Chapter 1 Functional Dependencies The topic of Functional Dependencies is very important for databases because it makes it possible to analyse

how the attributes in a given set of relations relate to each other. Analysing these relationships is essential if one wants to avoid problems when the information is actually stored in the database. Examples of these problems will be seen in more detail in the next Chapter, when we speak about Normalisation. 1.1 A simpli_ed view of the topic In its simplest form, a functional dependency (FD) in databases, as the name says, is a dependency of the value of one attribute of a relation on the value of another attribute (not necessarily distinct). To understand what a functional dependency is, it is useful to recall the concept of a function in mathematics. I will make a quick recap here stressing only the properties which are important for us. A function is simply an association of elements of two sets X and Y . The set X is called the domain of the function and the set Y is called its image. This association is such that \every element in X is associated with exactly one element in Y ". Note that di_erent elements in X can be associated to the same element in Y . What is not allowed is that the same element in X is mapped to di_erent elements in Y . Figure 1.1 shows examples of associations and whether they qualify as functions or not. The association on the far right end is not a function because x 2 is associated with both y 2 and y 3. Let us now bring the concept back to _eld of databases. The elements in X and Y are

simply the values in the attributes of the relation. The simplest case in databases occurs when X and Y are just one attribute each, but we will see that, in general, X and Y might be a composition of attributes. Consider the relation \Albums" below. 1

Notes on Logical Database Design & Normalization

Database design primarily consists of two parts: logical design and physical design. Logical database design requires following certain sets of rules called rules of normalization Normalization Conditions: Any database designer must address two fundamental issues: Storing data on the disk that is most efficient in disk space usage- resulting in low cost. Fetching and saving data with the fastest response time resulting in high performance.

For your better understanding I am giving you the live examples from Project with a clear explanation to the 3 important rules of Normalization: Scope of my project: To display all the Doctors information who gives lectures on my companys products. Rule I: The first rule of normalization requires removing repeating data values and specifies that no two rows can be identical in a database. This means that each entity must have a Primary key (One or more columns of the table) which uniquely identifies a row in the table. For e.g.: - Find below the table Professional that stores all the Doctors info. Each doctor is uniquely identified from this table because I have primary key constraints on the following columns. PROFESSIONAL PROF_ID DB2_SPKR_NUM LAST_NAME FIRST_NAME MIDDLE_INITIAL SUFFIX SALUTATION SSN EMAIL_ADDRESS ME_NUM CID REPSTRY_POST_DTM SPEC_ID

PRIMARY KEYS

SSN & CID are two unique numbers for a Professional (Doctor). So, at any point of time there cannot be a professional repeated twice in this table.

Rule II: An entity is in the second normal form if it confirms to the first normal form and all non-key attributes of the table are fully dependent on the entire primary key. If the primary key consists of multiple columns, then nonkey columns should depend on the entire key and not just the subset of the key. LAST_NAME FIRST_NAME MIDDLE_INITIAL Non Key columns SUFFIX SALUTATION These columns directly depend on the primary key ( SSN,CID) and not partially.

Rule III: An entity is in the third normal form if it already conforms the first two normal forms and none of none key elements are dependent on any other non-key attributes. All such attributes should be removed from the table. PROFESSIONAL PROF_ID DB2_SPKR_NUM LAST_NAME FIRST_NAME MIDDLE_INITIAL SUFFIX SALUTATION SSN EMAIL_ADDRESS ME_NUM CID REPSTRY_POST_DTM SPEC_ID FULL_NAME

If FULL_NAME = LAST_NAME + FIRST_NAME Existence of Full_Name column in the table violates third normal form because a non key attribute (Full_Name) is dependent on two other non key attributes (Last_Name and First_Name). Therefor, to confirm the third rule of normalization, you must remove the Full_Name column from Professional table. Advantages of Normalization: Because info. Is logically kept togeather, normalization provides better understanding of the system. With less redundant data, it is easier to maintain referential integrity for the system. Because tables are smaller with normalization, index creation and data sorts are much faster.

 Disadvantages of Normalization: The main goal of normalization is to reduce redundancy in the system. As a result of normalization data is stored in different tables. To retrieve or modify data, you usually have to establish several multiple joins across multiple tables which can have adverse impact on performance of the system.

CS3

Tutorial on Normalisation

Lucy Hederman

1. The determinacy diagram for table X(A,B,C,) is:

B A C

A B a 1 b1 ; An occurrence of the table is: a 2 b2 a 3 b2 a 4 b2

C c1 c1 c2 c2

Suppose a fifth row, starting with a2 as the value of A, were to be added. What must be the value of attribute B? What must be the value of attribute C? Why would the fifth row be illegal? Can attribute A contain duplicate values? Is attribute A a candidate key? 2. Shoes are sold in a variety of styles and sizes. A style is identified by a style#. Each style has a single description (e.g. men's slippers) and the same description may apply to several styles. The attribute weekly-sales represents the number of shoes of a particular size and style sold in the previous week (e.g. 25 pairs, style# 17 size 8). The attribute monthly-stylevalue represents the total sales value in the previous month for each style. Draw a determinacy diagram for the attributes style#, style-description, size, weekly-sales, monthlystyle-value. Identify the candidate key for the diagram. Derive a set of well-normalised tables for the determinancy diagram- i.e. remove to a new table any determinant that is not a candidate key. 3. Every room in a building is identified by a room# and has precisely one telephone. Each telephone has its own distinct extension#. There are two types of telephone, internal dialing only (type I), and external/internal dialing (type E). Information on rooms and telephones will be held in the table: Office (room#, number-of-occupants, telephone-extension#, telephone type). Making any further plausible assumptions necessary: (a) draw a determinacy diagram for Office; (b) write down the candidate key(s); (c) is the corresponding table well-normalised ? (d) if not derive a set ofwell-normalised tables. 4. Repeat question 3 but with the addition of the attributes employee# and employee-name. Values of employee# identify individual employees. Each employee has only one name and occupies only one room. 5. Repeat question 4 but allowing several telephones per room. All employees in a room share all the telephones in that room.

6. Tuples in the table EMP(ENAME, PNAME, DNAME) represent the fact that employee named ENAME works on project PNAME and has dependent DNAME. Plausible enterprise rules will result in multi-valued dependencies in this relation. State these rules and indicate the multi-valued dependencies. Give a sample relation instance which displays redundancy. Suggest fully normalised relations for the same information and show the resulting relation instances for your sample. 7. The figure below shows an invoice from a company called Bilbo and Baggins. a). Draw a determinancy diagram for the attributes in the invoice. Omit both "derivable" attributes, such as amount-due, and fixed attributes, such as VAT-regno, which are likely to be the same for all invoices. Assume plausible enterprise rules and state your assumptions. b) Derive a set of well-normalised tables from your determinancy diagram.

Normalisation
The theory of Relational Database Design

Introduction
Normalisation is a theory for designing relational schema that make sense and work well. Well-normalised tables avoid redundancy and thereby reduce inconsistencies. Redundancy is unnecessary duplication. In well-normalised DBs semantic dependencies are maintained by primary key uniqueness.
CS3/3ICT2 Normalisation 2

Goals of Normalisation
Eliminate certain kinds of redundancy avoid certain update anomalies good reresentation of real world simplify enforcement of DB integrity

CS3/3ICT2 Normalisation

Update anomalies
Undesirable side-effects that occur when performaing insertion, modification or deletion operations on badly designed relational DBs.
SSN 987 654 333 321 678 467 Name J Smith M Burke A Dolan K Doyle O ONeill R McKay Dept 1 2 1 1 3 2 DeptMgr Dept Name 321 467 ... 321 321 678 467
CS3/3ICT2 Normalisation

Representing Department info in the Employee table causes problems.

4

Sample anomalies
Modification when the manager of a dept changes we have to change many values. If we are not careful the DB will contain inconsistencies. There is no easy way to get the DB to ensure that a department has only one manager and only one name.

CS3/3ICT2 Normalisation

Anomalies continued
Deletion if O ONeill leaves we delete his tuple and lose
the fact that there is a department 3 the name of dept 3 who is the manager of dept. 3

Insertion
how would we create a new department before any employees are assigned to it ?
CS3/3ICT2 Normalisation 6

Better design
Separate entities are represented in separate tables.
SSN 987 654 333 321 678 467 Name J Smith M Burke A Dolan K Doyle O ONeill R McKay Dept 1 2 1 1 3 2 Dept 1 2 3 DeptMgr Dept Name 321 467 ... 678

Note that mapping from an ER model following the steps given will give a well-normalised DB.
CS3/3ICT2 Normalisation 7

Boyce-Codd Normal Form

After a lot of other approaches Boyce and Codd noticed a simple rule for ensuring tables are well-normalised. Tables which obey the rule are in BCNF (Boyce Codd Normal Form). BCNF rule: Every determinant in a table must be a candidate key for that table.
CS3/3ICT2 Normalisation 8

Determinants
A is a determinant of B if each value of A has precisely one (possibly null) associated value of B. Said another way A is a determinant of B if and only if whenever two tuples agree on their A value they agree on their B value.
A B
CS3/3ICT2 Normalisation 9

Determinants
Note that determinancy depends on semantics of data
cannot be decided from individual table occurences.

Alternative terminology
if A (functionally) determines B then B is (functionally) dependent on A

CS3/3ICT2 Normalisation

10

Example determinants
SSN determines employee name SSN determines employee department Dept. No. determines Dept. Name Dept. Name determines Dept. No.
assuming Dept. names are also unique

Emp. Name does not determine Emp. Dept

two John Smiths could be in difft. Depts.

Emp. Name does not determine SSN.

CS3/3ICT2 Normalisation 11

Determinancy Diagram
Name SSN Department Dept. Name

Dept. Mgr

In general key attributes of an entity determine all the single-valued attributes of the entity.

CS3/3ICT2 Normalisation

12

Composite Determinants
(SSN, Project#) together determine the hours that the employee works on the project. Suppose packsize of a part depends on the supplier.
Name SSN hours Project# PName

S# packsize P# PName

CS3/3ICT2 Normalisation

13

Superfluous Attrbiutes
Superfluous attributes
If SSN determines name, so does (SSN, Dept) and (SSN, Dept, salary), etc. Always remove superfluous attributes from determinants.

CS3/3ICT2 Normalisation

14

Transitive Dependencies
SSN actually determines DeptMgr but only because
DeptNo SSN

SSN determines DeptNo and DeptNo determines DeptMgr.

Dept. Mgr

Be careful to remove transitive dependencies.

They mess up normalisation.
CS3/3ICT2 Normalisation 15

Candidate keys
candidate key = any attribute or set of attributes which will be unique for a table (set of attributes).
As well as the primary key there may be other candidate keys. E.g. DNUMBER and DNAME are both candidate keys for the Department table.

Key = row identifier Candidate key = candidate identifier

CS3/3ICT2 Normalisation 16

Finding candidate keys

Every key is by definition a determinant of all other attributes in a relation.
So in a diagram, any attribute (or composite) from which all other attributes are reachable is a candidate key.
Name

(SSN, Project#) is a (composite) candidate key for a table containing these five attributes.

SSN hours Project# PName

17

CS3/3ICT2 Normalisation

What are the candidate keys ?

B E D F P Q R G H V S K L M T N U W Z Y B A C
CS3/3ICT2 Normalisation 18

student teacher E subject D F

Problems occur when ...

Redundancy and anomalies occur when there are determinants which are not candidate keys.
SSN Name

SSN is the only key for a table containing these attributes

all attributes are reachable from SSN.

DeptNo

Dept. Name

SSN, DeptNo and DeptName are determinants

they have arrows coming out of them.
CS3/3ICT2 Normalisation

Dept. Mgr

19

BCNF rule
In well-normalised relations (BoyceCodd normal form) every determinant is a candidate key.
SSN Name DeptNo Dept. Mgr The employee/dept table decomposed to BCNF. Note that both DeptNo and DeptName are candidate keys of the second table.
CS3/3ICT2 Normalisation 20

DeptNo

Dept. Name

Transformation to BCNF
Create new tables such that each non-key determinant is a candidate key in a new table. The new table contains the attributes which are directly determined by the new candidate key.
V V X W B A C Z A W Y V V W Z X Y B A C BCNF tables : (V, X) (A, B, C) (V, W, Z, A) (V, W, Y)
21

CS3/3ICT2 Normalisation

Other Normal Forms

First NF - no multi-valued attributes
all relational DBs are 1NF

2NF - every non-key attribute is fully dependent on the primary key G H J 3NF - eliminate functional Table is in 2NF dependencies between non-key but not 3NF attributes
all dependencies can then be enforced by uniqueness of keys.
CS3/3ICT2 Normalisation 22

BCNF vs. 3NF

BCNF goes further than 3NF, some say too far. A 3NF table that has no overlapping composite keys is in BCNF.
student teacher subject 3NF, not BCNF keys: (student, subject) (student, teacher) teacher is a determinant teacher subject student teacher

A teacher teaches only one subject. For a given subject a given student has only one teacher.

BCNF but tables are not independent

CS3/3ICT2 Normalisation 23

4NF : Multi-valued dependencies

If a course can have multiple teachers and multiple texts, blind mapping to 1NF will give
Subject Physics Physics Physics Physics Maths Maths Maths Teacher Green Brown Green Brown Green Green Green Text Basic Mechanis Basic Mechanics Principles of Optics Principles of Optics Basic Mechanics Vector Analysis Trigonometry
CS3/3ICT2 Normalisation

which clearly has redundancy.

24

Fully-normalised
BCNF relations are well-normalised Fully-normalised relations are those with no multi-valued dependencies (4NF) and no join dependencies (5NF).

CS3/3ICT2 Normalisation

25

3ICT2 Additional Normalisation Tutorial (Sample Exam Questions)

1. A database for a multi-branch bank is to record the following attributes {account#, customer-name, customer-address, branch#, branch-address, creditcode, credit-limit}. An account# is unique within a branch. A customer may have many accounts at one or many branches of the bank. Joint accounts are not supported. Assume that customers can be identified by name. Each account is assigned a credit code. The code determines the credit limit. For example accounts with code B have a credit limit of 500. (a) Draw a determinancy diagram for the seven attributes. Note any assumptions that you make. Derive a set of well-normalised (BCNF) relations from your determinancy diagram. Underline the primary key for each relation. Write down a sequence of relational algebra operations to get the names and addresses of all customers whose credit limit is less than 100.

(b)

(c)

2. An employee database is to hold information about employees, the department they are in and the skills which they hold. The attributes to be stored are emp-id, emp-name, emp-phone, dept-name, dept-phone, dept-mgrid, skill-id, skill-name, skill-date, skill-level An employee may have many skills, such as word-processing, typing, librarian, filing, ... The date on which the skill was last tested and the level displayed at that test are recorded for the purposes of assigning work and determining salary. An employee is attached to one department and each department has a unique manager. (a) Draw a dependency diagram for the above database, stating clearly any assumptions that you make. Derive a set of well normalised (BCNF) relations, indicating the primary key of each relation. Write down a sequence of relational algebra operations to get the names and phone numbers of employees who can both file and type to a level of 6 or more.

(b)

(c)

More Relational Algebra and SQL exercises

1. Consider the following relational database schema. It is intended to represent who will eat what kinds of sandwiches and the places which serve the various kinds of sandwiches. A sample database instance is also given.
TASTES Name Filling
LOCATIONS LName Phone Address

SANDWICHES Location Bread Filling Price

TASTES

NAME Brown Brown Brown Jones Green Green Green

FILLING Turkey Beef Ham Cheese Beef Turkey Cheese LOCATION BREAD FILLING PRICE Lincoln O'Neill's O'Neill's Old Nag Buttery O'Neill's Buttery Lincoln Lincoln Old Nag Rye White Whole Rye White White White Rye White Rye Ham Cheese Ham Beef Cheese Turkey Ham Beef Ham Ham 1.25 1.20 1.25 1.35 1.00 1.35 1.10 1.35 1.30 1.40

SANDWICHES

LOCATIONS

LNAME Lincoln O'Neill's Old Nag Buttery

PHONE 683 4523 674 2134 767 8132 702 3421

ADDRESS Lincoln Place Pearse St Dame St College St

(a) Give a series of relational algebra operations (select, project, join, difference, division, ...) to produce the following four relations : (i) Cheap_Places is the set of locations that do not have any sandwiches costing more than 1.30. Can_Eat is a set of (name, location) tuples indicating who can eat where (i.e. in what locations are there fillings which that person likes). Jones_Places is the set of locations where Jones can eat, along with their phone numbers.

(ii)

(iii)

(iv) All_Eat is the set of locations, if any, where everyone mentioned in the tastes relation can eat together. (Use the division operator.) (b) Write SQL statements to retrieve the following information: (i) places where Jones can eat (using a nested subquery). (ii) places where Jones can eat (without using a nested subquery). (ii) for each location the number of people who can eat there.

2.

Consider the following relational database schema. It is intended to represent the holdings of a multi-branch library. A sample database instance is also given.

Branch

BCode Librarian

Address

Titles

Title Author Publisher

Holdings

Branch Title #copies

BCode Librarian

Address

Branch Title

#copies

B1 B2 B3

John Smith 2 Anglesea Rd 34 Pearse St Mary Jones Francis Owens Grange X

Author Ann Brown Amy Fly David Little Blaise Pascal Ann Brown Publisher Macmillan Stop Press Wiley Applewoods Macmillan

Title Susannah How to Fish A History of Dublin Computers The Wife

B1 B1 B1 B2 B2 B2 B3 B3 B3 B3

Susannah How to A hist How to Computers The Wife A hist .. Computers Susannah The Wife

3 2 1 4 2 3 1 4 3 1

(a) Give a series of relational algebra operations (select, project, join, difference, division, ...) to produce the following four relations : (i) (ii) Librarians is a list of the names of the branch librarians. Brown_Branches is the set of branches which have holdings of books by Ann Brown. No_Browns is the set of branches which have no holdings of any books by Ann Brown. (Use the Brown_Branches relation from (ii). In_All_Branches is the set of book titles, if any, along with their authors, which are each held at all branches. (Use the division operator.)

(iii)

(iv)

(b) Write SQL statements to retrieve the following information: (i) (ii) (iii) (iv) the names of all library books published by Macmillan. branches that hold any books by Ann Brown (using a nested subquery). branches that hold any books by Ann Brown (without using a nested subquery). the total number of books held at each branch.

3.

A Library Database is to contain information about published papers in a number of selected subject areas; each paper is classified under any number of different subject areas. The following data is to be stored: subject classification number S1 S2 S3 S4 S5 paper number: title: author: journal: volume: paper number: title: author: journal: volume: paper number: title: author: journal: volume: paper number: title: author: journal: volume: name of subject relational databases theory data models security distributed databases

P1 subject areas: S1, S2, S3 A relational model for large shared data banks Codd, EF CACM 13 number: 6 P2 subject areas: S4 Data Security Denning, D ACM Computing Surveys 11 number: 3 P3 subject areas: S1, S2, S3 Formal aspects of the relational model Furtado, A Information Systems 3 number: 2 P4 subject areas: S5 Distributed deadlock detection algorithm Obermarck, R TODS 7 number: 2

(a) Show how this data could be represented in a relational database. (b) Draw a database schema for the relational library database, indicating primary and foreign keys. (c) Give a series of relational algebra operations (select, project, join, difference, union, intersection, division, ...) on your relational database schema to produce the following four relations : (i) (ii) CACM13 is a list of titles and authors of papers from volume 13 of the CACM journal. SECURITY is a list of the title, author, and paper identification number of papers whose subject is "security". (NB You must not use the fact that you know that the code for "security" is S4.) THE&SEC is a list of paper identification numbers for papers which are about both theory and security.

(iii)

(d) Provide SQL statements on your relational database schema for the following queries (i) (ii) the titles and authors of all papers from volume 13 of the CACM journal. the title, author, and paper identification number of papers whose subject is "security". (NB You must not use the fact that you know that the code for "security" is S4.) for each subject area, the subject name and the total number of papers in the database which are on that subject.

(iii)

http://www.dcs.kcl.ac.uk/teaching/units/1999/cs02db/pdf/lecture-notes-01.pdf http://www.dcs.kcl.ac.uk/teaching/units/1999/cs02db/ http://www.dcs.kcl.ac.uk/teaching/units/1999/cs02db/index.html#slides