D-Eclat Association Rules on Vertically Partitioned Dynamic Data to Outsourced Securely

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395 -0056
Volume: 04 Issue: 06 | June-2017 www.irjet.net p-ISSN: 2395-0072
© 2017, IRJET | Impact Factor value: 5.181 | ISO 9001:2008 Certified Journal | Page 2176
D-Eclat Association Rules on Vertically Partitioned Dynamic Data to
Outsourced Securely
Rasika Khairnar1, Prof. P. D. Lambhate2
1 Student, Department of Computer Engineering, JSPM College of Engineering, Pune, Maharashtra, India
2 Asst. Professor, Department of Computer Engineering, JSPM College of Engineering, Pune, Maharashtra, India
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Cloud computing is enhanced with the
information mining-as-a service model. This service
becomes popular choice among various kind of companies.
This service is the cost effective, secure, time efficient and
reliable. The various organizations does not have mining
abilities, therefore they can outsource their mining need on
cloud server. But the association rules and item sets of
database are the private properties of organization. It
suffers from the problems of security and authorization.
This paper focused on the problem of accurate association
rule mining over outsourced database with achieving the
security and privacy of data. For this, initially data owner
encrypt the data before outsourcing. After this, when client
requesting for mining to the cloud server, it returns the
results in encrypted format. For association rule mining
system makes use of D-Eclat algorithm and proves that it is
more time and memory efficient than Eclat algorithm. Also,
association rule mining is applied on both horizontally and
vertically partitioned data and prove that vertically
partitioned outperforms in terms of time efficiency and
system utilization. The performance of the system is tested
on dynamically created dataset.
Key Words: Data mining, data outsourcing, privacy
preserving, database partitions, association rule
mining.
1. INTRODUCTION
Paragraph comes content here. Paragraph comes content
Information mining method has emerged as methods for
identifying patterns as well as patterns from extensive
amounts of information[1]. Mining incorporates different
algorithms, for example, clustering, classification,
association rule mining as well as sequence detection.
Generally, every one of these algorithms have been
produced in a centralized model, with all information
being accumulated into a central site, as well as algorithms
being keep running against that information. Association
rule mining discovers all principles in the databases that
fulfill some minimum support as well as minimum
confidence constraints[2]. Numerous algorithms are
utilized to improve the protection and security of
information. Vertically partitioned imply that each site
contains a few components of an transaction. Utilizing the
conventional market basket case, one site may contain
basic supply buys, while another has apparel buys.
Utilizing a key, for example, MasterCard number and date,
we can join these to recognize connections between buys
of dress and basic supplies. In any case, this unveils the
individual buys at each site, perhaps damaging purchaser
security agreements. There are more reasonable cases. In
the process of sub-assembly manufacturing, diverse
makers give segments of the completed item. Cars
incorporate a few subcomponents; tires, electrical
hardware, and so on; made by independent producers.
Once more, we have restrictive information gathered by a
few parties, with a single key joining every one of the
informational sets, where mining would help
distinguish/foresee breakdowns. A real life example is the
cur-rent trouble in Ford Motor as well as Firestone Tire.
Ford Explorers with Firestone tires from a particular
factory had tread separation issues in specific conditions,
which have caused 800 injuries. Due to the tires did not
have any issues on other cars, as well as other tires on
Ford Explorers did not have any issue, neither one of the
sides felt capable. The time taken to find the main issue
prompted an advertising bad dream and the inevitable
substitution of 14.1 million tires. A large number of these
were most likely fine Ford Explorers represented just 6.5
million of the supplanted tires. Manufacturers had their
own particular information early era of association rules
in light of the greater part of the information may have
empowered Ford and Firestone to determine the security
issue before it turned into an public relations bad dream.
Casually, the issue is to mine association rules crosswise
over two databases, where the columns in the table are at
various sites, splitting each row. One database is assigned
the primary as well as is the initiator of the protocol. The
other database is the responder. There is a join enter
exhibit in both databases. The rest of the attributes are
available in one database or the other, however not both.
The objective is to discover association rules including
attributes other than the join key.
The main contribution of this systems enlists here:
 Dynamic data encryption and outsourcing.
 Horizontal and vertical database partitioning
 Association rule mining over encrypted data by
using Eclat and D-Eclat
 Secure Outsourcing of association rules over
cloud server.
In this paper we study about the related work done, in
section II, the proposed approach modules description,
mathematical modeling, algorithm and experimental setup

in section III .and at final we provide a conclusion in
section IV.
2. REVIEW OF LITERATURE
M. N. Kumbhar and R. Kharat, have the several of
technique for PPARM[1] is performed also their outcomes
are thought about. For fulfilling the privacy constraints in
vertically partitioned databases, algorithm in view of
cryptography methods, Homo-morphic encryption, Secure
Scalar product as well as Shamir’s secret sharing strategy
are utilized. For horizontal Partitioned databases,
algorithm that consolidates advantage standpoint of both
RSA public key cryptosystem as well as Homomorphic
encryption system as well as algorithm that utilizations
Paillier cryptosystem to calculate worldwide backings are
utilized.
In paper [2] , D. H. Tran, W. K. Ng and W. Zha, have
designed CRYPPAR. CRYPPAR is a full-fledged system for
privacy preserving association rule mining depending on a
cryptographic. Authors utilize secure scalar product
protocols as well as public key cryptosystems in CRYPPAR
for effectively mining of association rules on vertically
partitioned information. They also acquaint a partial
topology with lower correspondence cost however much
as could reasonably be expected. Also conducted several
test runs. Test outcomes demonstrate that the system is
proficient in privacy preserving association rules as well
as may turn into a general structure for PPDM
frameworks.
D. Trinca and S. Rajasekaran, have concentrated on the
issue present in of privately mining association rules in
vertically distributed Boolean databases [3]. At start, they
designed an efficient multiparty protocol for computing
item sets which provide privacy of the particular parties.
The designed protocol is algebraic as well as recursive in
nature, as well as depends on an as of late proposed two-
party protocol for a similar issue. It is not just appeared to
be considerably speedier than comparable protocols,
additionally more secure. Next, they exhibited a variation
of the extended protocol that is impervious to collusion
among parties. As future work, it is fascinating to plan as
well as test parallel variations of the developed multi-
party convention.
Yiqun Huang, Zhengding Lu and Heping Hu, gave the
secure scalar product of two parties from the perspective
of matrix computation[4]. They also provided a way of
security maximizing for both two parties. The securities in
the two groups would be adjusted. Sensitive factors that
impact the security of the two groups are additionally
broke down.
In paper [5], L. Li, R. Lu, K. K. R. Choo, A. Datta and J. Shao
have developed an efficient homomorphic encryption
technique also a secure comparison system. After that they
also developed a cloud-aided frequent itemset mining
resolution that is utilized to develop an association rule
mining arrangement. Developed method is intended for
outsourced databases that enable various information
owners to effectively share their information safely
without bargaining on information security. Developed
method release less data about the raw information than
most existing arrangements. In contrast with the main
known arrangement accomplishing a comparative
protection level as our proposed arrangements, the
execution of our proposed arrangements is 3 to 5 orders
of magnitude higher.
F. Liu, W. K. Ng and W. Zhang have developed a protocol
for outsourced rule mining known as PORM[6]. PORM
does association rule mining in supervision of the
outsourced model in which information is in encrypted
format as well as outsourced. They also confirmed that
PORM can return the frequent rules as well as check if a
rule holds properly. They also confirmed that PORM
satisfies two security properties: user server security as
well as user-user security.
F. Giannotti, L. V. S. Lakshmanan, A. Monreale, D. Pedreschi
and H. Wang, have analyses the issue of outsourcing the
association rule mining process in a corporate privacy-
preserving system[7]. They have developed an attack
model depending on back-ground information as well as
devise a method for protection preserving outsourced
mining. Designed method guarantees that each changed
thing is unclear concerning the aggressor’s background
information, from in any least k1 other changed items.
In paper [8], A. K. Sahu, R. Kumar and N. Rahim, have
utilized idea of distributed database that splits the
centralize database in distributed database environment,
database may be partitioned in various aspects like
horizontally partitioned, vertically partitioned as well as
mixed mode. The papers given privacy preserving
information mining algorithms working over vertically
partitioned database utilizing the ideas of distribution
privacy preservation as well as furthermore lessen the
time and space complexity nature with zero rate of
information leakage.
J. Ren and B. Zhang, [9] developed an efficient non-
deterministic one-to-n substitution encryption
transformation. Contrasted and the first algorithm,
developed algorithm accomplished the non-determinacy
by choosing appropriate E specifically and avoided doing
redundant repetitive operations that have over half
likelihood in the original one. They additionally promise it
sufficiently secure not to be secured by a balanced
mapping. Additionally, our E-era is unessential with the
source things set I which makes our algorithm more
adaptable to scramble diverse itemsets.
In paper [10], M. S. Joyce and V. Nirmalrani, developed a
novel method which minimizes the leakage of the data as
well as maximizes the security of the horizontally
distributed databases. The results of the developed system
gives technique that has no need of trusted third party, the
sites themselves communicate each other for a secure
mining and increase the security. This proposed
architecture covers every one of the disadvantages

happened in the past algorithms as well as has been
executed in the synthetic employment office database.
3. SYSTEM ARCHITECTURE/SYSTEM OVERVIEW
Detailed description of the proposed system is discussed
in this section.
The architectural view of proposed privacy preserving
association rule mining system is presented in figure 1.
The system consists of three entities named as, data
owner, system server and cloud server. Initially, multiple
data owners send their private data to server. For security
purpose, database is encrypted and then store on server.
After receiving databases, cloud server combined all
database. This combined version of database contains
either original or fake data. For performance point of view,
combined database is partitioned horizontally as well as
vertical manner. After this, association rule mining
procedure is applied on partitions of database. For
association rule mining two algorithms are used and
compare their performance. These algorithms are named
as E-clat and D-Eclat algorithm. These rules further
outsourcing to cloud server. Cloud server provides
encrypted rules only when user requesting for the same.
At the user side, rules are decrypted locally.
Fig -1: System Architecture
Module Description:
1. Dynamic databases creation by data owner To create
database, initially owners collect product list of
number of users. This product list is collected and
converted into arff file. The arff file is then encrypted.
For encryption three approaches are used named as,
hash function, probabilistic homomorphic encryption,
and substitution cipher. For rule mining purpose, arff
file is sending to cloud server.
2. Association rule mining at server: Browse and
combined the data: All data is collected from multiple
data owners and combined into single database. This
database contains either original or fake data.
Horizontal and vertical partitioning of data:
The combined database is partitioned as horizontal or
vertical manner. For vertical partitioned data, rule mining
is depend on support count of itemsets. And for horizontal
partitioning, transactions are distributed among item sets.
Association Rule Mining:
Association rule mining is applied on partitioned database.
For this purpose two algorithms are used names as Eclat
and D-Eclat algorithm. This rule mining approach is
applied on encrypted data files so that the generated rules
are also in the encrypted format. These rules are then
sending and storing at cloud server.
3. Fetching of rules from cloud server:
The data owners requesting to cloud server for association
rules. In returns, cloud server provides encrypted
association rules to data owners in encrypted format.
After receiving these encrypted rules, data owner decrypt
those association rules locally.
A. E-Clat Method
It takes a depth-first search and adopts a vertical layout to
represent databases, in which each item is represented by
a set of transaction IDs (called a tidset) whose
transactions contain the item.
However, using tidsets has an advantage that there is no
need for counting support, the support of an itemset is the
size of the tidset representing it.
The main operation of Eclat is intersecting tidsets, thus the
size of tidsets is one of main factors affecting the running
time and memory usage of Eclat.
The bigger tidsets are, the more time and memory are
needed
B. D-Eclat Method
The diffset format (the difference of two sets) has
drastically reduced the running time and memory usage of
the Eclat algorithm and the Eclat algorithm using diffset
format is called dEclat algorithm.
It was the same as the Eclat, except that it sorted tidsets in
ascending order and diffsets in descending order
according their size.
By sorting diffsets and tidsets the memory usage and
running time of dEclat could be reduced significantly.
C. Algorithm Used :
For association rule mining, Eclat and D-eclat algorithm
used to find frequent itemset and strong rule generation.
Following is algorithm

2: // add to create a new prefix
3: // initialize a new equivalence class
with the new prefix P
10:
11: then
12:
13:
14:
4. RESULT AND DISCUSSION
4.1 Experimental Setup
The system is built using Java (JDK Version 1.6)
framework on any Windows platform. The Net Beans
(Version 8.1) IDE are used as a development tool. The
system doesnt require any specific hardware to run; any
standard machine is capable of running the application.
4.2 Database
Database contain arff file of product list. This file is
dynamically created by data owners. This file contain
transaction ID, transactions and ERV. For each customer,
unique ID is allocated. Transaction contain list of all
products purchase by customer. ERV represent the reality
of data that is whether it is original or fake. 1 represent
original transaction and 0 represent fake transaction.
4.3 Result and Discussion
In this system, dataset is either horizontally or vertically
partitioned and for association rule mining Eclat or D-
Eclat algorithm is used. In proposed system we are used
the D-Eclat algorithm for association rule generation.
Figure 2 and 3 represent the graphical view of comparison
of Memory required to generate association rules with
Different Ts, constant Tc and Constant TS, different Tc,
described in table 2. D-Eclat on vertically portioned data is
more efficient that other system.
Ts=Threshold Support and
Tc=Threshold Confidence
Table -1: Memory Comparison
Association
rule
MEMORY IN BYTES
Sr No Ts Tc Ecalt Declat
1 3
60
78514088 52342726
2 4 73350648 48900482
3 5 100390256 66926838
4
3
50 131631231 87754150
5 60 83683016 55788678
6 70 76311336 50874224
0
20000000
40000000
60000000
80000000
100000000
120000000
MemoryinBytes
Ts Vary and Tc constant- Memory
Eclat
Declat
Fig -2: Memory Comparison when Ts in vary and Tc
constant
0
20000000
40000000
60000000
80000000
100000000
120000000
140000000
Ts=3
,Tc=50
Ts=3
,Tc=60
Ts=3
,Tc=70
Memory:Bytes
Ts Constantand Tc Vary: Memory
Eclat
Declat
Fig -3: Memory Comparison when Ts is constant and Tc
Vary.
Table II describes the Time analysis of existing system and
proposed system. Figure 4 and 5 represent the graphical
view of comparison of Time required to generate
association rules with Different Ts, constant Tc and
Constant TS, different Tc , described in table 3. Time
required for the proposed system is less as compare to
existing system.

Table -2: Time Comparison
Association rule TIME IN MILISECONDS
Sr No Ts Tc Ecalt Declat
1 3
60
29659 19733
2 4 35657 21769
3 5 16067 10712
4
3
50 23860 15907
5 60 14724 9816
6 70 15331 10221
0
5000
10000
15000
20000
25000
30000
Time:milisecond
Ts Vary and Tc constant-time
Eclat
Fig -4: Time Comparison when Ts in vary and Tc constant.
0
5000
10000
15000
20000
25000
30000
Time:milisecond
Ts Vary and Tc ConstantVary:Time
Eclat
Declat
Fig -5: Time Comparison when Ts in vary and Tc constant
is vary.
3. CONCLUSIONS
This paper solves the problem of privacy-preserving based
association rule mining over outsourcing data. This paper
presents the privacy preserving outsourcing of association
rule mining on dynamic dataset. For association rule
mining system makes use of EClat and D-Eclat Algorithm.
This association rule mining is applied on horizontal and
vertical partitioning database. To provide security,
combination of 3 different encryptions is used.
Cryptographic hash function, Substitution cipher and
probabilistic homomorphic encryption function, are used
to encrypt ID, Transaction, ERV value .Experimental
results prove that the combination on D-Eclat algorithm
on vertical partitioning data produces more accurate rules
in minimum amount of time..
REFERENCES
[1] M. N. Kumbhar and R. Kharat, “Privacy preserving
mining of Association Rules on horizontally and
vertically partitioned data: A review paper,” 2012
12th International Conference on Hybrid Intelligent
Systems (HIS), Pune, 2012, pp. 231-235.
[2] D. H. Tran, W. K. Ng and W. Zha, “CRYPPAR: An
efficient framework for privacy preserving
association rule mining over vertically partitioned
data,” TENCON 2009 - 2009 IEEE Region 10
Conference, Singapore, 2009, pp. 1-6.
[3] D. Trinca and S. Rajasekaran, “Towards a Collusion-
Resistant Algebraic Multi-Party Protocol for Privacy-
Preserving Association Rule Mining in Vertically
Partitioned Data,” 2007 IEEE International
Performance, Computing, and Communications
Conference, New Orleans, LA, 2007, 402-409
[4] Yiqun Huang, Zhengding Lu and Heping Hu, “A
method of security im-provement for privacy
preserving association rule mining over vertically
partitioned data,” 9th International Database
Engineering and Application Symposium (IDEAS’05),
2005, pp. 339-343
[5] L. Li, R. Lu, K. K. R. Choo, A. Datta and J. Shao,
”Privacy-Preserving-Outsourced Association Rule
Mining on Vertically Partitioned Databases,” in IEEE
Transactions on Information Forensics and Security,
vol. 11, no. 8, pp. 1847-1861, Aug. 2016.
[6] F. Liu, W. K. Ng and W. Zhang, “Encrypted Association
Rule Mining for Outsourced Data Mining,” 2015 IEEE
29th International Conference on Advanced
Information Networking and Applications, Gwangiu,
2015, 550-557
[7] F. Giannotti, L. V. S. Lakshmanan, A. Monreale, D.
Pedreschi and H. Wang, “Privacy-Preserving Mining
of Association Rules From Out-sourced Transaction
Databases,” in IEEE Systems Journal, vol. 7, no. 3,
385-395, Sept. 2013
[8] A. K. Sahu, R. Kumar and N. Rahim, “Mining Negative
Association Rules in Distributed Environment,” 2015
International Conference on Computa-tional
Intelligence and Communication Networks (CICN),
Jabalpur, 2015, 934-937.
[9] J. Ren and B. Zhang, “An Improvement on a Non-
deterministic One-to-n Substitution Scheme in
Outsourcing Association Rule Mining,” 2009 WRI
World Congress on Computer Science and
Information Engineering, Los Angeles, CA, 2009, pp.
43-47.

[10] M. S. Joyce and V. Nirmalrani, “Privacy in horizontally
distributed databases based on association rules,”
2015 International Conference on Circuits, Power
and Computing Technologies [ICCPCT-2015],
Nagercoil, 2015, pp. 1-6.
[11] Zhao, Chunye, et al. “Efficient association rule mining
algorithm based on user behavior for cloud security
auditing.” Online Analysis and Computing Science
(ICOACS), IEEE International Conference of. IEEE,
2016.
[12] Tran, Duc H., Wee Keong Ng, and Wei Zha. “CRYPPAR:
An efficient framework for privacy preserving
association rule mining over vertically partitioned
data.” TENCON 2009-2009 IEEE Region Conference.
IEEE, 2009.
[13] Ren, Jinghan, and Baowen Zhang. “An Improvement
on a Nondeterministic One-to-n Substitution Scheme
in Outsourcing Association Rule Mining.” Computer
Science and Information Engineering, 2009 WRI
World Congress on. Vol. 4. IEEE, 2009.
[14] Liu, Jie, Xiufeng Piao, and Shaobin Huang. “A privacy-
preserving mining algorithm of association rules in
distributed databases.” First International Multi-
Symposiums on Computer and Computational
Sciences (IMSCCS’06). 2006.
[15] Mr. Nitin J.Ghatge, Prof. Poonam D. Lambhate “An
Effective Use of Meta Information for Text Mining
International Journal of Advanced Research in
Computer Engineering Technology (IJARCET)
Volume 4, Issue 6, June 2015.
[16] Creighton, Chad, and Samir Hanash. “Mining gene
expression databases for association rules.”
Bioinformatics 19.1 (2003): 79-86., Nov. 1999.
BIOGRAPHIES
Rasika Khairnar, is currently
pursuing M.E (Computer) from
Department of Computer
Engineering, Jayawantrao Sawant
College of Engineering,Pune,
India. Savitribai Phule, Pune
University, Pune, Maharashtra
,India -411007. She received her
B.E. (Computer) Degree from
MET’S BKC IOE, Nashik, Savitribai
Phule Pune University, Pune,
Maharashtra, India - 422003. Her
area of interest is programming
languages & data mining.
Prof. P.D.Lambhate, received
her Degree from WIT, solapur,
ME(Comp) from BVCOE Pune,
Pursing PhD. In computer
Engineering. She is currently
working as Professor at
Department of Computer and IT ,
Jayawantrao Sawant College of
Engineering, Hadapsar, Pune,
India 411028, affiliated to
Savitribai Phule Pune University,
Pune, Maharashtra, India -
411007. Her area of interest is
Data mining, search engine.

D-Eclat Association Rules on Vertically Partitioned Dynamic Data to Outsourced Securely

More Related Content

What's hot (20)

Similar to D-Eclat Association Rules on Vertically Partitioned Dynamic Data to Outsourced Securely (20)

More from IRJET Journal (20)

Recently uploaded (20)

D-Eclat Association Rules on Vertically Partitioned Dynamic Data to Outsourced Securely