nach oben

2009 | Buch

Kapitel lesen Erstes Kapitel lesen

Ontology-Based Multi-Agent Systems

verfasst von: Maja Hadzic, Pornpit Wongthongtham, Tharam Dillon, Elizabeth Chang

Verlag: Springer Berlin Heidelberg

Buchreihe : Studies in Computational Intelligence

Enthalten in: Springer Professional "Wirtschaft+Technik" , Springer Professional "Technik" , Springer Professional "Wirtschaft"

Einloggen, um Zugang zu erhalten

Über dieses Buch

During the last two decades, the idea of Semantic Web has received a great deal of attention. An extensive body of knowledge has emerged to describe technologies that seek to help us create and use aspects of the Semantic Web. Ontology and agent-based technologies are understood to be the two important technologies here. A large number of articles and a number of books exist to describe the use individually of the two technologies and the design of systems that use each of these technologies individually, but little focus has been given on how one can - sign systems that carryout integrated use of the two different technologies. In this book we describe ontology and agent-based systems individually, and highlight advantages of integration of the two different and complementary te- nologies. We also present a methodology that will guide us in the design of the - tegrated ontology-based multi-agent systems and illustrate this methodology on two use cases from the health and software engineering domain. This book is organized as follows: • Chapter I, Current issues and the need for ontologies and agents, describes existing problems associated with uncontrollable information overload and explains how ontologies and agent-based systems can help address these - sues. • Chapter II, Introduction to multi-agent systems, defines agents and their main characteristics and features including mobility, communications and collaboration between different agents. It also presents different types of agents on the basis of classifications done by different authors.

Inhaltsverzeichnis

Frontmatter

Current Issues and the Need for Ontologies and Agents

Introduction

The internet had become a major source of information in many knowledge domains. The general public uses Google predominantly to obtain in-formation pertaining to a variety of knowledge domains. Generally, users will have different access to, and understanding of, the results they obtain from their ‘Google’ search. As Google is not built to separate authoritative from dubious in-formation sources, users may have to rely on specialized search engines.

The large volume of published information that is being accounted for is an additional problem that complicates the search. For example, biomedical re-searchers may use PubMed which is a service of the U.S. National Library of Medicine that includes over 16 million citations from life science journals for biomedical articles going back to the 1950s. Using the PubMed search engine, the user receives a list of journals related to the given keyword. It is then left to the user to read each journal individually and to try to establish links within this information. This would be easy if the journal list consisted of a small number of journals. But the journal list usually consists of thousands of journals, and medical researchers usually do not have time to go through these results thoroughly. There is a high chance that some important information will be omitted.

There is a need to design an intelligent search engine that performs searches not only on keywords, but also on the meaning of the information. The search engine would go through the available information, understand this information, and select highly relevant information; moreover, it would link this information and present it in a meaningful format to the users.

In this chapter, we will briefly introduce the technologies underpinning such meaningful representation of information and the use of an intelligent and supportive retrieval approach. We examine current issues related to information representation, information access and information retrieval on the web. We will introduce the meaning of web semantics and the role of ontologies and agent tech-nologies in the creation of semantically rich environments.

Maja Hadzic, Pornpit Wongthongtham, Tharam Dillon, Elizabeth Chang

Introduction to Multi-Agent Systems

Introduction

The body of available information is rapidly increasing every day. It is becoming impossible for humans to peruse all the available information and extract the data related to a specific topic. A great deal of important information is being neglected primarily for two major reasons. Firstly, highly relevant information is losing its true meaning within a pool of insignificant and unauthorized information. A lot of available information is published which is of no significance and sometimes may even be destructive and corruptive in nature. Secondly, new knowledge and discoveries are emerging quickly as we continue to progress and new technologies emerge. It is becoming more and more difficult each day to find specific information even within a body of authorized information.

As a solution to this problem, humans are designing software agents that will do the search for them. In this chapter, we will discuss agent features such as environment, internal structure, mobility, communication, and cooperative work with other agents.

Maja Hadzic, Pornpit Wongthongtham, Tharam Dillon, Elizabeth Chang

Introduction to Ontology

Introduction

It is important for agents to communicate and interact with each other, especially if they are part of the same multi-agent system. In most cases, different agents are working collaboratively towards the same goal. They need to talk to each other, share tasks, exchange results etc. Here, it is important that agents understand each other; for example, they need to speak the same language or be able to translate and understand the language spoken by other agents.

Ontologies are used to establish effective communication between different agents. Ontologies specify the terms used in agents’ communication and provide the exact meaning of those terms relative to other ontology terms and within a specific context. Ontologies provide the agent with the domain knowledge and enable it to function intelligently.

In this chapter, we will introduce ontologies. We will provide a definition of ontology and explain associated terminology such as ontology commitments, ontology representation, ontology classification; we will give a formal description of ontologies and ontology design criteria.

Maja Hadzic, Pornpit Wongthongtham, Tharam Dillon, Elizabeth Chang

Design Approaches for Multi-Agent-Based Systems

Introduction

In Chapter 2, we discussed agent-based systems, their characteristics and numerous advantages. We may feel inspired to design such agent-based systems but we need a methodology to guide us through the design process. There are many different multi-agent system design methodologies. In this chapter, we will discuss some of these design methodologies. We will give a brief overview of their different approaches and their important advantages. The chosen methodologies are different from each other and together cover various aspects of the multi-agent system design process.

Maja Hadzic, Pornpit Wongthongtham, Tharam Dillon, Elizabeth Chang

Ontology Design Approaches

Introduction

In Chapter 3 we discussed the concept of ontology and its characteristics, and the numerous advantages of ontologies. We may feel inspired to design an ontology for our information system but we still do not know: where to start, how to prepare ourselves, how many people are needed for the ontology design, how we ascertain that we are on the right track, whether similar ontologies already exist, how to determine that we are designing a valid ontology, what tool we can use to assist us with our design, and other similar issues.

There is not a single, consensual ontology-design methodology. Many different ontology design methodologies have been proposed. We will consider some of them in order to illustrate different approaches to ontology design. The methodologies presented are different from each other and together cover various aspects of the ontology design process.

Maja Hadzic, Pornpit Wongthongtham, Tharam Dillon, Elizabeth Chang

Significance of Ontologies, Agents and Their Integration

Introduction

In this chapter, we will discuss some advantages of ontology- and agent-based systems. Ontologies provide machine readable and understandable domain knowledge and play an important role in knowledge representation, sharing and management, data semantics, intelligent information retrieval, mediation, natural language applications, and the like. Two important characteristics of agents are their autonomous and collaborative behaviour and, as such, agent-based systems are used to support distributed computing, dynamic information retrieval, automated service discovery, computational intelligence etc. We also describe the advantages of the integrated approach i.e. ontology-based multi-agent systems. Ontology- and agent-based computing are two different but complementary technologies; ontologies give intelligence to the system while the agents provide the system dynamics.

Maja Hadzic, Pornpit Wongthongtham, Tharam Dillon, Elizabeth Chang

Design Methodology for Integrated Systems - Part I (Ontology Design)

Introduction

Ontologies are high expressive knowledge models and as such increase the expressiveness and intelligence of a system. Ontologies were introduced into the computer and information community to be used by various agents and for different applications addressing the problems of various knowledge domains.

There is not a single widely accepted ontology design methodology. We discussed some of the currently available methodologies in Chapter 5. We propose an Onto-Agent Methodology as the first methodology that unifies the different approaches of the existing ontology and multi-agent systems design methodologies. The five steps of the first part of the Onto-Agent Methodology (Onto Methodology) are described in this chapter and are accompanied by some illustrative examples. The second part of the Onto-Agent Methodology (Agent Methodology) also consists of five steps which are described with illustrative examples in the next chapter.

Maja Hadzic, Pornpit Wongthongtham, Tharam Dillon, Elizabeth Chang

Design Methodology for Integrated Systems - Part II (Multi-Agent System Design)

Introduction

A multi-agent system provides a distributed collaborative platform and as such determines the system dynamics. Ontologies represent the domain knowledge and can be used to support some important processes within a multi-agent system such as: problem decomposition and task sharing among different agents, result sharing and analysis, information retrieval, selection and integration etc.

There is not a single and consensual multi-agent system design methodology. We discussed some of these methodologies in Chapters 4. We introduce an Onto-Agent Methodology as the first methodology that unifies the different approaches of the existing ontology and multi-agent systems design methodologies. This methodology is composed of two interconnected processes. The five steps of the first part of the Onto-Agent Methodology (Onto Methodology) were presented in the previous chapter where we described how to design an ontology on which the multi-agent system will be based. The second part of the Onto-Agent Methodology (Agent Methodology) also consists of five steps and will be presented in this chapter. Using the Agent Methodology, we will define the multi-agent system which functions and operates using the ontology designed with the Onto Methodology as described in the previous chapter.

Maja Hadzic, Pornpit Wongthongtham, Tharam Dillon, Elizabeth Chang

Notations for the Integrated Ontology and Multi-Agent System Design

Introduction

In this chapter, we present the modelling of the integrated ontology and multi-agent system using well designed notations. Modelling is an important aspect of all methodologies. The modelling process analyses the range of problems that describe various aspects of the system being considered. With a precise definition of modelling, design can take place and the development process can then be transformed. The life cycle spans everything from modelling to evolution built up of possibly many autonomous, rational, proactive, and semantic entities. Particular modelling languages and notations have to be created in order to capture particular features of systems. It is insufficient and ineffective to use traditional software engineering modelling languages in the development of integrated ontology and multi-agent systems due to the specific characteristics of semantic and autonomous agents.

Maja Hadzic, Pornpit Wongthongtham, Tharam Dillon, Elizabeth Chang

Architecture of the Integrated Ontology and Multi-Agent System

Introduction

In this chapter, we explore the architecture of the integrated ontology and multi-agent system. The ontology provides an important mechanism to facilitate producing semantic information. Since the ontology has been used to express formally a shared understanding of information, it enables the sharing of an agreement among users by making assumptions explicit. The key idea is to have agreement explicitly interpreted by software tools rather than just being implicitly interpreted by a human. The representation of knowledge including ideas, tasks, models, processes as well as documentation using an ontology and sub-ontology, will provide intuitive, clear, precise concepts and ideas, knowledge and classified issues. Knowledge is organised into the ontology and used as the basis for classifying knowledge enabling questions and problem solving. Additionally, knowledge can be shared among users in community. A framework showing the use of ontologies in conjunction with multi-agent system is given in Figure 10.1.

Maja Hadzic, Pornpit Wongthongtham, Tharam Dillon, Elizabeth Chang

Case Study I: Ontology-Based Multi-Agent System for Human Disease Studies

Introduction

In this chapter, we shall put the Onto-Agents Methodology into practice and use it step by step to develop a Generic Human Disease Ontology (GHDO) and to design a multi-agent system that can use the designed GHDO for intelligent information retrieval.

In Section 2, we discuss the domain of human diseases, the purpose of the ontology, the community of users and agents for which this ontology is being developed, and applications based on the designed ontology. The aligning and merging of existing medical ontologies against the defined ontology structure is discussed in Section 3. Section 4 describes how to design the ontology base, while Section 5 explains the design of the ontology commitment layer. In Section 6, we discuss the evaluation of the GHDO.

In Section 7, we identify and describe different groups of agents according to their functions and responsibilities within the system. In Section 8, we describe a mechanism by which GHDO ontology is used in this system during the processes of problem solving, task and result sharing, and assembling of results. In Section 9, we focus on the structural organization of the agents within the system and define agents’ collaborations. Individual agent structure and agents’ components are described in Section 10. In Section 11, we discuss security issues within the multi-agent system. Examples of the use of the designed systems are given in Section 12.

Maja Hadzic, Pornpit Wongthongtham, Tharam Dillon, Elizabeth Chang

CASE STUDY II: Ontology-Based Multi-Agent System for Software Engineering Studies

Introduction

In this chapter, we explore a case study in the domain of software engineering. We will go through the development of Software Engineering Ontology (SE Ontology) and a multi-agent system in which agents are interacting and mediating with the SE Ontology. The purpose of such development is for multi-site software development as a communication framework.

In Section 2 we discuss on software engineering domain including purpose of the SE Ontology, the context of software engineering domain knowledge, communities of users and agents for which the SE Ontology is being used, and applications on the SE Ontology. In Section 3, we discuss how we design the formal SE Ontology. We then evaluate the SE Ontology in Section 4.

In Section 5, we identify and describe different type of agents according to functionalities in the domain. We discuss the need for the SE Ontology to support agents’ intelligence in Section 6 and describe agents’ collaboration in Section 7. The construction of each individual agent is explained in Section 8. We present examples of the practical uses in Section 9. We conclude the chapter in Section 10.

Maja Hadzic, Pornpit Wongthongtham, Tharam Dillon, Elizabeth Chang

Potential Applications of Ontology-Based Multi-Agent Systems

Overview

In this book, we introduced multi-agents systems in Chapter 2 and ontologies in Chapter 3. Many advantages of ontology-based, agent-based and the integrated ontology/agent-based systems are described in Chapter 6. We described a number of existing design methodologies for agent-based systems in Chapter 4 and ontology- based systems in Chapter 5. We have developed our own design methodology by examining the strengths and weaknesses of the existing design methodologies. This new design methodology consists of two parts: the ontology design part which is described in Chapter 7, and the multi-agent design part, described in Chapter 8. The notation for ontology-based multi-agent systems is defined in Chapter 9, while the system architecture and system implementation are described in Chapters 10 and 11. The implementation of ontology-based multi-agent system in the domain of human diseases and the software engineering domain is described in Chapters 12 and 13 respectively.

In this chapter, we will highlight the three main application areas (Collaborative Environments, Information Access and Retrieval; and Data Mining). We will also discuss some open issues related to the implementation of the ontologybased, multi-agent systems.

Maja Hadzic, Pornpit Wongthongtham, Tharam Dillon, Elizabeth Chang

Titel: Ontology-Based Multi-Agent Systems
verfasst von: Maja Hadzic
Pornpit Wongthongtham
Tharam Dillon
Elizabeth Chang
Verlag: Springer Berlin Heidelberg
Electronic ISBN: 978-3-642-01904-3
Print ISBN: 978-3-642-01903-6
DOI: https://doi.org/10.1007/978-3-642-01904-3