Agent based frameworks for distributed association rule mining an analysis

International Journal in Foundations of Computer Science & Technology (IJFCST), Vol.5, No.1, January 2015
DOI:10.5121/ijfcst.2015.5102 11
AGENT BASED FRAMEWORKS FOR DISTRIBUTED
ASSOCIATION RULE MINING: AN ANALYSIS
G. S. Bhamra1
, A. K. Verma2
and R. B. Patel3
1
M. M. University, Mullana, Haryana, 133207 - India
2
Thapar University, Patiala, Punjab, 147004- India
3
Chandigarh College of Engineering & Technology, Chandigarh- 160019- India
ABSTRACT
Distributed Association Rule Mining (DARM) is the task for generating the globally strong association
rules from the global frequent itemsets in a distributed environment. The intelligent agent based model, to
address scalable mining over large scale distributed data, is a popular approach to constructing
Distributed Data Mining (DDM) systems and is characterized by a variety of agents coordinating and
communicating with each other to perform the various tasks of the data mining process. This study
performs the comparative analysis of the existing agent based frameworks for mining the association rules
from the distributed data sources.
KEYWORDS
Knowledge Discovery, Association Rules, Intelligent Agents, Multi-Agent System .
1. INTRODUCTION
Data Mining (DM) is a process to automatically extract some interesting and valid data patterns
or trends representing knowledge, implicitly stored in large databases [1], [2]. The traditional
approach for knowledge discovery in distributed environment creates a single centrally integrated
data repository called Data Warehouse (DW) and then DM techniques are used to mine the data
and extract the knowledge [3]. The central DW based approach, however, is ineffective or
infeasible because of heavy storage and computational costs involved in managing data form the
ever increasing and updated distributed resources where data is produced continuously in streams.
Network communication cost is also involved while transferring huge data over the wired or
wireless network in a limited network bandwidth scenario. It is also not desirable to centrally
collect the privacy-sensitive raw distributed data of the business organizations like banking, and
telecommunication as they want only knowledge to be exchanged globally. Data from modern
business organizations are not only geographically distributed but also horizontally or vertically
fragmented making it difficult if not possible to combine them in a central location. Performance
and scalability of a DM application can be increased by distributing the workload among sites [4].
Intelligent software agent technology is an interdisciplinary technology inherited from Distributed
Computing (DC), Distributed Artificial Intelligence (DAI), advanced knowledge base systems,
and human computer interaction. The motivating idea of this technology is the development and
efficient utilization of autonomous software objects called agents, which have access to
geographically distributed and heterogeneous information resources to simplify the complexities
of DC. They are autonomous, adaptive, reactive, pro-active, social, cooperative, collaborative and
flexible. They also support temporal continuity and mobility (weak and strong) within the
network. An intelligent agent with mobility feature is known as Mobile Agent (MA). MA

12
migrates from node to node in a heterogeneous network without losing its operability. It can
continue to function even if the user is disconnected from the network. It carries its code,
execution state and other data while on the move. On reaching at a node MA is delivered to an
Agent Execution Environment (AEE) where its executable parts are started running. Upon
completion of the desired task, it delivers the results to the home node. With MA, a single
serialized object is transmitted over the network carrying the small amount of resultant data only
thus reducing the consumption of network bandwidth, latency (response time delay) and network
traffic. They are robust, fault-tolerant and useful for low cost, light weight, portable computing
devices having the low processing powers, memory constraints, and intermittent low bandwidth
connection. Agent’s strong mobility feature is helpful in load balancing of processor and memory
intensive tasks. Number of participating hosts can be increased without any significant impact on
the complexity of the application. MAs are self-contained and highly reusable. The parent agent
can also clone several child agents to implement concurrent operations, and improve the
efficiency. They also facilitate the rapid prototyping of distributed applications as software
components can be flexibly and dynamically deployed in the form of MAs.
A Mobile Agent Platform (MAP)/Agent Execution Environment (AEE)/Agent Development
Toolkit, is a middleware, distributed, server application that provides the appropriate functionality
to MAs to authenticate, execute, communicate (with other agents, users, and other platforms),
migrate to other platform, and use system resources in a secure way. A Multi Agent System
(MAS) is distributed application comprised of multiple interacting intelligent agent components
[5].
2. DISTRIBUTED DATA MINING
The issues discussed above for centralized DW based DM results into the development of
techniques for parallel knowledge discovery (PKD) and distributed knowledge discovery (DKD).
Distributed Data Mining (DDM) is the related pattern extraction problem in DKD. DDM is
concerned with application of classical DM procedures in a DC environment trying to make the
best of the available resources including communication networks, computing units and
distributed data repositories, human factors etc. In DDM, DM takes place both locally at each
geographically distributed site and at a global level where the local knowledge is merged in order
to discover global knowledge. DDM techniques are scalable where its performance does not
degrade much with the increase of data set size and the number of distributed sites involved. A
DDM system is a very complex entity that is comprised of many components; mining algorithms,
communication subsystem, resources management, task scheduling, user interfaces, etc. It should
provide efficient access to both distributed data and computing resources; monitor the entire
mining procedure; and present results to users in appropriate formats. A successful DDM system
is also flexible enough to adapt to various situations. It should dynamically identify the optimal
mining strategy under the given resources and provide an easy way to update its components [3],
[4], [6], [7], [8].
In one of the important general DDM architecture (Figure 1) proposed in [3], processing at the
different distributed nodes generates several local models which are then aggregated to form a
global model representing the global knowledge. Authors in [4] proposed another phase wise
DDM approach. In the first phase local distributed databases are analyzed. Then, the discovered
knowledge is transmitted to a merger site, where all the distributed local models are integrated.
The global knowledge is then transmitted back to update the distributed databases. In some cases,
instead of having a merger site, the local models are broadcasted to all other sites, so that each
site can compute the global model in parallel. Data replication, data fragmentation, adaptation,
interestingness property and privacy preservation of the local data are some of the issues that
need to be addressed designing DDM applications [7].

13
Figure 1. Distributed Data Mining Framework [3]
2.1. Why Agents for DDM?
Above mentioned problems and challenges of DDM and inherent features of software agents of
being autonomous, capable of adaptive and deliberative reasoning clearly indicate the use of MA
technology for the development of advanced DDM systems. Agent Mining or Agent enriched
Data Mining or Multi Agent driven Data Mining, is an emerging interdisciplinary area that
integrates MAS, DM and knowledge discovery, machine learning and other relevant areas such as
statistics and semantic web. The interaction and integration between agent technology, DM and
machine learning come from the intrinsic challenges, needs and opportunities faced by the
constituent technologies [9]. All of the agent based DDM systems employ one or more agents per
data site. These agents are responsible for analyzing local data and communicate with other
agents during the mining stage. A globally coherent knowledge is synthesized via exchanges of
locally mined knowledge. However, in an agent-based model, an efficient control over remote
resources is inherently difficult. The motivation for the use of agent technology in DDM stems
from two reasons. Firstly, there is the underlying basis of DDM being a technology that has
characteristics that are intuitively suited for an agent-based approach. These characteristics are
modular and well defined sub-tasks, the need to encapsulate different mining algorithms to
present a common interface, the requirement for interaction and co-operation between different
system and the ability to deal with the distribution. Thus, from this perspective, the focus is on the
collaborating and information aspects of agency. Secondly, agent technology is seen as
addressing the specific concerns of increasing scalability and enhancing performance by reducing
the communication overhead associated with the transfer of large volumes of data. In the second
criterion for using agents as the building blocks of DDM systems, the focus is on the mobility
aspects of agency in addition to the collaborating and information aspects. Usually, such systems
have one agent that acts as a controlling and coordinating entity for a task [10], [11].
2.2. Existing DDM Systems
There are predominantly three architectural frameworks for the development of DDM systems,
the client-server model, the agent-based model and the hybrid approach which integrates the two

14
former techniques. The important technologies used to develop client server DDM are Common
Object Request Broker Architecture (CORBA), Distributed Component Object Model (DCOM),
Enterprise Java Beans (EJB) , Remote Method Invocation (RMI) and Java Database Connectivity
(JDBC) [10]. The most prominent DDM systems developed using client-server architectural are
Kensington Enterprise Data Mining Decision Centre [12], IntelliMiner [13] and InterAct [14].
A number of DDM solutions are provided in recent years using various techniques such as,
distributed clustering, Bayesian learning, classification (regression), and compression, distributed
association rules but only a few of them make use of intelligent agents [15]. The agent based
model can be further classified into systems that use mobile agents and those that use stationary
agents [11]. These systems are generally Java based to support the need for heterogeneity and
platform independence. The most prominent DDM systems developed using agent-based
architectural are PArallel Data Mining Agents (PADMA)[16], Java Agents for Meta-Learning
(JAM) [17], Besiezing knOwledge through Distributed Heterogeneous Induction (BODHI) [18],
Papyrus [19], InfoSleuth [20], Distributed Knowledge Networks (DKN) [21], a mediator oriented
agent based DDM system [22], Optimized Incremental Knowledge Integration(OIKI) [23],
Extendible Multi-Agent Data mining System(EMADS)[24], [25], [26], [27].
Authors in [11] compared DecisionCentre, IntelliMiner, InterAct, PADMA, JAM, BODHI,
Papyrus and InfoSleuth DDM systems and proposed a hybrid model called Distributed Agent
based Mining Environment (DAME) integrating the client-server and mobile agent model for
delivering internet-based DDM services by incorporating cost metrics such as application run
time estimation and optimization of the DDM process. Authors in [28] proposed a FIPA-
compliant multi-agent platform based on mining-driven agent (Agent Academy) that offers
facilities for design, implementation and deployment of multi agent systems. The researchers
describe the agent academy as an attempt to develop a framework through which users can create
an agent community having the ability to train and retain its own agents using DM techniques.
3. ASSOCIATION RULE MINING
Let  , 1jDB T j D   be a transactional dataset of size D where each transaction T is
assigned an identifier (TID ) and  ,i 1iI d m   , total m data items in DB . A set of items in
a particular transaction T is called itemset or pattern. An itemset,  ,i 1iP d k   , which is a
set of k data items in a particular transaction T and P I , is called k-itemset. Support of an
itemset,  
No_of_T_containing_P
%s P
D
 is the frequency of occurrence of itemset P
in DB , where No_of_T_containing_P is the support count (sup_count) of itemset P . Frequent
Itemsets (FIs) are the itemset that appear in DB frequently, i.e., if   min_th_sups P  (given
minimum threshold support), then P is a frequent k-itemset. Finding such FIs plays an essential
role in miming the interesting relationships among itemsets. Frequent Itemset Mining (FIM) is the
task of finding the set of all the subsets of FIs in a transactional database. It is CPU and
input/output intensive task, mainly because of the large size of the datasets involved [2].
Association Rules (ARs) first introduced in [29], are used to discover the associations (or co-
occurrences) among item in a database. AR is an implication of the form
 support,confidenceP Q where, ,P I Q I  and P and Qare disjoint itemsets, i.e.,
P Q   . An AR is measured in terms of its support and confidence factor where:

15
 Support  s P Q =  p P Q =
No_of_T_containing_both_P_and_Q
%
D
: the
probability of both P and Q appearing in T , we can say that s % of the transactions
support the rule P Q , 0 1.0s  or 0% 100%s  .
 Confidence  c P Q =  |p Q P =
 
 
s P Q
s P

=
 
 
sup_count
sup_count
P Q
P

%: the
conditional probability of Q given P , we can say that when itemset P occurs in a
transaction there are c % chances that itemset Qwill occur in that transaction,
0 1.0c  or 0% 100%c  .
An AR P Q is said to be strong if   min_th_sups P Q  (given minimum threshold
support) and   min_th_confc P  (given minimum threshold confidence). Association Rule
Mining (ARM) today is one of the most important aspects of DM tasks. In ARM all the strong
ARs are generated from the FIs. The ARM can be viewed as two step process [30], [31].
1. Find all the frequent k-itemsets( kL )
2. Generate Strong ARs from kL
a. For each frequent itemset, kl L , generate all non empty subsets of l .
b. For every non empty subset s of l, output the rule “  s l s  ”, if
 
 
sup_count
min_th_conf
sup_count
l
s

3.1. Distributed Association Rule Mining
Distributed Association Rule Mining (DARM) generates the globally strong association rules
from the global FIs in a distributed environment. Because of an intrinsic data skew property of
the distributed database, it is desirable to mine the global rules for the global business decisions
and the local rules for the local business decisions.
3.1.1. Preliminaries and Definitions
Few preliminaries notations and definitions required for defining DARM and to make this study
self contained are as follows:
  ,i 1iS S n   , n distributed sites.
 CENTRALS , Central Site.
  , 1i j iDB T j D   , Horizontally partitioned data set of size iD at the local site iS ,
where each transaction jT is assigned an identifier (TID).
 1
n
ii
DB DB
  , the aggregated dataset of size 1
n
ii
D D
  , i jDB DB  
  ,i 1iI d m   , total m data items in each iDB .
 ( )
FI
k iL , Local frequent k-itemsets at site iS .
 ( )
FISC
k iL , List of support count ( )
FI
k iItemset L  .

16
 LSAR
iL , List of locally strong association rules at site iS .
 1
nTLSAR LSAR
ii
L L
  , List of total locally strong association rules.
 ( )1
nTFI FI
k k ii
L L
  , List of total frequent k-itemsets.
 ( )1
nGFI FI
k k ii
L L
  , List of global frequent k-itemsets.
 GSAR
CENTRALL , List of Globally strong association rule.
Local Knowledge Base (LKB), at site iS , comprises of ( )
FI
k iL , ( )
FISC
k iL and LSAR
iL which can provide
reference to the local supervisor for local decisions. Global Knowledge Base (GKB), at CENTRALS ,
comprises of TLSAR
L , TFI
kL , GFI
kL and GSAR
CENTRALL for the global decision making. If the raw data from
each of the individual databases were sent to a single database to generate the rules, certain useful
rules, which would aid in making decisions about local branches, would be lost. In such cases
organization may miss out certain rules that were prominent in certain branches and were not
found in other branches. The frequent patterns in distributed databases are divided into three
classes [32]: (a) Local patterns- Local branches need to consider the original data in their data sets
so they can identify local patterns for local decisions; (b) High-vote patterns- Patterns that are
supported by most of the branches and are used for making global decisions; (c) Exceptional
patterns- Such patterns are strongly supported by only a few branches and used to create policies
for specific branches. Like ARM, DARM task can also be viewed as two-step process [31]:
1. Find the global frequent k-itemset ( GFI
kL ) from the distributed Local frequent k-itemsets
( ( )
FI
k iL ) from the partitioned datasets.
2. Generate globally strong association rules ( GSAR
CENTRALL ) from GFI
kL .
4. COMPARATIVE ANALYSIS OF EXISTING AGENT BASED DARM SYSTEMS
The existing agent based systems specifically dealing with DARM task are: Knowledge
Discovery Management System (KDMS) [33], Efficient Distributed Data Mining using
Intelligent Agents [34], Mobile Agent based Distributed Data Mining [35], An Agent based
Framework for Association Rule Mining of Distributed Data (AFARMDD) [36], [37], Multi-
Agent Distributed Association Rule Miner (MADARM) [38]. All these systems are academic
research projects. Discussion of these and few others are given below.
A mobile-agent based distributed knowledge discovery system (MADKDS) is proposed in [33].
Various agents involved in the system are: Data Mining Mobile Agent (DMMA) encapsulated
with a novel incremental algorithm, IAA[39] for mining the local frequent itemsets to generate
the local knowledge base and return back this knowledge to Mining Process Manager, Data Pre-
processing Mobile Agent (PMA) to preprocess the local data and collect back at central data
warehouse which results into an increase in the storage cost at central site and Counter Mobile
Agent (CMA) to scan the local databases and collect the support count of some itemsets. A
central site is known as Knowledge Discovery Management System (KDMS) and distributed site
are called Knowledge Discovery sub Systems (sub-KDS) in this architecture. All the mobile
agents are dispatched by Mobile Agent Control Centre (MACC) at KDMS site and received and
handled by Mobile Agent Execution Environment (MAEE). MACC and MAEE components on
are designed on the top of IBM Aglet Workbench [44], [45]. Parallel itinerary is maintained for
mobile agents and this framework is implemented using Java and C++ as dynamic link library

17
through Java native Interface (JNI). No privacy preserving techniques are used for the local
knowledge. No user interface for the MAS is designed. No cost model for the overall DARM task
is discussed and experimental validation using a large size synthetic or real data set is also
required.
An Agent-Based Framework for Association Rules Mining of Distributed Data (AFARMDD) is
proposed in [36], [37]. The main aim of this study is to protect the privacy of the local data from
being exposed to other distributed sites encapsulating the existing techniques proposed in [42],
[43] into agents. Various agents involved in the system are: Encrypt Secure Union Agent (ESUA)
to encapsulate data mining operation and the encryption of Secure Union operation at each site,
Decrypt secure union Agent (DSUA) to encapsulate the decryption of secure union operation at
each site, Encrypt Sum Agent (ESA) to encapsulate the encryption of secure sum operation at
each site, Decrypt Sum Agent (DSA) to encapsulate the decryption of secure sum operation at
each site, Broadcast Agent (BA) to carry the global frequent k-itemsets to each site and Over
Agent(OA) to notify all the sites at the mining operation has terminated. Parallel as well as serial
itinerary is maintained for mobile agents. Agent Server and Local Host components are designed
as underlying AEE. Apriori [40] algorithm is used for mining the local frequent itemsets. Privacy
preserving techniques are discussed for the local sensitive data and these techniques are the core
area of this study. No user interface for the MAS is designed. No cost model for the overall
DARM task is discussed and experimental validation using a large size synthetic or real data set
is also required. Globally strong association rules are also not generated.
Authors in [38] proposed theoretical cost models for agent based ARM in distributed data using a
prototype model called Multi-Agent Distributed Association Rule Miner (MADARM). These
cost models serve as a basic model to estimate and predict the response time of a DARM task.
Various agents involved in the system are: Association Rule Mining Coordinating Agent
(ARMCA) for creating and coordinating other agents in agent zone, Mobile Agent-Based
Association Rule Miner (MAARM) for performing ARM task at each data source, Mobile Agent-
Based Result Reporter (MARR) created by MAARM agent for migrating the result to ARMCA
and Results Integration Coordinating Agent (RICA) for knowledge or result integration.
Knowledge integration is optimized on the data sources by using agent based distributed
knowledge integration (ADKI) as opposed to incremental knowledge integration proposed in
[23]. Theoretical cost models are the core area of this study. Apriori [40] and FP-growth [46]
algorithm are considered for mining the local frequent itemsets. Parallel itinerary for MAARM,
MARR and serial itinerary for RICA agent is maintained. No underlying AEE is discussed. Only
conceptual views are presented in the paper while the researchers concluded that the work still
needs improvement and experimental validation.
In an experimental setup, authors in [34] performed efficient DDM intelligent agents
incorporating standard Apriori [40] algorithm implemented in J#. Though authors claim that it is
an agent based setup but it has been observed that there is no AEE and related agents exist in the
study. So lot of work needs to be done designing an agent based framework where intelligent
agents are actually implemented comprising a MAS on the top of an AEE using a large synthetic
or real datasets.
Authors in [35] proposed agent based DDM approach which as an improvement over PMFI
algorithm proposed in [41]. The basic objective is to reduce the time required to compute Global
Frequent Itemset (GFI). The proposed algorithm performs two tasks parallel: (1) Local sites send
LFIs to central site and also to all their neighbours; (2) Calculation of GFI/Candidate GFI at
central site and counts of CGFI at local sites is done as an overlapped operation. That is, local
sites need not wait for central site to send CGFI. Thus total time taken is reduced drastically. No
information is given about which algorithm used by Mining Agent to generate Local Frequent

18
Itemset (LFI). No AEE exist in the study. Implementation, validation and the underlying AEE
required to actually perform the agent enabled DDM using a large synthetic or real datasets.
Qualitative comparison of some prominent current agent based DARM frameworks is provided in
Table 1 taking into account some of the features they provide. The features include the following
fields:
 Agents field shows the community of agents involved in the system.
 Itinerary indicates the serial or parallel travel plan followed by the mobile agents.
 AEE/MAP shows which underlying Agent Execution Environment or Mobile Agent
Platform is used for developing MAS for DARM.
 Impl filed indicates whether the MAS is implemented along with the language used for
implementation or it is just a Prototype framework.
 Algorithm indicates the FIM/ARM algorithm considered in the study.
 CM indicates whether any cost model is discussed in the study.
 PP points out whether any privacy-preserving mechanism is taken into account for the
sensitive local data.
 GUI is for Graphical User Interface feature of the MAS.
 DS indicates whether any dataset (synthetic or real) is used in experimental validation.
 Use indicates the use of MAS in practical applications, development projects, case
studies etc.
Table 1. Qualitative comparison of the three agent based DARM Frameworks.
Features
DARM Framework
MADKDS AFARMDD MADARM
Agents DMMA, PMA,
CMA
ESUA, DSUA,
ESA, DSA, BA,
OA
ARMCA, MAARM, MARR,
RICA
Itinerary Parallel Serial and Parallel Serial and Parallel
AEE/MAP Yes Yes No
Impl Yes Yes No
Algorithm IAA [39] Apriori[40] Apriori[40], FP-growth[46]
CM No No Yes
PP No Yes No
GUI No No No
DS No No No
Use No No No
This analysis reveals that AFARMDD [36], [37] and MADARM [38] is based on the parallel and
serial itinerary of the MAs whereas MADKDS[33] use parallel itinerary. Only MADKDS[33] has
an existing IBMs Aglet Workbench as the underlying AEE, others dont have any underlying AEE
to test and validate the DARM system. MADARM [38] is only a prototype model without any
implementation. Apriori [40] algorithm is mostly used for FIM in such systems. Only MADARM
[38] discuss the cost model involved in the entire DARM task, others don’t have it. Privacy
preserving mechanism for the sensitive local data is the core area of AFARMDD [36], [37]
system while others don’t have any such mechanism. None of these frameworks has any
Graphical user interface designed to work with the systems. None of these frameworks is used in
any real applications, development projects, case studies etc.
Researchers in this area should focus more on developing algorithms and architecture that will
reduce the massive data movement in global knowledge mining and integration thereby reducing
the response time. Further algorithms and methods should also consider the development of

19
adaptive, fault tolerant and easily extendable systems in the area of DARM. Such systems will
greatly reduce communication and interpretation costs, improve autonomy, efficiency and
scalability, collaboration, security and trustworthiness of the DARM system, all of which are
common issues with existing systems [8]. Agent based DARM framework must be designed on
the top of an effective AEE with a GUI and implementation of all the agents involved in the
system. It should effectively address and validate the cost model for the overall DARM task.
Such systems must be equipped with a case study usage.
5. CONCLUSION
Mobile agents strongly qualify for designing distributed applications. DDM, when clubbed with
the agent technology, makes a promising alliance that gives favourable results. In this study,
comparative analysis of existing agent based frameworks for DARM task is done. Most of the
existing agent based frameworks for DARM task are only prototype model and lacks the
appropriate underlying Agent Execution Environment(AEE), scalability, privacy preserving
techniques, global knowledge generation and implementation using a real datasets. With this
study, we expect to contribute to cover the need of an updated review and analysis of the role of
intelligent agents in designing DARM framework and also to encourage future work in this
domain.
REFERENCES
[1] U. M. Fayyad, G. Piatetsky-Shapiro, P. Smyth & R. Uthurusamy, (1996) Advances in Knowledge
Discovery and Data Mining, AAAI/MIT Press.
[2] J. Han & M. Kamber, (2006) Data Mining: Concepts and Techniques, 2nd ed. Morgan Kaufmann.
[3] B. -H. Park & H. Kargupta, (2002) “Distributed Data Mining: Algorithms, Systems, and
Applications,” Department of Computer Science and Electrical Engineering, University of Maryland
Baltimore County, 1000 Hilltop Circle Baltimore, MD 21250, Available:
http://www.csee.umbc.edu/_hillol/PUBS/review.pdf
[4] G. Tsoumakas & I. Vlahavas, (2009) “Distributed Data Mining”, Department of Informatics, Aristotle
University of Thessaloniki, Thessaloniki, Greece, Available:
http://talos.csd.auth.gr/tsoumakas/publications/E1.pdf
[5] G. S. Bhamra, R. B. Patel & A. K. Verma, (2014) “Intelligent Software Agent Technology: An
Overview”, International Journal of Computer Applications (IJCA), vol. 89, no. 2, pp. 19–31.
[6] H. Kargupta & P. Chan, (2000) Advances in Distributed and Parallel Knowledge Discovery.
AAAI/MIT Press.
[7] Y. Fu, (2001) “Distributed Data Mining: An Overview”, Department of Computer Science,
University of Missouri-Rolla, Available: http://academic.csuohio.edu/fuy/Pub/tcdp01.pdf
[8] A. O. Ogunde, O. Folorunso, A. S. Sodiya & G. O. Ogunleye, (2011) “A Review of Some Issues and
Challenges in Current Agent-Based Distributed Association Rule Mining”, Asian Journal of
Information Technology, vol. 10, no. 2, pp. 84–95.
[9] L. Cao, G. Weiss & P. S. Yu, (2012) “A brief introduction to agent mining”, Autonomous Agent and
Multi-Agent System, vol. 25, pp. 419–424.
[10] R. Orfali, D. Harkey & J. Edwards, (1995) The essential distributed objects survival guide, John
Wiley & Sons, USA.
[11] S. Krishnaswami, (2002) “A Hybrid Model for Delivering Internet-based Distributed Data Mining
Services”, Ph.D. dissertation, School of Computer Science and Software Engineering, Monash
University, Australia.
[12] J. Chattratichat, J. Darlington, Y. Guo, S. Hedvall, M. Kohler & J. Syed, (1999) “An architecture for
distributed enterprise data mining”, in High-Performance Computing and Networking, ser. Lecture
Notes in Computer Science, P. Sloot, M. Bubak, A. Hoekstra, & B. Hertzberger, Eds. Springer Berlin
- Heidelberg, vol. 1593, pp. 573–582.

20
[13] S. Parthasarathy & R. Subramonian, (1999) “Facilitating Data Mining on a Network of
Workstations”, in Advances in Distributed Data Mining, H. Kargupta and P. Chan, Eds. AAAI Press,
pp. 229–254.
[14] S. Parthasarathy & S. Dwarkadas, (2002) “Shared State for Distributed Interactive Data Mining
Applications”, Journal of Distributed and Parallel Databases, vol. 11, no. 2, pp. 129–155.
[15] M. Klusch, S. Lodi & M. Gianluca, (2003) “The role of agents in distributed data mining: issues and
benefits”, in Proceedings of the IEEE/WIC International Conference on Intelligent Agent
Technology(IAT 2003). IEEE, pp. 211–217.
[16] H. Kargupta, I. Hamzaoglu & B. Stafford, (1997) “Scalable, Distributed Data Mining Using An
Agent Based Architecture”, in Proceedings of the 3rd International Conference on the Knowledge
Discovery and Data Mining(KDD-97), D. Heckerman, H. Mannila, D. Pregibon & R. Uthurusamy,
Eds. AAAI Press, Menlo Park, California, pp. 211–214.
[17] S. Stolfo, A. L. Prodromidis, S. Tselepis, W. Lee, D. W. Fan & P. K. Chan, (1997) “JAM: Java
Agents for Meta-Learning over Distributed Databases”, in Proceedings of the 3rd International
Conference on the Knowledge Discovery and Data Mining(KDD-97), D. Heckerman, H. Mannila, D.
Pregibon, and R. Uthurusamy, Eds. AAAI Press, Menlo Park, California, pp. 74–81.
[18] H. Kargupta, B. Park, D. Hershberger & E. Johnson, (1999) “Collective Data Mining: A new
perspective toward Distributed Data Mining”, in Advances in Distributed and Parallel Knowledge
Discovery, H. Kargupta and P. Chan, Eds. AAAI/MIT Press, pp. 131–178.
[19] S. Bailey, R. Grossman, H. Sivakumar & A. Turinsky, (1999) “Papyrus: a system for data mining
over local and wide area clusters and super-clusters”, in Proceedings of the ACM/IEEE conference on
Supercomputing (SC’99). ACM New York, NY, USA, p. 63.
[20] G. L. Martin, A. Unruh & S. D. Urban, (1999) “InfoSleuth: An agent infrastructure for knowledge
discovery and event detection”, Microelectronics and Computer Technology Corporation (MCC),
Tech. Rep. MCC-INSL-003-99.
[21] V. Honavar, L. Miller & J. Wong, (1998) “Distributed Knowledge Networks”, in Proceedings of the
IEEE Information Technology Conference.
[22] S. W. Baik, J. Bala & J. S. Cho, (2005) “Agent Based Distributed Data Mining”, in Parallel and
Distributed Computing: Applications and Technologies, ser. Lecture Notes in Computer Science, K.-
M. Liew, H. Shen, S. See, W. Cai, P. Fan & S. Horiguchi, Eds. Springer Berlin – Heidelberg, vol.
3320, pp. 42–45.
[23] E. I. Ariwa, M. B. Senousy & M. M. Medhat, (2003) “Informatization and E-Business Model
Application for Distributed Data Mining Using Mobile Agents”, in Proceedings of the IADIS
International Conference WWW/Internet (ICWI 2003), pp. 85–92.
[24] K. A. Albashiri, F. Coenen, R. Sanderson & P. Leng, (2007) “Frequent Set Meta Mining: Towards
Multi-Agent Data Mining”, in Proceedings of the 27th SGAI International Conference on Artificial
Intelligence (AI 2007), pp. 139–151.
[25] K. A. Albashiri & F. Coenen, (2009) “Agent-Enriched Data Mining Using an Extendable
Framework”, in Agents and Data Mining Interaction, ser. Lecture Notes in Computer Science, L. Cao,
V. Gorodetsky, J. Liu, G. Weiss & P. S. Yu, Eds. Springer Berlin - Heidelberg, , vol. 5680, pp. 53–68.
[26] K. A. Albashiri, F. Coenen & P. Leng, (2009) “EMADS: An extendible multi-agent data miner”,
Knowledge-Based Systems, vol. 22, no. 7, pp. 523–528
[27] K. A. Albashiri, (2010) “An investigation into the issues of Multi-Agent Data Mining”, Ph.D.
dissertation, The University of Liverpool, Ashton Building, Ashton Street, Liverpool L69 3BX,
United Kingdom.
[28] A. L. Symeonidis & P. A. Mitkas, (2005) Agent Intelligence Through Data Mining , First ed., ser.
Multiagent Systems, Artificial Societies, and Simulated Organizations. Springer, vol. 14.
[29] R. Agrawal, T. Imielinski & A. Swami, (1993) “Mining association rules between sets of items in
large databases”, in Proceedings of the ACM-SIGMOD International Conference of Management of
Data, pp. 207–216.
[30] R. Agrawal & J. C. Shafer, (1996) “Parallel mining of association rules”, IEEE Transaction on
Knowledge and Data Engineering, vol. 8, no. 6, pp. 962–969.
[31] M. J. Zaki, (1999) “Parallel and distributed association mining: a survey”, IEEE Concurrency, vol. 7,
no. 4, pp. 14–25.
[32] X. Wu & S. Zhang, (2003) “Synthesizing high-frequency rules from different data sources”, IEEE
Transactions on Knowledge and Data Engineering, vol. 15, no. 2, pp. 353–367.

21
[33] Y.-L. Wang, Z.-Z. Li & H.-P. Zhu, (2003) “Mobile agent based distributed and incremental
techniques for association rules”, in Proceedings of the International Conference on Machine
Learning and Cybernetics(ICMLC 2003), vol. 1, pp. 266–271.
[34] C. Aflori & F. Leon, (2004) “Efficient Distributed Data Mining using Intelligent Agents”, in
Proceedings of the 8th International Symposium on Automatic Control and Computer Science, pp. 1–
6.
[35] U. P. Kulkarni, P. D. Desai, T. Ahmed, J. V. Vadavi & A. R. Yardi, (2007) “Mobile Agent Based
Distributed Data Mining”, in Proceedings of the International Conference on Computational
Intelligence and Multimedia Applications (ICCIMA 2007), IEEE Computer Society, pp. 18–24.
[36] G. Hu & S. Ding, (2009a) “An Agent-Based Framework for Association Rules Mining of Distributed
Data”, in Software Engineering Research, Management and Applications 2009, ser. Studies in
Computational Intelligence, R. Lee and N. Ishii, Eds. Springer Berlin - Heidelberg, vol. 253, pp. 13–
26.
[37] G. Hu & S. Ding, (2009b) “Mining of Association Rules from Distributed Data using Mobile
Agents,” in Proceedings of the International Conference on e-Business(ICE-B 2009), pp. 21–26.
[38]A. O. Ogunde, O. Folorunso, A. S. Sodiya, J. A. Oguntuase & G. O. Ogunleye, (2011) “Improved cost
models for agent based association rule mining in distributed databases”, Anale SEria Informatica,
vol. 9, no. 1, pp. 231–250, Available: http://anale-informatica.tibiscus.ro/download/lucrari/9-1-20-
Ogunde.pdf
[39] Y. Wang, Z. Li, J. Xue & Y. Zhao, (2002) “A Novel Incremental Algorithm for Mining Frequent
Itemsets”, in Proceedings of the 2000 International Symposium on Distributed Computing and
Application to Business Engineering and Science (DCABES 2002), pp. 60–64.
[40] R. Agrawal & R. Srikant, (1994) “Fast Algorithms for Mining Association Rules in Large
Databases”, in Proceedings of the 20th International Conference on Very Large Data Bases
(VLDB’94). Morgan Kaufmann Publishers Inc., pp. 487–499.
[41] Y.-L. Ruan, G. Liu & Q.-H. Li, (2005) “Parallel algorithm for Mining Frequent Item sets”, in
Proceedings of the 4th International Conference on Machine Learning and Cybernetics, vol. 4. IEEE,
pp. 2118–2121.
[42] D. W. Cheung, V. T. Ng, A. W. Fu & Y. Fu, (1996) “Efficient Mining of Association Rules in
Distributed Databases”, IEEE Transactions on Knowledge and Data Engineering, vol. 8, no. 6, pp.
911–922.
[43] M. Kantarcioglu & C. Clifton, (2004) “Privacy-Preserving Distributed Mining of Association Rules
on Horizontally Partitioned Data”, IEEE Transactions on Knowledge and Data Engineering, vol. 16,
no. 9, pp. 1026–1037.
AUTHORS
Gurpreet Singh Bhamra is currently working as Assistant Professor at Department
of Computer Science and Engineering, M. M. University, Mullana, Haryana. He
received his B.Sc. (Computer Sc.) and MCA from Kurukshetra University,
Kurukshetra in 1995 and 1998, respectively. He is pursuing Ph.D. from Department
of Computer Science and Engineering, Thapar University, Patiala, Punjab. He is in
teaching since 1998. He has published 10 research papers in International/National
Journals and International Conferences. He has received Best Paper Award for “An
Agent enriched Distributed Data Mining on Heterogeneous Networks”, in
“Challenges & Opportunities in Information Technology” (COIT-2008). He is a Life
Member of Computer Society of India. His research interests are in Distributed Computing, Distributed
Data Mining, Mobile Agents and Bio-informatics.
Dr. Anil Kumar Verma is currently working as Associate Professor at Department
of Computer Science & Engineering, Thapar University, Patiala. He received his
B.S., M.S. and Ph.D. in 1991, 2001 and 2008 respectively, majoring in Computer
science and engineering. He has worked as Lecturer at M.M.M. Engineering College,
Gorakhpur from 1991 to 1996. He joined Thapar Institute of Engineering &
Technology in 1996 as a Systems Analyst in the Computer Centre and is presently
associated with the same Institute. He has been a visiting faculty to many institutions.
He has published over 100 papers in referred journals and conferences (India and

22
Abroad). He is a MISCI (Turkey), LMCSI (Mumbai), GMAIMA (New Delhi). He is a certified software
quality auditor by MoCIT, Govt. of India. His research interests include wireless networks, routing
algorithms and securing ad hoc networks and data mining.
Dr. Ram Bahadur Patel is currently working as Professor and Head at Department
of Computer Science & Engineering, Chandigarh College of Engineering &
Technology, Chandigarh. He received PhD from IIT Roorkee in Computer Science &
Engineering, PDF from Highest Institute of Education, Science & Technology
(HIEST), Athens, Greece, MS (Software Systems) from BITS Pilani and B. E. in
Computer Engineering from M. M. M. Engineering College, Gorakhpur, UP. Dr.
Patel is in teaching and research since 1991. He has supervised 36 M. Tech, 7 M.
Phil. and 8 PhD Thesis. He is currently supervising 6 PhD students. He has published
130 research papers in International/National Journals and Refereed International
Conferences. He has written 7 text books for engineering courses. He is member of
ISTE (New Delhi), IEEE (USA). He is a member of various International Technical Committees and
participating frequently in International Technical Committees in India and abroad. His current research
interests are in Mobile & Distributed Computing, Mobile Agent Security and Fault Tolerance and Sensor
Network.

Agent based frameworks for distributed association rule mining an analysis

More Related Content

What's hot (16)

Viewers also liked (9)

Similar to Agent based frameworks for distributed association rule mining an analysis (20)

More from ijfcstjournal (20)

Recently uploaded (20)

Agent based frameworks for distributed association rule mining an analysis