PG6680/UG7156 Business Intelligence Systems: ES, DSS, and Models

This report provides an analysis of expert systems (ES) and decision support systems (DSS) within the context of business intelligence. It explores the applications of expert systems, their advantages in correcting mistakes compared to conventional programs, and their limitations, particularly in the domain of knowledge, assessment, and time pressure. The report also presents a rule-based expert system example concerning interest rates and bond prices, demonstrating forward chaining. Furthermore, it discusses the development of a DSS for a university, highlighting its use of expert systems for student guidance and course selection. Finally, it examines data-driven and knowledge-driven DSS models for error detection during university enrollment, focusing on verifying student data and prerequisites. This document was contributed by a student and is available on Desklib, a platform that offers a range of AI-based study tools and resources for students.

Running head: BUSINESS INTELLIGENCE SYSTEM
BUSINESS INTELLIGENCE SYSTEM
Name of the Student:
Name of the University:
Author Note

1
BUSINESS INTELLIGENCE SYSTEM
Answer to question number 1
A computer program which is able to mimic decisions made by human experts is known
as an expert system. These type of systems are capable of teaching, advising and taking expertise
decisions by themselves without eternal human interference. The environment for the
development of this type of this type of applications involve the actives which are acquired
necessarily and the representation of the knowledge for making explanations and the inferences.
The ES is applied in various type of programing conventions and has several other applications.
Among the applications the Rule-base ES deserve a special mention. This type of application is
represented along with a series of rules. The frame based systems have knowledge and the term
comes from a series of frames which are taken into considerations. In hybrid systems the
application of fuzzy logic and neural networks are done. In the model based system the system
under study is structured with model from the functions of the ES. An unutilized prepacked
software is known as the readymade software and the real times system are developed to produce
the users the real time updates.
The best application of the ES however remains the fuzzy network and neural network.
The neural networks are developed with the help of the fuzzy logic where the machines are made
to understand a particular set of instructions several times and then the machine provides the
users with the result just by reading the input. This applies ES to a huge extent.
As it is already known that the human expert is mimicked by an expert system it is worth
noting that the system is also capable of making errors that a human expert might do. However,
this occurs as there are wrong rules defined in the system and hence, a wrong diagnosis in the
system occurs. There is also situation when an ES is made to work with the incomplete

2
BUSINESS INTELLIGENCE SYSTEM
information and this makes them provide the probabilistic answers to the users. This is because
often the ES is often not sure about the results of the diagnosis and often they are unable to
interpret a particular rule as stated by the users. For instances it can be considered that the rate of
malfunction can be measures with the certainty factor of 0.8 as per confirmation. It can be then
easily inferred that the system can provide incorrect results 20% of the time to the users. In
addition to providing wrong answers and probabilistic answers the system can also ask the wrong
question to the users at times and this results in yielding the in correct result for the users.
However, the separation of the inference engine from the knowledge base is very
essential and it is important the knowledge is organized in a non-sequential manner and hence, it
is easier to correct the mistakes once they are detected provided that the rules applied for the
system are incorrect.
Answer to question number 2
Part 2.1
Fact: dollar is declining.
Forward: Then, interest rate is up;
[ Rule 2: IF interest rates increase,
THEN bond prices will decline.]
then, bonds price is down.
[ Rule 5: IF the dollar falls,
THEN interest rates will decrease.]

3
BUSINESS INTELLIGENCE SYSTEM
Conclusions: take loan.
Therefore, the client would be recommended to take the loan.
Part 2.2
Fact: unchanged interest rate.
Using forward chain:
[ Rule 3: IF interest rates are unchanged
THEN bond prices will remain unchanged.]
However,
[ Rule 6: IF bond prices decline,
THEN buy bonds.]
Conclusion: Since bonds price is not declining; do not buy bonds.
Therefore, the client would be advised not to buy bonds until the bond price declines.
Answer to question number 3
Part a
DSS or Decision support system is basically a system which is based on the interaction in
between the software so that fruitful result can be yielded effectively from the system. This type
of systems is generally used by the managers of the large organization who take decision after
accessing a large amount of information which are being generated from the different types of
information system which is connected to the central system. The summary of information,

4
BUSINESS INTELLIGENCE SYSTEM
patterns, exception and trends are used by an DSS algorithm for development of the model. The
system is used for the decision making however, the decision is shown by itself instead the user
is advised about the decision. The system compiles very useful data from the raw data, personal
knowledge and documents of the related business for the identification and the resolution to their
problems. The DSS for a university can be developed for the students, so that they can be
mentored and guided in to a course of their choice.
The decision system uses an Expert system except the fact that it lets its users make their
own choice. In a system developed for an university the decision support system would be
making use of expert system to analyses the merit of the student and offer a range of subject to
the students. The student who goes for an enrollment here would be helped in selecting a subject
and the course of the education would be suggest. The student themselves would then be doing
the enrolments accordingly.
Part b
The expert system which would be used for the education management of the student is
fed with data of the student such the past educations, details of the students age, number of hours
of study every day and score of every test that the student gives, Further the systems would be
using the infractions such as average time required by a particular student to complete a
homework, average time required by a student to complete a test. All this would be resulting in
testing the aptitude of the student and decide if the student is ready for the next lesson.
There are certain number of limitation to the Expert System designed for the university:
 Domain of knowledge: The ES works well within only a narrow and restricted domain of
knowledge. However, education system is huge therefore the use of ES can be limited.

5
BUSINESS INTELLIGENCE SYSTEM
 Frequently multiple correct assessments: A student can be excelling at many different
course hence, it might be difficult to select a particular course for the ES.
 Time pressure: The amount of time required to build such is system would be huge a
hence, it would be impossible to incorporate all the functionalities within a set of
allocated time.
Answer to question number 4
There is different type of models that can be deployed for the university in the process of
their enrollment. The task at hand currently is to verify the different type of errors that can be
detected from the enrollment process and two models which can be deployed for the error
detection process are:
Data-driven DSS: In the data driven DSS the administrators and the managers of any facilities
are being taken into consideration. It can be used for querying the data warehouse for the DSS
and specified result can be obtained from the queries as per the requirements. Deployments for
these type of models are generally done with the help of the main frame servers and the client
servers for the system. This model would be helpful in the data storage of the students who are
making the enrollments in the system. The previous study history and in some case the medical
histories of the students would also be checked. The prerequisites certifications and the
performance logs for the students can be fetched into this and thus detection in any type of error
during the enrollment can be detected. In situations where the student has to merit to study a
particular subject but doesn’t not pass all the criteria for an admission to that subject. The system
would then be able to query through the education history of the student and detect the error.

6
BUSINESS INTELLIGENCE SYSTEM
Knowledge-driven DSS: The knowledge based DSS cover a huge range of area and a large
number of students and universities are used as study objects for the system. They are able to
include additional objects such as the ones which are interacting with the university as well.
These objects are the potential students who are looking to get admission to the university and
the students who have already qualified from the university. The system can be used in the
enrollment system which would be useful for guiding the prerequisites of the students. The
system would be fetching data about the students who have taken enrolment for subjects within
the university and have given test on those subject. A link can be easily found in tread that the
student is taking up subject and the overall results achieved by them. For instance, student is able
to clear all the prerequisites and fail the test after enrollment an error takes place. This can be
detected easily by the DSS. The system would be able to refine the prerequisites and detect
errors early on during the merit test for the student aiming enrollments.