|
 |


Syllabus
33:010:458
Accounting Information Systems
Fall 1999
 |
|
MEMORANDUM
|
Expert Systems
Students
September 1, 1999
The syllabus for this course is attached to
this memorandum. It contains information about
grading, homework, examinations, and my course
philosophy and policies. A separate document
contains the timetable for the semester.
You
may, of course, contact me by e-mail or by
telephone either to discuss problems and issues
arising from the course, or to fix a mutually
convenient time for us to meet outside office
hours.
This is a new
course offering primarily designed for Accounting
Information System majors in the Ph. D. in
Management program, but expected to appeal also
to other Ph. D. students with an interest in
Information Systems. Expert Systems represent a
practical application of artificial intelligence
techniques to unstructured problems requiring
knowledge and expertise for their solution. A
very large number of Expert Systems have been
built, of which a small proportion (but still a
significant number) have been successfully
implemented in practice. Typically, successful
implementations have required large financial
outlays and considerable time expenditures. This
course, however, is a research-level course, and
although we will study the techniques involved,
the obstacles they have faced, and current
research offering solutions, we will not be
aiming to develop practical implementation
skills. The elicitation of expertise from human
experts, for example, is an important skill that
we will discuss but not practice. The main focus
of the course will be (a) a general introduction
to the subject; (b) study of certain mathematical
background materials including graph theory and
probability theory; (c) probabilistic and
quasi-probabilistic expert systems using various
methods for local computation (including belief
functions, possibility theory and epistemic
beliefs). I hope that you will find this material
as challenging and as interesting as I do.
The various
policies and procedures for the course will be
followed strictly and without exception. Every
effort will be made to grade homework, projects
and examinations promptly and accurately. Grades
will be posted anonymously on my Web Site. In
order to benefit from your efforts, please be
sure to check your homework answers. Otherwise
you will have made most of the effort for only a
fraction of the potential benefit.
My goal is that
if you make the effort required to succeed in
this class, you will develop a good understanding
of current Expert System methodologies and
techniques. This will require hard work on all
sides. I am sure you will expect me to be
prepared for every class, and I intend to be. In
return, I expect you to be both present and
prepared. In addition, I expect you to
participate in classroom activities to the
fullest extent possible given the number of
people present. Remembering and understanding the
material are both necessary; they will take time,
and will require regular study. It is most
unlikely that pre-examination cramming will be
successful as a study technique.
I will be giving
this course my best effort throughout the
semester, and I need you to do the same if I am
to achieve my goal. I am looking forward to
working with you. Expert Systems, and
particularly modern probabilistic and
quasi-probabilistic systems, are an exciting
research topic with potential practical
application of enormous significance and value to
the business community, and you will see that I
take a serious view of the commitment we will all
need to make if the course is to be a success.
However, nothing in this memorandum is meant to
suggest that we should not enjoy our semester
together. Studying can be both rewarding and fun
in itself, but if you see opportunities for
adding fun to our class that I have missed,
please do not hesitate to draw them to my
attention. I wish you every success in studying
Expert Systems this semester.
Yours sincerely
Peter R Gillett
|
 |

|
Room:
|
MEC 201
|
Time: |
Th 1.30 - 4.30
|
Instructor: |
Professor Peter R. Gillett
|
Office: |
Levin 231
Ackerson 302
|
Telephone: |
(732) 445 4765
|
Fax: |
(732) 445 3201
|
E-mail: |
gillett@everest.rutgers.edu
|
Office Hours: |
Th 4.30 - 5.30 or at other times
by appointment
|
Texts: |
Introduction to Expert
Systems. Peter Jackson. (Addison-Wesley
1999).
For additional readings, see the References
appended to the Timetable.
|
 |

|
OBJECTIVES
|
At the conclusion of
this course, students should have gained:
|
|
an enhanced understanding of the
nature and role of Expert Systems, their
advantages and their limitations
|
|
an understanding of the
principal components of expert systems and the
range of applicable technologies
|
|
an understanding of the
knowledge engineering process and applicable
techniques
|
|
an understanding of various
methods of machine learning
|
|
familiarity with ID3, C4.5 and
certain related algorithms for machine learning
|
|
familiarity with rule-based
Expert Systems, and practical experience of
issues related to building them
|
|
an understanding of uncertain
reasoning in Expert Systems, including Certainty
Factors as used in MYCIN, and the methods used in
PROSPECTOR, together with their limitations
|
|
an understanding of Pearl's
work, leading to Bayes' nets
|
|
familiarity with the
Lauritzen-Spiegelhalter, Aalborg and
Shenoy-Shafer algorithms for local propagation
|
|
an understanding of probability
theory, Dempster-Shafer belief functions,
possibility theory and Spohn's epistemic
calculus, and their application in Expert Systems
|
|
an introduction to current research in
probabilistic and quasi-probabilistic Expert
Systems.
|

|
BACKGROUND
|
The course is designed assuming
that students may not previously have taken any
prior course in Expert Systems. No specific
programming language is required, but it will be
assumed that students have had some prior
programming experience and will be able to learn
to use an Expert System tool through out-of-class
experiential learning. Although the course will
begin with a wide-ranging general introduction to
Expert Systems it will proceed rapidly to cover
more advanced topics at a research level.
Significant efforts will be expended in the
second part of the course in studying primary
research literature, including both classic and
more recent papers. Students should expect
classes to be a mixture of lecture, discussion,
demonstration and debate, and should anticipate
the need to make in-class presentations and to
write a research paper of publishable quality.
Prior preparation for class is expected.
Attendance and active participation are strongly
encouraged.
Any students who
consider themselves disabled should communicate
directly with the Dean's Office early in the
semester so that the nature of their disability
and any necessary accommodations can be
determined.
|
 |
|
|

|
GRADING
|
|
Total
Possible Points:
|
|
|
|
Homework and Classroom
Participation |
300
|
|
|
CLIPS Programming Project |
300
|
|
|
Mid-Term Examination |
300
|
|
|
Final Paper |
300
|
|
|
Final Examination |
300
____
|
|
|
TOTAL |
1500
____
|
|

|
HOMEWORK AND ASSIGNMENTS, ETC.
|
All homework and assignments are
due at the beginning of the class period shown on
the attached schedule.
Although
individuals must turn in their own work, which
should be the result of their own efforts,
students are encouraged to seek assistance from
each other and from the instructor if necessary
prior to turning in homework assignments. These
assignments are primarily designed as learning
opportunities rather than evaluation tools.
Collaboration and co-operative learning are here
regarded as positive aspects of scholarly
endeavor. Your worst homework assignment grade
will be dropped.
For the CLIPS
programming assignment, students are encouraged
to share ideas, approaches and solutions to
particular programming difficulties. However, the
exchange of actual program code is forbidden, and
students must each submit their own program
solutions, which should be their own work.
The final paper
should be original work of a publishable nature
and quality. However, co-authored papers will be
accepted (with advance approval by the
instructor) provided that students can show that
the topic addressed is commensurately more
complex, broader, or more time-consuming than an
individual paper would otherwise be. The paper
will be due immediately prior to the commencement
of the final examination, although it may be
turned in earlier. You should plan on developing
the paper as part of normal course work as the
semester progresses, although a week has been
allowed to finish the paper after classes are
over. It is not my practice to allow incompletes
for coursework not finished, except in the case
of medical or other serious problems.
Needless to say,
examinations must be the result of unaided
individual efforts.
|

|
PARTICIPATION
|
Participation grades will be
assigned based on my assessment of how
consistently and how effectively you contribute
to the learning experiences of the class by your
active participation. Factors assessed will
include, but will not necessarily be limited to,
demonstrated preparation of assigned discussion
papers, chapters or questions, posing or
answering questions during class, and
student-lead discussions and presentations. A
pre-condition of your participation, of course,
will be your physical presence in class.
Although late
arrival for class cannot always be avoided,
persistent lateness is a discourtesy to me and to
your fellow students, and will be penalized as
part of the class participation grade along with
absence. It is my policy to teach with the
classroom doors closed; if you arrive late please
close the door again quietly and avoid disturbing
those of us who are already at work. You are
expected to behave in class in a manner
appropriate for professional accounting students.
Quality of
participation is more important than quantity;
too much is no better than too little; the
ability or intention to participate is not a
substitute for actually doing so. Be adventurous:
wrong answers will gain you nothing - but they
will not lose you points you have already gained!
|

|
EXAMINATIONS
|
There will be a take-home
mid-term examination due on the day shown on the
timetable. The form of this examination will be
specified in advance, and the examination itself
will be made available one week in advance of the
due date. You may spend up to, but no more than,
four hours working on this examination, without
reference to notes, books, or other study aids.
The examination must not be shown to or discussed
with other students, or anyone else, until after
class is over on the due date. Graded
examinations will not be returned, but may be
reviewed in my office.
You must take
the Final Examination in order to pass the
course. The Final Examination will be
comprehensive, but greater emphasis will be
placed on new material covered since the
mid-term.
A single make-up
examination will be given if necessary for
students unable to take the Final Examination
during its scheduled period due to documented
medical problems, participation in other
university-sanctioned activities, or required
religious observances.
Examinations
will be closed book. Examination questions will
be a mixture of essay questions and problems, and
may cover any material in assigned
sections of the textbooks or handouts (whether or
not they surfaced in class discussions), any
matters discussed in class (whether or not they
are in the textbook or handouts), and matters
arising from assignments and projects.
|

|
ACADEMIC INTEGRITY
|
University policy on academic
integrity will be strictly enforced; penalties
for cheating are severe. All suspected violations
will be pursued and maximum penalties may be
expected to be imposed. Academic dishonesty is
always unacceptable, and never more so than in a
professional school. |

|
WITHDRAWAL POLICY
|
The last day to withdraw from
this course with a "W" grade is October
5. Late withdrawals will be given the appropriate
letter grade based on the actual number of points
accumulated at the time of withdrawal.
|

|
INCLEMENT WEATHER
|
The class will meet as scheduled
if the University is open. Should the University
close for any reason, any assignments due that
day will be accepted at the next class meeting,
but subsequent assignments will not be postponed.
If the University is closed on a day that an
examination is scheduled, then the examination
will be deferred until the next class meeting.
|

|
ABOUT THE INSTRUCTOR
|
Peter R. Gillett is an Associate
Professor in the Department of Accounting and
Information Systems of the Faculty of Management
at Rutgers. He holds B.A. and M.A. degrees in
Mathematics and Philosophy from the University of
Oxford, and a Ph. D. in Business from the
University of Kansas, where he was the Ronald G.
Harper Doctoral Fellow in the School of Business,
and held the Ernst & Young Doctoral
Fellowship in the Ernst & Young Center for
Auditing Research and Advanced Technology.
Prior to joining
Kansas, he spent from 1975 to 1992 in
professional practice as an auditor, EDP auditor,
and management consultant. After nine years with
Price Waterhouse in London, he joined Grant
Thornton's National office as National Computer
Audit Partner, and subsequently assumed roles as
Partner in charge of Advanced Audit Techniques,
and European Director of Audit Methods.
Dr. Gillett has
taught courses on Managerial Information Systems,
Accounting Information Systems, Auditing and
Advanced Auditing. He is a member of the
Editorial Board of Auditing: A Journal of
Practice and Theory, in which he has also
published. In addition to numerous articles in
professional journals in the UK, he has recently
contributed a chapter on audit judgment to a
monograph on Auditing Practice, Research and
Education published by the AICPA. He is a Fellow
of the Institute of Chartered Accountants in
England and Wales, and of the Institute of
Management; in addition, he is a Member of the
British Computer Society and of the Institute for
the Management of Information Systems.
Dr. Gillett's
dissertation "A Comparative Study of Audit
Evidence and Audit Planning Models Using
Uncertain Reasoning" won the 1997
Outstanding Dissertation Award at the University
of Kansas. His current research is studying the
use of uncertain reasoning techniques in audit
planning models, and the representation of
causality in auditing using event spaces.
|
|
 |