[Www.pkbulk.blogspot.com]dbms04

Relational Algebra
1
Fall 2001 Database Systems 1
Relational Algebra
• A relation is a set of tuples. Each relational
algebra operation takes as input a list of
relations and produces a single relation.
• General form:
OperatorArguments (List of Relations)

New Relation
• After each operation, the remaining attributes
are carried to the new relation. The attributes
may be renamed, but their domain remains the
same.
Set Theoretic Operations
• Regular set operations on relations
• A set operation requires two participating relations R and
S to be compatible
– R and S should have the same attributes
• R(Name:D1, Email:D2)
• S(Name:D1, Email:D2, Address:D3)
• T(Name:D1, Email:D4)
• V(Name:D1, Email:D2)
Which relations above are union (set operation)
compatible?
– Union compatibility may require type conversion
(casting).

Relational Algebra
2
Set Operations
¢¡¤£¦¥ §©¨£ ¥¡ ! #¤$%¥(')0¥$¡1$)
243©5 673
SR ∪
8 9 @BADC7E F©GIHQP©RDHSR©T¤F
F 5H QPUFVT 5 3 8 6 T 9
W 3 H¤FVTGQFVXYH 56 V3
SR ∩
8 9 @BADC7E F©GIHQP©RDHSR©T¤F
` 6 HDP 5 3 8aR 3Yb 9
c 5d Qd FVT¤F 3 XQF
SR −
8 9 @BADC7E F©GIHQP©RDHSR©T¤F 5 3
8e`¦AQH 36 H 5 3 9
ffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff
gggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggggg
hhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhhh
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii
pYq¢rs0t%u pQvw r¤x€y¤ ‚yx0vƒ
„Y… †(‡Yˆ‰ Smith@aol.com
„Y‘ ’“€”€•—– Johns@geocities.com
˜™ d0e “gf)• Brown@netmail.com
˜h i1e j—k Gray@microsoft.com
l0m)n)o0p0q l‚r s0tu vw‚xy o0x w0tz
{%| })~ € Gray@microsoft,com
{% ‚‚ƒ‚„g…‚~ † ‡ Robert432@aol.com
{0ˆ ‰ƒgŠ‚‹—‡ Johns@somewhere.com
{€Œ ‰—‡ ƒg‹ Jj@yahoo.com
{0 Ž(g—… Duke@jazz.com‘g’¤“0”Q• ‘0–— ˜Q–— ™š‚›œ —–ž—Ÿ g¡ œ ¢g£gž š€Ÿ—–£g¤
¥%¦ §Y¦ ¨)© ª‚«¬‚ ®€¯€ ° ¯g±
¥%² §Y² ³1´µgµ€´± ¶U®%´· ¸ ¹ ¯gº— ° ¯€±
¥0» §Y² ¼¬%«¬‚ © ¯gµ—¬ ½¤¬€¾¦¿¯€· À
¥€Á §» Â‚¯gÃ¤µ—ª— ³ª‚« «ª—
Ä0Å ÆÇ ÈÉ Ê€ËgÌ¤Í‚Ê‚Î—Ï‚Ð
É€ÑÒ Ó—Ô Ô Ê ÕÖ
Ó—× ×—Ø
Ö
Ä0Ù ÆYÚ Æ¤Û€ÑË—ÜÝÉ€ÑÛ
Ö Þ
Ê€Ô × Êgß
à‚áâ%ã à%äå â%æ€ç è á‚äå à0é—êgá€å ë1ì¤í—égî ç
ï0ðñ%ò ó0ô—õ€ö ÷%ò ïø òù
ï0ðñ%ø ó0ô—õ€ö ÷gú ï%ù û0ù
ï0ðñ0õ ó0ô—õ%ò ÷%ò ï€ú òó
ï0ðñgú ü%ôgò ÷0ù ï%ù ý%ù
ï0ðñ0ù ü%ôgò ÷%ò ïø øgö
ï0ðñ0û ü%ôgø ÷gú ï€ú ý‚ö
ï0ðñ‚ý ü%ôgø ÷%ø ïò ú
ï0ðñ0ó ü%ô—õ ÷%ø ï%õ ù

Relational Algebra
3
Assignment and Alias
• Given a relation R, head(R)=A1…An is the list of attributes
in R.
• For each attribute, dom(A) gives the domain of A,
name(A) gives the name of the attribute
• Suppose B1…Bn is a list of attributes such that for each i,
dom(Bi) = dom(Ai)
Then assignment (:=) changes the names of some or all of
the attributes in R and results in a new relation S,
S(B1…Bn) := R(A1…An)
and dom(B1) = dom(A1) … dom(Bn) = dom(An).
Set Operations
• Find all people who are registered as buyers or owners
¢¡¤£¦¥¨§© ! ! #
$% ')(10 243 Smith@aol.com
$65 798@3¤ACB Johns@geocities.com
DE% F¦GIHCP Gray@microsoft.com
QRQRQ QRQ@Q QRQRQ
BUYERS-2(oid,name,email) :=
BUYERS(buyid,name,email)
RESULT := BUYERS-2 ∪ OWNERS

Relational Algebra
4
Set Operations
• Find all people who are owners but not buyers
¢¡¤£¦¥¨§© ! !¤ #
$% ')(0 132 Smith@aol.com
$4 5768293@ Johns@geocities.com
$¨A B¦CED3F Gray@microsoft.com
G G G
RESULT:= OWNERS - BUYERS-2
Set Operations
• Find all people who are both owners and buyers
HPI7QSRUTV WYX ` acb7dUe e7dUbfXhg
RESULT := OWNERS ∩ BUYERS-2

Relational Algebra
5
Projection
• Select all tuples from relation R, restrict each tuple to the
input attributes
– written as R[A1…An] or
for attributes A1…An from R
– meaning
nAA ...1
π
for each tuple in R
construct a new tuple containing
only attributes A1…An
place the new tuple in the result
end-for
remove duplicate tuples to preserve
the set property
Projection
• Find the names and email addresses of people who are
both buyers and owners
¢¡¤£¦¥¨§© !#
$%('0) G ra y @ m ic ro s o ft.c o m
RESULT:= OWNERS[name,email] ∩ BUYERS[name,email]

Relational Algebra
6
Projection
• Find all the items on which at least one bid was placed
¢¡¤£¦¥¨§©

!
#
RESULT := BIDS[iid]
Selection
• Select certain tuples from a relation R based on a boolean
condition C
– written as R where C or
– meaning:
RCσ
for each tuple t in R
if t makes C true then
return t in the resulting relation
end-for

Relational Algebra
7
Selection
• The selection condition C is built of
– atomic conditions of the form Ai ∝ Aj or Ai ∝ c where
• Ai, Aj are attributes and c is a constant of the same domain of
values, and
• ∝ is a comparison predicate of the form: =, , , = , = , etc.
• If C1, C2 are selection conditions, then so are
• C1 AND C2
• C1 OR C2
• NOT C1
Selection
• Find the bid identifiers of bids for item number “I1”:
¢¡¤£¦¥¨§©
!
$#
¤%
I1-BIDS := BIDS WHERE iid = “I1”
RESULT := I1-BIDS[bid]

Relational Algebra
8
Selection
• Find the bid identifiers of bids for item number “I1” that are
strictly more than $15.
I1-BIDS := BIDS WHERE
iid = “I1” AND amount 15
RESULT := I1-BIDS[bid]
¢¡¤£¦¥¨§©
!
#
Selection
• Find the dates in which a bid for item number “I1” or
“I5” was placed
I1orI5-BIDS := BIDS WHERE
iid = “I1” OR iid = “i5”
RESULT := I1orI5-BIDS[date]
$%('#)¤021 3542687
9A@2BDC
9A@2BFE
G(@HE

Relational Algebra
9
Cartesian Product
• Multiply the relations R and S, producing |R|x|S| many
tuples with attributes head(R)+head(S)
• Cartesian product combines two relations into one “big”
relation
• meaning
for each tuple r in R do
for each tuple s in S do
compute a new tuple containing
all attributes in r and s
end-for
end-for
R := Bids WHERE Amount 20 T := Items WHERE oid=“O2”
S := T[iid,name]
P := R X S
¡£¢ ¤ ¥§¦©¨ ¢ ¤ ¡¤ !#©$¨
% '( )0213 465 %7 8£7
% '5 9@0A 47 %7 BC7
% '@8 9@02( 465 %5 B3
D EF G HPIQSR
T@U VXW Y2YW `badc£W egf
h@i p£q£r qstuY6q
v w§x€y‚ ƒ w§x „£…†b‡ w§x ˆ ƒ w‰x€y2g‘2ˆ6ƒ w§x ’§“X”•–† —Sx ˆ ƒ —Sx ˜#…©“X‡
™d e@f g2h6i6j kl ™m n‚m k2f oCd pp©d qgr
™d ste
™d e©l u2h©v k2m ™m w@m k2f oCd pp©d qgr
™d ste
™d en u2h©f kl ™l wj k2f oCd pp©d qgr
™d ste
™d e@f g2h6i6j kl ™m n‚m k©i x‚y@zy©{b|–}2pgy
™d e©l u2h©v k2m ™m w@m k©i x‚y@zy©{b|–}2pgy
™d en u2h©f kl ™l wj k©i x‚y@zy©{b|–}2pgy

Relational Algebra
10
Relational Algebra - Join
• For all pairs of tuples r and s from relations R and S where
Head(R)=A1…AnB1…Bk and Head(S)=B1…BkC1…Cm
such that r[B1…Bk] = s[B1…Bk],
include a tuple in R
¢¡ S with
attributes A1…AnB1…BkC1…Cm
• The attributes in common between the two relations (B1…Bk) are
called the “join attributes”
• The join is a cartesian product followed by a selection on the join
attributes, and a projection to remove the duplication of the join
attributes [formally: you have to rename the join attributes in one
relation so that you can express the selection/projection
operations]
Also called equijoin since
the main criteria is the
equality of join attributes
• Find all people with the same name and email address from the
“buyers” and “owners” relations
– join attributes are “Name” and “Email” from the buyer and
owners relations
– the joined relation will have attributes “Buyid, Name, Email, Oid”
– there are 5 × 4 pairs of tuples to compare on the attributes
“Name” and “Email”
£¥¤§¦©¨
¥ ¨¥
£!$#%' )(1032 ¨ 03( %54 %'
67 8@9BADC G r a y @ m ic ro s o ft,c o m E3F

Relational Algebra
11
¢¡¤£¦¥¢§¢¨ ¢© Gray@microsoft,com
!# $% Smith@aol.com
#')(10323
45 6¤798A@B Johns@geocities.com
4DC)E#F¢GIH
PDQ RASUTWVYX Brown@netmail.com `badcDegf3h
iqp r¤sUtvu Gray@microsoft.com
ixw€ybA‚¢ƒ
„¢…¤†¦‡¢ˆ¢‰ „¢ ‘’“” Gray@microsoft,com
„¢…¤†¦‡¢ˆ¢‰ „¢ ‘’“” Gray@microsoft,com
•¢–¤—¦˜¢™¢d •¢e fghi Gray@microsoft,com
Example below shows four of the twenty pairs of tuples
considered for the join
• Find the names of items with bids over $20
• Find the names of the owners of items with bids over $20
and are located in Boston
Temp1 := Bids WHERE Amount 20
Temp2 := Items jlk Temp1
Result := Temp2[Name]
Temp1 := Bids WHERE Amount 20
Temp3 := Items WHERE Location = “Boston”
Temp2 := Temp3 jlk Temp1
Temp4 := Temp2 jlk Owners
Result := Temp4[Owners.Name]

Relational Algebra
12
Relational Algebra - Division
• Let head(R)=A1…AnB1…Bk and head(S)= B1…Bk
• R ÷ S is the “maximal” set of tuples from R[A1…An] such that
R contains all tuples in (R ÷ S) x S.
• Division is often
used for “for-all”
queries, e.g. “Find
tuples in R that are
in a relation with
‘all’ tuples in S.”
• All the attributes in the dividing relation S must exist in the
divided relation R!
for all tuples r in R[A1…An]
if for all tuples s in S, r+s is in R
place r in R ÷ S
end-if
end-for
¡ ¢ £
¤¦¥ §¨¥ ©¥
¤ §¨¥ ©¥
¤¦¥ § ©¥
¤¦¥ § ©
¤ §¨¥ ©
¤¦¥ § ©
¤¦¥ § ©
¤¦¥ §¨¥ ©
!
$#
$%
' (
)10 230
4 5 6
798 @38
7BA @38
Find:
• R ÷ S
• R ÷ V
• R ÷ Y
C DEGFIH¦PQSR TUR VXW
÷ S

Relational Algebra
13
¡ ¢ £
¤¦¥ §¨¥ ©¥
¤ §¨¥ ©¥
¤¦¥ § ©¥
¤¦¥ § ©
¤ §¨¥ ©
¤¦¥ § ©
¤¦¥ § ©
¤¦¥ §¨¥ ©
!
$#
$%
' (
)10 230
4 5 6
798 @38
7BA @38
Find:
• R ÷ S
• R ÷ V
• R ÷ Y
• Find identifiers of buyers who placed a bid on “all” the items
• Find the identifiers of items that have a bid from “all” the buyers
Temp := Items[Iid]
Temp1:= Bids[Buyid,Iid]
Result := Temp1 ÷ Temp
A := Buyers[BuyId]
B := Bids[Iid, BuyId]
Result := B ÷ A

Relational Algebra
14
Relational Algebra Operations
– Union R ∪ S
– Intersection R ∩ S
– Difference R - S
– Assignment S(B1…Bn) := R(A1…An)
– Selection R where C
– Projection R[A1…An]
– Cartesian Product R x S
– Join R
¢¡ S
– Division R ÷ S
£¥¤§¦¥¨© £¥ ¦ !!#
$¥% (')021 Smith@aol.com
$¥3 465718@9 Johns@geocities.com
$A BC@5EDF8 Brown@netmail.com
$ G H(CPIRQ Gray@microsoft.com
S!TFU(VWX SY2`bac d¥ef g Vfe!ah
ibp qsrut2v Gray@microsoft,com
iw xy7€7!r@‚Pƒ Robert432@aol.com
i„ …6y7†‡Pƒ Johns@somewhere.com
i7ˆ …6tRƒRy!‡ Jj@yahoo.com
i‰ (‘E’@ Duke@jazz.com “R”F•–— “˜™ de˜™ f
gEh(i j2kRl2g2m
˜
kon
p7q req sbtuuvwRx yEz@x@{Pz7|
p7} re} ~6€R€o|u
y‚„ƒ …
zo†PxP{Pz7|
pE‡ re} ˆ!wvwox2t@zR€@w ‰FwoŠ
‹zR‚@Œ
2Ž ‘7’o“s”@•2–P– —!•˜˜•@–
E™ FŽ š› 2’Rœ2
ž’oŸR 7¡
›R¢£u•2–P–@’
¤!¥ •@¦P¦o˜¥
o§ e¨ ›R¢” ¥ © ’@–@¦P’7ª
«!¬(® «b¯° (±E²@³ ¬!¯° «´2µ¬E° ¶s· ¸7´E¹2²
º»¼b½ ¾!¿7À7Á Â½ ºÃ ½uÄ
º»¼bÃ ¾!¿7À7Á ÂRÅ ºÄ ÆÄ
º»¼À ¾!¿7À½ Â½ º7Å ½„¾
º»¼Å Ç!¿½ ÂEÆ ºÄ ÈbÄ
º»¼Ä Ç!¿½ Â½ ºÃ ÃEÁ
º»¼Æ Ç!¿Ã ÂRÅ º7Å È!Á
º»¼!È Ç!¿Ã ÂÃ º½ Å
º»¼¾ Ç!¿7À ÂÃ ºÀ Ä
Auction Database

Relational Algebra
15
Problem 1
Find the name and email of owners of items located in
“Boston”
A := Items WHERE Location = “Boston”
B := A
¢¡ Owners
Result := B[Owners.Name, Owners.Email]
Problem 2
Find the identifiers and amount of bids placed by
buyer “Roberts”
A := Buyers WHERE Name = “Roberts”
B := A
¢¡ Bids
Result := B[Bid, Amount]

Relational Algebra
16
Problem 3
Find the names of buyers who placed a bid on an item
owned by “Brown”
A := Owners WHERE Name = “Brown”
B := (A
¢¡ Items) [Iid]
C := B
¢¡ Bids
Result := (C
¢¡ Buyers) [Buyers.Name]
Problem 4
Find the identifier of items with more than one bid
A := Bids[Bid, Iid]
B := A x Bids[Bid,Iid]
C := B WHERE (A.Bid Bids.Bid)
AND (A.Iid = Bids.Iid)
Result := C[Iid]

Relational Algebra
17
Problem 5
Find the names of items all buyers placed a bid on
A := (Bids[Iid, BuyId]) ÷ (Buyers[Buyid])
B := A
¢¡ Items
Result := B[Name]

[Www.pkbulk.blogspot.com]dbms04

More Related Content

Viewers also liked (7)

Similar to [Www.pkbulk.blogspot.com]dbms04 (20)

More from AnusAhmad (20)

Recently uploaded (20)

[Www.pkbulk.blogspot.com]dbms04