COMP 421: Files and Databases

Lecture 16: Schema Refinement

1
Back to Database Design
• Do we just translate the ER diagram?

In HW1, we saw that, we don’t really need a separate table for the relationship PLACES

2
Back to Database Design
• Is it a good idea to store the same information multiple times?

An instance of the ORDERS table

3
Back to Database Design
• Decompose

.
.
An instance of the ORDERS table . 4
Anomalies due to Redundant Storage
• Update anomaly: If one copy of a repeated data is updated, an
inconsistency is created unless all copies are similarly updated
• Rating determines the hourly wage

ssn name lot rating hourly_wages hours_worked

123-22-3666 Attishoo 48 8 10 40
231-31-5368 Smiley 22 8 10 30
131-24-3650 Smethurst 35 5 7 30
434-26-3751 Guldu 35 5 7 32
612-67-4134 Madayan 35 8 10 40
An instance of the Hourly_Emps table

Suppose, we update hourly_wages for Attishoo, what will happen to Smiley and Madayan?
5
Anomalies due to Redundant Storage
• Insertion anomaly: It may not be possible to store certain information
unless some other, unrelated, information is stored
• Rating determines the hourly wage

ssn name lot rating hourly_wages hours_worked

123-22-3666 Attishoo 48 8 10 40
231-31-5368 Smiley 22 8 10 30
131-24-3650 Smethurst 35 5 7 30
434-26-3751 Guldu 35 5 7 32
612-67-4134 Madayan 35 8 10 40
An instance of the Hourly_Emps table

Suppose, we want to insert a new employee whose rating is 6. what should be the hourly_wage?
6
Anomalies due to Redundant Storage
• Delete anomaly: It may not be possible to store certain information
without losing some other, unrelated, informatio
• Rating determines the hourly wage

ssn name lot rating hourly_wages hours_worked

123-22-3666 Attishoo 48 8 10 40
231-31-5368 Smiley 22 8 10 30
131-24-3650 Smethurst 35 5 7 30
434-26-3751 Guldu 35 5 7 32
612-67-4134 Madayan 35 8 10 40
An instance of the Hourly_Emps table

Suppose, we delete Shemthusrt and Guldu’s informaiton from the table. How to keep the
association between rating and hourly_wages? 7
Decompositions
• Replace the relation with a collection of smaller relations
ssn name lot rating hourly_wages hours_worked
123-22-3666 Attishoo 48 8 10 40
231-31-5368 Smiley 22 8 10 30
131-24-3650 Smethurst 35 5 7 30
434-26-3751 Guldu 35 5 7 32
612-67-4134 Madayan 35 8 10 40

ssn name lot rating hours_worked rating hourly_wages

123-22-3666 Attishoo 48 8 40 8 10
231-31-5368 Smiley 22 8 30 5 7
131-24-3650 Smethurst 35 5 30
434-26-3751 Guldu 35 5 32
612-67-4134 Madayan 35 8 40
8
Eliminate Redundancy
• Redundancy is the root of several problems associated with the schema
• Redundant storage
• Insert/delete/update anomalies
• Functional Dependencies, FD can be used to identify schemas with
such problems
• rating → hourly_wage
• cid → shippingAddress
• Main refinement technique: decomposition
• Another table with rating, hourly_wage
• Another table with cid, ShippingAddress

9
Functional Dependency
• A functional dependency is a kind of Integrity Constraint (IC)
• X → 𝑌 holds over relation R if the following holds for every pair of
tuples 𝑡1 and 𝑡2
• If 𝑡1 . 𝑋 = 𝑡2. 𝑋 then 𝑡1 . 𝑌 = 𝑡2 . 𝑌

A B C D
a1 b1 c1 d1
Do these FDs hold over the given relation?
a1 b1 c1 d1
A→𝐷
a1 b2 c2 d1 AB → 𝐶
a2 b1 c3 d2
a1 b2 c2 d1

10
Primary Key
• A special case of an FD beccause it uniquely determines the value of
other attributes
• Since no two tuples can have the same primary key, we never
encounter a situation where 𝑡1 . 𝑋 = 𝑡2. 𝑋, so no chance of violation the
FD condition: If 𝑡1 . 𝑋 = 𝑡2. 𝑋 then 𝑡1 . 𝑌 = 𝑡2 . 𝑌
What are the FDs in Relation?
P A B C D
P→𝐴 A→𝐷
p1 a1 b1 c1 d1 P→𝐵 AB → 𝐶
p2 a1 b1 c1 d1 P→𝐶
p3 a1 b2 c2 d1 P→𝐷
p4 a2 b1 c3 d2
P → 𝐴𝐵
P → 𝐴𝐵C
p5 a1 b2 c2 d1
P → 𝐴𝐵𝐶𝐷
11
Primary Key
• FD X → 𝑌 does not require the X to be minimal
• If X is minimal, then it’s a key, candidate key
• If not, then it’s a super key

P A B C D P→𝐴 𝑃𝐴 → 𝐴𝐵𝐶𝐷
p1 a1 b1 c1 d1 P→𝐵 𝑃𝐵 → 𝐴𝐵𝐶𝐷
p2 a1 b1 c1 d1 P→𝐶
P→𝐷
p3 a1 b2 c2 d1
P → 𝐴𝐵
p4 a2 b1 c3 d2 P → 𝐴𝐵C
p5 a1 b2 c2 d2 P → 𝐴𝐵𝐶𝐷

𝑃𝐴, 𝑃𝐵 are superkeys, 𝑃 is a primary key

12
Closure of a set of FDs
• Set of FD, S: A → 𝐷, AB → 𝐶, P → 𝐴𝐵𝐶𝐷
• Closure of S includes set of all FDs implied by S
• Apply Armstrong’s Axioms 3 rules repeatedly to infer all FDs
• Reflexivity: if X ⊇ 𝑌, then X → 𝑌
• Augmentation: If X → 𝑌, then XZ → 𝑌𝑍 for any 𝑍
• Transitivity: If X → 𝑌 and Y → 𝑍, then X → 𝑍
• Additional rules:
• Union: If X → 𝑌 and X → 𝑍, then X → 𝑌𝑍
• Decomposition: If X → 𝑌𝑍 then X → 𝑌and X → 𝑍

If XZ → 𝑌 then X → 𝑌 and Z → 𝑌

You can add attributes to LHS, but cannot remove anything from LHS
13
Closure of a set of FDs
• Suppose, we have a relation names Contracts
• Contracts(contractid, supplierid, taskid, deptid, partid, qty, value)
• Schema for Contracts: CSTDPQV
• This schema holds following ICs:
• The contractid C is a key: 𝐶 → 𝐶𝑆𝑇𝐷𝑃𝑄𝑉
• A task purchases a given part using a single contract: 𝑃𝑇 → 𝐶
• A department purchases at most one part from a supplier: 𝑆𝐷 → 𝑃

14
Closure of a set of FDs
• Suppose, we have a relation names Contracts
• Contracts(contractid, supplierid, taskid, deptid, partid, qty, value)
• Schema for Contracts: CSTDPQV
• This schema holds following ICs:
• The contractid C is a key: 𝐶 → 𝐶𝑆𝑇𝐷𝑃𝑄𝑉
• A task purchases a given part using a single contract: 𝑃𝑇 → 𝐶
• A department purchases at most one part from a supplier: 𝑆𝐷 → 𝑃

From 𝑃𝑇 → 𝐶 and 𝐶 → 𝐶𝑆𝑇𝐷𝑃𝑄𝑉, using transitivity we get 𝑃𝑇 → 𝐶𝑆𝑇𝐷𝑃𝑄𝑉

From 𝑆𝐷 → 𝑃, using augmentation we get 𝑆𝐷𝑇 → 𝑃𝑇

From 𝑃𝑇 → 𝐶𝑆𝑇𝐷𝑃𝑄𝑉 and 𝑆𝑇𝐷 → 𝑃𝑇, using transitivity we get 𝑆𝑇𝐷 → 𝐶𝑆𝑇𝐷𝑃𝑄𝑉
You can not conclude 𝑆𝐷 → 𝐶𝑆𝐷𝑃𝑄𝑉 by removing T from both sides
15
Attribute Closure
• Given a set of functional dependencies (FDs) in a relation, the total set
of FDs that can be inferred is called the closure of the FDs.
• Computing the closure can be very expensive
• Typically we just want to check if a given FD X→Y is in the closure
of a set of FDs
• Suppose we have the following FDs:
• 𝐹 = {𝐴 → 𝐵, 𝐵 → 𝐶, 𝐶 → 𝐸}
• Is 𝐴 → 𝐸 in the closure 𝐹?
• If A’s closure contains E, then we can say 𝐴 → 𝐸

16
Attribute Closure
Algorithm to find closure

1. Closure_X: Start with the initial set of attributes 𝑋.

2. For each functional dependency 𝑋 → 𝑌
• If the left-hand side 𝑋 is a subset of the current closure, add the right-hand side 𝑌 to
the closure.
3. Repeat this process until no more attributes can be added to the closure.

𝐹 = {𝐴 → 𝐵, 𝐵 → 𝐶, 𝐶 → 𝐸}
Is 𝐴 → 𝐸 in the closure 𝐹?
so we have to find A’s closure

17
Attribute Closure
Algorithm to find A’s closure

1. Closure_X: Start with the initial set of attributes 𝑋.

2. For each functional dependency 𝑋 → 𝑌
• If the left-hand side 𝑋 is a subset of the current closure, add the right-hand side 𝑌 to
the closure.
3. Repeat this process until no more attributes can be added to the closure.

𝐹 = {𝐴 → 𝐵, 𝐵 → 𝐶, 𝐶 → 𝐸} Closure = {𝐴}
Is 𝐴 → 𝐸 in the closure 𝐹?

18
Attribute Closure
Algorithm to find A’s closure

1. Closure_X: Start with the initial set of attributes 𝑋.

2. For each functional dependency 𝑋 → 𝑌
• If the left-hand side 𝑋 is a subset of the current closure, add the right-hand side 𝑌 to
the closure.
3. Repeat this process until no more attributes can be added to the closure.

𝐹 = {𝐴 → 𝐵, 𝐵 → 𝐶, 𝐶 → 𝐸} Closure = {𝐴}
Is 𝐴 → 𝐸 in the closure 𝐹? LHS of 𝐴 → 𝐵 (=A) is a subset of the current closure
we add RHS of 𝐴 → 𝐵 (=B) to closure set
Closure = {𝐴, 𝐵}

19
Attribute Closure
Algorithm to find A’s closure

1. Closure_X: Start with the initial set of attributes 𝑋.

2. For each functional dependency 𝑋 → 𝑌
• If the left-hand side 𝑋 is a subset of the current closure, add the right-hand side 𝑌 to
the closure.
3. Repeat this process until no more attributes can be added to the closure.

𝐹 = {𝐴 → 𝐵, 𝐵 → 𝐶, 𝐶 → 𝐸} Closure = {𝐴}
Is 𝐴 → 𝐸 in the closure 𝐹? LHS of 𝐴 → 𝐵 (=A) is a subset of the current closure
we add RHS of 𝐴 → 𝐵 (=B) to closure set
Closure = {𝐴, 𝐵}

20
Attribute Closure
Algorithm to find A’s closure

1. Closure_X: Start with the initial set of attributes 𝑋.

2. For each functional dependency 𝑋 → 𝑌
• If the left-hand side 𝑋 is a subset of the current closure, add the right-hand side 𝑌 to
the closure.
3. Repeat this process until no more attributes can be added to the closure.

𝐹 = {𝐴 → 𝐵, 𝐵 → 𝐶, 𝐶 → 𝐸} Closure = {𝐴, 𝐵}
Is 𝐴 → 𝐸 in the closure 𝐹? LHS of 𝐵 → 𝐶 (=B) is a subset of the current closure
we add RHS of 𝐵 → 𝐶 (=C) to closure set
Closure = {𝐴, 𝐵, 𝐶}

21
Attribute Closure
Algorithm to find A’s closure

1. Closure_X: Start with the initial set of attributes 𝑋.

2. For each functional dependency 𝑋 → 𝑌
• If the left-hand side 𝑋 is a subset of the current closure, add the right-hand side 𝑌 to
the closure.
3. Repeat this process until no more attributes can be added to the closure.

𝐹 = {𝐴 → 𝐵, 𝐵 → 𝐶, 𝐶 → 𝐸} Closure = {𝐴, 𝐵, 𝐶}
Is 𝐴 → 𝐸 in the closure 𝐹? LHS of C→ 𝐸 (=C) is a subset of the current closure
we add RHS of 𝐶 → 𝐸 (=E) to closure set
Closure = {𝐴, 𝐵, 𝐶, 𝐸}

Now that A’s closure contains E, then we can say 𝐴 → 𝐸

22
Exercise
• Consider a relation R with 5 attributes ABCDE. You are given the
following dependencies:
• 𝐴→𝐵
• 𝐵𝐶 → 𝐸
• 𝐸𝐷 → 𝐴

1. List all keys for R (try all combinations: A, B, AB, AC …) -- difficult

2. Lets see if ACD is a key?
• Find closure of ACD, check if all 5 attributes are in the closure
3. Lets see if ADE is a key?
4. Lets see if BCD is a key?
5. Lets see if CDE is a key?

23
Exercise
• Consider a relation R with 5 attributes ABCDE. You are given the
following dependencies:
• 𝐴→𝐵
• 𝐵𝐶 → 𝐸
• 𝐸𝐷 → 𝐴

2. Lets see if ACD is a key?

• Find closure of ACD, check if all 5 attributes are in the closure
• Closure = {A, C, D}
• Using 𝐴 → 𝐵, LHS = A which is in the current closure, so add RHS = B to the closure,
closure = {A, B, C, D}
• Using 𝐵𝐶 → 𝐸, LHS = BC which is in the current closure, so add RHS = E to the closure,
closure = {A, B, C, D, E}

24