Announcements

Midterm is Nov. 22 (8:15am – 9:45am)

Covers until the end of Normalization

You answer in the same exam paper

Closed-Book, One sheet is allowed for your notes

1

Revision

Instructor: Mohamed Eltabakh meltabakh@cs.wpi.edu

2

Covered Topics

Entity-Relationship Model & ERD

Relational Model

Relational Algebra

Functional Dependencies & Normalization

3

Relational Algebra

4

Key Points Understand the meaning of each operator and what it does, E.g.,

Selection (σc): selects certain tuples based on a condition (all columns) Projection (πk): selects certain columns (All tuples) Join (⋈C): joins tuples from one relation with another relation based on a

condition

Practice how to combine these operators to answer queries

Performance rules: Apply selection earlier whenever possible Use joins instead of Cartesian product Avoid un-necessary joins

5

Relational Algebra: Exercise

Ships having the same design belong to the same class. The class information is stored in “ShipClass”. The ship information is stored in “WarShips”

Each ship may participate in multiple missions. The mission information is stored in “Missions”

Each ship in a mission has a status as either ‘damaged’, ‘sunk’, or ‘ok’. This information is stored in “Results”

6

ShipClass(name, type, country, numGuns, designYear, weight)

WarShips(shipName, className, builtYear)

Missions(missionName, date)

Results(shipName, missionName, status)

Write the algebraic expression for the following queries

Query 1

7

For classes with at least 10 guns, report the class name and country;

πname,country (σnumGuns >= 10(ShipClass))

Query 2

8

List the warShips built after 1932

σbuiltYear > 1932(WarShips)

Query 3

9

Report the names of ships damaged in the mission of “Sun Rising”.

πshipName (σmissionName= ‘Sun Rising’ AND status=‘damaged’ (Results))

Query 4

10

Suppose there was an agreement prohibiting war-ships heavier than 40,000 tons. Report the ship names and built year of ships violating this agreement.

πshipName,builtYear (σweight > 40,000(ShipClass ⋈name=className WarShips))

πshipName,builtYear (σweight > 40,000(ShipClass) ⋈name=className WarShips)

Better

Query 5

11

Continue from the previous query, Find the ships that violate the agreement the least (i.e., the lightest ships violating the agreement)

R1 πID(σFirstName = ‘John’ AND LastName=‘Smith’(Doctor) ⋈SSN=Doctor_SSN Prescription)

R2 πIDPrescription_id(σTradeName=‘Aspirin’(Prescription_Medicine)) ∩ R1

Result (πPateint_SSN( R2 Prescription))⋈

Query 6

12

In the mission of “Sun Rising”, find the name and weight of the ships involved

R1 πshipName(σmissionName = ‘Sun Rising’(Results))

Result πR1.shipName, weight (R1 ⋈R1.shipName=Warships.shipName WarShips

⋈className=name ShipClass)

Query 7

13

Report ship name for ships that were ‘ok’ in one mission, but later ‘sunk’ in another.

πshipName(σstatus = ‘sunk’(Results)) ∩ πshipName(σstatus = ‘ok’(Results))

ER & Relational Models

14

Key Points Remember the notations used in ERD

Relationships cardinalities (1-1, 1-M, M-M) Entity sets and weak entity sets Primitive, derived, and composite attributes

Mapping rules from ERD to relational model, E.g., M-M relationship separate table 1-M relationship take the key from the one-side to the many-side 1-1 relationship take the key from either sides to the other side ISA relationship Many choices depending on its type

Remember to indicate the integrity constraints, most important are: Primary keys, Foreign keys

15

Airline Application We have a set of planes, each plane can make many flights.

Each plane has properties such as: ID (unique identifier), model, capacity, year in which it is built, and weight

Each flight has a departure city, arrival city, and it can make transit in many other cities. Also each flight has the departure date, arrival date, and the transit date in each of the transit cities (assume the ‘date’ captures both the date and exact time).

Each flight has some properties such as: FlightNum (unique identifier), number of transit stops, number of passengers

For each city, we have an CityID (unique identifier), name, and country

16

Question 1 Design an ER diagram for the given application. The

diagram must include: The entity sets, attributes, primary keys The relationships with the correct cardinalities

State any assumptions that you make and affect your design

17

18

Plane

year

ID

model

capacity

weight

City

CityID countryname

Flight

FNum

NumStops

NumPassengers

makes

transit

transitdate

arrival

Arrivaldate

departure

departuredate

Assumption: The entire flight is done by a single plane

Question 2

Extend your ERD in the previous question to capture the following additional requirements:

Each flight has one captain, one assistant captain, and between 2 to 4 flight attendants

Each of the captain, assistant captain, and attendants have a common set of properties such as: EmployeeID (unique identifier), name, salary, and DoB

The same person can be captain on some flights and assistant captain on other flights (basically same person can have different roles (captain or assistant captain) in different flights).

19

20

Question 3 Map the ERD from the previous question (Question 2) to

the corresponding relational model Provide the CREATE TABLE commands with the field names and

appropriate data types

Identify the primary keys and foreign keys in the relations

21

Relational Model

22

Create Table Plane (ID: int Primary Key,Year: int,Capacity: int,Weight: int,Model: varchar(50));

Create Table City(ID: int Primary Key,Name: varchar(100),Country: varchar(100));

Create Table Flight(FNum: varchar(8) Primary Key,NumStops: int,NumPassengers: int,PlaneID: int Foreign Key References Plane(ID),ArrivalCityID: int Foreign Key References City(CityID),DepartCityID: int Foreign Key References City(CityID),ArrivalDate: date,DepartDate: date,CaptainID: int Foreign Key References Employee (ID),AssistID: int Foreign Key References Employee (ID));

Create Table Employee (ID: int Primary Key,name: varchar(100),DOB: int,Salary: int,JobTitle: varchar(100) check JonTitle in (‘Captain’, ‘Attendant));

Relational Model (Cont’d)

23

Create Table Transit (FNum: int Foreign Key References Flight(FNum),CityID: int Foreign Key References City(CityID),TransitDate: date,Primary Key (Fnum, CityID));

Create Table Flight_Attendant(FNum: int Foreign Key References Flight(FNum),AttendantID: int Foreign Key References Employee(ID),Primary Key (Fnum, EmpID));

Relational Model (Cont’d) Assumptions:

The ISA relationship is modeled using one table “Employee” Whether the employee is Captain or Attendant is modeled using a new

attribute “JobTitle”

Business Constraints: Attributes CaptainID and AssistID in “Flight” must correspond to

employee with job title “Captain” Attribute AttendantID in “Flight_Attendant” must correspond to

employee with job title “Attendant” Attribute NumPassengers in “Flight” must be less than or equal to

attribute Capacity in “Plane”

24

Functional Dependencies & Normalization

25

Key Points Identifying the candidate keys of relations

Use of Transitive, and Union properties to generate more FDs Computing the attribute closure

A is a key if all attributes are in its closure A candidate key is also a super key (But it is minimal)

Remember the rules for BCNF and 3NF Test which FD is violating the rules Decompose based on this rule

26

Question 1 Given relation R= (A, B, C, D, E) with the following FDs:

F = {AB C, C D, BD, CD E, AB E}

Compute the canonical (minimal) cover of F = G

What are the candidate keys of R?

27

G = {AB C, C DE, BD}

AB

Question 2 Given relation R = (A, B, C, D) with the following FDs:

F = {B C, B D}

What are the candidate keys of R?

Report FDs violating BCNF (if any)?

Decompose R to be in BCNF (if not already)?28

AB

B CB D

Question 2 (Cont’d) Given relation R = (A, B, C, D) with the following FDs:

F = {B C, B D}

What are the candidate keys of R?

Report FDs violating 3NF (if any)?

Decompose R to be in 3NF (if not already)?

29

AB

B CB D