SRELS Journal of Information Management

Vol. 42, No. 1, March 2005, Paper A. p3-11.

EXPERT SYSTEMS IN LIBRARY AND INFORMATION

SCIENCE
Pratibha A. Gokhale*
Expert Systems have evolved as one of the most successful branch of Artificial
Intelligence during last two decades. The fifth generation computers developed a
potential for natural language processing and reasoning capabilities in a much better way
than their precursors. Areas in man - machine interfaces developed into knowledge based
expert systems. The present paper gives a short overview and a background of Expert
Downloaded From IP - 210.212.129.125 on dated 31-May-2013

Systems, their features, components involved, and applications in Library and

Information Science. It brings out the future challenges which are likely to be faced by
information professional with the intelligent use of Expert Systems.
Members Copy, Not for Commercial Sale
www.IndianJournals.com

KEYWORDS/DESCRIPTORS: Expert systems, Library and information science

1 INTRODUCTION
Expert systems have evolved from a long tradition of Artificial Intelligence.
They are devoted to programming computers to carry out tasks that would require
intelligence if carried out by human beings. Computers have long been known for
their incredible capacity to carry out huge and complicated calculations, different
types of combinations, logical and sequential operations in a very short time. But
when it was realized that computers are not just giant calculators but they can be
made to think as intelligent systems, Artificial Intelligence (AI) evolved as a
separate discipline in computer science. Thus AI is the computer system that
exhibits the characteristics associated with human intelligence such as
understanding language, learning, reasoning, solving, problems etc. Robotics is
one example of AI.
Expert systems belong to broader discipline of Artificial Intelligence. They
are the computer based systems that use knowledge and reasoning techniques to
solve problems that would normally require human expertise. An expert is
supposed to have knowledge and hence expert systems are also knowledge based
systems. However, they are broader systems. An expert system is a computer
system designed to stimulate the problem solving idea of a human, who is an

* Reader, Department of Library and Information Science, University of Mumbai, Kalina,

Vidyanagari, Santacruz (East), Mumbai 400 098, INDIA. E-mail: pratibha_gokhale@yahoo.
com.

Vol. 42, No. 1, March 2005 3

Gokhale

expert in a narrow domain or, it is a computer system that incorporates human

expertise. If knowledge on a particular subject is to be developed, computer will
process all the available knowledge and the Expert System in the computer will
analyse the knowledge and provide probable solutions to a problem. An expert
system is regarded as the system in the computer to express knowledge based
component in such a way that it can offer intelligent advice or can take an
intelligent decision about something. It is the capacity of the system to justify its
reasoning in a manner, the enquirer needs it.
The expert systems are limited to specific 'domain' or areas. They are drawing
conclusions and inferences in the form of advices or probable solutions, having a
clear separation of knowledge and inference mechanism. They are capable of
Downloaded From IP - 210.212.129.125 on dated 31-May-2013

solving problems of categories such as interpretation, predication, diagnosis,

planning, instructions etc.
Members Copy, Not for Commercial Sale
www.IndianJournals.com

The database construction in an expert system is carried out by a trial and

error process known as Knowledge Engineering. Expert systems bring together
many skills and knowledge from both computer science and the experts' speciality
domain. It is codified and stored.
2 EVOLUTION
The information processing behaviour of human beings was studied by
mathematicians. It was realized that people use specific knowledge based on
experience, intuition, observation, as a basis for solving problems. People also
simultaneously think forward and backward and try to keep track with the current
status of problem. This type of thinking forward and backwards from the goal
they are trying to reach, is very typical to human intelligence. This technique is
used today by knowledge engineers to create expert systems to solve complex
problems with the help of computer.
The first expert system developed was DENDRAL at Stanford University in
1965. DENDRAL was able to determine the chemical structure of unknown
organic compounds by analysing the mass spectrometry data. DENDRAL has
been successfully used by many chemists. Feigenbaum and Lederburg of Stanford
University developed a combination of rules and heuristics (intuition + rules of
the thumb + guess techniques) for interpreting spectrographic data. The result was
the program DENDRAL, which is still being used today in laboratories all over
the world. It marked the beginning of new branch of enquiry for researchers in AI
- that is Expert Systems.

4 SRELS Jl. Info. Manage.

Expert Systems in Library and Information Science

Following the success of DENDRAL, the same research team produced

MYCIN. MYCIN is an expert system designed to deal with problems in the
diagnosis and treatment of infectious blood diseases.
Some other Expert Systems are:
PIP: To record history of patients with oedema
CASNET: For diagnosis and treatment of glaucoma
MYCIN: For diagnosis and treatment of blood infections
PROSPECTOR: To assist geologists in the discovery of mineral deposits.

3 REASONS FOR DEVELOPING EXPERT SYSTEMS

Downloaded From IP - 210.212.129.125 on dated 31-May-2013

• Human experts may retire, join another firm;

Members Copy, Not for Commercial Sale

• They can only be in one place at one time;

www.IndianJournals.com

• They require considerable time in training;

• They may be difficult to interact; and
• They may be emotion bound, subjective, forgetful.

4 FEATURES OF EXPERT SYSTEMS

• The expert system is always and instantly available. Skills of several
experts can be combined;
• It is logical, objective and consistent;
• It does not forget or make mistakes in calculations;
• It is directly and instantly available for consultation. Access is not
limited by time and location;
• It is a knowledge bank of sustainable value;
• It is useful in training programmes;
• Human beings understand and appreciate the difference in certain
cultural factors;
• Human experts are generally aware of the scope and limitations of
their knowledge. E.S.'s do not know what they do not know;
• When faced with an unusual situation - human experts develop entirely
new, original approaches; and
• Human experts are far more flexible than computers.

Vol. 42, No. 1, March 2005 5

Gokhale

Expert Systems are not alternative to computer programming, but they are an
approach to decision making or a methodology in the support of decision making.
The rule based formats for knowledge representation are used in expert systems.
The rules within such a format are represented by a set of 'If'-'Then' statements. 'If'
a certain condition exists- 'then' take a certain action.
For Example:
• 'If' you must avoid whole milk.
• 'and' you dislike skimmed milk.
• 'or' you wish to reduce cholesterol.
• 'Then' consider soya milk.
Downloaded From IP - 210.212.129.125 on dated 31-May-2013

The text includes a software in the form of a floppy disk or a CD and a users'
Members Copy, Not for Commercial Sale

manual. It may be used on an IBM PC or compatible machine.

www.IndianJournals.com

5 LANGUAGES OF EXPERT SYSTEMS

The four general categories of programming languages used in expert systems
are:
• List Processing Language : LISP
• Logic Programming Language : PROLOG
• Object Oriental Language : Small talk (SMALLTALK)
• Conventional Programming Language : Pascal, Fortran, C, Basic
The present trend in E.S.'s is towards conventional ones- 'C'++ in particular.
6 COMPONENTS OF EXPERT SYSTEMS
The following elements are involved:
1. Knowledge base: It contains the knowledge associated with a specific
domain-knowledge like a human expert may have about a problem
domain. It is the heart of any expert system. It is represented in the form of
rules, frames, scripts, decision trees and semantic nets.
2. Interface or user interface: A reasoning mechanism or interface - which
enables the system to apply the knowledge in the knowledge base to reach
a decision. It provides a link between the developer and the external client.
3. Inference mechanism: This mechanism is responsible for explaining the
user how to actually solve a problem. If a particular question is asked by

6 SRELS Jl. Info. Manage.

Expert Systems in Library and Information Science

the user - it provides answer. It does this by using a set of algorithms or

decision making strategies to generate inference from the facts and
heuristics held in the knowledge base control strategies or reasoning
strategies are applied at each stage to decide what operators should be
applied. This is a software which will process the knowledge, analyse it
and draw conclusions.
4. Global database: It is the section of the program that keeps track of the
problem by storing data, facts obtained from external sources. It offers
links with other systems and will support the main expert system and will
have interaction with outside world.
7 CONSTRUCTION OF EXPERT SYSTEMS
Downloaded From IP - 210.212.129.125 on dated 31-May-2013

1. Identification: It involves determining the nature of the problem domain,

Members Copy, Not for Commercial Sale

clarifying its scope, contacting the experts to define framework.

www.IndianJournals.com

2. Conceptualisation: It is making the key concepts and relations in the

domain explicit and elaborate. Experts and knowledge engineers co-
operate to describe the problem solving process.
3. Formalisation: At this stage, appropriate languages and expert system
tools are selected. Concepts and relations are mapped together for
knowledge representation.
4. Implementation: A prototype system, combining knowledge base,
inference engine and associated control structures is developed.
5. Testing: The prototype system is evaluated. Failures are analysed and
necessary corrections are done.
8 APPLICATIONS OF EXPERT SYSTEMS
Expert Systems have been made to use in many areas such as:
• Industrial applications: Large industrial and manufacturing companies
have introduced expert systems for:
∗ Fault diagnosis of the machinery, plants, circuits etc.
∗ Designing
∗ Military operations
∗ Trouble shooting mechanisms, air traffic control, production control.
∗ Software
∗ Construction activity

Vol. 42, No. 1, March 2005 7

Gokhale

• Financial applications: Banks, financial institutions, insurance companies

use expert system:
∗ to give advice on investment, stock exchange market, shares and
securities regulations
∗ to assess insurance claims
∗ to assess and predict business opportunity, etc.
• Social Applications: For the benefit of common lay persons, to give advice
and assistance for:
∗ Employee's rights
Downloaded From IP - 210.212.129.125 on dated 31-May-2013

∗ Guidance on legislation and practice relating to social security

∗ Assistance on matters related to property, housing, travelling
Members Copy, Not for Commercial Sale
www.IndianJournals.com

• Global News agencies like Reuter, CNN have to handle huge news
reports. They have developed topic identification systems.
9 EXPERT SYSTEMS IN LIBRARY AND INFORMATION SCIENCE
Every Knowledge Based System is a software. The system can directly
communicate with the database vendors. National Library of Medicine - USA,
Library of Congress have done lot of work in the area of E.S.'s for library
science. AACR II contains more than 2000 rules.
9.1 In Cataloguing
Cataloguing rules are very much adapted to expert systems. The AACR II
rules are structured on a comprehesive basis and they are coded. There are nearly
2000 rules in AACR II revised edition. The cataloguing activity is a rule based
activity. Some expert systems developed in cataloguing are:
• Prototype expert cataloguing system developed at the Exerter University
on the PROLOG language for rules of bibliographic description and for
determination of access points. It also converts the catalogued records
on MARC format for standardization.
• AutoCat prototype for cataloguing physics journal articles was
developed in Germany.
• Prototype Expert System MAPPER: for cataloguing of maps based on
AACR2. This is developed at University of California at Los Angeles.

8 SRELS Jl. Info. Manage.

Expert Systems in Library and Information Science

•CATALYST: developed by researchers at Strathclyde University to

advice on the choice and form of access points.
• MITINET for converting records into MARC format.
9.2 In Reference Service
• Answerman: Developed by the National Agricultural Library. The
knowledge base is composed of facts and documents from agriculture. It
also links the user to standard online databases like AGRICOLA.
• POINTER: Developed by Library of Congress for reference service.
• COMIT: Cherie B. Weil developed a system by using symbols and
LISP to get answers to bibliographic reference questions.
Downloaded From IP - 210.212.129.125 on dated 31-May-2013

9.3 In Referral Service

• PLEXUS is a prototype, micro computer based expert system for referral
Members Copy, Not for Commercial Sale

service developed by Central Information Service of University of

www.IndianJournals.com

London as a part of a project funded by British Library.

9.4 For Online search
• TOMSEARCH: the above team of University of London in co-operation
with European Space Agency has developed an expert system to serve
as a tool for retrieval of information from online databases. It is claimed
as world's first programme for information retrieval to use the
technology of natural language processing in PC environment.
There are expert systems available for selection of database, to help frame the
query for search formulation. Cataloguing rules are very much adaptible to expert
systems. Catalogue records are converted to MARC format by E.S.'s. The
computer automatically recognises author, title, imprint. It scans the title page - if
there is a difference in font, style -it recognizes the changes, realizes the gaps. It
then applies next rule. Therefore, it is rule based system.
10 OTHER POTENTIAL AREAS IN LIBRARY AND INFORMATION
SCIENCE
The expert systems have been increasingly used in industry sector and now
they are used in service sector (ISO 9004). There exists lot of potential for their
use in many library activities, such as:
Administration - They can answer complex administrative problems,
workout, holidays/leave management etc.

Vol. 42, No. 1, March 2005 9

Gokhale

Staff management - Placement, for deciding staff ratio, hiring or

contracting norms for people, preparation of duty
charts etc.
Collection development - They can help to form decide policies with respect
to gifts, donations, weeding out practices, collection
building.
Classification - Useful in classifying multi volume works,
encyclopedias, monographs, serial publications.
- It can avoid inconsistency in classifying documents,
useful while classifying series publications.
SDI - SDI profile can be done with USERS’ profile, their
Downloaded From IP - 210.212.129.125 on dated 31-May-2013

subject interest and selection of documents.

Membership - Library membership rules for different categories
Members Copy, Not for Commercial Sale

such as students, faculty, institutions can be

www.IndianJournals.com

subjected to expert system to streamline the

procedures. This can take care of rules of fines,
deposits, returns.
Stock-taking - Policies and rules may be explicitely framed for
stock taking activity using expert systems.
11 LIMITATIONS OF EXPERT SYSTEMS
• They are rule based systems. The rules are represented by a set of 'If'-
'then' statements. e.g. If a certain condition exists, then take a certain
action;
• They are not alternative to computer programming but they are used as a
methodology in decision making;
• They work in restricted domain;
• Expertise needs to be gathered and codified which is in the form of facts
and rules; and
• They work on principle of knowledge base, user interface and inference
mechanism.
12 CONCLUSION
Many automated systems have helped in reducing time for repetitive or
routine work. If used properly, mundane queries can be answered quickly by
such systems. There will be less backlogs as the processes will be speeded up.

10 SRELS Jl. Info. Manage.

Expert Systems in Library and Information Science

Standardization has an important role to play in developing expert systems. Users

will increasingly expert adoption of de facto (italics) standards in interface design.
Interfaces in expect systems for multimedia systems including text, graphics,
sound are likely to appear in near future. Advances in hardware, processing power
of the software will support the interfaces. The major challenge for the
information professionals will be to customize the interfaces to suit the needs of
their users and modify an expert system to meet the local needs. It has certainly
opened up new avenues in the field of Library and Information Science.
13 SELECT READING
1. Bawden (D). IT Interfaces. In Handbook of Special Librarianship and Information Work. 6
Ed. Edited by Petti Dossett. London, Aslib, 1992, p441-468.
Downloaded From IP - 210.212.129.125 on dated 31-May-2013

2. Davis (R); James (B. Chichester). Intelligent Information Systems, Progress and Prospects.
Ellis Harwood, 1987.
Members Copy, Not for Commercial Sale

3. Morris (A) Ed. The Application of Expert Systems in Libraries and Information Centres.
www.IndianJournals.com

London, Bowker - Saur, 1992.

4. Rashid (A). Library and information applications of knowledge based systems. Library
Science. Vol. 32(3); September 1995; p122-132.
5. Sowell (S L). Expanding horizons in collection development with expert systems. Special
Libraries. Vol. 80; 1989; p45-50.

Vol. 42, No. 1, March 2005 11