Transcript Document 7200555
The Relational Data Model (Based on Chapter 5)
1
1. Relational Model Concepts BASIS OF THE MODEL
• The relational Model of Data is based on the concept of a
Relation.
•
A Relation is a mathematical concept based on the ideas of sets.
• The strength of the relational approach to data management comes from the formal foundation provided by the theory of relations.
2
INFORMAL DEFINITIONS •
RELATION:
A table of values • A relation may be thought of as a
set of rows.
• A relation may alternately be though of as a
columns.
set of •
Each row of the relation may be given an identifier.
• Each column typically is called by its column name or column header or attribute name.
3
FORMAL DEFINITIONS
• A
Relation
may be defined in multiple ways.
• The
Schema
of a Relation: R (A1, A2, .....An) Relation R is defined over
attributes
A1, A2, .....An
For Example CUSTOMER (Cust-id, Cust-name, Address, Phone#) Here, CUSTOMER is a relation defined over the four attributes Cust-id, Cust-name, Address, Phone#, each of which has a domain or a set of valid values.
4
For example, the
domain
• A
tuple
of Cust-id is 6 digit numbers.
is an ordered set of values • Each value is derived from an appropriate domain.
• Each row in the CUSTOMER table may be called as a tuple in the table and would consist of four values.
<632895, "John Smith", "101 Main St. Atlanta, GA 30332", "(404) 894-2000"> is a triple belonging to the CUSTOMER relation.
• A relation may be regarded as a set of tuples (rows).
• Columns in a table are also called as attributes of the relation.
5
FORMAL DEFINITIONS (contd..)
• The relation is formed over the cartesian product of the sets; each set has values from a domain; that domain is used in a specific role which is conveyed by the attribute name.
• For example, attribute Cust-name is defined over the domain of strings of 25 characters. The role these strings play in the CUSTOMER relation is that of the name of customers.
• Formally, Given R(A1, A2, .........., An) r(R) subset-of dom (A1) X dom (A2) X ....X dom(An) • R: schema of the relation r of R: a specific "value" or population of R.
6
R is also called the
intension
of a relation r is also called the
extension
of a relation Let S1 = {0,1} Let S2 = {a,b,c} Let R be a subset-of S1 X S2 for example: r(R) = {<0.a> , <0,b> , <1,c> } 7
Informal Terms Table Column Row Values in a column Table Definition Populated Table
DEFINITION SUMMARY
Formal Terms Relation Attribute/Domain Tuple Domain Schema of Relation Extension 8
Figure 7.1
The attributes and tuples of a relation STUDENT.
9
2 Characteristics of Relations Ordering of tuples in a relation r(R)
: The tuples are
not
considered to be ordered, even though they appear to be in the tabular form.
Ordering of attributes in a relation schema R
(and of values within each tuple): We will consider the attributes in R(A1, A2, ..., An) and the values in t=
ordered
.
10
Values in a tuple
: All values are considered
atomic
(indivisible). A special
null
value is used to represent values that are unknown or inapplicable to certain tuples.
Notation: We refer to
component values
of a tuple t by t[Ai] = vi (the value of attribute Ai for tuple t).
- Similarly, t[Au, Av, ..., Aw] refers to the subtuple of t containing the values of attributes Au, Av, ..., Aw, respectively.
11
Figure 7.2
The relation STUDENT from Figure 7.1, with a different order of tuples 12
3 Relational Integrity Constraints
Constraints are
conditions
that must hold on
all
valid relation instances. There are three main types of constraints: • • •
Key
constraints
Entity integrity
constraints,
Referential integrity
constraints 13
3.1 Key Constraints Superkey
of R: A set of attributes SK of R such that no two tuples
in any valid relation instance r(R)
will have the same value for SK. That is, for any distinct tuples t1 and t2 in r(R), t1[SK]
<>
t2[SK].
Key
of R: A "minimal" superkey; that is, a superkey K such that removal of any attribute from K results in a set of attributes that is not a superkey.
14
Example: The CAR relation schema: CAR(State, Reg#, SerialNo, Make, Model, Year) has two keys Key1 = {State, Reg#}, Key2 = {SerialNo}, which are also superkeys. {SerialNo, Make} is a superkey but
not
a key.
If a relation has several candidate keys , one is chosen arbitrarily to be the
primary key
. The primary key attributes are
underlined
.
15
Figure 7.4
The CAR relation with two candidate keys: LicenseNumber and EngineSerialNumber.
16
Figure 7.5
Schema diagram for the COMPANY relational database schema; the primary keys are underlined.
17
Figure 7.5
Schema diagram for the COMPANY relational database schema; the primary keys are underlined.
18
Figure 7.6
(continued) 19
3.2 Entity Integrity Relational Database Schema
: A set S of relation schemas that belong to the same database. S is the
database
.
name
of the S = {R1, R2, ..., Rn}
Entity Integrity
: The
primary key attributes
PK of each relation schema R in S cannot have null values in any tuple of r(R). This is because primary key values are used to tuples.
identify
the individual t[PK]
<>
null for any tuple t in r(R) Note: Other attributes of R may be similarly constrained to disallow null values, even though they are not members of the primary key.
20
3.3 Referential Integrity
A constraint involving
two
relations (the previous constraints involve a
single
relation).
Used to specify a
relationship
referencing relation
among tuples in two relations: the and the
referenced relation
.
Tuples in the
referencing relation
foreign key
R1 have attributes FK (called attributes) that reference the primary key attributes PK of the
referenced relation
R2. A tuple t1 in R1 is said to
reference
t2[PK].
a tuple t2 in R2 if t1[FK] = A referential integrity constraint can be displayed in a relational database schema as a directed arc from R1.FK to R2.
21
Figure 7.7
Referential integrity constraints displayed on the COMPANY relational database schema diagram.
22
Figure 7.6
One possible relational database state corresponding to the company schema.
23
-
4 Update Operations on Relations
INSERT a tuple.
DELETE a tuple.
MODIFY a tuple.
Integrity constraints should not be violated by the update operations.
Several update operations may have to be grouped together.
Updates may
propagate
to cause other updates automatically. This may be necessary to maintain integrity constraints.
24
In case of integrity violation, several actions can be taken: - cancel the operation that causes the violation (REJECT optiom) - perform the operation but inform the user of the violation - trigger additional updates so the violation is corrected (CASCADE option, SET NULL option) - execute a user-specified error-correction routine 25
5 The Relational Algebra
Operations to manipulate relations.
Used to specify retrieval requests (queries).
Query result is in the form of a relation.
Relational Operations:
5.1 SELECT s and PROJECT P operations.
5.2 Set operations: These include UNION U, INTERSECTION | |, DIFFERENCE -, CARTESIAN PRODUCT X.
5.3 JOIN operations X.
5.4 Other relational operations: DIVISION, OUTER JOIN, AGGREGATE FUNCTIONS.
26
5.1 SELECT
s
and PROJECT
P
SELECT
operation (denoted by
s
): Selects the tuples (rows) from a relation R that satisfy a certain
selection condition
c Form of the operation: s c(R) The condition c is an arbitrary Boolean expression on the attributes of R Resulting relation has the
same attributes
as R Resulting relation includes each tuple in r(R) whose attribute values satisfy the condition ‘c’ 27
Examples: s DNO=4 s SALARY>30000 (EMPLOYEE) (EMPLOYEE) s (DNO=4 AND SALARY>25000) OR DNO=5 (EMPLOYEE) 28
PROJECT
operation (denoted by P ): Keeps only certain attributes (columns) from a relation R specified in an
attribute list
L Form of operation: P L (R) L Resulting relation has only those attributes of R specified in Example: P FNAME,LNAME,SALARY (EMPLOYEE) The PROJECT operation
eliminates duplicate tuples
resulting relation so that it remains a mathematical set (no in the duplicate elements) 29
Example: P SEX,SALARY (EMPLOYEE) If several male employees have salary 30000, only a single tuple
Duplicate tuples are eliminated by the
P
operation
.
30
Sequences of operations
: Several operations can be combined to form a
relational algebra expression
(query) Example: Retrieve the names and salaries of employees who work in department 4: P FNAME,LNAME,SALARY ( s DNO=4(EMPLOYEE) ) -
Alternatively
, we specify explicit intermediate relations for each step: DEPT4_EMPS < s R < P DNO=4(EMPLOYEE) FNAME,LNAME,SALARY(DEPT4_EMPS) 31
Attributes can optionally be
renamed
in the resulting left-hand-side relation (this may be required for some operations that will be presented later): DEPT4_EMPS < s DNO=4 (EMPLOYEE) R(FIRSTNAME,LASTNAME,SALARY) < P FNAME,LNAME,SALARY (DEPT4_EMPS) 32
(a) s
Figure 7.8
Results of SELECT and PROJECT operations.
(DNO=4 AND SALARY>25000) OR (DNO=5 AND SALARY>30000) (EMPLOYEE).
(b) LNAME, FNAME, SALARY (EMPLOYEE). (c) SEX, SALARY (EMPLOYEE).
33
Class Number – CS 304 Class Name DBMS
Instructor – Sanjay Madria Lesson Title – Relational Algebra – 3rd July
34
5.2 Set Operations
Binary operations from mathematical set theory:
UNION
: R1 U R2,
INTERSECTION
: R1 | | R2,
SET DIFFERENCE
: R1 - R2,
CARTESIAN PRODUCT
: R1 X R2.
For U, | |, -, the operand relations R1(A1, A2, ..., An) and R2(B1, B2, ..., Bn) must have the same number of attributes, and the domains of corresponding attributes must be compatible; that is, dom(Ai)=dom(Bi) for i=1, 2, ..., n. This condition is called
union compatibility
.
The resulting relation for U, | |, or - has the same attribute names as the
first
operand relation R1 (by convention).
35
Figure 7.10
Query result after the UNION operation: RESULT RESULT1 RESULT2 36
Figure 7.11
Illustrating the set operations union, intersection and difference.
(b) STUDENT (a) Two union compatible relations.
INSTRUCTOR. (c) STUDENT INSTRUCTOR (d) STUDENT - INSTRUCTOR (e) INSTRUCTOR - STUDENT 37
CARTESIAN PRODUCT
R(A1, A2, ..., Am, B1, B2, ..., Bn) < R1(A1, A2, ..., Am) X R2 (B1, B2, ..., Bn) A tuple t exists in R for each combination of tuples t1 from R1 and t2 from R2 such that: t[A1, A2, ..., Am]=t1 and t[B1, B2, ..., Bn]=t2 - If R1 has n1 tuples and R2 has n2 tuples, then R will have n1*n2 tuples.
CARTESIAN PRODUCT can
combine related tuples
from two relations
if followed by the appropriate SELECT operation
.
38
Example: Combine each DEPARTMENT tuple with the EMPLOYEE tuple of the manager.
DEP_EMP <-DEPARTMENT X EMPLOYEE DEPT_MANAGER < s MGRSSN=SSN (DEP_EMP) 39
5.3 JOIN Operations THETA JOIN
: Similar to a CARTESIAN PRODUCT followed by a SELECT. The condition c is called a
join condition
.
R(A1, A2, ..., Am, B1, B2, ..., Bn) <-R1(A1, A2, ..., Am) X c (B1, B2, ..., Bn) Here c can be <, >, =, <=, >= R2
EQUIJOIN
: The join condition c includes one or more
equality comparisons
involving attributes from R1 and R2. That is, c is of the form: (Ai=Bj) AND ... AND (Ah=Bk); 1
join attributes
of R1 Bj, ..., Bk are called the
join attributes
of R2 40
Example of using EQUIJOIN: Retrieve each DEPARTMENT's name and its manager's name: T <-DEPARTMENT X MGRSSN=SSN RESULT < P EMPLOYEE DNAME,FNAME,LNAME (T) 41
NATURAL JOIN
(*): In an EQUIJOIN R <- R1 X c R2, the join attribute of R2 appear
redundantly
in the result relation R. In a NATURAL JOIN, the
redundant join attributes
of R2 are
eliminated
from R. The equality condition is
implied
and need not be specified.
R <- R1 *(join attributes of R1),(join attributes of R2) R2 Example: Retrieve each EMPLOYEE's name and the name of the DEPARTMENT he/she works for: T<- EMPLOYEE *(DNO),(DNUMBER) DEPARTMENT RESULT < P FNAME,LNAME,DNAME (T) 42
If the join attributes
have the same names
relations, they
need not be specified
write R <- R1 * R2.
in both and we can Example: Retrieve each EMPLOYEE's name and the name of his/her SUPERVISOR: SUPERVISOR(SUPERSSN,SFN,SLN)< P SSN,FNAME,LNAME(EMPLOYEE) T<-EMPLOYEE * SUPERVISOR RESULT < P FNAME,LNAME,SFN,SLN (T) 43
Figure 7.13
Illustrating the JOIN operation.
44
Figure 7.14
An illustration of the NATURAL JOIN operation. (a) PROJ_DEPT PROJECT * DEPT.
(b) DEPT_LOCS DEPARTMENT * DEPT_LOCATIONS.
45
Note: In the
original definition
join attributes were
required
both relations.
of NATURAL JOIN, the to have the same names in There can be a
more than one set of join attributes different meaning
example: with a between the same two relations. For JOIN ATTRIBUTES EMPLOYEE.SSN= RELATIONSHIP EMPLOYEE
manages
DEPARTMENT.MGRSSNthe DEPARTMENT EMPLOYEE.DNO= DEPARTMENT.DNUMBER
EMPLOYEE
works for
the DEPARTMENT 46
A relation can have a
set of join attributes itself
: to join it
with
JOIN ATTRIBUTES EMPLOYEE(1).SUPERSSN= EMPLOYEE(2).SSN
RELATIONSHIP EMPLOYEE(2)
supervises
EMPLOYEE(1) One can
think of this
as joining
two distinct copies
relation, although only one relation actually exists of the In this case,
renaming
can be useful 47
Figure 7.15
Illustrating the division operation.
(a) Dividing SSN_PNOS by SMITH_PNOS. (b) T R S.
48
Complete Set of Relational Algebra Operations
: All the operations discussed so far can be described as a sequence of
only
the operations SELECT, PROJECT, UNION, SET DIFFERENCE, and CARTESIAN PRODUCT. Hence, the set { s , P , U, - , X } is called a
complete set
of relational algebra operations. Any query language
equivalent to
these operations is called
relationally complete
.
For database applications, additional operations are needed that were not part of the
original
relational algebra. These include: 1. Aggregate functions and grouping.
2. OUTER JOIN.
49
5.4 Additional Relational Operations AGGREGATE FUNCTIONS
Functions such as SUM, COUNT, AVERAGE, MIN, MAX are often applied to sets of values or sets of tuples in database applications
F
F
AVERAGE SALARY (EMPLOYEE) 50
Example 2: For each department, retrieve the department number, the number of employees, and the average salary (in the department): R(DNO,NUMEMPS,AVGSAL) < DNO
F
COUNT SSN , AVERAGE SALARY (EMPLOYEE) DNO is called the
grouping attribute
in the above example 51
Figure 7.16
SALARY An illustration of the AGGREGATE FUNCTION operation.
(a) R(DNO, NO_OF_EMPLOYEES, AVERAGE_SAL) (EMPLOYEE). (b) DNO (C) COUNT SSN, AVERAGE SALARY DNO (EMPLOYEE).
COUNT SSN, AVERAGE COUNT SSN, AVERAGE SALARY (EMPLOYEE).
52
OUTER JOIN
In a regular EQUIJOIN or NATURAL JOIN operation, tuples in R1 or R2 that do not have matching tuples in the other relation
do not appear in the result
Some queries require all tuples in R1 (or R2 or both) to appear in the result When no matching tuples are found,
null
s are placed for the missing attributes 53
LEFT OUTER JOIN
: R1 X R2 lets every tuple in R1 appear in the result -
RIGHT OUTER JOIN
: R1 X R2 lets every tuple in R2 appear in the result -
FULL OUTER JOIN
: R1 X R2 lets every tuple in R1 or R2 appear in the result 54
Figure 7.18
The LEFT OUTER JOIN operation.
55
RENAME Operator S(B1, B2,…Bn) (R) – Renaming relation R as S and renaming attributes of R as Bi’s. S (R) – Renaming R as S (B1, B2,…Bn) (R) – Renaming attributes of R as Bi’s. 56
Some Queries Q. Retrieve the SSNs of all the employees who either work in dept. 5 or supervise an employee who works in dept 5.
Dept-Emps s DNO = 5 (Employee) Result1 SSN (Dept-Emps) Result 2(SSN) SuperSSN (Dept-Emps) Result = Result 1 Result 2 57
Q. Find for each female employee, a list of names of her dependents.
Female-Emp s SEX = F (Employee) Empname FNAME, LNAME, SSN (Female-Emp) Emp-dep Empname Dependent Actual-dep s SSN = ESSN (Emp-dep) Result FNAME, LNAME, Dependent-name (Actual-dep) Q. Find the name of the manager of each department.
Dept-mgr Department MGRSSN = SSN (Employee) Result DNAME, LNAME, FNAME (Dept-mgr) 58
• List names of managers who have atleast one dependent • Find names of employees who have no dependents • List names of all employees with two or more dependents 59
Figure 7.12
An illustration of the CARTESIAN PRODUCT operation.
60
Figure 7.12 (continued) 61
The
Insert
Operation
1. Insert <‘Cecilia’, ‘F”, “Kolonsky’, null, ‘1960-04 05’, ‘6357 Windy Lane, Katy, TX’, F, 28000, null, 4> into EMPLOYEE This insertion violates the entity integrity constraint (null for the primary key SSN), so it is rejected.
2. Insert <‘Alicia’, ‘J’, ‘Zelaya’, 999887777, 1960 04-05’, ‘6357 Windy Lane, Katy, TX’, F, 28000, 987654321’, 4> into EMPLOYEE This insertion violates the key constraint because another tuple with the same SSN Value already exists in the EMPLOYEE relation, and so it is rejected.
62
The
Insert
Operation (contd.)
3. Insert <‘Cecilia’, ‘F’, ‘Kolonsky’, ‘67768989’, 1960 04-05’, ‘6357 Windswept, katy, TX’, F, 28000, ‘987654321’, 7> into EMPLOYEE This insertion violates the referential integrity constraint specified on DNO because no DEPARTMENT tuple exists with DNUMBER = 7 4. Insert <‘Cecilia’, ‘F’, ‘Kolonsky’, ‘67768989’, 1960 04-05’, ‘6357 Windy Lane, Katy, TX’, F, 28000, null, 4> into EMPLOYEE.
This insertion satisfies all constraints, so it is acceptable.
63
The
Delete
Operation
1. Delete the WORKS_ON tuple with ESSN = ‘999887777’ and PNO = 10.
This deletion is acceptable 2. Delete the EMPLOYEE tuple with SSN = ‘999887777’ This deletion is not acceptable, because tuples in WORKS_ON refer to this tuple. Hence, if the tuple is deleted, referential integrity violations will result.
64
The
Delete
Operation (contd.)
3. Delete the EMPLOYEE tuple with SSN= ‘333445555’ This deletion will result in even worse referential integrity violations, because the tuple involved is referenced by tuples from the EMPLOYEE, DEPARTMENT, WORKS_ON and DEPENDENT relations.
• Options • Reject deletion • Propagate deletion by deleting other tuples • Modify the referencing attributes that cause the violations • Or combination of three above 65
The
Update
Operation
1. Update the SALARY of the EMPLOYEE tuple with SSN = ‘999887777; to 28000 Acceptable 2. Update the DNO of the EMPLOYEE tuple with SSN = ‘999887777’ to 1 Acceptable 3. Update the DNO of the EMPLOYEE tuple with SSN = ‘999887777’ to 7 Unacceptable, because it violates referential integrity.
66
The
Update
Operation (contd.)
4. Update the SSN of the EMPLOYEE tuple with SSN = ‘999887777’ to ‘987654321’ Unacceptable, because it violates primary key and referential integrity constraints.
Updating an attribute which is neither primary key or foreign key usually causes no problems, new value should be of correct data type and domain. 67