1. Developing Relational Data Models
Assoc.Prof.Dr. Erkan TIN
Week-4
This presentation contains copyrighted material excerpted from
Database Management with Web Site Development Applications, G. Riccardi, Addison Wesley, 2003.

2. Relation is not Relationship
Be careful of these two words
Relation
Relationship
A relation is a table that contains a set of entities
A relationship is an association between two entities
We must be very careful to use the correct word
Listen closely to the lectures and correct me if I get it wrong!

3. Basics of the Relational Model
The relational model represents information in tables (called relations)
Each table represents a set of entities
Each column of a table represents the attribute values of the entities
Each row of a table represents a single entity
A database schema is a collection of table definitions (relation schemas)
A relational database is a collection of tables
Each table stores objects for a single relation schema

4. Relation Schemas and Keys
The rows of a relational table are unique
No 2 rows have the same values for all of the attributes
A key is a collection of attributes in which
No 2 rows have the same values for all attributes in the key
Every table must have a key
Why?
A relation schema is the specification of the structure of a table
Name of the table
Name and type of each attribute
Declaration of the key
A key declaration is a constraint
A table is not allowed to have 2 different rows that have the same value for the
key
Database systems enforce key constraints
By blocking any attempt to modify a table that will result in a violation of the key
constraint

5. Translating E-R Diagrams
We create a relational model from an E-R model
For each component of the E-R diagram, create a representation in the
relational model
This lecture describes how to systematically translate an E-R
model into a database schema
Entity classes
Simple attributes
Composite attributes
Key attributes
Relationship types of different cardinalities
Weak entity classes
Multi-valued attributes
Inheritance

6. Representing Entity Classes
For each strong entity class in your E-R model create a relation schema:
Rule 1a: Define a relation schema by the same name.
Rule 1b: For each single-valued attribute of the entity class
Create an attribute by the same name in the relation schema and specify a
type for the attribute
Rule 1c: Define the key of the new relation schema as the key of the
entity class
If the entity class key consists of multiple simple attributes, the key of the
relation schema will be that set of attributes.
Underline your selected key attribute in each schema in order to identify the
key.

7. Representing Entity Classes
purchase
Id
Date lineNumber
M
1 PurchaseOrder quantity
PurchaseOrder Has Detail
M M unitCost
Buys Orders
From
date
videoId
Acquired
1 movieId 1
Is
Supplier title Movie Copy Video
1 Of M
genre
name address
length rating

8. Composite Attributes
Rule 2. For each composite attribute of a lastName firstName accountId
strong entity class balance
Create an attribute in the relation otherUsers Customer
schema for each component Multi-
valued
attribute Attribute
Composite
numberRentals address
If appropriate, use the name of the Attribute
composite attribute as a prefix for
street zipcode
each of the component attribute Derived Component
Attribute Attribute
names city state
Schema: Customer (accountId number, lastName string, firstName string, street string, city
string, state string, zipcode string, balance number)
accountId lastName firstName street city state zipcode balance
101 Block Jane 1010 Main St. Apopka FL 30458 0.00
102 Hamilton Cherry 3230 Dade St. Dade City FL 30555 4.47
103 Harrison Kate 103 Dodd Hall Apopka FL 30457 30.57
104 Breaux Carroll 76 Main St. Apopka FL 30458 34.58

9. Representing Relationship Types as Attributes
A relationship type may be represented by attributes in a relation
schema
Add the key attributes of one table to the related table
Consider the IsCopyOf relationship type between Video and Movie
Key of Movie is movieId
Add attribute movieId to table Video
Schema Video (videoId number, [other attributes omitted] movieId
number references Movie)
Attribute movieId of relation Video is called a foreign key
Because its value is the value of the key of an entity in another (foreign)
table.
Referential Integrity Constraint: The value of a foreign key attribute of
an entity must be the key of an entity in the related table.

10. One-to-Many RelationshipTypes
For a one-to-many relationship
type
Add the key attributes of one entity
class to the other entity class
(foreign key attributes).
Add the foreign key attributes to
the class whose role is to-one.
Rule 3: For each one-to-many
Schema: Video(videoId number, dateAcquired date, movieId number
relationship type R between subject references Movie, storeId number references Store)
class S and target class T
Add the key attributes of class S to
class T as foreign keys
Name the attributes using the role
that S plays in relationship type R
Add the attributes of the relationship
type R to target class T.

11. One-to-Many RelationshipTypes
(How Do You Represent “Takes” Relationship? Part-I)
Child Takes
T-shirt
C1 Vedat Başaran T1 Black O
T2 White V
C2 Berrrak Temizsoy
T3 Yellow O
C3 Leyla Durukan
T4 Grey O
T5 Green O
Child
ChildId Name Surname
C1 Vedat Başaran
TShirt
C2 Berrak Temizsoy
ShirtId Color Collar
C3 Leyla Durukan
T1 Black O
T2 White V
We would like our database to answer the T3 Yellow O
question “Who Takes Which T-shirt?.” T4 Grey O
T5 Green O

12. One-to-Many RelationshipTypes
(How Do You Represent “Takes” Relationship? Part-II)
1. Adding ShirtId column to Child table introduces multiple-values.
Multi-valued attributes are not allowed in RDBMs!
Multiple values TShirt
Child
ShirtId Color Collar
ChildId Name Surname ShirtId
T1 Black O
C1 Vedat Başaran T1, T2
T2 White V
C2 Berrak Temizsoy T4
T3 Yellow O
C3 Leyla Durukan NULL
T4 Grey O
T5 Green O
2. Adding ChildId column to TShirt table solves the problem and
represents “Takes” relationship in the database.
TShirt
Child
ShirtId Color Collar ChildId
ChildId Name Surname
C1 Vedat Başaran T1 Black O C1
T2 White V C1
C2 Berrak Temizsoy
T3 Yellow O NULL
C3 Leyla Durukan
T4 Grey O C2
T5 Green O NULL

13. One-to-One Relationship Types
The foreign key attributes may be added to either schema
Each entity class is to-one in the relationship type
Choose which class to include the foreign key attributes
One option is to try to minimize the number of null values
Rule 4: For each one-to-one relationship type between two
classes, choose one class to be the subject and one to be
the target
Add the key attributes of the subject class to the target schema as
foreign key attributes
Add the attributes of the relationship type to the target schema, just
as in Rule 3

14. One-to-One RelationshipTypes
(How Do You Represent “Takes” Relationship? Part-I)
Child Takes
T-shirt
C1 Vedat Başaran T1 Black O
T2 White V
C2 Berrrak Temizsoy
T3 Yellow O
C3 Leyla Durukan
T4 Grey O
T5 Green O
Since the relationship “Takes” has cardinality ratio 1-to-1, one can add key attribute columns of one table to
the other.
However, key attribute columns of T-shirt table should be added to Child table (mandatory
participation side) in order to eliminate NULL values. The solutions are illustrated below.

15. One-to-One RelationshipTypes
(How Do You Represent “Takes” Relationship? Part-II)
1. Adding ShirtId column to Child table.
No NULL Value
TShirt
Child
ShirtId Color Collar
ChildId Name Surname ShirtId
T1 Black O
C1 Vedat Başaran T1
T2 White V
C2 Berrak Temizsoy T4
T3 Yellow O
C3 Leyla Durukan T2
T4 Grey O
T5 Green O
2. Adding ChildId column to TShirt table.
(Number of T-shirts) minus
(Number of Children) NULL values
TShirt
Child
ShirtId Color Collar ChildId
ChildId Name Surname
C1 Vedat Başaran T1 Black O C1
T2 White V C3
C2 Berrak Temizsoy
T3 Yellow O NULL
C3 Leyla Durukan
T4 Grey O C2
T5 Green O NULL

16. Many-to-Many Relationship Types
Many-to-Many relationship types between 2 classes cannot be represented as
simple attributes in either related table
Rule 5: For each many-to-many relationship type R between classes S and T
Create a new relation schema R
Add attributes to represent the key of S and the key of T as foreign key
attributes
The key of schema R is the combination of those attributes
Add the relationship attributes to schema R, as in Rule 3
Schema: WorksIn (name string references Employee, surname string references
Employee, storeId number references Store, startDate date)
WorksIn
name surname storeId startDate
Cevdet Narin 3 01.04.2005
Gamze Bulut 5 07.09.2006
Veli Hancı 5 03.02.2003

17. Weak Entity Classes (Part-I)
• Weak Entity classes have no keys of their own.
Weak Entity Owner Entity
Customer Class Class
1
1 1
Has Rental Has Video
M
dateDue dateRented cost
Identifying Schema: Rental(videoId number references Video,
Relationship
Type dateDue date, dateRented date, cost currency, accountId
number references Customer)
videoId dateRented dateDue cost accountId
• Create keys from
– Foreign keys of identifying relationship types 101 1/3/99 1/4/99 $1.59 103
– Partial keys of the weak class
113 2/22/99 2/25/99 $3.00 101
• Relation Rental, shown above, still
must add foreign key for Customer. 114 2/22/99 2/25/99 $3.00 101
123 12/1/98 12/31/98 $10.99 103

18. Weak Entity Classes (Part-II)
Defining
relationship
type
Rule 6: For each weak entity
class W
1 M M 1 Create a new relation
Employee Has TimeCard Has Store schema with the same name
For each identifying
Weak entity relationship,
discriminator class
startTime endTime Add the key attributes of
the related class to the new
schema as foreign key
attributes
Schema: TimeCard (ssn string references Employee, startTime
Declare the key of the
date, endTime date, storeId number references Store, paid boolean)
schema to be the
combination of
ssn startTime endTime storeId paid the foreign key attributes
and
145-09-0967 01/14/99 8:15 01/14/99 12:00 3 yes
the partial key attributes of
245-11-4554 01/14/99 8:15 01/14/99 12:00 3 yes the weak entity class
376-77-0099 02/23/99 19:00 02/24/99 2:00 5 yes Add the simple and
composite attributes of class
145-09-0967 01/16/99 8:15 01/16/99 12:00 3 yes
W to the schema, as in
376-77-0099 01/03/99 10:00 01/03/99 14:00 5 yes Rules 1b and 2
376-77-0099 01/03/99 15:00 01/03/99 19:00 5 yes

19. Multi-valued Attributes
lastName firstName accountId
balance
otherUsers Customer
Multi-
valued
Attribute
Composite
numberRentals address
Attribute
street zipcode
Derived Component
Attribute Attribute
city state
Schema: Customer (accountId number, lastName string, firstName string, street
string, city string, state string, zipcode string, otherUser string)
Can Store Singe Value.
For this reason cannot model our world.

20. Multi-valued Attributes
Representing multi-valued attributes directly in a relational table is not possible.
accountId lastName firstName street city state zipcode otherUser
104 Breaux Carroll 76 Main St. Apopka FL 30458 Judy Breaux
Cyrus Lambeaux
Jean Deaux
Multiple values
Dividing the entity into as many rows as the number of multiple-values is not a solution.
accountId lastName firstName street city state zipcode otherUser
104 Breaux Carroll 76 Main St. Apopka FL 30458 Judy Breaux
104 Breaux Carroll 76 Main St. Apopka FL 30458 Cyrus Lambeaux
104 Breaux Carroll 76 Main St. Apopka FL 30458 Jean Deaux
This schema conflicts with two basic principles of information systems:
1. A data model should correspond closely to the real situations it represents.
2. Data models should keep duplication of values to a minimum.

21. Multi-valued Attributes
Represent each multi-valued attribute as if it were a weak entity class with
Identifying relationship with owner class
All composite attributes of multi-valued attribute are partial keys
lastName firstName accountId
1 is a M
Customer OtherUser
part of
numberRentals
address otherUser
street city state zipcode
• Rule 7: For each multi-valued attribute M of an entity class C
– Define a new relation schema M
– Add the components of attribute M to the new schema
– Add the key attributes of the schema that contains the other attributes of C to M as a
foreign key
– Define the key of the new schema to be the combination of all of its attributes.
• The diagram above shows attribute otherUsers represented as a weak entity
class and as a schema
– Do not modify the E-R diagram. This diagram is included to illustrate the methodology.

22. Derived Attributes
A derived attribute is assigned a value which is computed by an external
function.
Derived attributes are in fact computed columns in relations.
In the example below, value of Age attribute is derived from the
DateOfBirth attribute by a function.
Child
ChildId Name Surname DateOfBirth Age
C1 Vedat Başaran 07.06.2001 =ToDay() – [DateOfBirth]
C2 Berrak Temizsoy 01.09.2000 =ToDay() – [DateOfBirth]
C3 Leyla Durukan 03.05.2003 =ToDay() – [DateOfBirth]
Or, one may call a procedure which goes over each record and compute the
child’s age similarly and update the corresponding Age value.

23. Inheritance
First representation strategy
Represent the superclass and each subclass as individual tables
Each table has the attributes of the corresponding class
Each subclass table has the key of the superclass as both key and
foreign key
Second representation strategy
Represent the superclass and all subclasses as a single table
The table has the attributes of superclass and all of the subclass
attributes
Third representation strategy
Represent each subclass as individual tables
Each table has the attributes of that subclass and the attributes of the
superclass

24. Example of First Strategy
movieId
Movie
date
videoId
Acquired
is 1
Superclass M
Video Copy
Of
Defining media Inheritance
attribute relationship
d type
Subclass
"dvd" "tape"
DVD Videotape
video
languages captioning format soundtrack
Format

25. Example of Second Strategy
• Schema Definition and Sample Table Representing Specialization as a
Single Table with Attributes from the Superclass and Subclasses
Schema: Video:(videoId number, dateAcquired date, media string, movieId number references Movie,
videoFormat string, languages string, captioning string, format string, soundtrack string)
videoId date media movieId video languages captioning format sound
Acquired Format track
101 1/25/98 dvd 101 letterbox English, yes
Spanish
111 2/5/97 tape 123 VHS English
112 12/31/95 tape 123 VHS Spanish
113 4/5/98 dvd 123 letterbox English, none
Russian,
French

26. Example of Third Strategy
Schemas Representing Specialization as One Table for Each Subclass
Schema: DVD:(videoId number, dateAcquired date, media string, movieId number references Movie, videoFormat string,
languages string, captioning string)
videoId date media movieId video languages captioning
Acquired Format
101 1/25/98 dvd 101 letterbox English, Spanish yes
113 4/5/98 dvd 123 letterbox English, Russian, none
French
Schema: Videotape:(videoId number, dateAcquired string, media string, movieId number references Movie, format
string, soundtrack string)
videoId date media movieId format sound
Acquired track
111 2/5/97 tape 123 VHS English
112 12/31/95 tape 123 VHS Spanish

27. Inheritance
Rule 8a:
Create a relation schema for each superclass C using rules 1 and 2.
For each subclass of C that has a defining attribute, add that attribute
to the schema for C.
For each subclass S, create a new relation schema.
Add the simple and composite attributes of class S to the schema, as
in Rules 1b and 2.
Add the key of the superclass C as a foreign key referencing relation
C.
Declare the key of the subclass relation for S to be this foreign key

28. Case in Point
Relational Model for Video Sales for BigHit Online
Process
Evaluate E-R model for BigHit Online
Transform E-R diagram into relation schemas
Apply rules 1-8
Evaluate relation schemas for clarity, accuracy and completeness

29. E-R Diagram for BigHit Online
1-to-M
Add primary key
attributes of to-many
Multi-valued side to to-one side
Attribute (i.e., accountId to
Needs to be Sale relation)
transformed
into a weak-
entity
1-to1
Adding cartId
attribute to Customer
relation instead of
ShoppingCart relation
to represent Select
relationship type
results in the
introduction of Null Inheritance: Overlaying
values. Quantity and Cost
attributes of subclasses
cannot be moved to
superclass Movie.
M-to-M
Create a new relation.

30. Creating Relation Schemas
After applying Rule 1, before adding relationship types
and inheritance
Customer: (accountId number, lastName string, firstName string)
Sale: (saleId number, dateSold date, totalCost number)
Movie: (movieId number, title string, genre string)
ShoppingCart: (cartId number, dateCreated date)
Adding composite attributes to Sale
Sale: (saleId number, dateSold date, totalCost number,
creditCardType string, creditCardExpiration date,
creditCardAccountNumber number)
Adding relationship type Purchases to Sale
Sale: (saleId number, dateSold date, totalCost number,
creditCardType string, creditCardExpiration date,
creditCardAccountNumber number, accountId number references
Customer)

31. Adding Relationship Types
Add Selects to ShoppingCart
ShoppingCart: (cartId number, dateCreated date, accountId number
references Customer)
Add schemas for many-to-many relationship types
SaleItem: (saleId number references Sale, movieId number references
Movie, quantity number)
CartItem: (cartId number references ShoppingCart, movieId number
references Movie, quantity number)
Add schemas for multi-valued attributes
CreditCards: (accountId number references Customer, accountNumber
number, type string, expiration date)
ShippingAddresses: (accountId number references Customer, street string,
city string, state string, zipcode string)
Add schemas for subclasses of Movie
DVD: (movieId number references Movie, videoFormat string, languages
string, captioning string, cost currency, quantity number)
Videotape: (movieId number references Movie, format string, soundtrack
string, cost currency, quantity number)

32. Resulting Database Schema for BigHit Online
Customer: (accountId number, lastName string, firstName string)
Sale: (saleId number, dateSold date, totalCost number, creditCardType string,
creditCardExpiration date, creditCardAccountNumber number, accountId
number references Customer)
Movie: (movieId number, title string, genre string)
ShoppingCart: (cartId number, dateCreated date, accountId number references
Customer)
SaleItem: (saleId number references Sale, movieId number references Movie,
quantity number)
CartItem: (cartId number references ShoppingCart, movieId number
references Movie, quantity number)
CreditCards: (accountId number references Customer, accountNumber
number, type string, expiration date)
ShippingAddresses: (accountId number references Customer, street string, city
string, state string, zipcode string)
DVD: (movieId number references Movie, videoFormat string, languages
string, captioning string, cost currency, quantity number)
Videotape: (movieId number references Movie, format string, soundtrack
string, cost currency, quantity number)