to achieve both BCNF and dependency
preservation, we consider a weaker normal form, known as third normal
form.
Third Normal Form

❖ A relation schema R is in third normal form (3NF) if for all:

+
α → β in F
at least one of the following holds:
• α → β is trivial (i.e., β ∈α)
• α is a superkey for R
• Each attribute A in β – α is contained in a candidate key for R.
(NOTE: each attribute may be in a different candidate key)
❖ If a relation is in BCNF it is in 3NF (since in BCNF one of the first two
conditions above must hold).
❖ Third condition is a minimal relaxation of BCNF to ensure dependency
preservation (will see why later).

Goals of Normalization

❖ Let R be a relation scheme with a set F of functional dependencies.

❖ Decide whether a relation scheme R is in ―good‖ form.
❖ In the case that a relation scheme R is not in ―good‖ form, decompose it
into a set of relation scheme {R , R , ..., R } such that
1 2 n
• Each relation scheme is in good form
• The decomposition is a lossless-join decomposition
• Preferably, the decomposition should be dependency preserving.

How good is BCNF?

❖ There are database schemas in BCNF that do not seem to be sufficiently

normalized
❖ Consider a relation
inst_info (ID, child_name, phone)
• Where an instructor may have more than one phone and can
have multiple children
How good is BCNF? (Cont.)

❖ There are no non-trivial functional dependencies and therefore the

relation is in BCNF
❖ Insertion anomalies – i.e., if we add a phone 981-992-3443 to 99999, we
need to add two tuples
(99999, David, 981-992-3443)
(99999, William, 981-992-3443)

Functional-Dependency Theory

❖ We now consider the formal theory that tells us which functional

dependencies are implied logically by a given set of functional
dependencies.
❖ We then develop algorithms to generate lossless decompositions into
BCNF and 3NF
❖ We then develop algorithms to test if a decomposition is dependency-
preserving

Closure of a Set of Functional Dependencies

❖ Given a set F set of functional dependencies, there are certain other

functional dependencies that are logically implied by F.
• For e.g.: If A → B and B → C, then we can infer that A → C
❖ The set of all functional dependencies logically implied by F is the closure
of F.
+
❖ We denote the closure of F by F .
Closure of a Set of Functional Dependencies
+,
❖ We can find F the closure of F, by repeatedly applying Armstrong’s
Axioms:
• if β ⊆α, then α → β (reflexivity)
• if α → β, then γ α → γ β (augmentation)
• if α → β, and β → γ, then α → γ (transitivity)
❖ These rules are
• sound (generate only functional dependencies that actually hold),
and
• complete (generate all functional dependencies that hold).
Example

❖ R = (A, B, C, G, H, I)
F={ A→B
A→C
CG → H
CG → I
B → H}
+
❖ some members of F
• A→H
✓ by transitivity from A → B and B → H
• AG → I
✓ by augmenting A → C with G, to get AG → CG
and then transitivity with CG → I
• CG → HI
✓ by augmenting CG → I to infer CG → CGI,
and augmenting of CG → H to infer CGI → HI,
and then transitivity

+
Procedure for Computing F

❖ To compute the closure of a set of functional dependencies F:

+
F = Frepeat
+
for each functional dependency f in F
apply reflexivity and augmentation rules on f
+
add the resulting functional dependencies to F
+
for each pair of functional dependencies f 1and f 2 in F
if f1 and f2 can be combined using transitivity
+ +
then add the resulting functional dependency to F until F does
not change any further
NOTE: We shall see an alternative procedure for this task later
Closure of Functional Dependencies (Cont.)

❖ Additional rules:
• If α → β holds and α → γ holds, then α → β γ holds (union)
• If α → β γ holds, then α → β holds and α → γ holds
(decomposition)
• If α → β holds and γ β → δ holds, then α γ → δ holds
(pseudotransitivity)
The above rules can be inferred from Armstrong’s axioms.

Closure of Attribute Sets

+
❖ Given a set of attributes α, define the closure of α under F (denoted by α )
as the set of attributes that are functionally determined by α under F
+
❖ Algorithm to compute α , the closure of α under F
result := α;
while (changes to result) do
for each β → γ in F do
begin
if β ⊆result then result := result 𝖴 γ
end

Example of Attribute Set Closure

❖ R = (A, B, C, G, H, I)
❖ F = {A → B
A→C
CG → H
CG → I
B → H}
+
❖ (AG)
1. result = AG
2. result = ABCG (A → C and A → B)
3. result = ABCGH (CG → H and CG ⊆AGBC)
4. result = ABCGHI (CG → I and CG ⊆AGBCH)

❖ Is AG a candidate key?
1. Is AG a super key?
+
1. Does AG → R? == Is (AG) ⊇R
2. Is any subset of AG a superkey?
+
1. Does A → R? == Is (A) ⊇R
+
2. Does G → R? == Is (G) ⊇R
Uses of Attribute Closure

There are several uses of the attribute closure algorithm:

❖ Testing for superkey:
+, +
• To test if α is a superkey, we compute α and check if α contains
all attributes of R.
❖ Testing functional dependencies
• To check if a functional dependency α → β holds (or, in other
+ +
words, is in F ), just check if β ⊆α .
+
•
That is, we compute α by using attribute closure, and then check
if it contains β.
• Is a simple and cheap test, and very useful
❖ Computing closure of F
+ +
• For each γ ⊆R, we find the closure γ , and for each S ⊆γ , we
output a functional dependency γ → S.
Canonical Cover

❖ Sets of functional dependencies may have redundant dependencies that

can be inferred from the others
• For example: A → C is redundant in: {A → B, B → C, A→ C}
• Parts of a functional dependency may be redundant
✓ E.g.: on RHS: {A → B, B → C, A → CD} can be
simplified to
{A → B, B → C, A → D}
✓ E.g.: on LHS: {A → B, B → C, AC → D} can be
simplified to
{A → B, B → C, A → D}
❖ Intuitively, a canonical cover of F is a ―minimal‖ set of functional
dependencies equivalent to F, having no redundant dependencies or
redundant parts of dependencies

Extraneous Attributes

❖ Consider a set F of functional dependencies and the functional

dependency α → β in F.
• Attribute A is extraneous in α if A ∈α
and F logically implies (F – {α → β}) 𝖴{(α – A) → β}.
• Attribute A is extraneous in β if A ∈β
and the set of functional dependencies
(F – {α → β}) 𝖴{α →(β – A)} logically implies F.
❖ Note: implication in the opposite direction is trivial in each of the cases
above, since a ―stronger‖ functional dependency always implies a weaker
one
❖ Example: Given F = {A → C, AB → C }
• B is extraneous in AB → C because {A → C, AB → C} logically
implies A → C (I.e. the result of dropping B from AB → C).
• A+= AC in {A → C, AB → C }
❖ Example: Given F = {A → C, AB → CD}
• C is extraneous in AB → CD since AB → C can be inferred
even after deleting C
• AB+=ABCD in {A → C, AB → CD}
Testing if an Attribute is Extraneous
❖ Consider a set F of functional dependencies and the functional
dependency α → β in F.
❖ To test if attribute A ∈α is extraneous in α
+
1. compute ({α} – A) using the dependencies in F
+
2. check that ({α} – A) contains β; if it does, A is extraneous in α
❖ To test if attribute A ∈β is extraneous in β
+
1. compute α using only the dependencies in
F’ = (F – {α → β}) 𝖴{α →(β – A)},
+
2. check that α contains A; if it does, A is extraneous in β

Canonical Cover
❖ A canonical cover for F is a set of dependencies F such that
c
• F logically implies all dependencies in F and
c,
• F logically implies all dependencies in F, and
c
• No functional dependency in F contains an extraneous attribute,
c
and
• Each left side of functional dependency in F is unique.
c

❖ To compute a canonical cover for F:repeat

Use the union rule to replace any dependencies in F
α → β and α → β with α → β β
1 1 1 2 1 1 2
Find a functional dependency α → β with an
extraneous attribute either in α or in β
/* Note: test for extraneous attributes done using F c, not F*/
If an extraneous attribute is found, delete it from α → β until F
does not change

❖ Note: Union rule may become applicable after some extraneous attributes
have been deleted, so it has to be re-applied
Computing a Canonical Cover

❖ R = (A, B, C)
F = {A → BC
B → C
A → B
AB → C}
❖ Combine A → BC and A → B into A → BC
• Set is now {A → BC, B → C, AB → C}
❖ A is extraneous in AB → C
• Check if the result of deleting A from AB → C is implied by the other
dependencies
4 Yes: in fact, B → C is already present!
• Set is now {A → BC, B → C}
❖ C is extraneous in A → BC
• Check if A → C is logically implied by A → B and the other dependencies
4 Yes: using transitivity on A → B and B → C.
– Can use attribute closure of A in more complex cases
❖ The canonical cover is: A → B
B→C
Lossless-join Decomposition
❖ For the case of R = (R , R ), we require that for all possible relations r on schema R
1 2

r = ∏R1 (r ) ∏R2(r )
❖ A decomposition of R into R and R is lossless join if at least one of the following
1 2
+
dependencies is in F :
• R ∩R →R
1 2 1
• R ∩R →R
1 2 2
❖ The above functional dependencies are a sufficient condition for lossless join
decomposition; the dependencies are a necessary condition only if all constraints are
functional dependencies

Example
Dependency Preservation
+
❖ Let F be the set of dependencies F that include only attributes in R .
i i
✓ A decomposition is dependency preserving, if
(F1 𝖴F2 𝖴… 𝖴Fn ) + = F+

✓ If it is not, then checking updates for violation of functional

dependencies may require computing joins, which is
expensive.

Testing for Dependency Preservation

❖ To check if a dependency α → β is preserved in a decomposition of R into
R , R , …, R we apply the following test (with attribute closure done with
1 2 n
respect to F)
• result = α while (changes to result) do
for each R in the decomposition
i
+
t = (result ∩ R i) ∩ R i
result = result 𝖴 t
•
If result contains all attributes in β, then the functional dependency
α → β is preserved.
❖ We apply the test on all dependencies in F to check if a decomposition is
dependency preserving
❖ This procedure takes polynomial time, instead of the exponential time
+ +
required to compute F and (F 𝖴F 𝖴… 𝖴F )
1 2 n

Example
❖ R = (A, B, C )
F = {A → B
B → C}
Key = {A}
❖ R is not in BCNF
❖ Decomposition R 1 = (A, B), R 2 = (B, C)
• R and R in BCNF
1 2
• Lossless-join decomposition
• Dependency preserving
Testing for BCNF
❖ To check if a non-trivial dependency α →β causes a violation of BCNF
+
1. compute α (the attribute closure of α), and
2. verify that it includes all attributes of R, that is, it is a superkey of R.

❖ Simplified test: To check if a relation schema R is in BCNF, it suffices to

check only the dependencies in the given set F for violation of BCNF,
+
rather than checking all dependencies in F .

• If none of the dependencies in F causes a violation of BCNF, then

+
none of the dependencies in F will cause a violation of BCNF
either.

❖ However, simplified test using only F is incorrect when testing a relation

in a decomposition of R

• Consider R = (A, B, C, D, E), with F = { A → B, BC → D}

✓ Decompose R into R = (A,B) and R = (A,C,D, E)
1 2
✓ Neither of the dependencies in F contain only attributes
from
(A,C,D,E) so we might be mislead into thinking R satisfies
2
BCNF.
+
✓ In fact, dependency AC → D in F shows R is not in BCNF.
2

❖ To check if a relation R in a decomposition of R is in BCNF,

i
• Either test R for BCNF with respect to the restriction of F to R
i i
+
(that is, all FDs in F that contain only attributes from R )i
• or use the original set of dependencies F that hold on R, but with
the following test:
+
– for every set of attributes α ⊆R i, check that α (the
attribute closure of α) either includes no attribute of
R - α, or includes all attributes of R .
i i

✓ If the condition is violated by some α → β in F, the

dependency
+
α → (α - α ) ∩ Rican be shown to hold on R i, and R i
violates BCNF.
✓ We use above dependency to decompose R
i
BCNF Decomposition Algorithm

Example of BCNF Decomposition

❖ R = (A, B, C )
F = {A → B
B → C}
Key = {A}

❖ R is not in BCNF (B → C but B is not superkey)

❖ Decomposition
• R = (B, C)
1
• R = (A,B)
2

❖ class (course_id, title, dept_name, credits, sec_id, semester, year, building,

room_number, capacity, time_slot_id)

❖ Functional dependencies:
• course_id→ title, dept_name, credits
• building, room_number→capacity
• course_id, sec_id, semester, year→building, room_number,
time_slot_id

❖ A candidate key {course_id, sec_id, semester, year}.

❖ BCNF Decomposition:

• course_id→ title, dept_name, credits holds

✓ but course_id is not a superkey.
• We replace class by:
✓ course(course_id, title, dept_name, credits)
✓ class-1 (course_id, sec_id, semester, year,
building, room_number, capacity,
time_slot_i
BCNF Decomposition (Cont.)

❖ course is in BCNF
• How do we know this?
❖ building, room_number→capacity holds on class-1
• but {building, room_number} is not a superkey for class-1.
• We replace class-1 by:
✓ classroom (building, room_number, capacity)
✓ section (course_id, sec_id, semester, year, building,
room_number, time_slot_id)
❖ classroom and section are in BCNF.

BCNF and Dependency Preservation

It is not always possible to get a BCNF decomposition that is

dependency preserving
❖ R = (J, K, L )
F = {JK → L
L→K}
Two candidate keys = JK and JL
❖ R is not in BCNF
❖ Any decomposition of R will fail to preserve
JK → L
This implies that testing for JK → L requires a join

Third Normal Form: Motivation

❖ There are some situations where

• BCNF is not dependency preserving, and
• Efficient checking for FD violation on updates is important
❖ Solution: define a weaker normal form, called Third Normal
Form (3NF)
• Allows some redundancy (with resultant problems; we will see
examples later)
• But functional dependencies can be checked on individual
relations without computing a join.
• There is always a lossless-join, dependency-preserving
decomposition into 3NF.
3NF Example

❖ Relation dept_advisor:
• dept_advisor (s_ID, i_ID, dept_name)
F = {s_ID, dept_name → i_ID, i_ID → dept_name}
• Two candidate keys: s_ID, dept_name, and i_ID, s_ID
• R is in 3NF
✓ s_ID, dept_name → i_ID s_ID
– dept_name is a superkey
✓ i_ID → dept_name
– dept_name is contained in a candidate key
Redundancy in 3NF

Testing for 3NF

+
❖ Optimization: Need to check only FDs in F, need not check all FDs in F .
❖ Use attribute closure to check for each dependency α → β, if α is a
superkey.
❖ If α is not a superkey, we have to verify if each attribute in β is contained
in a candidate key of R
• This test is rather more expensive, since it involve finding
candidate keys
• Testing for 3NF has been shown to be NP-hard
• Interestingly, decomposition into third normal form can be
done in polynomial time.
3NF Decomposition Algorithm

3NF Decomposition Algorithm (Cont.)

❖ Above algorithm ensures:

• Each relation schema Ri is in 3NF
• Decomposition is dependency preserving and lossless-join
• Proof of correctness is at end of this presentation
3NF Decomposition: An Example

❖ Relation schema:
cust_banker_branch = (customer_id, employee_id, branch_name, type )
❖ The functional dependencies for this relation schema are:
1. customer_id, employee_id → branch_name, type
2. employee_id → branch_name
3. customer_id, branch_name → employee_id
❖ We first compute a canonical cover
st
1. branch_name is extraneous in the r.h.s. of the 1 dependency
2. No other attribute is extraneous, so we get F =
C

customer_id, employee_id → type

employee_id → branch_name
customer_id, branch_name → employee_id
3NF Decompsition Example (Cont.)
❖ The for loop generates following 3NF schema:
(customer_id, employee_id, type )
(employee_id, branch_name)
(customer_id, branch_name, employee_id)

• Observe that (customer_id, employee_id, type ) contains a candidate

key of the original schema, so no further relation schema needs be
added

❖ At end of for loop, detect and delete schemas, such as (employee_id,

branch_name), which are subsets of other schemas

• Result will not depend on the order in which FDs are considered

❖ The resultant simplified 3NF schema is:

(customer_id, employee_id, type)
(customer_id, branch_name, employee_id)

Comparison of BCNF and 3NF

❖ It is always possible to decompose a relation into a set of relations that

are in 3NF such that:
• The decomposition is lossless
• The dependencies are preserved
❖ It is always possible to decompose a relation into a set of relations that are
in BCNF such that:
• The decomposition is lossless
• It may not be possible to preserve dependencies.

Design Goals
❖ Goal for a relational database design is:
• BCNF.
• Lossless join.
• Dependency preservation.
❖ If we cannot achieve this, we accept one of
• Lack of dependency preservation
• Redundancy due to use of 3NF
❖ Interestingly, SQL does not provide a direct way of specifying functional
dependencies other than superkeys.
Can specify FDs using assertions, but they are expensive to test, (and
currently not supported by any of the widely used databases!)

❖ Even if we had a dependency preserving decomposition, using SQL we

would not be able to efficiently test a functional dependency whose left
hand side is not a key.
Multivalued Dependencies

❖ Suppose we record names of children, and phone numbers for

instructors:

• inst_child(ID, child_name)
• inst_phone(ID, phone_number)

❖ If we were to combine these schemas to get

• inst_info(ID, child_name, phone_number)

• Example data:
(99999, David, 512-555-1234)
(99999, David, 512-555-4321)
(99999, William, 512-555-1234)
(99999, William, 512-555-4321)

❖ This relation is in BCNF

• Why?

Multivalued Dependencies (MVDs)

❖ Let R be a relation schema and let α ⊆R and β ⊆R. The

multivalued dependency
α →→ β
holds on R if in any legal relation r(R), for all pairs for tuples t and t in r
1 2
such that t [α] = t [α], there exist tuples t and t in r such that:
1 2 3 4

t [α] = t [α] = t [α] = t [α]

1 2 3 4
t [β] = t [β]
3 1
t [R – β] = t [R – β]
3 2
t [β] = t [β]
4 2
t [R – β] = t [R – β]
4 1

MVD (Cont.)

❖ Tabular representation of α →→ β
Example

❖ Let R be a relation schema with a set of attributes that are partitioned

into 3 nonempty subsets.
Y, Z, W
❖ We say that Y →→ Z (Y multidetermines Z )if and only if for all possible
relations r (R )
< y , z , w > ∈r and < y , z , w > ∈r
1 1 1 1 2 2

then
< y , z , w > ∈r and < y , z , w > ∈r
1 1 2 1 2 1
❖ Note that since the behavior of Z and W are identical it follows that
Y →→ Z if Y →→ W
Example (Cont.)

❖ In our example:
ID →→ child_name
ID →→ phone_number
❖ The above formal definition is supposed to formalize the notion that given
a particular value of Y (ID) it has associated with it a set of values of Z
(child_name) and a set of values of W (phone_number), and these two sets
are in some sense independent of each other.
❖ Note:
• If Y → Z then Y →→ Z
• Indeed we have (in above notation) Z = Z The claim follows.
1 2
Use of Multivalued Dependencies

❖ We use multivalued dependencies in two ways:

1. To test relations to determine whether they are legal under a given
set of functional and multivalued dependencies
2. To specify constraints on the set of legal relations. We shall thus
concern ourselves only with relations that satisfy a given set of
functional and multivalued dependencies.
❖ If a relation r fails to satisfy a given multivalued dependency, we can
construct a relations r′ that does satisfy the multivalued dependency by
adding tuples to r.
 Theory of MVDs

❖ From the definition of multivalued dependency, we can derive the

following rule:
• If α → β, then α →→ β
That is, every functional dependency is also a multivalued dependency
+
❖ The closure D of D is the set of all functional and multivalued
dependencies logically implied by D.
+
• We can compute D from D, using the formal definitions of
functional dependencies and multivalued dependencies.
• We can manage with such reasoning for very simple multivalued
dependencies, which seem to be most common in practice
• For complex dependencies, it is better to reason about sets of
dependencies using a system of inference rules (see Appendix C).

Fourth Normal Form

❖ A relation schema R is in 4NF with respect to a set D of functional and

+
multivalued dependencies if for all multivalued dependencies in D of the
form α →→ β, where α ⊆R and β ⊆R, at least one of the following hold:
• α →→ β is trivial (i.e., β ⊆α or α 𝖴β = R)
• α is a superkey for schema R
❖ If a relation is in 4NF it is in BCNF

Restriction of Multivalued Dependencies

n The restriction of D to R is the set D consisting of

i i
+
l All functional dependencies in D that include only attributes of R
i
l All multivalued dependencies of the form
α →→ (β ∩ R )
i
+
where α ⊆R i and α →→ β is in D
4NF Decomposition Algorithm
+
result: = {R};done := false;compute D ;
+
Let D denote the restriction of D to R
i i

while (not done)

if (there is a schema R in result that is not in 4NF) then
i
begin
let α →→ β be a nontrivial multivalued dependency that holds
on Ri such that α → R i is not in D i, and α∩β=φ;
result := (result - R ) 𝖴(R - β) 𝖴(α, β); end else done:= true;
i i
Note: each R is in 4NF, and decomposition is lossless-join
i
Example
❖ R =(A, B, C, G, H, I) F
={ A →→ B
B →→ HI
CG →→ H }
❖ R is not in 4NF since A →→ B and A is not a superkey for R
❖ Decomposition
a) R = (A, B) (R is in 4NF)
1 1
b) R = (A, C, G, H, I) (R is not in 4NF, decompose into R and R )
2 2 3 4
c) R = (C, G, H) (R is in 4NF)
3 3
d) R = (A, C, G, I) (R is not in 4NF, decompose into R and R )
4 4 5 6

• A →→ B and B →→ HI ➔ A →→ HI, (MVD

transitivity), and
• and hence A →→ I (MVD restriction to R )
4

e) R = (A, I) (R is in 4NF)
5 5
(R is in 4NF) f)R = (A, C, G)
6 6

Further Normal Forms

❖ Join dependencies generalize multivalued dependencies

• lead to project-join normal form (PJNF) (also called
fifth normal form)
❖ A class of even more general constraints, leads to a normal form
called domain-key normal form.
❖ Problem with these generalized constraints: are hard to reason
with, and no set of sound and complete set of inference rules exists.
Hence rarely used

Overall Database Design Process

❖ We have assumed schema R is given
• R could have been generated when converting E-R diagram to a
set of tables.
• R could have been a single relation containing all attributes that
are of interest (called universal relation).
• Normalization breaks R into smaller relations.
• R could have been the result of some ad hoc design of relations,
which we then test/convert to normal form.
ER Model and Normalization

❖ When an E-R diagram is carefully designed, identifying all entities

correctly, the tables generated from the E-R diagram should not need
further normalization.
❖ However, in a real (imperfect) design, there can be functional
dependencies from non-key attributes of an entity to other attributes of
the entity
• Example: an employee entity with attributes
department_name and building,
and a functional dependency
department_name→ building
• Good design would have made department an entity
❖ Functional dependencies from non-key attributes of a relationship set
possible, but rare --- most relationships are binary

Denormalization for Performance

Other Design Issues

❖ Some aspects of database design are not caught by normalization

❖ Examples of bad database design, to be avoided:
Instead of earnings (company_id, year, amount ), use
• earnings_2004, earnings_2005, earnings_2006, etc., all on the
schema (company_id, earnings).
✓ Above are in BCNF, but make querying across years
difficult and needs new table each year
• company_year (company_id, earnings_2004, earnings_2005,
earnings_2006)
✓ Also in BCNF, but also makes querying across years
difficult and requires new attribute each year.
✓ Is an example of a crosstab, where values for one attribute
become column names
✓ Used in spreadsheets, and in data analysis tools
Modeling Temporal Data

❖ Temporal data have an association time interval during which the data
are valid.
❖ A snapshot is the value of the data at a particular point in time
❖ Several proposals to extend ER model by adding valid time to
• attributes, e.g., address of an instructor at different points in time
• entities, e.g., time duration when a student entity exists
• relationships, e.g., time during which an instructor was associated
with a student as an advisor.
❖ But no accepted standard
❖ Adding a temporal component results in functional dependencies like
ID → street, city
not to hold, because the address varies over time
❖ A temporal functional dependency X → Y holds on schema R if the
functional dependency X → Y holds on all snapshots for all legal
instances r (R).
Modeling Temporal Data (Cont.)

❖ In practice, database designers may add start and end time attributes to
relations
• E.g., course(course_id, course_title) is replaced by
course(course_id, course_title, start, end)
✓ Constraint: no two tuples can have overlapping valid times
– Hard to enforce efficiently
❖ Foreign key references may be to current version of data, or to data at a
point in time
• E.g., student transcript should refer to course information at the
time the course was taken
End of Chapter

Proof of Correctness of 3NF Decomposition Algorithm

Correctness of 3NF Decomposition Algorithm

❖ 3NF decomposition algorithm is dependency preserving (since there is a

relation for every FD in F )
c
❖ Decomposition is lossless
• A candidate key (C ) is in one of the relations R i in decomposition
• Closure of candidate key under F must contain all attributes in R.
c
• Follow the steps of attribute closure algorithm to show there is
only one tuple in the join result for each tuple in R
i
Correctness of 3NF Decomposition Algorithm (Cont’d.)
Claim: if a relation R is in the decomposition generated by the
i
above algorithm, then R satisfies 3NF.
i

❖ Let Ri be generated from the dependency α → β

❖ Let γ → B be any non-trivial functional dependency on R i. (We need only
consider FDs whose right-hand side is a single attribute.)
❖ Now, B can be in either β or α but not in both. Consider each case
separately.

❖ Case 1: If B in β:

• If γ is a superkey, the 2nd condition of 3NF is satisfied

• Otherwise α must contain some attribute not in γ
+
• Since γ → B is in F it must be derivable from F , by using
c
attribute closure on γ.
• Attribute closure not have used α →β. If it had been used, α must
be contained in the attribute closure of γ, which is not possible,
since we assumed γ is not a superkey.
• Now, using α→ (β- {B}) and γ → B, we can derive α →B
(since γ ⊆α β, and B ∉ γ since γ → B is non-trivial)
• Then, B is extraneous in the right-hand side of α →β; which is not
possible since α →β is in F c.
• Thus, if B is in β then γ must be a superkey, and the second
condition of 3NF must be satisfied.

Correctness of 3NF Decomposition (Cont’d.)

❖ Case 2: B is in α.
• Since α is a candidate key, the third alternative in the definition of
3NF is trivially satisfied.
• In fact, we cannot show that γ is a superkey.
• This shows exactly why the third alternative is present in the
definition of 3NF.
Q.E.D.
Figure 8.02

Figure 8.03

Figure 8.04
Figure 8.05

Figure 8.06

Figure 8.14

Figure 8.15

Figure 8.17
31Fatimah AL-Shaikh