1. Database Design ERD and Normalisation
FdSc in ICT

2. Problems with a flat file design
The same ID number can appears in multiple rows. You cannot retrieve a unique record with the ID number
Entities may not have an ID number.
Details are repeated – DATA REDUNDANCY, wasting space.
Data can be INCONSISTENT as it is stored more than once
Multiple records have to be updated. This is an AMENDMENT ANOMALY
Multiple records have to be deleted. This is a DELETION ANOMALY
More than one record may need to be added. This is an ADDITION ANOMALY
Data may not be ATOMIC.

3. Improving flat files
Split into smaller tables with a defined relationship between them
At the design stage we can use an Entity Relationship Diagram (E-R)

4. E-R diagram A dog can only have one owner One owner can have many dogs
The relationship is that:
A dog can only have one owner
One owner can have many dogs
This is a “one to many” relationship
OWNER
DOG

5. Result of E-R diagram ID Initial Surname Title Address Postcode
Dog Name
Gender
DOB
Breed 1 A
Smith
Mrs
4 High St, Hereford
HR1 1ZX
Ant M
21/08/02
Alsatian
Dec
Jordan F
08/08/04
Terrier Ms
Rooney
23/10/05
Poodle 2 C
Miles Mr
72 Castle Road, Ledbury
HR7 AA
Fang
14/03/07
Chihuahua ID
Initial
Surname
Title
Address
Postcode 1 A
Smith
Mrs
4 High St, Hereford
HR1 1ZX 2 C
Miles Mr
72 Castle Road, Ledbury
HR7 AA
DogID
Dog Name
Gender
DOB
Breed ID 1
Ant M
21/08/02
Alsatian 2
Dec 3
Jordan F
08/08/04
Terrier 4
Rooney
23/10/05
Poodle 5
Fang
14/03/07
Chihuahua

6. Keys Primary Key Primary Key Foreign Key ID Initial Surname Title
Address
Postcode 1 A
Smith
Mrs
4 High St, Hereford
HR1 1ZX 2 C
Miles Mr
72 Castle Road, Ledbury
HR7 AA
DogID
Dog Name
Gender
DOB
Breed ID 1
Ant M
21/08/02
Alsatian 2
Dec 3
Jordan F
08/08/04
Terrier 4
Rooney
23/10/05
Poodle 5
Fang
14/03/07
Chihuahua

7. Why is this better? Member’s details are only stored once
Each dog has a unique ID
New owners only entered once
Changes only made in one place
Details need only be deleted from one record ID
Initial
Surname
Title
Address
Postcode 1 A
Smith
Mrs
4 High St, Hereford
HR1 1ZX 2 C
Miles Mr
72 Castle Road, Ledbury
HR7 AA
DogID
Dog Name
Gender
DOB
Breed ID 1
Ant M
21/08/02
Alsatian 2
Dec 3
Jordan F
08/08/04
Terrier 4
Rooney
23/10/05
Poodle 5
Fang
14/03/07
Chihuahua

8. Further improvements OWNER DOG BREED
Now many dogs can belong to a breed
A dog can only be of one breed

9. The tables now look like thisID
Initial
Surname
Title
Address
Postcode 1 A
Smith
Mrs
4 High St, Hereford
HR1 1ZX 2 C
Miles Mr
72 Castle Road, Ledbury
HR7 AA
DogID
Dog Name
Gender
DOB
BreedID ID 1
Ant M
21/08/02 2
Dec 3
Jordan F
08/08/04 4
Rooney
23/10/05 5
Fang
14/03/07
Breed ID
Breed 1
Alsatian 2
Terrier 3
Poodle 4
Chihuahua

10. The benefits are Member’s details are only stored once
Each dog has a unique ID
New owners only entered once
Changes only made in one place
Details need only be deleted from one record
We only need to store the details of each breed once
A new breed can be added without an owner

11. Many to many relationships
Many students can take many exams This will be a problem for the tables
STUDENT
EXAM ID
Name
Exam 1
Exam 2
Exam 3
Exam 4
Exam n
TOS
Tosker
4046
3545
6354
2875
9756

12. Many to many problems How many columns do we need for exams?
A record in an exam table won’t relate to one student
What is the solution?

13. A link table STUDENT TAKES EXAM What can we add here? ExamID Exam 1
French 2
German 3
Italian
StudentID
Name 1
Carl 2
Adam 3
Jane
What can we add here?
StudentID
ExamID 1 3 2

14. to identify unique records
A better solution
STUDENT
TAKES
EXAM
Primary Key
to identify unique records
StudentID
Name 1
Carl 2
Adam 3
Jane
TakesID
StudentID
ExamID 1 2 3 4 5 6
ExamID
Exam 1
French 2
German 3
Italian

15. The benefits are We only store the student details once
We only store the exam details once
Each primary key is unique
We can recombine data from the three tables into one record

16. Exercise Draw an ERD for:
A company which employs sales staff who take orders from customers
DVDs which contain films with scenes in which actors appear

17. Optionality
The name given to a relationship which may or may not exist
A company employs sales staff
It may also use agency sales staff
One company employs one or more sales staff
One sales staff is employed by the company or no company
COMPANY
SALES STAFF
COMPANY
SALES STAFF

18. Attributes
An attribute is an item whose properties are associated with an entity
Car – colour, model, VIN
Student – Name, address,
Attributes can only have one value at a particular point in time
An address may change, but for anybody retrieving the data at a specific time, they only get one value

19. Attributes Attributes added under the entity name COMPANY SALES STAFF
Address
Tel no
Fax no
VAT no
Name
Address
Salary
Age

20. Exercise
Add attributes to the ERD diagram you drew for the DVD.

21. Exercise Look again at the coal merchant’s requirements
Assume that the solution will involve a database
Create an ERD diagram for the database
Draw a diagram and describe each step you take
Clearly identify how you resolve many to many relationships
Include optionality and attributes
Underline the primary keys

22. Normalisation
We have just down the top down Entity Relationship diagram method
We now need to try the bottoms up approach called normalisation

23. Draw up a typical paper record for the project you are converting to a database
In this example we are using records of potential students enquiring about courses
Course Name
AVCE in ICT
Length
2 years
Tutor
Bob Higgie
Location
Main campus
Location code
001
Level 3
Date
Name
Address
DOB
Action
10/10/04
W Rooney
Manchester
5/08/87
Send details
19/10/04
A Wenger
London
5/3/47
Tutor to call
20/10/04
A Ferguson
28/4/42

24. Draw up a typical paper record for the project you are converting to a database
In this example we are using records of potential students enquiring about courses
Course Name
AVCE in ICT
Length
2 years
Tutor
Bob Higgie
Location
Main campus
Location code
001
Level 3
Date
Name
Address
DOB
Action
10/10/04
W Rooney
Manchester
5/08/87
Send details
19/10/04
A Wenger
London
5/3/47
Tutor to call
20/10/04
A Ferguson
28/4/42
Repeating group

25. Draw up a landscape table with 5 columns
0NF
1NF
2NF
3NF
Table names

26. List all the attributes from the paper record in the 0NF column
Table names

27. List all the attributes from the paper record in the 0NF column
Table names
Course Name
Length
Tutor
Location
Location Code
Level
Date
Name
Address
DOB
Action

28. Course Name Length Tutor Location Location Code Level (Date Name
Identify the group or groups of attributes that are repeated on the paper record. Put a bracket round all of them
0NF
1NF
2NF
3NF
Table names
Course Name
Length
Tutor
Location
Location Code
Level
(Date
Name
Address
DOB
Action)

29. Add a primary key for the repeating group and for the group only entered once.
0NF
1NF
2NF
3NF
Table names
CourseID
Course Name
Length
Tutor
Location
Location Code
Level
(EnquirerID
Date
Name
Address
DOB
Action)

30. Copy the non-repeating group across into the 1NF column
Copy the non-repeating group across into the 1NF column. This group is already in 1NF as it doesn’t have any brackets around it.
0NF
1NF
2NF
3NF
Table names
CourseID
Course Name
Length
Tutor
Location
Location Code
Level
(EnquirerID
Date
Name
Address
DOB
Action)

31. Move the repeating group across into the 1NF column, making sure it is separated from the group already there. Add the key from the non-repeating group to it and remove the brackets
0NF
1NF
2NF
3NF
Table names
CourseID
Course Name
Length
Tutor
Location
Location Code
Level
(EnquirerID
Date
Name
Address
DOB
Action)
EnquirerID
Action

32. This is now in first normal form
There are no repeating groups
The second group has a compound primary key (EnquirerID/CourseID)
This corresponds to 2 tables
All records can be recombined into one record by using the keys
To put the database into 2NF we need to make sure it is in 1NF and all the non-key attributes depend entirely on the primary key

33. Move groups which only have a simple primary key into the 2NF column
Table names
CourseID
Course Name
Length
Tutor
Location
Location Code
Level
(EnquirerID
Date
Name
Address
DOB
Action)
EnquirerID
Action

34. CourseID Course Name Length Tutor Location Location Code Level
Next we identify those attributes in the other group that do not depend on both keys.
0NF
1NF
2NF
3NF
Table names
CourseID
Course Name
Length
Tutor
Location
Location Code
Level
(EnquirerID
Date
Name
Address
DOB
Action)
EnquirerID
Action

35. Move the non-key attributes that depend on only one key across to the 2NF column. Then copy the attribute they do depend on into the group with them.
0NF
1NF
2NF
3NF
Table names
CourseID
Course Name
Length
Tutor
Location
Location Code
Level
(EnquirerID
Date
Name
Address
DOB
Action)
EnquirerID
Action

36. The remaining group (EnquirerID, CourseID, Date and Action are then moved into the 2NF column and given a unique ID. In our case we call it the EnquiryID.
0NF
1NF
2NF
3NF
Table names
CourseID
Course Name
Length
Tutor
Location
Location Code
Level
(EnquirerID
Date
Name
Address
DOB
Action)
EnquirerID
Action
EnquiryID

37. Look at attributes that are not part of a key and work out if they depend on another attribute and not the primary key of the group they are in.
0NF
1NF
2NF
3NF
Table names
CourseID
Course Name
Length
Tutor
Location
Location Code
Level
(EnquirerID
Date
Name
Address
DOB
Action)
EnquirerID
Action
EnquiryID

38. Move Location across into its own group and copy Location Code into it as a primary key. Then Location Code is also left as a foreign key in the CourseID group.
0NF
1NF
2NF
3NF
Table names
CourseID
Course Name
Length
Tutor
Location
Location Code
Level
(EnquirerID
Date
Name
Address
DOB
Action)
EnquirerID
Action
EnquiryID

39. The final task is to give meaningful names to each group, which are now the tables to go into the database.
0NF
1NF
2NF
3NF
Table names
CourseID
Course Name
Length
Tutor
Location
Location Code
Level
(EnquirerID
Date
Name
Address
DOB
Action)
EnquirerID
Action
EnquiryID
Course
Enquirer
Enquiry

40. Notation
Standard notation starts with the name of the table, followed the fieldnames separated by commas, the primary key field is underlined
NAME (FIELD 1, FIELD 2, FIELD 3)
Course (CourseID, Course Name, Length, Tutor, Location Code, Level
Location (Location Code, Location)
Enquirer (EnquirerID, Name, Address, DOB )
Enquiry (EnquiryID, CourseID, EnquirerID, Date, Action)

41. The End The database is now in third normal form
It is possible that the design arising from the two methods differs.
It is then up to the database designer to decide which solution to implement.

42. Examples Normalise a bank account Customer ID 43543 Surname Williams
Postcode
BJ42 8AD
Landline
Account Number
Account Type
Description
Balance
Gold
Savings
£4560
Platinum
Current
£826.42
Silver
ISA
£3000

43. Example with 2 repeating groups
Applicant ID
2098
Address
Case officer ID
Name
Internal phone
Surname
Di Canio
Landline
231
Ferguson
417
First Name
Paolo
Date of birth
9/7/1968
Employer
Job
Start date
End date
Final salary
Swindon
County Ground
Manager
20/5/2011
18/02/2013
£250,000
Sunderland
Stadium of Light
31/03/2013
22/12/2013
£750,000
Subject
Level
Grade
Year
English
Functional Skills
Pass
2011
Sports
Coaching
Merit
2009

44. Assignment Task 5 (20%): From the coal merchant’s brief:
Identify the attributes for a bottom up model.
Normalise the database to third normal form.