UML MODELING AND SYSTEM ARCHITECTURE FOR AGENT BASED INFORMATION RETRIEVAL

International Journal of Computer Science & Information Technology (IJCSIT) Vol 7, No 6, December 2015
DOI:10.5121/ijcsit.2015.7604 51
UML MODELING AND SYSTEM ARCHITECTURE
FOR AGENT BASED INFORMATION RETRIEVAL
D. Muhammad Noorul Mubarak, Philomina Simon
Department of Computer Science, University of Kerala
Kariavattom, Thiruvananthapuram, Kerala, India
ABSTRACT
In this current technological era, there is an enormous increase in the information available on web and
also in the online databases. This information abundance increases the complexity of finding relevant
information. To solve such challenges, there is a need for improved and intelligent systems for efficient
search and retrieval. Intelligent Agents can be used for better search and information retrieval in a
document collection. The information required by a user is scattered in a large number of databases. In this
paper, the object oriented modeling for agent based information retrieval system is presented. The paper
also discusses the framework of agent architecture for obtaining the best combination terms that serve as
an input query to the information retrieval system. The communication and cooperation among the agents
are also explained. Each agent has a task to perform in information retrieval.
KEYWORDS
Intelligent Agents, Crawling, Agent based information retrieval, Object Oriented modeling, Unified
Modeling Language, Ontology, Agent Architecture
1. INTRODUCTION
The web contains abundant data and it is getting updated on an hourly basis. Usually, the
retrieved results are not relevant to the input query. There is a drastic need for improved search
strategies in an information system. Current search engines are primarily passive tools. The
retrieval of the information has become a tedious task now days because of the enormous amount
of structured and unstructured data. Web users find it difficult to retrieve a relevant result based
on the input query. The relevant information can be obtained in the form of textual documents,
video files and audio files and so on. The use of intelligent agents [4] in information retrieval for
efficient search and retrieval techniques. Lot of research works are going on in the area of
information retrieval and intelligent agents to obtain more relevant information.
In our method, the relevant information is obtained in the form of textual documents. Documents
are represented by document descriptors. Document descriptors are defined as keywords or terms
in a document collection. The system architecture is proposed to derive the combination terms ie,
best query terms. Combination terms are obtained from the list of keywords in the document
collection. Combination terms are keywords which help to frame the query terms which doesn’t
have a related meaning. For example, “Genetic Algorithms for Information Retrieval” [8] here,
genetic algorithm and information retrieval are not co-occurrence terms and hence they do not
belong to the same class. Co-occurance terms are keywords which have a related meaning. Hence
use of co-occurrence data for document retrieval is not always considered as an efficient method.
This method gives limited performance when the keywords have unrelated keywords. To
overcome the limitations, a method for finding combination of terms is proposed. This concept of

52
combination terms work better in document retrieval since the keywords does not have any
related meaning.
The proposal describes an approach in which an intelligent agent can be incorporated with
information retrieval to retrieve the relevant information.
A. Intelligent Agents
In agent based architectures, agents act as mediators between user specific task and a particular
application [6]. They are often decomposed into functional modules that refer to perception,
modeling, planning, coordination, and task or plan execution. The strong point is that the agents
provide the ability to execute and coordinate independently and autonomously sub-tasks, which
lead to completing the major user task. In this context, computers have to assist users in planning,
coordinating and delegating a specific task. Software agents play a crucial role in the human-
computer interaction and performing the coordination of the processes of the retrieval system.
Based on the functionality, Agents have different goals to perform based on the context and the
application. Conceptual structures enhance the functional operationalism of the agent by applying
it for knowledge representation and its manipulation mechanisms, navigation, and problem
definition. Ontology plays a key role in describing and defining a particular context. Agent-based
intelligent systems can provide an enhanced learning and understanding for a specific
functionality of a system.
Intelligent agents have different features such as social ability, proactiveness and task oriented
and task delegation behavior. The agent itself collects the request from the user and the task
delegation is performed. The user who searches information from the document collections is an
agent. The user agent may teach the intelligent agent by example or by employing a set of criteria
for the intelligent agent to follow. Intelligent agent sense and act autonomously in its
environment. To some extend, intelligent agents are designed to be adaptive to their environments
and to the changes in their environments. Three major approaches for building agents for the
document collections are discussed. Firstly, integrate agents into existing search techniques.
Secondly, intelligent agent follows rule based filtering approaches. With this approach, an agent
is given information about the application. A knowledge engineer generates the rules and
knowledge for the agent. Thirdly, In the training approach, the agent collects the information
around it and actions of its user (Jensen, 2002).
This proposed work describe the architecture of an intelligent agent that searches the document
collection and generate a combination terms which serve as an input query to the information
retrieval system.
The paper is organized as follows. Section 1 explains the introduction to information retrieval.
Section 2 discusses the different strategies and methodologies for performing information
retrieval. Section 3 presents the proposed architecture for Agent based Information Retrieval.
Section 4 deals with the cooperation and communication among different agents. Section 5
describes the UML modeling of the agent based system architecture. Section 6 presents the results
and discussion. Section 6 summarizes and concludes the paper.
2. LITERATURE REVIEW
In this section, a study on different agent based information retrieval is performed, also the survey
of different strategies and methodologies for performing information retrieval is also presented in
Table 1.

53
SAIRE [2]-A Scalable Agent-Based Information Retrieval Engine.
It is a multi agent based search system which incorporates agents, language processing and
heuristic search techniques to support Earth and Space Science data over the web. Agents in
SAIRE form seven modules, in which an agent manager controls several local agents. Agent
managers communicates and sends messages through master agent which act as a central
coordinating agent. If the central agent fails, then it may have huge impact on other agents and it
may lead to system failure. A natural language parser extracts a frame containing actions and
concepts from the natural language input. Two dictionaries are used to specify a user’s domain of
interest. SAIRE performs several functions such as communication activities, directory services,
knowledge archiving and retrieval services, data storage access services and information
processing services. SAIRE is a Goal-oriented, autonomous intelligent software that is developed
to achieve intelligent information retrieval.
SARI [1] The SARI (Software Agents for Retrieval of Information) system act as a mediator
between human users or computers at one end, and heterogeneous information sources with
various search engines at the other. In SARI, different agents work corporately and coordinately
to perform a specified task. Application Agents represent the users or stake holders to the SARI
system. They send agent messages for performing information retrieval to Control Agents. Search
Agents gather queries from Control Agents and send the results back to the Control Agents.
Control Agents act as mediators in the SARI system. Control Agent receives agent messages
containing query requests from Application Agents. Search Agents forwards the requests, sends
messages containing the retrieval request and receives search results from the Search Agents,
incorporates them into information retrieval results and retrieved results are sent to Application
Agents .Ontology Agent contains metadata that describe the structure of the information present
in the information sources.
MARS [3] (Multi-Agent Retrieval System) is discussed to test the integration of multiple
retrieval methods in a single system for more user flexibility and complementary retrieval. The
retrieval method incorporated in MARS is an intelligent conceptual retrieval based on the use of a
thesaurus, semantic retrieval and conceptual browsing. The system is developed in client-server
architecture. In the system architecture, the telecommunication facilities enables distributed
agents to exchange messages , standards for message formats and server options and a name
server so that agents can locate each other and also the security considerations. Since term co
occurrences have disadvantages, the author used the semantic retrieval strategy. Concept retrieval
is used in this system compared with keyword retrieval, since it offers more flexibility to restrict
the retrieval by replacing a concept in a query. With the rapid growth of information and
multiplicity of retrieval techniques, it is neither reasonable to store all information in a single
database nor gather all evaluation methods in a single program.
XSAR [5] a dynamic, DTD independent search agent. XML repositories help to generate
structured data so that we can process easily. Bajaj et.al presented a novel search tool: XSAR
(XML Based Search Agent For Information Retrieval) to access the information repositories.
XML build the information in a structured manner much like in a relational database. Now a days
XML format is the best way of sending information content across web. Agents access
information and XSAR reformulates the input query as an XQL representation, and then launches
a spider, a crawler that traverses the database repository and executes the XQL query for each
XML page in the database.

54
Table 1. Survey on different strategies for Information Retrieval
Authors Type of
information
retrieval
Methodology Purpose Evaluatio
n Metrics
Comments/
Features
R. Dhanapal
(2008) [10]
Document
Information
Retrieval
Edge Index
Graph based
Perform web
document
clustering
and contextual
indexing
Indexing
time
Scalable ,
Dynamic and
Contextual
retrieving system
Suriati
Akmal et.al
(2014) [11]
Product
Information
Retrieval from
Web
Ontology based
Formal concept
analysis (FCA)
used to design
ontologies
Generate
ontology-based
Semantic
Similarity
measures
Feature
based
Similarity
Measure
Composite
similarity
Efficient
Shaw et.al
(2002) [12]
Integrated
Intelligent
Information
Retrieval for
Distributed
Heterogeneous
Data sources
Agent-based
architecture
Design of an
Intelligent
Information
Retrieval and
Filtering system
Intelligent
system is
developed
and tested
Modular
Flexible and
Scalable
Seyfi et.al
(2016) [13]
Document
Information
Retrieval based
on Focused
Crawler
T-Graph
Principles and
Scoring function
Finding
relevant
documents and
prioritizing
them for
effective
download
Overall
Similarity
Measure
Priority
Score
Accuracy of
prediction of an
unvisited page
Reduce
processing
overhead
D. Yoo
(2012) [14]
Personalized
Information
Retrieval
on the
Semantic Web
Hybrid Query
Processing
Method on
Semantic Web
Querying for
inferred
knowledge
base,
Processing
time
Quick response
time
Efficient
retrieval of
personal
information.
Addis et.al
(2012) [15]
Information
Retrieval with
X.MAS
Multi agent
systems
Retrieving,
Filtering and
Reorganizing
information
according to
user interests
Degree of
pairwise
similarity
under the
threshold
of 25%.
Effective in the
design and
concrete
realization of IR
applications
Ropero et.al
2012 [16]
Fuzzy Logic
Intelligent
Agent for
Information
Extraction
Fuzzy Logic-
based Term
Weighting
Scheme
Intelligent
agent to manage
any set of
knowledge
where
information is
vague or
imprecise
Web page
identified
by
questions
Term
weighting
Flexible
Better Extraction
of information
Pai et.al
(2013) [17]
Semantic and
Ontology
based
information
retrieval
Semantic-based
Content Mapping
Extracts
semantic
features and
structure of the
documents to
form a content
Precision
Recall
F-Measure
Improved
accuracy

55
map and then
performs the
search based on
this.
Khan et.al
(2012) [18]
Instructional
Video
Information
Retrieval
E-Learning
management
systems with
Natural
Language
Processing
Semantic
engine helps to
derive the
meaning of the
input
words
Question
and
Answer
Module
Good
understandability
of the user
content
Belsis et.al
2014 [19]
Agent
architecture
benchmark
Agent based developed
implements a
distributed
knowledge base
management
system
Round trip
time
Round
Trip
delays
Test bed for
benchmarking
and performance
monitoring
JADE used
Teng-yang
et.al 2012
[20]
Ontology
based
information
retrieval
Vegetable e-
commerce
ontology
Domain
ontology helps
to index
information and
information
inference.
Recall
Precision
Usage of
Semantic
Information
Annotation
model
Attia
et.al[21]
2014
Fuzzy
Information
Retrieval
Model
Linguistic based
multi-view Fuzzy
Ontology
Information
Retrieval model
Rranking
algorithm that
ranks the set of
relevant
documents
Document
matching
degree
Linguistic based
user query
system
3. PROPOSED ARCHITECTURE FOR AGENT BASED INFORMATION
RETRIEVAL
Even though searching several types of information in the internet, user is not satisfied with the
results obtained. Here a different approach is proposed to overcome the limitations of information
retrieval. So this concept of combination terms [7][8] is useful for retrieving those documents
from the document collection in which the query term does not have any related meaning. The
different agents and their features is discussed in the following sections. Crawling agent
automatically gathers the keywords and documents which is available in the document collection.
The indexing agent collects the information from the crawler agent and stores into database.
Search agent searches for the more relevant data in the database and it retrieves more relevant
results based on the combination of terms. Indexing agent indexes and stores the information
which is collected from the crawler. Query agent processes the query send by the user agent and
this agent delegate the search task to search agent. Retrieval agent checks that the pages retrieved
are relevant in nature. If the user query and the retrieved results are matches by the retrieval
agent, then the results are retrieved. Combination agent: learn from the keywords stored by the
index agent and generates a combination of the keywords which does not have any related
meaning.
The components involved in the agent based information retrieval are Crawling agent, Search
agent, Indexing agent, Query Agent and user agent. A crawler is a type of software agent.
Crawling agent is used to crawl through the document collections and extract the keywords

56
A. Ontology
Ontology represents the formal naming, definition of underlying concepts, properties and
relationships among different modules in a system design. Ontologies help to perform
communication and perform decision tasks in relation to the information stored in the document
collection. Ontologies are also used to represent the syntax, semantics and the messages
exchanged among agents. An ontology is a set of classes, entities, properties and relationships,
describing a standardized terminology (W3C, 2009). Users apply ontology in all applications
sharing domain specific knowledge. Communication is possible only if agents share a common
ontology, or if there are associations between logically separated ontologies (FIPA, 2000). In our
approach, two ontologies are discussed. First, Document Ontology which represents the syntax ,
semantics and relationships of document collections is discussed. Second one is domain-specific
ontology which deals with information retrieval and logical relationships between the modules.
Domain ontologies, which represent the content and the specific structure of data sources
connected to the system. Document ontology deals with the keywords extraction from the
document crawler and finds the best combination terms from a set of keywords extracted.
Combination terms represents the terms which does not have related meaning.
4. COOPERATION AND COORDINATION AMONG AGENTS FOR INFORMATION
RETRIEVAL
The proposed architecture for the agent based information retrieval is shown in Fig 1. The crawler
agent gathers and collects information from document collections. The intelligent agent
automatically collects the information and keywords and sends that to indexing agent. The
crawler agent is so flexible to collect any type of textual information as per the user request given
by the user agent.
The index agent collects the information from the crawler agent and stores into database. The
index agent collects different types of information such as documents, keywords, hyperlinks and
URLs. So the database consists of the above information. The user agent processes the user
request .This request is passed to the crawler agent. The concept of the task delegation, goal
oriented nature and flexibility is well inbuilt in this proposed architecture. The agents interact
intelligently and autonomously to fulfill the functionalities delegated to them.
The combination agent generates a combination of different keywords and these combination
keywords are stored in the database. The best combination terms are generated by applying the
genetic algorithm [8]. When the user agent sends a query, the query agent forwards the query to
search agent. Search agent performs the search operation and it is send to retrieval agent.
Retrieval agent analyses the search results and gives the result to the user agent.

57
Fig1. Proposed Architecture for retrieving documents using intelligent agents.
Fig 2: Agent Communication: Cooperation and Coordination
5. UML MODELING FOR AGENT BASED INFORMATION RETRIEVAL
The components involved in the sequence diagram are Crawling agent, Search agent, Indexing
agent, Query Agent and user agent. The sequence diagram is represented in Fig 3.The UML
representation for the Search agent is shown in Fig 4.
Interface
User agent
Crawler
agent
Database
Indexer
Agent
Combination
Agent
Internet
Query
Agent
Retrieval
agent
Search
Agent Database
Information Retrieval System
Crawling
agent
-------
-------
-------
-------
-------
Combination
Agent
Indexing
Agent
Retrieval
Agent
User
Agent
Search
Agent
Query
Agent
Finding
Combination
terms
Key words
extracted
Document
Database
Document
Collection
Information Retrieval Ontology
Document Ontology

Fig 3: Sequence diagram for Agent based
6. RESULTS AND DISCUSSION
The challenges that an information retrieval system face is that
documents or they do not retrieve all the relevant documents. The
retrieval system can be calculated
defined as the proportion of the retrieved documents that are actually relevant. R
as proportion of all relevant documents that are actually retrieved. If searchers want to raise
precision, then they have to narrow their queries.
broaden their query. It represents many expected docs.
program for the crawler is tested to extract different number of documents and the keywords are
stored. From the keywords, combination agent generates terms of different combinations and it is
stored separately. When the user
data that is best suited for the input query
retrieval is implemented.
7. CONCLUSION
Agent architecture and object oriented modeling
proposed. A review on different agent based information retrieval systems
proposed system, combination agent generates a good number of probable search queries and
save it in database. This will serve as an input query for the information retrieval system. A
system is developed which generated the best combination terms
information retrieval is performed.
Fig 3: Sequence diagram for Agent based Information Retrieval System
ISCUSSION
information retrieval system face is that they retrieve some irrelevant
do not retrieve all the relevant documents. The performance of the
etrieval system can be calculated by two evaluation measures: Precision and Recall.
defined as the proportion of the retrieved documents that are actually relevant. Recall
precision, then they have to narrow their queries. If searchers want to raise recall, then they
It represents many expected docs. The input given is a set of documents.
crawler is tested to extract different number of documents and the keywords are
keywords, combination agent generates terms of different combinations and it is
stored separately. When the user input a search query, the search agent retrieve the relevant set of
that is best suited for the input query. The system architecture for agent based information
and object oriented modeling for document based information retrieval is
proposed. A review on different agent based information retrieval systems is discussed.
combination agent generates a good number of probable search queries and
This will serve as an input query for the information retrieval system. A
system is developed which generated the best combination terms for query processing and
information retrieval is performed.
58
they retrieve some irrelevant
performance of the information
evaluation measures: Precision and Recall. Precision is
ecall is defined
If searchers want to raise recall, then they
The input given is a set of documents. The
crawler is tested to extract different number of documents and the keywords are
keywords, combination agent generates terms of different combinations and it is
, the search agent retrieve the relevant set of
agent based information
information retrieval is
discussed. In the
combination agent generates a good number of probable search queries and
This will serve as an input query for the information retrieval system. A
for query processing and

Fig
REFERENCES
[1] Kuldar Taveter ,”SARI- System for Semantic
and Metadata, 1998
[2] Jide.B.Odubiyi, David J.Kocur et.al,
Proceedings of First International Conference on Autonomous agents,ACM,
[3] J.Nie,Anquetil,J.Vaucher,”A Multi
Proceedings ,1993
[4] Masoud Mohammadian ,”Intelligent Agents for data
[5] Bajaj, Akhilesh; Tanabe, Hideaki; and Wang, Chao
Information Retrieval" (2002).
[6] Lora Aroyo , Piet Kommers,
Journal of Interactive learning Research
[7] Philomina Simon, A Two Stage approach to Document Retrieval using Genetic Algorithm,
International Journal of Recent Trends in Engineering, Vol 1, No 1, 526
[8] S.Siva Sathya , Philomina Simon, A Document Retrieval System with
Genetic Algorithm, International Journal of Computer and Electrical Engineering, Vol. 2, No. 1,
pp 1793-8163, February, 2010
[9] http:// www.w3.org/ TR/ rdf-sparql
[10] R. Dhanapal,"An intelligent information retrieval agent", Knowledge
70
[11] Suriati Akmal,Li-Hsing Shih, Rafael Batre, "Ontology
retrieval", Computers in Industry, Vol 65, Issue 1,
[12] Neal G. Shaw, Ahmad Mian, Surya B. Yadav,"A comprehensive agent
intelligent information retrieval in a distributed heterogeneous environment", Decision Support
Systems 32 (2002) 401– 415
[13] Ali Seyfi, Ahmed Patel,"A focused crawler combinatory link and content model based on T
principles", Computer Standards & Interfaces 43 (2016) 1
[14] Donghee Yoo,"Hybrid query processing for personalized information retrieval on the Semantic Web",
Knowledge-Based Systems 27 (2012) 211
[15] Andrea Addis, Giuliano Armano, Eloisa Vargiu,"Multiagent systems and information retrieval our
experience with X.MAS",Expert Systems with Applications 39 (2012) 2509
[16] Jorge Ropero, Ariel Gómez, Alejandro Carrasco,
Information Extraction: Introducing
Systems with Applications 39 (2012) 4567
[17] Mao-Yuan Pai a, Ming-Yen Chen b, Hui
based content mapping mechanism for
(2013) 2447–2461
Fig 4: Search Agent: Activity diagram
System for Semantic Information Retrieval”, ERCIM, Advanced databases
David J.Kocur et.al, SAIRE- a scalable agent based information retrieval,
International Conference on Autonomous agents,ACM, 292-299,1997
J.Nie,Anquetil,J.Vaucher,”A Multi-Agent Distributed Retrieval System”, CAIS-ASCI
Masoud Mohammadian ,”Intelligent Agents for data Mining and Information Retrieval”, 2004
Bajaj, Akhilesh; Tanabe, Hideaki; and Wang, Chao-hin, "XSAR: Xml Based Search Agent For
(2002). AMCIS 2002 Proceedings. Paper 197
Lora Aroyo , Piet Kommers,”Intelligent agent for educational and computer-Aided systems”,
Journal of Interactive learning Research ,Special issue,, 10(3/4), 235-242, 1999.
Philomina Simon, A Two Stage approach to Document Retrieval using Genetic Algorithm,
International Journal of Recent Trends in Engineering, Vol 1, No 1, 526-528, 2009
S.Siva Sathya , Philomina Simon, A Document Retrieval System with Combination Terms Using
February, 2010
sparql-query/ W3C. (2009)
Dhanapal,"An intelligent information retrieval agent", Knowledge-Based Systems 21 (2008) 466
Hsing Shih, Rafael Batre, "Ontology-based similarity for product information
rs in Industry, Vol 65, Issue 1, Pages 91-107, 2014
Neal G. Shaw, Ahmad Mian, Surya B. Yadav,"A comprehensive agent-based architecture for
,"A focused crawler combinatory link and content model based on T
principles", Computer Standards & Interfaces 43 (2016) 1–11
Donghee Yoo,"Hybrid query processing for personalized information retrieval on the Semantic Web",
ems 27 (2012) 211–218
Andrea Addis, Giuliano Armano, Eloisa Vargiu,"Multiagent systems and information retrieval our
experience with X.MAS",Expert Systems with Applications 39 (2012) 2509–2523
Jorge Ropero, Ariel Gómez, Alejandro Carrasco, Carlos León,"A Fuzzy Logic intelligent agent for
Information Extraction: Introducing a new Fuzzy Logic-based term weighting scheme", Expert
Systems with Applications 39 (2012) 4567–4581
Yen Chen b, Hui-Chuan Chu c, Yuh-Min Chen, "Development of a semantic
based content mapping mechanism for information retrieval", Expert Systems with Applications 40
59
Retrieval”, ERCIM, Advanced databases
a scalable agent based information retrieval,
299,1997
ASCI Conference
nformation Retrieval”, 2004
: Xml Based Search Agent For
Aided systems”,
Philomina Simon, A Two Stage approach to Document Retrieval using Genetic Algorithm,
Combination Terms Using
Based Systems 21 (2008) 466–
based similarity for product information
based architecture for
,"A focused crawler combinatory link and content model based on T-Graph
Donghee Yoo,"Hybrid query processing for personalized information retrieval on the Semantic Web",
Andrea Addis, Giuliano Armano, Eloisa Vargiu,"Multiagent systems and information retrieval our
Carlos León,"A Fuzzy Logic intelligent agent for
based term weighting scheme", Expert
Development of a semantic-
information retrieval", Expert Systems with Applications 40

60
[18] Emdad Khan, Adel AlSalem,"Ivia: Interactive Video Intelligent Agent Framework for Instructional
Video Information Retrieval", International Educational Technology Conference, Procedia - Social
and Behavioral Sciences 64 ( 2012 ) 186 – 191
[19] Petros Belsisa*, Stefanos Gritzalisb, Costas Lambrinoudakisc, Christos Skourlasa,Vassilis
Tsoukalasa, Dimitris Vassisa,"An Agent Based Architecture Benchmark", ICININFO, Procedia -
Social and Behavioral Sciences 147 ( 2014 ) 429 – 435
[20] TAO Teng-yang and ZHAO Ming,"An Ontology-Based Information Retrieval Model for Vegetables
E-Commerce", Journal of Integrative Agriculture 2012, 11(5): 800-807
[21] Zeinab E. Attiaa, Ahmed M. Gadallaha and Hesham M. Hefnya, "An Enhanced Multi-View Fuzzy
Information Retrieval Model based on Linguistics", 2013 International Conference on Applied
Computing, Computer Science, and Computer Engineering , IERI Procedia 7 ( 2014 ) 90 – 95

UML MODELING AND SYSTEM ARCHITECTURE FOR AGENT BASED INFORMATION RETRIEVAL

Recommended

More Related Content

What's hot (20)

Similar to UML MODELING AND SYSTEM ARCHITECTURE FOR AGENT BASED INFORMATION RETRIEVAL (20)

Recently uploaded (20)

UML MODELING AND SYSTEM ARCHITECTURE FOR AGENT BASED INFORMATION RETRIEVAL