Requirement Analysis & Conceptual Database Design · Conceptual Database Design Problem analysis Entity Relationship notation ... Distinguish data and operations Requirement analysis

Requirement Analysis &Requirement Analysis &Conceptual Database DesignConceptual Database Design

Problem analysisEntity Relationship notationIntegrity constraintsGeneralization

1.2

FU-Berlin, D

BS I 2006, Hinze / Scholz

Introduction: LifecycleLifecycle

Requirement analysis -> Text

Conceptual Design -> ER-Model

Name

SID

Email

KID

Name

Dauer

Name DID

EmailVName VName

Studentin

Kurs

Dozentin

anmelden

voraussetzen

Vorgänger Nachfolger

(1,*)

(0,*)

(0,*) (0,*)

halten

(1,*)

(1,1)

Administration

Logical Schema DesignCREATE TABLE Studentin(SID INTEGER PRIMARY KEY,VName CHAR(20)Name CHAR(30) NOT NULL,Email Char(40));

CREATE TABLE Kurs(KID CHAR(10) PRIMARY KEY,Name CHAR(40) NOT NULL,Dauer INTEGER);

CREATE TABLE Anmelden(SID INTEGER PRIMARY KEY FOREIGN KEY REFERENCES Studentin (SID),KID INTEGER PRIMARY KEY FOREIGN KEY REFERENCES Studentin (KID));

-> Database schema

Physical Schema Design -> Access paths

1.3

FU-Berlin, D


Introduction: Database Design TerminologyIntroduction: Database Design Terminology

Different from Software Engineering !Database Design

process of defining the structure of a database layers of abstraction: Conceptual, logical, physical level

Includes "Analysis" and "Design" from SE

Database Software EngineeringRequirements RequirementsConceptual modelingLogical modelingPhysical modeling

AnalysisDesign

1.4

FU-Berlin, D


Requirement AnalysisRequirement Analysis

Customer communication!

Identify essential "real world" information Remove redundant, unimportant detailsClarify unclear natural language statementsFill remaining gaps in discussionsDistinguish data and operations

Requirement analysis & Conceptual Design aims at focusing thoughts and discussions !

1.5

FU-Berlin, D


Requirement Analysis: ExampleRequirement Analysis: Example

“I’m the owner of a medium size video store. We have over 10.000video tapes that we need to keep track of. Since a few weeks, we also have DVDs. Each of our video tapes has a tape number. For each movie, we need to know its title and category (e.g. comedy, suspense, drama, action, or SciFi), director and year. Yes, we do have multiple copies of many of our movies. We give each movie a specific id, and thentrack which movie a tape contains.

A tape may be either Beta or VHS format. We always have at leastone tape for each movie we track, and each tape is always a copy of a single, specific movie. Our tapes are adapted to the movielengths, so we don’t have any movies which require multiple tapes. The movies are stored on shelf according to their category sorted by movie title. We are frequently asked for movies starring specific actors, John Wayne and Katherine Hepburn are always popular. So we’d like to keep track of the star actors appearing in each movie.....

1.6

FU-Berlin, D


Conceptual Design: ExampleConceptual Design: Example

.... Not all of our movies have star actors. Customers like to know each actor’s “real” birth name and age. We track only actors whoappear in the movies in our inventory. We have lots of customers. We only rent videos to people who havejoined our “video club”. To belong to our club, they must have good credit. For each club member, we’d like to keep their first andlast name, current phone number, and current address. And, of course each club member has a membership number. Then we need to keep track of what video tapes each customer currently has checked out. A customer may check out multiple video tapes at any given time. Rentals are for one or more days, each movie with an individual price per day. Furthermore we additionally charge 1 $ per beta format tape, 2 $ for a DVD andanother $ for movies longer than 2 hours. Maximum rental time is 4weeks. The customer gets a bill with movie titles, individual prices and total amount, when he /she returns movies."

1.7

FU-Berlin, D



Note: what is important depends on the applications Essential information

customers, tapes, movies, ...movies have many copiestape (or DVD) contains exactly one moviecustomers have customer identification (id) four weeks maximal rental time…

Redundant, unimportant details"...DVDs since a few weeks“"... John Wayne..“Names of the categories (but categories are important!) Tapes on shelf (since we don't design a tape robot)…

1.8

FU-Berlin, D



Clarify unclear statements Video club: admission / annual fee? Charge per tape: is price for one day the minimum?...

Fill gaps Any discounts?...

Distinguish data/operationsProcessing a billBecoming a member of the club …

1.9

FU-Berlin, D


Conceptual Design: Modeling primitives Conceptual Design: Modeling primitives

Entitysomething which exists Examples: tape, movie

Attributeproperty of an entityExamples: tape has id, movie has title

Relationship: connects two or more entities Examples: tape contains a movie.

Attributes have valuesExample: “John Wayne is starring”Not modelled!

Tape

contain

id

Movie

Tape

titleMovie

id

Tape

title

Movie

Important terms

1.10

FU-Berlin, D


Conceptual Design: Structuring the problemConceptual Design: Structuring the problem

Modelling / Conceptual Designtypically: Entity-Relationship-Model, data-oriented1976 introduced by P.P. Chen (Peter P. Chen: The Entity-Relationship Model - Toward a Unified View of Data. TODS 1(1): 9-36, 1976)

sometimes: UML (problematic!)

StudentSID: NumberName: StringFName: StringEmail: String

LectureLID: NumberTitle: Stringhours: Number

attend

Lecture

attend

Name

Email

LID

Title

hours

FNameStudent

SID

1.11

FU-Berlin, D


Conceptual Design: EConceptual Design: E--R notation R notation

Entities & Attributes:

Entities have identifying (or key) attributes or sets of attributes

Tape

id

keyMovie

title director category

These attributes together identify a movie

Entity

attribute attribute attribute

1.12

FU-Berlin, D


Conceptual Design: EConceptual Design: E--R notation R notation

Relationships

Always have a nameDo not have a key!Not always reflected by the relationship namesee example: "rents" is equivalent to "is-rented_by"

Entity_1 Entity_2relationship

Customer rents Tape

attribute

attribute

attribute

attribute

1.13

FU-Berlin, D



“[…] We have […] video tapes that we need to keep track of. Since a few weeks, we also have DVDs. Each of our video tapes has a tape number. For each movie, we need to know its title and category, director and year. We do have multiple copies of many of our movies. We give each movie a specific id, and then track which movie a tape contains. A tape may be either Beta or VHS Format. We always have at least one tape for each movie we track, and each tape is always a copy of a single, specific movie. Our tapes are adapted to the movie lengths, so we don’t have any movies which require multiple tapes. The movies are stored on shelf according to their category sorted by movie title. We are frequently asked for movies starring specific actors. […] Customers like to know each actor’s “real” birth name and age. We track only actors who appear in the movies in our inventory. […]Rentals are for one or more days, each movie with an individual price per day. Furthermore we additionally charge 1 $ per beta format tape, 2 $ for a DVD and another $ for movies longer than2 hours.” [abbreviated]

1.14

FU-Berlin, D



MovieTape hold

Actor

play

Format

belong_to

chargebirthday

name stage_name

real_name

directortitlecategory year

id

Price_per_day

id

length

(without customers)

1.15

FU-Berlin, D


Conceptual Design: EConceptual Design: E--R notationR notation

Multiple relationships

Recursive relationships

Lecture

require

predecessor successor

Person

have

parent child

Lectureattend

Student

hold

role name

1.16

FU-Berlin, D


Conceptual Design: EConceptual Design: E--R notationR notation

Relationships with attributes

Different semantics in separate entity:

Lecture Lecturerhold

rating

Lecture Lecturerhold

rating

Lecture_datebelong_to

1.17

FU-Berlin, D


Conceptual Design: Integrity constraintsConceptual Design: Integrity constraints

Integrity constraints (=Invariants)Database must always meet these invariants

Explicit state restrictions from requirements Examples:

"There is always at least one tape for each movie …, and each tape is always a copy of a single, specific movie”"Not all of our movies have star actors"

Implicit invariants commonExamples:

Tape cannot be loaned by more than one customer a timeActor may be starring in more than one movie

Important concept

1.18

FU-Berlin, D



Constraints:

Attribute restrictionsAttribute contain at most one value: unstructured attribute restriction of E-R M.Key attributes are unique

Attribute must / may have a valuee.g., movie must have title,

director is not necessarily known Value may be restricted e.g., movies are made after 1900 : movie.year > 1900Special restrictions not in E-R model!

Attribute restrictionsCardinality restrictions

1.19

FU-Berlin, D



Cardinality restrictions# entities-instances in relationship (UML: multiplicity)

Moviehold

Tape

1:None movie on a tape

Movieplay

Actor

N:Mseveral actors in many movies

Personown

drivers license

1:1one licence per driver

Examples: one movie on a tape, tape loaned by at most one customer at a time, #rented tapes per customer ≥ 0

1.20

FU-Berlin, D


Conceptual Design: CardinalitiesConceptual Design: Cardinalities

Simple Notation E1 R E2k1 k2

Tape rent CustomerN 1

“N tapes are rented by 1 customer”

E1 R E2(min1,max1)

(min2,max2)

Min-Max Notation:

Tape rent Customer(0,1) (0,*)

“Each tape is rented by 0-1 customers”“Each customer has rented 0-N tapes”

1.21

FU-Berlin, D


Conceptual Design: CardinalitiesConceptual Design: Cardinalities

1:1 relationship

1:N relationship

N:M relationship

person have driversLicence(0,1) (1,1)

Movie hold Tape(1,*) (1,1)

Lecture hold Lecturer(1,1) (0,*)

User_account has emailMessage(0,*) (1,*)

Do not mix notations!

1.22

FU-Berlin, D


Conceptual Design: Semantics N:M, (Conceptual Design: Semantics N:M, (min,maxmin,max))

R is 1:N ⇔ R is a partial function R: E2 → E1⇔ for all extensions of R ∀e2 ∈ E2:

|{e1 | e1 ∈E1 ∧ (e1,e2) ∈ R }| ≤ 1

R is 1:1 ⇔ E2 → E1 is a injective partial functionR is M:N ⇔ R is a relation, but not a function

Important conceptLet R ⊆ E1 x E2 be a relationship between entity sets E1 and E2

(R,E1) has (min1, max1) ⇔ for all extensions of R and for all x ∈E1

min1 ≤ |{y| y ∈ E2 ∧ (x,y) ∈ R }| ≤ max1

(R,E2) has (min2, max2)⇔ for all extensions of R and for all y ∈E2

min2 ≤ |{x | x ∈ E1 ∧ (x,y) ∈ R }| ≤ max2

1.23

FU-Berlin, D



MovieTape hold

Actor

play

Format

belong_to

chargebirthday

name stage_name

real_name

directortitlecategory year

id

Price_per_day

id

length

(without customers)

(1,1)

(1,1)

(0,*) (1,*)

(0,*)

(1,*)

1.24

FU-Berlin, D


Conceptual Design: notational overviewConceptual Design: notational overview

10 .. 1 0 ..*

*0 .. k

1 ..*k..jk

UML

11N or MN or ME-RM1:N

(1,1)(0,1)(0,*)(1,*)E-RM

(min,max)

mandatory/ single

optional/ single

optional/ multiple

mandatory/ multiple

k,j,N,M are natural numbers

More notations in use, e.g., oracle-special notation...

1.25

FU-Berlin, D


Conceptual Design: Weak entitiesConceptual Design: Weak entities

Existentially dependent entitiesNot globally identifiableExample:

Notation:

min2 = 1 : at least one e1 for each e2, otherwise e2 could not be identified

max2 = 1

Building has Room(1,*) (1,1)

id

E1 R E2(min1,max1) (min2, max2)

id

1.26

FU-Berlin, D



Example: Movies and their scenes

Example: Order and Delivery

Movie Scene(0,*) (1,1)

has

has(1,*) (1,1)

Item IidOrder

Delivery Did

contains

Oid

(1,1)

(1,*)

1.27

FU-Berlin, D



Modelling decision:

or

employee has child(0,*) (1,1)

eid

cid

employee has child(0,*) (1,1)

eid

cid

Advantages and disadvantages?

1.28

FU-Berlin, D


Conceptual Design: Temporal dataConceptual Design: Temporal data

1. Condition: tape only rented by one customer at the same time:

2. Condition: tape rented by many customers one after the other

customer videotaperent(0,*) (0,*)

customer

videotape

cid

vid

from

untilrentalhas

is_in

(0,*)

(0,*)

(1,1)

(1,1)

customer videotaperent(0,*) (0,1)

cid vid

from until

1.29

FU-Berlin, D



Movie

Tape

hold

Actor

play

First_name

birthday

Address

Telephone

Last_name

stage_name

real_name

directortitle category

yearid

Customer

Mem_no

Price_per_day

id

Format

belong_to

charge

namelength

Rentalfrom

is_in

have

(1,1)

(1,1)

(0,*)

(1,*)

(0,*)

(1,*)

(0,*)(1,1)

(1,1)

(0,*)

until

1.30

FU-Berlin, D


Conceptual Design: Conceptual Design: nn--aryary relationshipsrelationships

Example:Lecturer recommends textbook for lecture

Cardinalities:A particular lecturer recommends zero or more "(textbook,lecture) pairs" For each lecture at least one textbook is recommended by lecturershere, (min,max)-constraints not very expressive

Lecturer recommend Textbook(0,*) (1,*)

Lecture

(1,*)

1.31

FU-Berlin, D



1:N notation in n-ary relationships Remember: R is 1:N ⇔ R is function R:E2 → E1

Generalization: R: E1 x…x Ej-1 x Ej+1 x…x En → Ej

recommend: Textbook x Lecturer → Lecturerecommend: Textbook x Lecture → Lecturer

E1 E21 NR

Lecturer recommend Textbook1 N

Lecture

M

1.32

FU-Berlin, D


Conceptual Design: Conceptual Design: nn--aryary relationships relationships (Example 2)(Example 2)

Student is allowed to have only one topic for each Prof.Student is allowed to have only one Prof. for each topicProfs. are allowed to use the same topic againA topic can be used be used by many Profs., but for different students

Prof. attend Topic1 1

Student

M

1.33

FU-Berlin, D



N-ary relationships expressed by binaries

1. Connect pairs of entities

Less constraints expressible!

2. New “central” entity, binary relationships to old ones

Lecturer Lecture

Textbook

Lectureplan ……

…

Lecturer LectureTextbook ……

1.34

FU-Berlin, D


Conceptual Design: GeneralizationConceptual Design: Generalization

Consider lecturing students:similar objects in different entitiesRedundancy ⇒ errors!

Generalization (specialization) helps

LectureattendStudent

hold2

Lecturer hold1

SID Email

FName

Name

Namelid Email

FNameContract

LID

title

hours

1.35

FU-Berlin, D



Lecture

attend

Student

hold

LecturerSID

pid

Contract

LID

title

hours

Person

is-a is-a

Name

Email

FName

DB Interpretation of Generalization differs from OO!instances of Student and Lecturer also instances of PersonStudent ⊆ Person, Lecturer ⊆ Person

1.36

FU-Berlin, D



OO-interpretation: Instances of A, B and C different but share some attributes

DB interpretation: Instances of B and C also instances of A: B ⊆ A and C ⊆ AKey-inheritance"is-a" different from ordinary relationships

Special cases: Disjoint specialization: Instances of B and C are disjointComplete specialization: A = B ∪ C, no extra tupel in A

B C

A

is-a is-a

1.37

FU-Berlin, D


Conceptual Design: Short summaryConceptual Design: Short summary

E-R ModellingStandard semi-formal language for conceptual modellingDifferent notations: we use E-R and (min,max)

Important termsEntity (type, set): set of objects with the same attributesAttribute: property of an entity, at most one value per attribute. Relationship (type): association between entity typesRole: name of one side of relationship Cardinality (multiplicity): number of entity objects that participate in relationship Weak entity: existentially dependent entityGeneralization

Documents

Requirement Analysis & Conceptual Database Design · Conceptual Database Design Problem analysis Entity Relationship notation ... Distinguish data and operations Requirement analysis