An%20Experimental%20Study%20of%20Association%20Rule%20Hiding%20Techniques

1
An Experimental Study of Association Rule Hiding
Techniques

• Emmanuel Pontikakis
• pontikak_at_ceid.upatras.gr
• Dept. of Computer Engineering and Informatics
• University of Patras
• Patra, Greece
• Vassilios Verykios
• verykios_at_cti.gr
• Dept. of Computer and Communication
  EngineeringUniversity of ThessalyVolos, Greece
• Computer Technology Institute
• Research Unit 3
• Athens, Greece

2
Outline

• Introduction - Related Work
• Distortion-based Techniques
• Blocking-based Techniques
• Comparison and Analysis
• Conclusions

3
Introduction
Database
Changed Database
4
Related Work

• Association Rule Hiding
• Blocking-based Technique (Saygin, Verykios,
  Clifton)
• Distortion-based (Sanitization) Technique
  (Oliveira, Zaiane, Verykios, Dasseni)

5
Outline

• Introduction - Related Work
• Distortion-based Techniques
• Blocking-based Techniques
• Comparison and Analysis
• Conclusion

6
Distortion-based Techniques
Sample Database
Distorted Database
A B C D
1 1 1 0
1 0 1 1
0 0 0 1
1 1 1 0
1 0 1 1
A B C D
1 1 1 0
1 0 0 1
0 0 0 1
1 1 1 0
1 0 0 1
Rule A?C has Support(A?C)80 Confidence(A?C)10
0
Rule A?C has now Support(A?C)40 Confidence(A?C
)50
7
Side Effects
Before Hiding Process After Hiding Process Side Effect
Rule Ri has had conf(Ri)gtMCT Rule Ri has now conf(Ri)ltMCT Rule Eliminated (Undesirable Side Effect)
Rule Ri has had conf(Ri)ltMCT Rule Ri has now conf(Ri)gtMCT Ghost Rule (Undesirable Side Effect)
Large Itemset I has had sup(I)gtMST Itemset I has now sup(I)ltMST Itemset Eliminated (Undesirable Side Effect)
8
Distortion-based Techniques

• Challenges/Goals
• To minimize the undesirable Side Effects that the
  hiding process causes to non-sensitive rules.
• To minimize the number of 1s that must be
  deleted in the database.
• Algorithms must be linear in time as the database
  increases in size.

9
Our Proposal Weight-based Sorting Distortion
Algorithm (WSDA)

• High Level Description
• Input
• Initial Database
• Set of Sensitive Rules
• Safety Margin (for example 10)
• Output
• Sanitized Database
• Sensitive Rules no longer hold in the Database

10
WSDA Algorithm

• High Level Description
• 1st step
• Retrieve the set of transactions which support
  sensitive rule RS
• For each sensitive rule RS find the number N1 of
  transaction in which, one item that supports the
  rule will be deleted

11
WSDA Algorithm

• High Level Description
• 2nd step
• For each rule Ri in the Database with common
  items with RS compute a weight w that denotes how
  strong is Ri
• For each transaction that supports RS compute a
  priority Pi, that denotes how many strong rules
  this transaction supports

12
WSDA Algorithm

• High Level Description
• 3rd step
• Sort the N1 transactions in ascending order
  according to their priority value Pi
• 4th step
• For the first N1 transactions hide an item that
  is contained in RS

13
WSDA Algorithm

• High Level Description
• 5th step
• Update confidence and support values for other
  rules in the database

14
Experimental Results of WSDA algorithm
15
Experimental Results of WSDA algorithm
16
Outline

• Introduction - Related Work
• Distortion-based Techniques
• Blocking-based Techniques
• Comparison and Analysis
• Conclusion

17
Quality of Data

• Sometimes it is dangerous to delete some items
  from the database (etc. medical databases)
  because the false data may create undesirable
  effects.
• So, we have to hide the rules in the database by
  adding uncertainty without distorting the
  database.

18
Blocking-based Techniques
Initial Database
New Database
A B C D
1 1 1 0
1 0 1 1
0 0 0 1
1 1 1 0
1 0 1 1
A B C D
1 1 1 0
1 0 ? 1
? 0 0 1
1 1 1 0
1 0 1 1
Support and Confidence becomes marginal. In New
Database 60 conf(A ? C) 100
19
Modification of Association Rule Definition

• A rules A?B confidence and support becomes
  marginal
• sup(A?B) minsup(A?B), maxsup(A?B)
• conf(A?B) minconf(A?B), maxconf(A?B)
• minsup(A?B)
• maxsup(A?B)

20
Modification of Association Rule Definition

• minconf(A?B)
• maxconf(A?B)

21
Negative Border Rules Set (NBRS) Definition

• When a rule R has either
• sup(R)gtMST AND conf(R)ltMCT
• OR
• sup(R)ltMST AND conf(R)gtMCT,
• then we say that R belongs to NBRS.

22
Side Effects Definition Modification in
Blocking-based Techniques
Before Hiding Process After Hiding Process Side Effect
Rule Ri has had conf(Ri)gtMCT Rule Ri has now minconf(Ri)ltMCT Rule Eliminated (Undesirable Side Effect)
Rule Ri has had conf(Ri)ltMCT Rule Ri has now maxconf(Ri)gtMCT Ghost Rule (Desirable Side Effect)
Large Itemset I has had sup(I)gtMST Itemset I has now minsup(I)ltMST Itemset Eliminated (Undesirable Side Effect)
Itemset I has had sup(I)ltMST Itemset I has now maxsup(I)gtMST Ghost Itemset (Desirable Side Effect)
23
Privacy Breaches Definitions

• If an item i, some values of which, are hidden by
  ?s, is contained in a sensitive rule, a privacy
  breach will occur if the adversary can assume
  that with c confidence.
• For a rule R with maxconf(R)gtMCT, a privacy
  breach occurs if it can be estimated, with c
  confidence, that R is either a sensitive or a
  ghost rule.
• For a blocked item i in a specific transaction T,
  a privacy breach occurs if the adversary can
  estimate with c confidence that its original
  value is either 0 or 1.

24
Blocking-Based Techniques

• Goals that an algorithm has to achieve
• To put a relatively small number of ?s and
  reduce significantly the confidence of senstitive
  rules.
• To minimize the undesirable side effects (rules
  and itemsets lost) by selecting the items in the
  appropriate transactions to change, and maximize
  the desirable side effects.
• To modify the database in a way that an adversary
  cannot recover the original values of the
  database.

25
Our Proposal Blocking Algorithm (BA)

• High Level Description
• 1st step
• For each sensitive rule RS (Rule RS has left
  itemset IL and right itemset IR) compute how many
  0s and 1s you have to block, in order to reduce
  the confidence of RS.
• 2nd step
• Find the set of transactions TR that support RS
  or the set of transactions TLpR that support
  partially RS (support partially the left itemset
  and do not support the right itemset).
• For each transaction in TR find the rules Rcommon
  with at least one common item with IR and for
  each transaction in TLpR find the Rcommon?NBRS
  with at least one common item with IL.
• Assign a weight w for each Rcommon and a weight
  w for each Rcommon.
• Assign a PT for each transaction in T such as PT
  is large if transaction Ti has many Rcommon rules
  with large w, and a priority value PT for each
  Ti such as PT is small if transaction T has
  many Rcommon rules with large w.

26
Blocking Algorithm

• High Level Description
• 3rd step
• Sort T?TR starting from them with lowest PTi. and
  sort T?TLRp starting from them with highest
  PTi.
• 4th step
• For the first N1 sorted T?TR block an item i?IR
  and for the first N0 sorted T?TLRp block an item
  i? IL
• 5th step
• Update values minconf(Ri), minsup(Ri), for all
  other rules that have been affected.

27
Blocking-Based Techniques

• Main Problems of blocking technique
• The maximum confidence of a sensitive rule cannot
  be reduced.
• An adversary can infer the hidden values if he
  applies a smart inference technique, if the
  blocking algorithm does not add much uncertainty
  in the database.
• Both 0s and 1s must be hidden, because if only
  1s were hidden the adversary would simply
  replace all the ?s with 1s and would restore
  easily the initial database.
• Many ?s must be inserted, if we dont want an
  adversary to infer hidden data.

28
Experimental Results of Blocking Algorithm
29
Experimental Results of Blocking Algorithm (2)
30
Experimental Results of Blocking Algorithm (3)
31
Outline

• Introduction - Related Work
• Distortion-based Techniques
• Blocking-based Techniques
• Comparison and Analysis
• Conclusions

32
Comparison and Analysis
Distortion-based Techniques Blocking-based Techniques
Privacy Breaches No privacy breaches Many kinds of privacy breaches
Simplicity of algorithms Simpler More complicated
Database Modification Database contains false information Many ?s must be inserted in the Database
33
Outline

• Introduction - Related Work
• Distortion-based Techniques
• Blocking-based Techniques
• Comparison and Analysis
• Conclusions

34
Conclusions

• There are open research problems in Blocking
  Technique
• A) What techniques must be used in order to
  reduce the privacy breaches?
• B) In what other ways can we prevent an adversary
  from inferring the association rules in the
  database?
• C) Maybe applying a chi-square test to the final
  database reveal some correlations between the
  items

35
References

• Evfimienski et.al Alexandre Evfimievski,
  Ramakrishnan Srikant, Rakesh Agrawal, Johannes
  Gehrke. Privacy Preserving Mining of Association
  Rules. SIGKDD 2002, Edmonton, Alberta Canada.
• Murat Kantarcioglou and Chris Clifton, Privacy
  Preserving Distributed Mining of Association
  Rules on Horizontally Partitioned Data, In
  Proceedings of the ACM SIGMOD Workshop on
  Research Issues in Data Mining and Knowledge
  Discovery (2002), 2431.
• Jaideep Vaidya and Chris Clifton, Privacy
  Preserving Association Rule Mining in Vertically
  Partitioned Data, In the 8th ACM SIGKDD
  International Conference on Knowledge Discovery
  and Data Mining (2002), 639644.

36
References

• Stanley R. M. Oliveira and Osmar R. Zaïane.
  Algorithms for Balacing Privacy and Knowledge
  Discovery in Association Rule Mining.  In Proc.
  of the Seventh International Database Engineering
  Applications Symposium (IDEAS'03), pp. 54-63,
  Hong Kong, July 16-18, 2003.
• Yucel Saygin, Vassilios Verykios, and Chris
  Clifton, Using Unknowns to Prevent Discovery of
  Association Rules, SIGMOD Record 30 (2001), no.
  4, 4554.
• S. Verykios, Ahmed K. Elmagarmid, Bertino Elisa,
  Yucel Saygin, and Dasseni Elena, Association Rule
  Hiding, IEEE Transactions on Knowledge and Data
  Engineering (2003).