SLIDE 1
Security in Outsourcing of Association Rule Mining
- !"
Security in Outsourcing of Association Rule Mining - - PowerPoint PPT Presentation
Security in Outsourcing of Association Rule Mining - - PowerPoint PPT Presentation
Security in Outsourcing of Association Rule Mining
SLIDE 2
SLIDE 3
Introduction and motivation
"$
$%)&()%
$(
- $
+%$ $((($
SLIDE 4
Security concerns in outsourcing
'%&$, %$
- $% . %
/%&0$$ $$
- $% . %
12 $
SLIDE 5
Outsource model
- 3
'( "$ 4/430 "$ 4/40
Data Owner
Data Miner Outsourcing
SLIDE 6
Item mapping - encryption
SLIDE 7
Example item mapping (one-to-one)
,+567 $$+5896
:,$$5+5:896675
:678965
:$,5:,$$5;
*,$%
$&%()
"&$;;;;
SLIDE 8
One-to-n item mapping
"++%%
<( <#+5
)%#=>,$?
=>8@76?
/0=>876? /,0=>? /$0=>@6?
#++$;
SLIDE 9
Itemset mapping using one-to-n item mapping
<#+5<++%% <# +5<%% /A0=B) A /)0=C +8/C0=A(/A0=C )%<
/:$50=:8@765 /:,$50=:@65 +8/:8@7650=:$5 +8/:8@7650=:,$5
<%%
%%
< +5>876? ,+5>? $+5>@6?
SLIDE 10
Correctness – restrictions on one- to-n mapping
:,5=5:8@5 :,$5=5:8@5
D $&%(D
:5=5:85
EEE
:,5=5:8@5
F
:,$5=5:8@
75
< +5>8? ,+5>@? $+5>8@? 3< +5>? ,+5>? $+5>?
", %%<%%($$ G 4<+8//A00=A/$$$&%0(( ,
SLIDE 11
Is one-to-n mapping more secure?
'=
>? >,? >$? >,? >$? >,$? >,$?
'3 =
>876? >? >@6? >876? >8@76? >@6? >8@76?
< +5>876? ,+5>? $+5>@6? 3< +5>8? ,+5>? $+5>@?
Todecrypttransactionsencryptedby,wecanuse! (misnotmoresecurethanm’)!!!!
SLIDE 12
Function coverage
M1:2I !>2D1 M2:2I !>2D2 M1 coversM2iff
forallX⊆ I,letY=M2(X)
M2
!1(Y)=M1 !1(Y∩ D1)
M1 coversM2
IfanytransactionencryptedbyM2canbe
decryptedbyusingtheinverseofM1
SLIDE 13
One-to-n is not more secure than
- ne-to-one mapping
H/%0
"&,++
%%$,&/$, $&%,&0 ++ %%
DDD
H++%%H'
$++%%
SLIDE 14
One-to-n vs one-to-one
++ ++;
#&++,$
B(*$<
++ I %%
=++ I %% H<
"$%$
(
#++&,/
$0++
SLIDE 15
One-to-n Transformation
++%%
+5>8?,+5>?F =>,?
3 =>8?
++%%
+5>8@?,+5>@?F =>,?
3 =>8@?
++(
+5>8@?,+5>@?F =>,?
3 =>8@79 ?
Randomly generated
SLIDE 16
Transaction transformation
<# +5 ,++
%%
<$(
%(# BJ
3 =/0=/0B
,(BJKJ
(
+8/30=>)L/)03?
SLIDE 17
Example transformation
'=
>? >,? >$? >,? >$? >,$? >,$?
'3 =
>876? >8@? >@67? >876? >8@76? >@68? >8@76?
< +5>876? ,+5>? $+5>@6? 3< +5>8? ,+5>? $+5>@? /0=/0B
!
- !
SLIDE 18
Necessary properties of transformation N
'$&%$$ +8//00=
%/,0
J&$/0&
%,3 /=/0B0 $+8/30=
- <++
%%$$$% $((++ %%
SLIDE 19
Generating E for valid and complete transformation N
N(<c>)=<3,5>UE
Form:a!>{1,4,5}
E={1}or{4},butnot{1,4}
Form:b!>{2}
E=Φ
ThetransformationNisvalidifEiseither{1}or
{4}orΦ ;
Niscompleteifitispossibletogenerateallofthe
threecases,i.e.,E={1}or{4}orΦ.
< +5>876? ,+5>? $+5>@6?
E=?
SLIDE 20
Algorithm – valid and complete transaction transformation
SLIDE 21
Algorithm to perform valid and complete transformation
t=<…>
Start
Meet quota? a!>… b!>… … !>… Mappings No N(t) Pickone x!>x1,…,xn History Storesitemswe mustnotadd xi,…,xj Filter E E=Ø atstart Someadd toE Othersto history Next AddEto result
SLIDE 22
Important Property
'$(
%$,&" $%E
SLIDE 23
Experiments
SLIDE 24
Design
- %
*&$&(($$&(
*$&
"$
(G$
#%$((G$&
&
(($$&
'( H
SLIDE 25
Background knowledge
- %<$%$$
$;
/"%0
#&$$(,
&%,$/,$0
$/%0
%<%M(
,
,<%%
- %%N/8,0
SLIDE 26
20 40 60 80 100 100% 90% 80% 70% 60% 50% Mapping accuracy (%) 0% 10% 20% 30% 40% 50%
alpha beta
Mapping accuracy
&%%$$
H&,$
$((%%
SLIDE 27
Efficiency
796 @O@ P@ 7 O
H $
88 P76 !@O 7OO 8P6
- 88E
7
8E6 PE6 6E6 EO
- /(
$&0
6 @ 8
SLIDE 28
Summary
'(,$%
%,($&%( $,
H++%%$,
$&%%$(($& ++%%
'$(
%%, $%
)%,
($
SLIDE 29