Documente Academic
Documente Profesional
Documente Cultură
*1
Mr. Sivabalan M, *2 Dr. Arulanandam K.,
*1
Guest Lecturer, PG and Research Department of Computer Science, Government Thirumagal Mills
College, Gudiyattam, Tamilnadu, India.
*2
Assistant Professor & Head of Department, PG and Research Department of Computer Science,
Government Thirumagal Mills College, Gudiyattam, Tamilnadu, India
---------------------------------------------------------------------***---------------------------------------------------------
Abstract: INTRODUCTION
The main objectives of this research is to Multi Agent Systems (MAS) often deal with
conduct a comparative study of the architectural complex Applications that require distributed
designs of Multi-Agent development toolkits, problem solving. In many applications the
studying the individual methodology used in the individual and collective behavior of the agents
designing the architectures within Multi-Agent depends on the observed data from distributed
development environments. The study aims to sources. In a typical distributed environment
research widely into the various design analyzing distributed data is a non-trivial problem
methodology used in implementing various Multi- because of many constraints such as limited
Agent design environments, analysis different bandwidth (e.g. wireless networks), privacy-
aspects relating to the components of the sensitive data, distributed compute nodes, only to
development tools.The FIPA (Foundation for mention a few. The field of Distributed Data Mining
Intelligent Physical Agents) a part of the analysis (DDM) deals with these challenges in analyzing
and evaluation of the specification providing the distributed data and offers many algorithmic
necessary requirements for agent systems. solutions to perform different data analysis and
mining operations in a fundamentally distributed
This analysis is to compare and contrast
manner that pays careful attention to the resource
each agent against and other properties. To make an
constraints. Since multi-agent systems are also
effective comparison as well as evaluate individual
distributed systems, combining DDM with MAS
agent tools using Multi-Agent development tool to
for data intensive applications is appealing.
compare their performance and a high level
understanding of a particular agent–oriented The aim is to understand the paradigm of
development methodologies put it use. The main agent oriented development tools, studying the
objective can allow in the analysis agent methodology used in Multi-Agent design models
architecture entirely. The analysis of Multi-Agent and the applications developed. This investigation
development tools aims to give insight into the will be carried out in the form of a comparative
many theoretical aspects of software engineering study that will include agent development tools to
and the drive for greater understanding of software understand each agent by carrying out a detailed
engineer methodology as well as the concept analysis of each Multi-Agent System (MAS)
behind agent design for an agent development building tools.
environment such as Multi-Agent systems.
The objective of this study is widely to fill in the
Keywords: Distributed Data Mining, Agent gap in knowledge as to agent-oriented
mining, Multi Agent System (MAS),Weka, developments tools and the methodology each has
Knime, Tanagra. chosen to use individually then carrying out a
comparison on many levels of its abstraction,
making an evaluation from a comparison
collectively reviewing the agent development depending on the environment of the system it is
toolkit thoroughly as well as effectiveness. visiting. The agents can interact with other agents
or with the user, as needed [15]. But FTP agents do
Existing Agent Simulation Tools not require any user interaction—based on push
technology, they can move from system to system,
In this section we would like to discuss the popular respond to events, and perform tasks according to
agent simulation tools like Aglets, JATLite, FTP criteria predefined by the user. An Agent Manager
Software Agent and Voyager. Aglets are Java-based is responsible for launching the agent. Voyager,
autonomous agents developed by IBM, which from Object Space, Inc., is an agent-enhanced
provide the basic capabilities required for mobility Object Request Broker (ORB) coded in Java. An
and has a globally unique name. A travel itinerary is ORB provides the capability to create objects on a
used to specify the destinations to which the agent remote system and invoke methods on those objects
must travel and what actions it must take at each [16]. Voyager augments the traditional ORB with
location. In order for an aglet to run on a particular agent capabilities. Voyager agents have mobility
system, the target system must be running an aglet and autonomy which is provided in the base class,
host application which provides a platform-neutral Agent. An Agent can move itself from one location
execution environment for the aglet. The aglet to another and can leave behind a forwarding
workbench includes a configurable Java security address with a secretary so that future message scan
manager. Aglets can communicate using a be forwarded to its new location. Specialized
whiteboard that allows agents to collaborate and agents, called Messengers, are used to deliver
share information asynchronously. Synchronous messages.
and asynchronous message passing is also Need of Agents:
supported for aglet communication. Aglets are
streamed using standard Java serialization or In Data mining perspective, an agent can be defined
externalization. A network agent class loader is as an object whose behavior is described by a
supplied which allows an aglet’s byte code stream "script", with its own calculation ways, and can
and state to travel across a network [13]. Java Agent move from place to place to communicate with
Template Lite (JATLite) is a set of light-weight other agents. With its "script", the agent is able to
Java packages being developed at Stanford follow a life behavior that will be instilled at the
University that can be used to build multiagent time of implementation and that will allow him to
systems. It is a layered architecture which provides have as main feature to be fully autonomous. One
a different communication protocol at each layer. of the discriminate characteristics of the agents is
The JATLite framework is intended for developing the representation and reasoning on the
typed-message, autonomous agents that environment (the external world and other agents),
communicate using a peer-to-peer protocol. Both based on this feature, we find two different classes,
synchronous and asynchronous message passing are which are Cognitive agents, Reactive agents.
supported. Messages can be delivered through
polling or message queuing. The framework A cognitive agent is an agent that has an explicit
provides additional security which checks the agent representation of its purpose and its environment.
name and password for a more secure connection The actions it performs to achieve its goal are the
[14]. FTP Software Agent Technology is Java-based result from a reasoning on the state of the
software designed to manage heterogeneous environment. Usually a cognitive system includes a
networks across the Internet using agent small number of agents; each is similar to a more or
technology. The agents are autonomous and mobile, less complex expert system. In this case we speak
and can move to any system in the network which of high granularity agent.
has an Agent Responder installed. As the agent
moves from system to system, its tasks may change,
Challenges in Distributed Data mining and 3.1.1Weka
Mining Agents:
Weka is the tool most commonly used due to its
Agents can enhance data mining through involving vast functionality and supported features. This java
agent intelligence in data mining systems, while an based data mining tool provides user with both GUI
agent system can benefit from data mining via and simple CLI for performing and managing tasks
extending agents’ knowledge discovery capability to be performed. It supports all data mining tasks
[12]. Nevertheless, the agent mining interaction from preprocessing, classification, and clustering to
symbiosis cannot be established if mutual issues are visualization and feature selection.[9]
not solved [13]. These mutual issues involve
fundamental challenges hidden on both sides and
particularly within the interaction and integration.
Issues in agent-mining interaction highlighting the
existence of mutual issues. Mutual issues
constraining agent-mining interaction and
integration consist of many aspects such as
architecture and infrastructure, constraint and
environment, domain intelligence, human
intelligence, knowledge engineering and
management, and nonfunctional requirements.
Figure 1: Weka GUI
Architecture and infrastructure Data mining always
3.1.2 Knime
faces a problem in how to implement a system that
can support those brilliant functions and algorithms KNIME (Konstanz Information Miner) is an open
studied in academia. API workflow based data mining tool that provides
easy accessibility to new nodes to be added into the
Nonfunctional requirements Nonfunctional requests
workflow. It provides its user with the GUI which
are essential in real-world mining and agent
aid with the simplification of workflow generation
systems. The agent-mining simians may more or
by the user. It also provides with features to modify
less address nonfunctional requirements such as
a particular node accordingly a nd execution of
efficiency, effectiveness, action ability, and user and
partial data flow[10].
business friendliness.
3.1.3 Tanagra
Constraint and environment both agent and mining
systems need to interact with the environment, and This extension of SIPINA provides the users an
tackle the constraints surrounding a system [15]. In easy to use interface for the analysis of either real
agent communities, environment could present or artificial data. It allows the researchers to easily
characters such as openness, accessibility, add their own data mining research methodology or
uncertainty, diversity, temporality, spatiality, and/or any newly identified data mining processing
evolutionary and dynamic processes. These factors technique and also supports byproviding them with
form varying constraints on agents and agent architecture and a means to compare their
systems. methodology performances. It provides the
beginners or naives with a platform where they can
Human intelligence both agent and mining need to
carry out their experimental procedures[6].
consider the roles and components of human
intelligence. Many roles may be better played by
humans in agent-mining interaction.
Data Set:
Preliminaries:
An agent may have, depending on the domain it is fx i, xj ∈ same IND(D) = {c ∈ C: f(c, xi) ≠ f(c, xj)}
situated in, some or all of the properties listed if xi, xj ∈ different IND(D)
below (Symeonidis & Mitkas, 2005): 2.2. Find the CORE from discernibility matrix:
• Autonomy (considered a must-have feature by For any c ∈ C, c ∈ CORE(C) if and only if there
many researchers in the field of agents) exists i, j, 1 ≤ j < i ≤ N such that mij = {c}. Note
that a core may be empty
• Interactivity: Reactivity or Pro-activeness
2.3. Determine the attribute set UA which user
• Adaptability prefers to emphasize. If UA is empty that means
that the user does not have preference for any
• Sociability
attribute.
• Cooperativity
2.4. Let RED = CORE ∪ UA
• Competitiveness
2.5. AR = AR – RED
• Mobility
2.6. Find attribute a in AR which has the
• Learning maximum SGF(a, RED, D)
The Rough Sets algorithm implemented in 2.7 RED = RED ∪ {ai}, AR = AR – {ai} (i =1, 2,
MASCE can be summarized as follows: …m)
This algorithm takes as input a decision table S = 2.8 If k(RED, D) = 1, then stop, otherwise go to
(U, C∪ D, V, f) and produces as output the set of step
decision rules {τij} (Attia at al., 2004).
Step 3: Generate the reduced relation by removing
Step 1: Vertical reduction: those attributes which are not in the best reduct
RED.
The vote value is calculated for all the tuples
(similar tuples are collapsed into one and their Step 4: Combine similar tuples in the reduced
number is added to the vote). relation.
Then tuples, with vote values less than the noise Step 5(a): Transform tuples in the reduced relation
filter threshold, are removed from the database into decision rules for each class in D.
table.
Step 5(b): For the same class in the reduced table,
Step 2: Horizontal reduction: Attributes reduction two tuples can be combined if the values of the
is made by calculating the best reduct RED as condition attributes differ in only one attribute, thus
follows, let all attributes be called AR and the user obtaining a more general set of decision rules. Or
preferred attributes if any be UA. Begin instead of steps 5(a) and 5(b) we can use the
following alternative method for generation of
2.1. Construct the modified discernibility matrix decision rules:
M(C): Each entry mij contains the condition
attributes whose values are not identical on both xi Step 6(a): Extract the decision rule which is called
and xj where xi, xj belong to different classes of τij as follows: τij = DesC(Xi) ⇒ DesD(Yj) such that
Xi ∩ Yj ≠ Φ for Xi ∈ C* and Yj ∈ D*.
Step 6(b): Call A rule (deterministic) if Xi ⊆ Y j i.e. give a partition of the set of values of the given
Xi ∩ Y j = Xi, otherwise a rule is nondeterministic. characteristic.
The set of all decision rules {τij} for all classes Yj
Conclusion:
∈ D * is called the decision algorithm of the
information system S. The above study was conducted by using four
algorithms over a data set: Zero Rule (ZeroR), One
/* End of the algorithm*/ The significance of an
Rule (OneR), decision tree (C4.5), and k-nearest
individual attribute {a} added to the set A with
neighbor (KNN). Tools were run over the data set
respect to the dependency between A and D
and results were observed for each algorithm.
(Decision set) is represented by significant factor
Accuracy percentage served as performance
SGF, given by:
measure. Weka was indentified a better performer
SGF (a, A, D) = k(A+{a}, D) – k(A, D) where k(A, with the specified algorithms, followed by KNIME
D) = card(POSA(D)) / card(U) . and Tanagra. This performance ranking based on
the type of data set used and how the classifier is
Decision trees mining in MASCE implemented within the tool, as task of
Decision tree induction is a well-known discipline classification is affected by so. But Weka still
in Machine Learning presented by Quinlan in 1986 proved to be better as it provided with the
(Quinlan, 1986). The basic algorithm for decision implementation of ZeroR and OneR over data types
tree induction is a greedy algorithm that constructs where other tools did not. The functionality offered
decision trees in a top-down recursive divide-and- by these tools, like API support and graphical
conquer manner. In the process of constructing a presentation along with other features aid with the
tree, the criteria of selecting test attributes selection of tool best suitable according to the usage
influences the classification accuracy of the tree. by different users. The methodologies discussed for
Presently, there are many criteria for choosing the UDMT formulation offer better approach towards
test attribute in building decision tree, such as ID3 data mining but still are left with some issues due to
(Quinlan, 1986) and C4.5 (Quinlan, 1993) which an incomplete theory for their correct formulation.
use an entropy-based measure known as The MAS suffer from the problem of selection of
information gain as a heuristic for selecting the appropriate set of algorithms for classification,
attribute. Decision trees represent a supervised clustering and visualization. The correct application
approach to classification. A decision tree is a of an algorithm as a function is also an issues
simple structure where non terminal nodes The Inductive Databases theory addresses the need
(internal) represent tests on one or more attributes for a language to design inductive queries and
and terminal (leaf) nodes reflect decision outcomes. generation of databases including both data and
The ordinary tree consists of one root, branches, patterns. All the defined patterns are then required
nodes (places where branches are divided) and to satisfy the constraints imposed upon. This results
leaves. In the same way the decision tree consists of in need of better algorithms for constraint based
nodes which stand for circles, the branches stand data mining.
for segments connecting the nodes. A decision tree
is usually drawn from left to right or beginning
from the root downwards, so it is easier to draw it.
Future Work:
The first node is a root. The end of the chain “root -
branch - node-...- node” is called “leaf”. From each Based on our result we concluded that there
internal node (i.e. not a leaf) may grow out two or is enough claim for Agent – oriented methodology
more branches. Each node corresponds with a varies regarding the different methodology applied;
certain characteristic and the branches correspond three out of the five architectures studied has
with a range of values. These ranges of values must applied the BDI theory into practices. Other method
such as O-MaSE used in Agent Tool3 and the
Cognitive method used in Courage has been applied
only within limited architectures. We found
significant performance boost that proofs our agent-
oriented methodology. Enhance comparison of
agent-oriented tools.
References:
[10]KNIMEAvailableat:http://www.knime.org/downloa
d-desktop[11]“Open-Source Tools for Data Mining”
Blaz Zupan, PhD, Janez Demsar, PhD.[12]“Tanagra: An
Evaluation” Jess