Appunti per l'esame di Data Bases 2

I

1b10) must if

- New go : s

on = max

you

Hobowall ,

.

IBardl

thrashold Otherwise

I min

the mir

I

T =

Comete the con

ten

down .

found

that ore

fac

Svenbound we

highest

with re-consente

the and

obsects bound

first score

lower

their

set

n by

must objects

the

After

stor you

ocen

every

con

sor . position

the ser

objects .

of

accessed

the

to

according new

CownBounds

the LB stesso 0

sullo ,

Example +

masca

zona se

: volare colema

della

stesso ultimo

sullo volare

UB manca

zona se

= b

, ↓

. = 0

15 o to =

cos :

Awas pub

best 08 2

2+ .

6

+ =

=

0

& core 1

: balb

chaiches

↑ corrent os o r

:

tog

J ,

2

> Ib 13

- home

they

because

must T

you go on

Someub !

↳ (scorel

2 )

T

=>

Some = .

2

T Stal

to

= 0 ,

,

7

T 3,5 +

= + a

0 95

-

T , =

9 7Sta6

= 0 + a =

. 1

, 77

/

2 23

, Pr

P

2

lihtbit P1

best

6

96 0

97 +0

75

+ +

0

+ 0 90

0 ,

. . .

wort com b recent

cos Rient

6

019tal to

5

970

astafto 8 0 , present

+

to , ,

0

. not a

- I

dias

6

7340 973 ↓

5 +

,

1 0.

, +

aizstasta7 ↓

2

0 ,

5

1 +0]

0

+ ,

w

& D

e Somma

Pinet bsoma

1902

↓ na rigay

considerati h)

(dopo

The MaxOptimal algorithm

Input: integer k, a monotone function S combining ranked lists R1, …, Rm;

Output: the top k pairs:

1. Do a sorted access on each object o

2. For each object o, do random accesses in the other lists Rj , thus extracting score sj

3. Compute overall score S(s1, …, sm), if you don’t have o on the other lists compute always

the max with the same value. If the value is among the k highest seen so far, remember o

4. Let sLi be the last score seen under sorted access for Ri

5. Define threshold T=S(sL1, …, sLm)

6. If the score of the k-th object is worse than T, go to step 1

7. Return the current top-k objects (with the highest max)

EXAMPLE Summary rach

sonted an

access

Do a its

of

the

compute max

and

element ,

volul

the

it's seme

done

they

(if ore

score timel ar

the

infinite

thrashold T

define

then a find

until son

T

deceose

end

other

vedate access

. Do

it elements

tagn

the

have

ser

so

the tay sare ,

h max

Comparing top-k query algorithms

Sated Random

accesses : accesses

:

medRank TA

Bo FA

Optimal

Max

NRA

Skylines

Tuple t dominates tuple s, indicated t s, iff

≺

– 1≤i≤m t[Ai ] ≤ s[Ai ] (t is nowhere worse than s)

→

∀i.

– 1≤j≤m∧ t[Aj ] < s[Aj ] (and better at least once)

∃j.

(lower values are better. Opposite convention wrt. top-k queries)

Skyline of a relation: set of its non-dominated tuples, set of potentially optimal tuples,

set of all top 1 objects according to some monotone scoring function

Skyline ≠ Top-k query because it is based on the notion of dominance and there is no

scoring function that, on all possible instances, yields in the first k positions the skyline

points.

k-skyband = set of tuples dominated by less than k tuples

Skylines - Block Nested loop (BNL)

Input: a dataset D of multi-dimensional points;

Output: the skyline of D

1. Let W = Ø

2. for every point p in D

3. if p not dominated by any point in W

4. remove from W the points dominated by p

5. add p to W

6. return W

Skylines - Sort Filter Skyline (SFS)

Input: a dataset D of multi-dimensional points;

Output: the skyline of D Summary accending

1. Let S = D sorted by a monotone function of D’s attributes points

the

onder

Sat accending

in first

Add the

sated

2. Let W = Ø accesses

do .

and

dominate f

if to

A B they

points if

all s

ed

its volus in

Then

.

I

ore intos

point

Of

3. for every point p in S , element

Bone at lest

and by

(NoI

dominated ore

! not

is

one ar menshoted the

all

mot

4. if p not dominated by any point in W if

point is

A

S

in in

the

. point enes

are

the

of

5. add p to W colus ·,

6. return W dominated

not sinline

EXAMPLE Ma

Novatel add it

because empty to

int

↳ s

s in

: o

un

Gillion

·

sort Novotel

by because of

the values

D all

:

25

f 0

Die . Willion Nevotel's

- th not itto

38 ore add

one

0 5

-

, 4

0 . 1 bis Novotel not

because Ibis

the

by all values of

No

475 : on

0 . it

oo

Nevotel's to

the

=> one 5

Straten Novotel

by because of

the values

D all

:

steaton

points Nevotel's

the

the not itto

or add

one

gray

in 5

-

dominated

ore

acc Hilton Novotel

by

offen peints

by because of

the

, values

some D all

:

it

thow

so away

son on Hilton Nevotel's

He not itto

or add

one 5

-

SEY

Se

E - : SE Gh

G dominato da Se

E

e

non =

e ,

Ap dominato do SE

GS2 Al

E G

SoRT mon e ,

= ,

,

dominato

>

- da

T A

e GAEDS 24

dominato SE

b E

da G A

ma el =

, , ,

,

dominato de A

e dominato

M de A

ri (Artidh

L 6

, . GEIG 53

dominato

3 do

ven Se

e E via = ,

A 4

,

D

, , ,

E6 EXL

ALT

GXA

EXA G4L AXL ALD

,

, ,

. , , ,

, 243

LCm 643

A A43

G2M E

CM 43

,

# ,

,

, , ,

1 SEY

Se

domina

E :

Eras -

G SE Gh

-

EXO Se

2 =

>

- ,

GXA

EXA SE

An Sa Al

G

, => ,

= ,

GXT

EXT ALT

+ ,

, S 24

SE

LEXL G4L G

ALL A

= = , ,

,

, , dominato

perché

LLD-drancese qualamo

de

CXDIALD da

DELD e

già so a

, ,

M

MEX GLM

, BEIG 53

A45 24 Se

E43 042 = 1

3 A 4

, , ,

, ,

2 SEY

Se

E :

-

GELG

EXA SE

An Al

Sa

=> ,

:

ELT

T ELL

↳

DELD

M ELM

ELJ

the

In

.

3 da

poide dominati

SE

2-sayBond punti

A3 di

di Da meno

: ,

Cin questo 0

da

casa

Fontil of

MR Come

1 the helf lists

the

a

S of

to more

. .

U max

BO a

S row

. , 8

now

until

FA trandom

Coman of

1

a

s

. missing

occhs

di

S "I f

fls least

time

. T

if

TA missime story and

scare

nouge

~ =

, , fleat

T

Irawleron 41b

if

stag

Sat offen and,

NRA =

!

a

eneg ,

not

Ib f(PE) pa mesent

o

=

= , f (B) present

03 =

f(PE) ment

ub Py o

=

= , mot

lauer vole

Pr proget

=

MAXO ofte

sot

42 T

if

stop Cultima

f Cestial every occer

= ,

sety points

SFS the da

point

20 ins S ND massimo

,

, punto tutti

da quaero voloni

Un

(se in quelli

D in

5 è

mol I

sai

a

e se

.

line befor

but

BNL adding that

Points volue

by

fam

sfs in s D

remove S

,

,

metze

(volare im

ediese

MR sal st

usa

Bo Isolomaxl

santminime

FA fllst

Sa vow)

mining

TA U2

+ -

NRA

MAXD

SFS

BNL integration

data

for

architectures

Application to

data

relational

mapping

for

interface

an

of

specification

(JPA):

API

Persistent

Java •

Java

in

data

oriented

object Java

in

transactions

managing

for

API

(JTA):

API

Transaction

Java

JPA provides a POJO (Plain Old Java Object) that is an object with methods gets and

-used to

set (as Javabean) menge

objects

more

design

Database

EXAMPLE

Application design Components:

Database connection in web tier

Repetitive code in the controllers

At initialization:

At termination:

Persistent data: extraction

Persistent data: creation

Persistent data: modification

Transaction

Data extraction with JPA

Data creation with JPA

Data modification with JPA

Transaction management in JPA

JPA object relational mapping

Object-relational mapping (ORM): technique of bridging the gap between the object

model and the relational mode, it tries to map the concepts from one model onto another

Impedance mismatch: challenge of mapping one model to the other lies in the concepts

in one model for which there is no logical equivalent in the other

Differences between ORM and RM

JPA main concepts to mapping:

Entity: a class (JavaBean) representing a collection of persistent objects mapped onto a

relational table;

Persistence Unit: the set of all classes that are persistently mapped to one database

(analogous to the notion of db schema);

Persistence Context: the set of all managed objects of the entities defined in the

persistence unit (analogous to the notion of db instance);

Managed entity: an entity part of a persistence context for which the changes of the

state are tracked;

Entity manager: the interface for interacting with a Persistence Context;

Client: a component that can interact with a Persistence Context, indirectly through an

Entity Manager (e.g., an EJB component). EXAMPLE enth

A

of spa

Entity manager operations

Entity EXAMPLE

Properties:

• Identification (primary key);

• Nesting;

• Relationship;

• Referential integrity (foreign key);

• Inheritance.