Relational Algebra

1
 Introduction
• Query languages are specialized languages for
asking questions or queries, that involve the data
in a database.
• Queries in algebra are composed of a collection of
operators.
• Every operator in relational algebra accepts (one
or two) relation instances as arguments and returns
a relation instance as the result.
– A relational algebra expression is recursively defined to
be a relation.
– Relational algebra is a procedural query language
• Define a step-by-step procedure for computing the desired
answer.
2
 Basic Operators
• There are six basic operators in relation
algebra:
– select (  )
– project (  )
– union ( U )
– set different ( - )
– Cartesian product ( x )
– rename (  )

3
• Selection
– The selection operation  specifies the tuples to
retain through a selection condition.
– Selection condition is a Boolean combination
(an expression using logical connectives  and
) of terms that have the form
• attribute op constant
• attribute1 op attribute2
(op is one of the comparison operators , , , ,
, ).

4
 Relation R
 a=b d>5 ( R )
a b c d
  1 7 a b c d
  5 7   1 7
  12 3   23 10
  23 10

5
• Projection
– The projection operator  allows us to extract
columns from a relation.
– The subscript specifies the fields to be retained.
(The other fields are ‘projected out’).
– The schema of the result of a projection is
determined by the fields that are projected.
– Duplicated row will be eliminated in the final
result.
• This follows from the definition of a relation as a set
of tuples.

6
Relation R  a,c ( R )

a b c a c a c
 10 1  1  1
 20 1  1 Eliminate
 1
Duplicated
 30 1  1 rows
 2
 40 2  2

7
 Employee
f_name l_name id sex salary superid dno
Joseph Chan 999999 M 29500 654321 4
Victor Wong 001100 M 30000 888555 5
Carrie Kwan 898989 F 26000 654321 4
Joyce Fong 345345 F 12000 777888 4

Find all employees who works in department 4 and whose

salary is greater than 25000.

 dno=4 salary>25000 (Employee)

f_name l_name id sex salary superid dno
Joseph Chan 999999 M 29500 654321 4
Carrie Kwan 898989 F 26000 654321 4

8
1. Find the employee names and department number of all
employees

2. Find the sex and department number of all employees

 f_name,l_name,dno ( Employee )  sex,dno ( Employee )

f_name l_name dno sex dno
Joseph Chan 4 M 4
Victor Wong 5 M 5
Carrie Kwan 4 F 4
Joyce Fong 4

9
– Compositing operation gives relational-algebra
expression.
– The result of a relational-algebra expression is
always a relation.

 f_name,l_name (  dno=4^salary>25000 ( Employee ) )

f_name l_name
Joseph Chan
Carrie Kwan

10
• Union
– R  S returns a relation instance containing all tuples
that occur in either relation R or relation instance S (or
both).
– The union operation is commutative: R  S = S  R
– Duplicate tuples are eliminated.
– R and S must be union-compatible.
• They have the same number of fields,
• The corresponding fields have the same domains.
– The field names are not important
– The schema of the result is defined to be identical to the
schema of R.
• The field of R  S inherit names from R.

11
 a b
a b
a b  1
 1
R S  2 RUS  2
 2
 3  1
 1
 3

12
• Set Difference
– Set difference R – S returns a relation instance
containing all the tuples that occur in R but not
in S.
– The set difference is not commutative: in
general, R – S  S – R.
– The relations R and S must be union-
compatible.
– The schema of the result is defined to be
identical to the schema of R.

13
 a b
a b a b
 1
R S  2 R-S  1
 2
 3  1
 1

14
• Intersection
– R  S returns a relation instance containing all
tuples that occur in both R and S.
– The relations R and S must be union-
compatible.
– The schema of the result is defined to be
identical to the schema of R.
– Intersection is not considered a basic operation,
as it can be derived from the basic operations as
shown below:
R  S R  ( R  S )
15
 a b
a b a b
 1
R S  2 RS  2
 2
 3
 1

16
• Rename ()
–  ( R(F), E ) or  ( R, E )
• where E is an arbitrary relation algebra expression
– result: relation named R,
– R = E except that the fields may be renamed
according to F.
– F is called the renaming list:
– oldname  newname or position  newname.

17
Example
 ( C ( sid  identity ), E )

We may rename more fields:

 ( C ( sid  identity, child  dependent ), E )

Example
 ( C ( 3  identity ), E )

Result is C
3rd attribute in E is renamed as “identity” in C

18
• Cartesian product (Cross product)
– R  S returns a relation instance whose schema
contains all the fields of R followed by all the
fields of S.
– The result contains one tuple <r,s>
(concatenation of tuples r and s) for each pair
of tuples r  R, s  S.

19
 RS
a b c d e
S  1  10 +
R c d e  1  10 +
a b  10 +  1  20 -
 1  10 +  1  10 -
 2  20 -  2  10 +
 10 -  2  10 +
 2  20 -
 2  10 -

20
Query: To retrieve for each female employee a list
of the names of her dependents.

EMPLOYEE

f_name l_name id bdate addr sex salary super_id dno

DEPENDENT

eid dependent-name sex bdate relationship

21
 Female_emps   sex=‘F’ ( Employee )
Empnames   f_name,l_name,id ( Female_emps )
Female_emps
f_name l_name id bdate address sex salary superid dno
Alicia Chan 998877 2-Jul-70 231, Cai Road, HK F 9500 654321 4
Jennifer Wong 654321 20-June-60 342, Cheung Road, HK F 30000 888555 4
Joyce Fong 345345 19-Dec-80 23, Young Road, HK F 12000 777888 5

f_name l_name id
Empnames
Alicia Chan 998877
Jennifer Wong 654321
Joyce Fong 345345

Dependents eid dep_name sex bdate relationship

334455 Alice F 5-Apr-90 Daughter
334455 Theodore M 3-Mar-92 Son
654321 Abner M 29-Feb-94 Son
123456 Alice F 2-Nov-97 Daughter

22
 Emp_dependents  Empnames x Dependents
Emp_dependents
f_name l_name id eid dep_name sex bdate relationship
Alicia Chan 998877 334455 Alice F 5-Apr-90 Daughter
Alicia Chan 998877 334455 Theodore M 3-Mar-92 Son
Alicia Chan 998877 654321 Abner M 29-Feb-94 Son
Alicia Chan 998877 123456 Alice F 2-Nov-97 Daughter
Jennifer Wong 654321 334455 Alice F 5-Apr-90 Daughter
Jennifer Wong 654321 334455 Theodore M 3-Mar-92 Son
Jennifer Wong 654321 654321 Abner M 29-Feb-94 Son
Jennifer Wong 654321 123456 Alice F 2-Nov-97 Daughter
Joyce Fong 345345 334455 Alice F 5-Apr-90 Daughter
Joyce Fong 345345 334455 Theodore M 3-Mar-92 Son
Joyce Fong 345345 654321 Abner M 29-Feb-94 Son
Joyce Fong 345345 123456 Alice F 2-Nov-97 Daughter

23
Query: To retrieve for each female employee a list of the names of her dependents

Actual_dependents   id=eid ( Emps_dependents )

Result   f_name,l_name,dep_name ( Actual_dependents )

Actual_dependents

f_name l_name id eid dep_name sex bdate relationship

Jennifer Wong 654321 654321 Abner M 29-Feb-94 Son

Result

f_name l_name dep_name

Jennifer Wong Abner

24
• Join
– Because the sequence of operations, × followed by  is
quite common, a special operations, called the “join”
operation ( ) was created to specify these as a single
operation.
– Join can be defined as a cross-product followed by
selections and sometimes with projections.
– The result of a cross-product is typically much larger
than the result of a join, and it is very important to
recognize joins and implement them without
materializing the underlying cross-product.

25
• Condition Join
– The most general version of join operation
accepts a join condition c.
– The join condition is identical to a selection
condition in form.
– The operation is defined as follows:
R  C S  C ( R S )

26
 S R
sid sname rating age sid bid day
22 Dustin 7 45.0 22 101 10/10/96
31 Lubber 8 55.5 58 103 11/12/96
58 Rusty 10 35.5

S S .sid [Link] R

(sid) sname rating age (sid) bid day
22 Dustin 7 45.0 58 103 11/12/96
31 Lubber 8 55.5 58 103 11/12/96

27
 S R
sid sname rating age sid bid day
22 Dustin 7 45.0 22 101 10/10/96
31 Lubber 8 55.5 58 103 11/12/96
58 Rusty 10 35.5

S S .sid [Link] R

(sid) sname rating age (sid) bid day
22 Dustin 7 45.0 58 103 11/12/96
31 Lubber 8 55.5 58 103 11/12/96

28
Example

Query: Find the names of employees with the highest salary.

 l_name,f_name ( Employee ) -
 Employee.l_name, Employee.f_name (

Employee [Link]<[Link]  ( F, Employee ) )

Note: We assume that <l_name,f_name> is a key in this relation.

29
• Equi-join
– A special case of the join operation is when the
join condition consists solely of equalities
(connected by ) of the form
R.name1 = S.name2
– In the resulting relation, S.name2 will be
dropped by an additional projection operation.

30
 R S R R.b S .b S

a b c b e f a b c e f
 1  1 X   1  X 
 5  3 X   1  X 
 4  1 X   1  X 
 1  2 Y   1  X 
 2  3 Y   2  Y 

31
 R S R R.b S .b S

a b c b e f a b c e f
 1  1 X   1  X 
 5  3 X   1  X 
 4  1 X   1  X 
 1  2 Y   1  X 
 2  3 Y   2  Y 

32
• Natural Join
– A further special case of the join operation R S
is an equijoin in which equalities
are specified on all fields having the same
names in R and S.
– We can simply omit the join condition.
– The resulting schema contains the attributes of
R followed by the attributes in S that are not in
R.
– If the two relations
R S have no attributes in
common, is simply the cross-product.
33
 R S
a b c d b d e R S
 1  X 1 X  a b c d e
 2  X 3 X   1  X 
 4  Y 1 X   1  X 
 1  Y 2 Y   2  Y 
 2  Y 3 Y 

34
 R S
a b c d b d e R S
 1  X 1 X  a b c d e
 2  X 3 X   1  X 
 4  Y 1 X   1  X 
 1  Y 2 Y   2  Y 
 2  Y 3 Y 

35
• Division
– The division operation is useful for expressing
certain kinds of queries, for example, “find the
names of sailors who have reserved all boats.
– Example
• Consider two relations A and B.
• A has exactly two fields x and y.
• B has just one field y, with the same domain as in A.
• The division operation A/B is the set of all x values
(in the form of unary tuples) such that for every y
value in a tuple of B, there is a tuple <x,y> in A.

36
 x y y x
 1 B 1 A/B 
 2 2 
 3
 1
A  1 The division operation A/B is the set of all x
 1 values (in the form of unary tuples) such that
 3 for every y value in a tuple of B, there is a
tuple <x,y> in A.
 4
 1
 2

37
 x y y x
 1 B 1 A/B 
 2 2 
 3
 1
A  1 Another way to understand division is as
 1 follows:
 3 For each x value in A, consider the set of y
values that appear in tuples of A with that
 4 x value. If this set contains all y values in
 1 B, the x value is in the result of A/B.
 2

38
 x y y x
 1 B 1 A/B 
 2 2 
 3
 1 An analogy with integer division may also help
to understand division. For integers A and B,
A  1 A/B is the largest integer Q such that Q * B  A .
 1 for relation instances A and B, A/B is the
 3 largest relation instance Q such that Q B  A

 4 In fact we can compute the disqualified tuples

 1 using the following algebra expression:

 2  x (( x ( A) B )  A)
Thus, A/B is
 x ( A)   x (( x ( A) B)  A)
39
The division above can be written in terms of basic operations:
Attributes in A but
A B not in B. Note that it is not a
x y y Temp1   A-B ( A )valid notation in relational algebra.
 1 1 Temp2   A-B ( ( Temp1  B ) – A )
 2 2 Result = Temp1 – Temp2
 3
 1 ( Temp1  B ) – A
x
Temp1 x x y
 1 A/B 

 1  2


 3  2

 4  2

 1

 2 Temp2
40
Employee fname lname id bdate address salary sid dno

Works_on id pno

Query: Retrieve the names of employees who work on all the projects
that `John Sung’ works on.

Sung   fname=“John” ^ lname=“Sung” ( Employee )

Sung_pnos   pno ( Works_on Sung )
fname lname id bdate address salary sid dno pno

Result_id  Works_on / Sung_pnos

Result   fname,lname ( Result_id 
Employee )

41
 4.8 More examples
Sailors ( sid, sname, age )

Boats ( bid, bname, color )

Reserves ( sid, bid, date )

Consider the above schemas, the primary key fields are

underlined.

42
Sailors ( sid, sname, age )
Boats ( bid, bname, color )
Reserves ( sid, bid, date )

Query 1: Find the names of sailors who have

reserved boat with bid = 103.
More - Solution 1:
efficient
 sname ( (  bid=103 Reserves ) Sailors )
- Solution 2:
 ( Temp1,  bid=103 Reserves )
 ( Temp2, Temp1 Sailors )
 sname ( Temp2 )

- Solution 3:
 sname (  bid=103 ( Reserves Sailors ) )
43
 Sailors ( sid, sname, age )
Boats ( bid, bname, color )
Reserves ( sid, bid, date )

Query 2: Find the names of sailors who have reserved at

least a red boat.

- Solution 1:
 sname ( (  color=‘red’ Boats) Reserves Sailors )

- Solution 2:
 sname (  sid ( (  bid  color=‘red’ Boats ) Reserves ) Sailors )

44
 Sailors ( sid, sname, age )
Boats ( bid, bname, color )
Reserves ( sid, bid, date )

Query 3: Find the names of sailors who have reserved at

least a red or a green boats.
We can identify all red or green boats, then find sailors who have
reserved one of these boats.

-Solution 1:
 ( Tempboats, (  color=‘red’ V color=‘green’ Boats ) )
 sname ( Tempboats Reserves Sailors )

What happens if V is replaced by

^ in this query?
45
 Sailors ( sid, sname, age )
Boats ( bid, bname, color )
Reserves ( sid, bid, date )

- Solution 2:
 sname ( (  color=‘red’ Boats ) Reserves ) Sailors ) U
 sname ( (  color=‘green’ Boats ) 
Reserves ) Sailors )

Query 4: Find the names of sailors who have reserved at least a

red boat and at least a green boat.

The previous solution 1 would not work. We must identify sailors

who have reserved red boats, sailors who have reserved green
boats, then find the intersection.

46
 Sailors ( sid, sname, age )
Boats ( bid, bname, color )
Reserves ( sid, bid, date )

Note that sid is a key for Sailors, but sname is not a key.

 ( Tempred,  sid ( (  color=‘red’ Boats )  Reserves ) )

 ( Tempgreen,  sid ( (  color=‘green’ Boats ) 
Reserves ) )
 sname ( ( Tempred  Tempgreen ) Sailors )

47
Sailors ( sid, sname, age )
Boats ( bid, bname, color )
Reserves ( sid, bid, date )

Query 5: Find the names of sailors who have reserved

all boats.

 ( Tempsids, (  sid,bid Reserves ) / (  bid Boats ) )

 sname ( Tempsids Sailors )

What if we simply do
Reserves / (  bid Boats ) ?
date sid bid
2-3-2002 007 A
3-7-2002 007 B

48
Sailors ( sid, sname, age )
Boats ( bid, bname, color )
Reserves ( sid, bid, date )

Query 6: To find sailors who have reserved all red boats:

 ( Tempsids, (  sid,bid Reserves ) / (  bid (  color=‘red’ Boats ) )

 sname ( Tempsids Sailors )

49
 Remarks
• Only standard relational algebra is covered
in the lecture
– Relation is assumed to be a set of records (no
duplicated records)
– Extended operators such as sorting operators
and aggregation operators are not covered.

50