Session S2G

INTEGRATING AGENT-ORIENTED ENTERPRISE SOFTWARE

ENGINEERING INTO SOFTWARE ENGINEERING CURRICULUM
Gilda Pour'
Abstracl ~ Agent-oriented enterprise software engineering
(AOESE) has emerged as a promising approach to
developing Web-based enterprise software system. AOESE
is based on developing and evolving enterprise software
systems from selected pre-engineered and pre-tested
software agents, and then assembling them within
appropriate software architectures. We view agent-oriented
software engineering as an extension of conventional
component-based software engineering, and software agents
as next-generation software components. Software agents
offer greater flexibility and adaptability than traditional
components. However, making transition to agent-oriented
enterprise software engineering requires software engineers
to learn a new set of technical ski l l s. To provide such a
learning opportunity, the author has created a new course
sequence that integrates both agent-oriented enterprise
software engineering and component-based enterprise
software engineering into software engineering curriculum.
A new course in the sequence is intended to provide students
with the opportunity to acquire o good understanding of t he
key concepts and principles of agent-oriented enterprise
software engineering, and the new Opportunities and
challenges involved in developing Web-based multi-agent
systems. The course is designed to help build a solid
foundation f or integration of research into education in the
area ofagent-oriented enterprise software engineering. The
course has speciol emphasis on developing architectures
and frameworks f or Web-based multi-agent enterprise
systems rather than developing single agent. In this paper,
the author shares her experience ofdeveloping the course,
and presents the course organization, its components, and
thefutureplans f or the course.
Index Terms - Agent-oriented enterprise software
engineering, component-based enterprise software
engineering, software agents, software engineering
curriculum. workflows.
AGENT-ORIENTED ENTERPRISE SOFTWARE
ENGINEERING
Rapidly rising demand for more flexible, adaptable,
extensible, and robust Web-based enterprise application
systems has motivated the search for new software
engineering methodologies and development strategies.
The new strategies must be based on developing
enterprise software systems by assembling flexible and
adaptable reusable software components built at different
times by various developers. Traditional soilware
development strategies and engineering methodologies,
which require development of software systems from
scratch, fall short in this regard.
Agent-oriented enterprise software engineering
(AOESE) has emerged as a promising approach to building
Web-based enterprise application systems. Agent-oriented
enterprise software engineering has the potential to
reduce significantly the cost and time-to-market of
enterprise application systems by allowing the
systems to he built through assembling a set of
software agents rather than from scratch;
enhance the reliability of enterprise application
systems by allowing the systems to be developed
through assembling a set of reusable software agents;
improve the maintainability and flexibility of
enterprise application systems by allowing
replacement of old agents with quality agents in the
system; and
enhance the quality of enterprise application systems
by allowing applicationdomain experts to develop
software agents and software engineers to build the
systems by assembling the agents.
Agent-oriented enterprise software engineering is based
on developing and evolving software systems from selected
pre-engineered and pre-tested software agents, and then
assembling them within appropriate software architectures.
Agent-oriented software engineering allows developers
to use a set of high-level, flexible abstractions to represent
enterprise software systems. Rapid integration of distributed
agents provides opportunities to build Web-based multi-
agent application system for a wide variety of application
domains; for instance, information (e.g. health care,
education, missing children, automobiles, travel, real estate,
employment), banks (e.g. virtual banks), finance, E-
business, media, government (e.g. space exploration),
personal assistants (PAS), wearable computers, smart
clothes, virtual aquarium, virtual pets, and manufacturing.
I Gilda Pour, Department of Computer Engineering, San Jose State Universily, San Jose, CA 95192-0180, U.S.A. gpour@sjsu.edu
0-7803-74444/02/$17.00 0 2002 I EEE November 6 - 9,2002, Boston, MA
32" d ASEEKEEE Frontiers in Education Conference
S2G8
Session S2G
SOFTWARE AGENTS
Agent-oriented enterprise software engineering has emerged
as an extension of conventional component-based software
engineering (CBSE), and software agents as next-generation
software components, and [ I]-[3].
A software agent is a specialized software component
that interacts as a surrogate for its user with its environment
and other agents, and reacts to changes in the environment.
There is a general agreement that an agent is a reusable
component that exhibits a combination of several of the
following characteristics:
Autonomous. The agent proactively initiates activities
according to its goals, acts on its users behalf, and
exercises control over its own actions.
Adaptable. The agent changes its behavior after
deployment as a result of its learning and acquired
information, user customization, andor downloading
new capabilities.
Mobile. The agent moves from one executing context
to another, continues execution in a new context, and
retains its state to continue its work.
Knowledgeable. The agent reasons about its goals,
acquired information, and knowledge about other
agents and users.
Collaborative. The agent communicates and
cooperates with other agents to form dynamic or static
societies of agents, and performs a task in
collaboration with some other agents.
Persistence. The infrastructure enables agents to
retain knowledge and state over extended periods,
including robustness at possible runtime failures [I].
Intelligent agents are autonomous agents that have their
own goals and beliefs and can reason about their present and
future behavior-offer many oppormnities for rapid,
incremental development of Web-based enterprise
application systems. Developers can apply these systems to a
variety of complex, dynamic domains, ranging from e-
commerce to human planetary exploration [ 1][2][5].
MAKING TRANSITION TO AGENT-ORIENTD
ENTERPRISE SOFTWARE ENGINEERJNG
While agent-based systems are becoming increasingly
understood, multiagent systems development is not [6].
This is due to several major differences between agent-
oriented software development lifecycle and the traditional
software development lifecycle. For instance,
Design phase in agent-oriented software development
lifecycle includes selection and customization of
software architectures, as well as selection and
customization of a set of software agents. It is critical
to design and architect the extensibility and scalability
into a multi-agent enterprise system and all its parts.
Implementation phase in agentariented software
development lifecycle is not about extensive coding
to build agentariented enterprise systems from
scratch. It rather involves developing enterprise
software systems that support interaction and
cooperation among a set of agents within appropriate
software architectures.
Testingphase in agent-oriented software development
lifecycle involves testing an enterprise system that is
an assembly of a set of software agents built by
various developers. The system integration testing has
to address a series of new issues raised by the lack of
Software agents offer greater flexibility and adaptability
than traditional components. There are three categories of
agents based on the maior function of agents:
confidence in and understanding of agents built by
others [3][5].
-
Personal agents interact directly with a user,
present some personality or character, monitor and
adapt to the users activities, learn the users style
and preferences, and automate or simplify certain
tasks.
Mobile agents visit remote sites to collect
information before returning with the results, and
aggregate and analyze data or perform local control.
Collaborative agents communicate and interact
with some other agents as they represent their users,
organizations, and services. Multiple collaborative
agents exchange messages to negotiate or share
information [ I ] .
Due to several major differences between the agent-
oriented enterprise software development lifecycle and the
traditional software development lifecycle, software
engineers need to acquire a new set of software engineering
skills for developing, maintaining, and evolving multi-agent
enterprise software systems.
The author has developed a new course that integrates
agent-oriented enterprise software engineering into the
software engineering curriculum. The course is designed to
provide students with the opportunity to learn the key
concepts and principles of agent-oriented enterprise software
engineering, software agents, and workflows for Web-based
enterprise application systems, agent technologies, and
multi-agent systems.
0-7803-7444-4/02/$17.00 0 2002 I EEE
November 6 ~ 9,2002, Boston, MA
320d ASEElIEEE Frontiers in Education Conference
S2G9
Session S2G
The availability and growing popularity of J ava, XML,
XML schema, and XMGbased technologies helps make
transition from traditional software engineering to agent-
oriented enterprise software engineering.
ENTERPRISE SOFTWARE A ~ c m m m u ~ ~
Software architecture refers to systems and components that
must be structured for the support of independent
development, and integration and evolution of components
and systems [4]. Agents are specialized components.
Welldesigned software architecture with flexibility to
adapt and extend is required for development of agent-
oriented enterprise software systems. Multi-agent enterprise
software systems must be architected, designed, packaged,
and supported for effective adoption of AOESE.
Agents reside and execute in a conceptual and physical
location called an agency. The agency provides facilities for
collecting knowledge about other agents and for locating and
messaging agents.
INTEGRATING AGENT-ORIENTED ENTERPRISE
SO~AREENGINEERING~OTHE S O ~ A R E
ENG"G CURRICULUM
To prepare software engineering students for a challenging
software engineering career, the author has created a course
sequence that integrates both agent-oriented enterprise
software engineering and component-based and reuse-driven
software engineering into the software engineering
cumculnm [7][8].
The course sequence includes a new major course that is
focused on agent-oriented enterprise software engineering,
and particularly, on the integration of research into education
in this area.
The new course has special emphasis on developing
architectures and frameworks for Web-based multi-agent
enterprise systems rather than developing single agent.
Because agent-oriented enterprise software engineering
deals with different types of agents for developing enterprise
software systems, the course is not limited to intelligent
agents as an extension of the work done in Artificial
Intelligence (AI) research and development.
The author has developed a new laboratory at San J ose
State University with the external funding. The students use
the lab and a variety of software tools different computing
environments available in the lab for their course projects.
Course Description
The course on agent-oriented enterprise software
engineering is developed for students who are interested in
pursuing software engineering profession. It is a graduate
course, Senior level software engineering students who have
met the prerequisites for the course are permitted to take the
course as one of their technical electives.
The course is designed to provide students with the
opportunity to
learn key concepts and principles of agent-oriented
enterprise software engineering, software agents
and workflows for Web-based enterprise
application systems, agent technologies, multi-
agent systems; and
enhance their professional software engineering
skills, particularly, teamwork and effective
comnunication skills (both written and verbal).
The course emphasizes both the theoretical and practical
aspects of the topics covered in the course. I t covers several
related case studies and industrial examples. The course also
includes a series of assignments and a team-based term
project.
Course Prerequisites
Students are expected to have a good understanding of the
key concepts and principles of object-oriented and
component-based software engineering, and some
experience working with J ava and extensible Markup
Language (XML) prior to taking the course. At San J ose
State University, we offer courses on the subjects that
constitute the prerequisites of this course.
Course Organization
The course is divided into four major parts as shown in
Figure I .
Part I Introduction to Agent-Oriented
Enterprise Software Engineering
Part I1 Multi-Agent Systems
Part III Software Agents and Workflows
Part N Application Areas
Figure 1. Course Organization
Part I- In the first part of the course, we introduce the key
concepts and principles of agent-oriented enterprise software
engineering in the first part of the course. We also provide a
common terminology required for understanding of the
0-7803-7444-4/02/$17.00 0 2002 IEEE
November 6- 9,2002, Boston, MA
32" ' ASEElIEEE Frontiers in Education Conference
S2 G1 0
Session S2G
subject. We discuss the motivation for agent-oriented
development of Web-based enterprise application systems,
and new opportunities and challenges involved in making
transition from traditional software engineering to agent-
oriented software engineering. We also introduce agent-
oriented software development lifecycle, and various types
of software agents with a different mixes of characteristics
such as intelligence, mobility, autonomy, collaboration,
persistence, and adaptability.
Part Il -I n the second part of the course, we present the key
concepts and principles of multi-agent systems. The
following topics are covered in this part of the course:
agent-based system architecture, agent interactions and
cooperation, multi-agent organizations, agent actions and
behaviors, agent communications, agent communication
languages, Knowledge Query and Manipulation Language
Agent technology will continue to he integrated into the
Internet with increasing intelligence and complexity. We
also discuss new opportunities (e.g. a combination of agents
and avatars, virtual cooperation), new challenges (e.g.
privacy and legal issues), and the role of standards. Future
research directions include developing next-generation
agents with new capabilities such as
activating and inhabiting real world robotics and
pursuing goals beyond the virtual world;
particularly, nanotechnology-based mobile entities
(miniature places and submarines, drugs,
computers).
self-replicating and developing other agents to meet
specific needs. These manager agents will be
independent and self-motivating, and in many
respects demonstrate human capabilities.
.
Course Resources
(KQML), the Foundation for Intelligent Physical Agents
(FIPA), and Web-based multi-agent system modeling. We
also discuss the use of Web-based technologies such as J ava,
XML, XML schema, XMLbased technologies, 2nd HTTP
for development of multi-agent systems.
Pad m ~ In the third part Of the course, We discuss the
combination of software agents and workflows as it provides
major benefits beyond those associated with components and
scripting. study of the subject
Part lV - In the fourth part of the course, we discuss
development of a wide variety of agent-based systems and
their applications for
The for the are csInte,ligent software
Agents [9] and Multi-Agent Systems [I O]. The hooks
contain an extensive list of other useful resources on related
topics. The students are also provided with additional notes
that the author has provided for the course,
students with a list of
related pub~ications for our class discussion and further
In addition, we provide
Course Assignments
To help balance the understanding of the key concepts and
network operations; principles of agent-oriented enterprise software engineering,
the application of the agent-oriented enterprise software
engineering theories and the best software engineering
practices, the Course includes a series of individual and
team-based assignments.
Furthermore. the course includes Droiects that are
e-business;
information (e.g. health care. education, missing
children, automobiles, travel, real estate, partners,
employment);
finance;
banks (e.g. virtual banks);
retail stores;
media;
government (e.g. space exploration);
information technology;
manufacturing;
personal assistants (PAS);
wearable computers;
smart clothes;
virtual aquarium; and
virtual pets.
. _
designed to provide students with opportunity to develop a
demonstration prototype of a Web-based multi-agent system.
The class is divided into project teams of three or four
students. Each team is assigned a different project topic. The
examples of project topics include developing Web-based
multi-agent systems for medical record management, air
navigation, supply chain management, and collaborative
review of software development.
Cooperative Learning Styles
We have adopted cooperative and hands-on, inquiry-based,
and active learning styles in this new course on agent-
oriented enterprise software engineering. We plan to
0-7803-744441021S17.00 0 2002 I EEE November 6 - 9,2002, Boston, MA
32 ASEE/IEEE Frontiers in Education Conference
S2Gl l
Session S2G
continue to enhance the course to take full advantage of the
pedagogical possibilities.
Course Evaluation
We have evaluated the effectiveness of various aspects of
the course. Students and industry representatives have
participated in the course evaluation. They have ranked the
course materials, course organization, student assignments
and projects, and the overall effectiveness of the course very
good to excel l ent .
CONCLUDING REMARKS AND
We have created and developed a new course that integrates
the agent-oriented enterprise software engineering into the
software engineering curriculum. The course is design to
help build a solid foundation in agent-oriented enterprise
software engineering for software engineering students. The
emphasis is on developing enterprisewide multi-agent
systems rather than a single agent.
The course provides students with a great opportunity to
acquire a good understanding of the key concepts and
principles of agent-oriented enterprise software engineering,
and the new challenges and opportunities involved in
developing Web-based multi-agent systems.
The course is designed to emphasize both the theoretical
and practical aspects of the course topics. To help students
gain hands-on experience in this new software engineering
area, the course includes a require term project that is
focused on developing a demonstration prototype of a Weh-
based multi-agent system as term project.
The demand for this course has been high. There is also
a great demand for an online version of the course. Those
who will benefit most Froman online version of the course
are the working adults who are seeking the opportunities for
learning new skills relevant to the software engineering
profession for their professional growth and career
advancement, but their schedules and commitments do not
permit the conventional campus experience.
Developing an online version of this course to provide a
learning oppomnity far such working adults will be
considered. We believe that development of an online
version of the course has to ensure the human dynamics of
the online course follow the practices and skills of the
contemporary workplace. Students should work on
individual as well as teambased projects: collaborate
through a variety of electronic media, and complete projects.
The online version of the course should emphasize the
combination of software engineering, writing and
interacting, as this will be a powerful combination for
students who must not only master the technical skills, but
also develop ability to communicate and collaborate
0-7803-7444-4/02/S17.00 0 2002 IEEE
professionally to use their technical skills in the actual
workplace.
The online version of this course will not onlv offer life- ~~~~~ ~~
long learners the access to this new valuable course, but it
will also serve as a pedagogical template for other courses
offering similar or related technical skills, using similar
technologies, or reaching out to the same audience.
REFERENCEs
Griss, M. andPour, G., Accelerating Development with Agent
Components, Cover F ~ ~ N w Article, IEEE Computer,VolM,%.
5 , May 2001, pp. 37-43.
Pour, G., Internet-Based MultCMobile-Agent Framework for
Planetarv Emlaration. Proc. Internoti onol Conference on
, .
Intelligent Agents. Web Technology and lnrerner Commerce,Ny
2001,pp. 153-163.
Pour, G., WcbBased Architecture for Component-Based
Application Generators, Pmc. Intemationol Conference on Inremei
Computing, June2002.
Shaw, M. andGarlan, D., Sofiare Architecrure: Perspclives on on
Emerging Di sci pl i ne, Prentice Hall, 1996.
Pour, G., Jini-Based MobileAgent Architecturefor Human
Planetary Exploration, Proe. l nrernori onol ConJerence on
Technology ojObject-Oricnted Systems and Languages (TOOLS
USA), IEEE Computer Society Press, August 2001, pp. 337-346.
Woodidge M. andJennings, N., Software Engineering with
Agents: Pitfalls and Pratfalls, IEEE htemetCom@g,Vd3,No.
3, MayiJune1999, pp. 20-27.
Pour, G., Component Technologies: Expanding the Possibilities for
Component-Based Development of WebBased Enterprise
Applications, Handbook qflnternet Compuring,CRC F ? q 2aS0,
pp. 133-156.
Pour, G., Integrating Component-Based and Reuse-IhivensOfn+m
Engineering into the SoAware and Information Engineering
CUmculum, Proc. Frontiers in Educorion (FIE), IEEE Computer
Society Press, 2000, Section TC2, pp. 18-23
Murch, R. and Johnson, T., lnrelligenr SoJware Agenis, Prentice
Hall, 1999.
Ferber, J ., Multi-Agent Systems, Addison-Wesley, 1999.
November 6 - 9.2002. Bostbn, MA
3Znd ASEEEEEE Frontiers in Education Conference
S2G12
/