THE DEVELOPMENT OF A KNOWLEDGE SYSTEM FOR
ISO 9001 QUALITY MANAGEMENT
Li-Yen Shue
Information Management Department
National Kaohsiung First University of Science and Technology
1 Uni. Rd. YenChao 824 Kaohsiung, Taiwan
Sheng-Tun Li
Information Management Department
National Cheng Kung University
No.1, Ta-Hsueh Road, Tainan 701, Taiwan
Hsun-Cheng Hu
Institute of Information Science
Academia Sinica Taiwan
128, section 2, Ein-Gio Rd, Nan-Kang, Taipei,Taiwan
Keywords: Ontology, Knowledge Management, Ontology Engineering, Document Management, Semantic Search.
Abstract: Many researchers in knowledge management point out that the first step toward knowledge management is
the management of documents. However, the complexity imbedded in some documents could present great
difficulty for most methodologies to deal with. The knowledge content for building an excellent quality
management system that complies with ISO 9001 falls into this category; this knowledge is characterized by
multi-dimensionality and knowledge embedment through various procedures and forms. We applied
Ontology, which is a new approach in AI for better presenting knowledge structure of a domain, to develop
a system structure that will facilitate the development of a knowledge-based ISO 9001 quality management
system. We use the real case of a Taiwanese chemical company that has a total of 175 ISO manuals to refer
to when needed. This system is built with Protégé 2000 as the knowledge platform, and we follow the
development process recommended by Ontology Engineering of Toronto Virtual Enterprise. One main
feature of the system is its capability of understanding the semantic of documents, which is a vital part of
the inference mechanism in answering user’s queries.
1 INTRODUCTION
Many researchers in knowledge management
indicate that the first step toward knowledge
management of an organization is the management
of documents. The conventional document
management system is capable of finding lots of
related information, but it may not be able to provide
user with information that can satisfy users’ exact
needs. A great deal of efforts has been made on
developing methodologies to overcome this
problem. What seems to be lacking, however, is the
ability for systems to understand the semantic of
documents. Researchers in AI have recently tuned to
ontology as a better approach for expressing shared
conceptualization of a domain through knowledge
structure and contents. The complexity and difficulty
for business to manage and control procedures and
documents to meet requirements of ISO 9001 has
262
Shue L., Li S. and Hu H. (2004).
THE DEVELOPMENT OF A KNOWLEDGE SYSTEM FOR ISO 9001 QUALITY MANAGEMENT.
In Proceedings of the Sixth International Conference on Enterprise Information Systems, pages 262-268
DOI: 10.5220/0002627402620268
Copyright
c
 SciTePress

prompted us to study the application of ontology in
this area.
ISO 9001 (Goetsch and Davis, 1998) contains
quality documents that represent the standards
required by the International Standard Organization
for building an excellent quality management
system, and its accreditation is sought by most
organizations worldwide; especially those in
international trade. These documents define
requirements and standard procedures that are well
recognized as the best practices in quality
management. In a sense, these procedures represent
the knowledge contents necessary for building an
excellent quality management system. Due to its
cross-departmental nature and the depth in
penetrating various levels of management, these
documents can be overwhelming for most
management to remember. In addition, the amount
of ISO 9001 contents may increase year after year; it
adds a further problem to management to keep track
of them.
The objective of this study is to develop an
ontology-based system structure that will facilitate
the development of an ISO 9001 knowledge
management system. This system could help
enterprises to overcome the difficult in managing
and control the contents of an ISO-based quality
system. We used information of a Taiwanese
chemical company for the development, and applied
TOVE (Toronto Virtual Enterprise) ontology
engineering approach to construct the ISO ontology
on Protégé 2000 (Gennari, 2003), (Noy, 2001),
(Grosso, 1999), which was developed by Stanford
Medical Informatics. The rest of this paper is
arranged as follows. Section 2 provides a brief
introduction of ontology and ontology engineering
approaches. Section 3 examines the details of ISO
9001 documents and look into the issues associated
in implementation. Section 4 introduces the system
architecture. Section 5 explains the development of
the ISO 9001 ontology, and is followed by the last
section of conclusion.
2 ONTOLOGY AND ONTOLOGY
ENGINEERING
Ontology is a “formal explicit expression of a shared
conceptualization of a domain” (
Uschold and
Gruninger
, 1996), it is one of the latest research
frontiers of AI in searching for better knowledge
representation of a domain, so that it can be easily
shared and reused. In actual application, ontology
consists of a set of vocabulary and the content theory
to express entities and relationships between entities
in a domain, which are normally expressed with:
classes, slots, instances, and axioms. Classes
represent the conceptual items of the domain. Slots
are the relations or attributes of classes. Instances are
the data that belong to classes and describe the
objects in real world. Axioms are inference logics
that serve as the reasoning mechanism.
There are two major approaches in developing
enterprise ontology. The first one is proposed by M.
Ushold et al. (
Uschold and King, 1995), their ontology
engineering is based on the experience of the
Enterprise Ontology. The other is proposed by M.
Gruninger et al. (
Gruninger and Fox, 1995), (Kim and
Fox, 2002
), (Kim, 2002), (Kim, Fox, and Gruninger,
1999
), which is based on the experience of the
TOVE (TOronto Virtual Enterprise). Based on the
evaluation by Fernandez Lopez (
Fernandez, 1999), the
former approach is more for the modelling of the
operation of an enterprising, and the latter is more
for the modelling of knowledge content of a domain,
hence, one could conclude the latter can better
accommodate needs of knowledge management
users for our case. As a result, we decided to apply
M. Gruninger et al.’s approach in this research; it
will be termed TOVE ontology engineering in the
later sections.
TOVE ontology engineering has six design
phases: motivation scenario, informal competency
question, terminology, formal competency question,
axiom, and completeness theorem. The motivation
scenario describes problems in application of a
domain that motivate the application of ontology; it
may also provide an expected solution to the
problems. The informal competency questions
transform motivation scenario into question forms,
which the system must answer when completed. The
terminology phase defines vocabularies and their
meaning that are to be used in the ontology, which
include all terms used for expressing knowledge
content. The formal competency question applies
terminology to formalize informal competency
questions in natural language forms the system can
understand. The axiom phase defines the inference
logics to lay the foundation to facilitate search
mechanism when the ontology is completed. The
last phase, completeness theorem, is to demonstrate
that the ontology can correctly answer all of the
competency questions. Kim (
Kim, Fox, and Gruninger,
1999
) redefines the six phases into Motivational
Scenario, Informal Competency Questions, and
Ontology; with the Ontology phase being further
broken down into 5 sub-phases: terminology,
hierarchical model, predicate model, formal
competency question, and axiom. While the
terminology, formal competency question and axiom
remain the same as before, hierarchical model
describes the relationships of all terminologies in a
hierarchical scheme, and Predicate model defines
THE DEVELOPMENT OF A KNOWLEDGE SYSTEM FOR ISO 9001 QUALITY MANAGEMENT
263

terms and relations of the hierarchical model in the
format of first-order logic to further facilitate system
development.
3 ISO 9001
The ISO 9001 standards are a set of international
quality management system standards and
guidelines, which was developed by International
Organization for Standard for specifying
requirements to achieve product conformance by a
quality management system. While these
requirements are generic and independent of any
specific industry, they serve as the goals for the
functioning of the overall quality management
system and its various subsystems, which are to be
developed and implemented by the management of
enterprises. Each of these systems must be well
documented in terms of its quality objectives,
processes it serves, procedures required for each
process and between them, and forms needed to
verify performances. Basically, the documents of a
quality management system of an enterprise consist
of four major categories: quality policy, quality
manual, process procedures, and quality records.
Figure 1 shows the relationships of the four
categories inside the triangle.
Figure 1: The four ISO 9001 document components
From top to bottom, they are quality policy, quality
manual, process procedure, and quality records
(Goetsch and Davis, 1998). Quality policy sits at the
top level, it defines commitments to quality by top
management, authorizes the organization to comply
with requirements and continually improve, and
provides framework for setting quality objectives of
various levels of management. Quality manual is
level 2, it defines and describes the scope of the
quality management system, describes how
processes interact to form the quality management
system to insure quality assurance of the process,
and achieve stated objectives in level 1. The process
procedure, based on the scope and process
interactions stated in quality manual, describes the
procedures for planning, operation, and control of
each process as well as that for the interaction of
processes. Quality record consists of sets of forms,
tables, or records, which serves to record actual data
of process procedure of level 3 as a proof of the
quality.
In terms of functionality, ISO 9001 covers five
major functions: quality management system,
management responsibility, resource management,
production realization, and measurement, analysis
and improvement. As shown in figure 2, these five
functions are closely interrelated.
Quality Management System
Management Responsibility
Resource Management
Product RealizationMeasurement, Analysis and Improvement
Organization
Structure
Figure 2: Interrelationships among five functions
A quality management system is built upon the
structure of a company and must take into
consideration responsibility of various levels of
management. Management responsibility, on the
other hand, must refer to the managerial structure
and take into account system requirements at the
same time. These two functions will in term affect
the other three directly and/or indirectly as shown in
the figure. These inter-webbed relationships and the
fact that each individual unit must develop its own
ISO documents, especially those that are related to
quality records, process procedures, and quality
assurance, could lead to the serious problem of
inconsistency in contents and terminologies, because
different departments may use different terms for the
same meaning and vice versa. In our case, the
company has developed a total of 175 ISO
documents that must be referred to from time to time
when needed and the inconsistency is not effectively
dealt with. Another problem that is fronting the
management is the fact that new issues and
requirements are continually being developed by
ISO, and that is becoming a major burden for many
enterprises to maintain up-to-date requirements, let
alone implementation. It is no wonder that Goetsch
ICEIS 2004 - ARTIFICIAL INTELLIGENCE AND DECISION SUPPORT SYSTEMS
264

and Davis’s study concluded that it is very hard to
understand all ISO 9001 requirements for most
enterprises; because there are too many
interrelationships among all requirements.
4 SYSTEM ARCHITECTURE
We employ Protégé 2000 as the platform for
building the ontology-based quality management
system (OQMS). Protégé 2000 is an ontology-based
knowledge acquisition platform that can better deal
with the issues of implicit relations, terminologies,
document management, and unit integration. In
addition, Protégé 2000 complies with Open
Knowledge Base Connectivity Protocol, so that
future integration with other systems would not be a
problem. The proposed system architecture is shown
in figure 3, which is composed of eight modules:
User Interface, KB maintenance Interface, PAL
Inference, Query Tab Query, Unit Inference Engine,
Protégé 2000, Company Documents, and ISO
ontology. Initially, knowledge engineers analyze
knowledge contents from various ISO documents,
and identify knowledge entities, their attributes, and
their relationships with others to build the ISO
ontology, which will be maintained by the
knowledge engineers through the KB maintenance
interface. The relevant company information and
forms are collected and encoded as instances of the
ontology. OQMS provides two types of query: PAL-
based and Query Tab-based. The PAL-based query
allows users to issue queries using the PAL (Protégé
Axiom Language) logic form (Hou, Noy and Musen,
2002); its editor will translate those queries into
PAL language for processing. The Query Tab is for
general users, it provides a template query mode to
allow users to fill out some specific fields. The PAL
inference engine is the inference engine of our
system. When performing reasoning instructions
upon requests, it will refer to entities and their
relationships of the ISO ontology for reasoning, and
retrieve related documents from ISO Document.
Protégé 2000 is the core module of our system. It
provides a platform for saving identified knowledge
structure of ISO documents and relevant company
information; including various forms and detailed
operation information. Users have the option of
using the default interface of Protégé 2000 for
maintaining ISO ontology and Company documents.
ISO ontology represents the overall knowledge
structure of all ISO 9001 documents. This structure
contains knowledge identities, their attributes, and
their interrelationships. Finally, ISO document
consists of original contents that support the
knowledge structure of the ISO Ontology. The
contents may include company documents, various
forms, detailed operating information, and other
relevant documents.
5 ISO 9001 ONTOLOGY
The ISO 9001 Ontology is the core of the whole
system, we follow TOVE ontology engineering for
Figure 3: Architecture of the quality management system
THE DEVELOPMENT OF A KNOWLEDGE SYSTEM FOR ISO 9001 QUALITY MANAGEMENT
265

the development, and the following explains the
various phases:
5.1 Motivation Scenario
It is the complexity and multi-dimensionality that is
making ISO documents difficult to comprehend and
manage while trying to build an excellent quality
management system. The ontology we propose must
be able to satisfy the needs of management in
meeting requirements of ISO 9001, which include:
(1) Identifying ISO requirements at different
levels, so that necessary procedure knowledge can
be provided to meet these requirements.
(2) Defining terminologies of the overall system
through a multi-naming mechanism, so that each
term can have its precise meaning to achieve
vocabulary consistency.
(3) Managing large number of documents to aid
the needs of implementation and references.
5.2 Informal Competency Questions
From motivation scenario, we identify the following
six informal competency questions for our case:
(CQ1) For a given job, what are the QA procedures
that are needed to complete the job?
(CQ2) For a given job, what are the departments that
are involved with the job?
(CQ3) For a given job, what are the quality records
that are needed to complete the job?
(CQ4) For a given QA procedure, what are the
departments that are involved with it?
(CQ5) For a given QA procedure, what are the
quality records that needed to support it?
(CQ6) For a given unit, what are the QA procedures
that are under its administration?
5.3 Ontology
As stated above, there are five steps in establishing
the ontology: terminology, hierarchical model,
predicate model, formal competency question, and
axiom. However, due to space limit, we will only
present part of hierarchical model and sample of
formal competency question and axiom.
5.3.1 Hierarchical Model
Hierarchical model defines relationships among
terms of the system. The hierarchical model of the
organization of our case is shown in Figure 4, which
indicates that there are five types of grouping:
executives, division, department, team, and group. In
this case, we classify all relations into two
conventional relationships: is-a and has-slot. The is-
a relation is applied to relate class and subclass, and
has-slot relation is applied to all other relations.
There are five “is-a” relationships, which are in
dotted lines, to relate groupings to Organization, and,
there are six “has-slot” relationship, which are in
solid lines, to relate among themselves. Figure 5 is
the hierarchical model for QA Procedure, Process,
and Quality Record. A QA Procedure may contain
more than one process; hence the “is-a” relation is
applied between them. The same is true between
Process and Operation Process and Decision Process.
In addition, the “Operation Process” is related to
“Process” with “Preceding Process” and
“Succeeding Process” to indicate the processes
before and after the given one of the process chain.
Similarly, the “Decision Process” is related to
“Process” with “Preceding Process”, “Succeeding
Process without modification”, and “Succeeding
Process with modification”. When performing a
process, employees may need to use one or more
quality forms as records, we use meta-class
“QUALITY-DOC” to relate Process and Quality
Record with “Use” and “Used by” as the two has-
slot relationships. All quality records are instances
of “QUALITY-DOC”, and each quality record may
belong to more than one group of measurement
records, which are subclasses of Quality Record, and
has an “is-a” relationship with “QUALITY-DOC”.
5.3.2 Predicate Model
Predicate Model is the formal definition of classes
and their relationships in First-Order Logic. Our
system has in total 213 classes, and 723
relationships. In the following, we present only
sample predicates.
(1) R: Filled Quality Record
Figure 4: The hierarchical model of the organization
ICEIS 2004 - ARTIFICIAL INTELLIGENCE AND DECISION SUPPORT SYSTEMS
266

Every filled quality record is an instance of
the quality record that is a subclass of
“Quality Record”.
(2) T: Empty Quality Record
The “Quality Record” is an abstract class;
every R is an instance of “QULAITY-DOC”,
and R has three common own-slots:
preservation period, responsible department,
and processes that need it.
(3) P: QA Procedure
P is an instance of “QA Procedure” class, it
has four own-slots: name, related process, ISO
code, version, and content.
(4) U: Organization
The “Organization” is an abstract class and
has five classes: Executives, Division,
Department, Team, and Group. U can be an
instance of each of the five classes.
Some of their related expressions in predicate
model are:
PT-1 Manage(U
1
, U
2
)
This predicate is valid if U1 is the superior of
U2.
PT-2 QA-Procedure-Including-Process(P, OP)
This predicate is valid if a given QA procedure
P includes the process OP.
PT-3 Process-Requiring-Quality-Record(OP, T)
This predicate is valid if the process OP requires
filling quality record T.
5.3.3 Formal Competency Question
The above definition of predicate model can be
expressed with first-order logic language to
formulate the formal competency questions, some
examples are:
(CQ1) When performing a job, which QA
procedures are related to this job?
P; QA-Procedure-is-related-to- Keyword (P,
“Job”)
(CQ 2) When performing a job, which departments
are related to this job?
U; Department-is-related-to- Keyword (U,
“Job”)
5.3.4 Axiom
We present following axioms that could assist in
discovering relationships among QA processes,
quality record, and department in order to solve the
two competency questions:
Defn-1 QA-Procedure-is-related-to- Keyword(P, K)
(
P:QA-Procedure [ ( (substring-of (K, (name
P))) ( OP:Process
( (QA-Procedure-Having-Process (P, OP))
( (substring-of (K, (name OP)))
( T:QUALITY-DOC
( (Process-Using-Quality-Record(OP, T))
(substring-of(K, (:NAME T)))))))))
QA-Procedure-is-related-to-Keyword(P, K)] )
Figure 5: The hierarchical model of the QA procedure, process and quality record
THE DEVELOPMENT OF A KNOWLEDGE SYSTEM FOR ISO 9001 QUALITY MANAGEMENT
267

Defn-2 QA-Procedure-is-related-to-Quality-Record
(P, T)
(
T:QUALITY-DOC
( QA Procedure
( Process
( (Process-Using-Quality-Record(OP, T))
(QA-Procedure-Having-Process(P, OP)))))
QA-Procedure-is-related-to-Quality-Record(P,
T)])
Defn-3 QA-Procedure-is-related-to- Department (P,
U)
(
U: Organization (
a
Organization
( QA Procedure
( ( Process
( (QA-Procedure-Having-Process(P, OP))
(Process-Execution-Department(OP, U))
( (Process-Execution-Department (OP, U
a
))
(Manage(U, U
a
))))))
( QUALITY-DOC
( (QA-Procedure-is-related-to-Quality-Record(P,
T))
(Quality-Record-Reserving-Department(T, U))
( (Quality-Record-Reserving-Department (T, U
a
))
(Manage(U, U
a
))))))))
QA-Procedure-is-related-to-Department(P, U)])
6 CONCLUSION
ISO 9001 certification is sought worldwide as a
proof of excellence in quality of products and
services provided by an enterprise. However,
because of the involvement of many units and too
much imbedded relationships in the system, it is a
complicate and difficult job for management. We
deal with this problem by designing an ontology-
based system structure, and use a real case of a
Taiwanese Chemical Company to develop an ISO
9001 quality knowledge system. We analyze the
content of 175 ISO9001 documents to develop ISO
ontology for the company, which contains 213
classes and 723 relationships. The system
development process followed the TOVE ontology
engineering and use Protégé 2000 as the platform for
the knowledge base. The system is capable of
modelling the metadata of ISO documents and
relationships between documents, identifying units
of management with various quality procedures, and
supporting semantic search to allow for meaning-
search rather than keyword based search for ISO
documents.
REFERENCES
M. Uschold & M. Gruninger, June 1996, “Ontologies:
Principles, Methods and Application”, Knowledge
Engineering Review, Vol. 11, No. 2.
J. H. Gennari, et. al., 2003, “The evolution of Protégé: an
environment for knowledge-based Systems
Development,” Int. J. Human-Computer Studies, Bol
58, pp. 89-123.
N. F. Noy, M. Sintek, S. Decker, M. Crubézy, R. W.
Fergerson, and M. A. Musen, March/April 2001,
“Creating Semantic Web Contents with Protégé-
2000”, IEEE Intelligent Systems, Vol. 16, No. 2, pp.
60-71.
W.E. Grosso, H. Eriksson, R. W. Fergerson, J. H. Gennari,
S.W. Tu and M.A. Musen, 1999, “Knowledge
Modeling at the Millennium: The Design and
Evolution of Protg-2000”, SMI Technical Report,
SMI-1999-0801.
Uschold, M. and King, M., 1995. “Towards a
Methodology for Building Ontologies”, Workshop on
Basic Ontological Issues in Knowledge Sharing.
Gruninger, M., and Fox, M.S., 1995 "Methodology for the
Design and Evaluation of Ontologies", Workshop on
Basic Ontological Issues in Knowledge Sharing,
IJCAI-95, Montreal.
Kim, Henry M. and Fox, Mark S., 2002 "Using Enterprise
Reference Models for Automated ISO 9000
Compliance Evaluation", Proceedings of the 35th
Hawaii International Conference on Systems Science
(HICSS-35 ‘02), USA.
Kim, Henry M., 2002 "XMLhoo!: A Prototype
Application for Intelligent Query of XML Documents
using Domain-Specific Ontologies", Proceedings of
the 35th Hawaii International Conference on Systems
Science (HICSS-35 ‘02), USA.
Kim, H.M., Fox, M.S., and Gruninger, M., (1999), "An
ontology for quality management - enabling quality
problem identification and tracing", BT Technology
Journal, Vol. 17, No. 4, pp. 131-140.
Fernandez Lopez, M., August 2, 1999. “Overview of
Methodologies for Building Ontologies”, Proceedings
of the IJCAI-99 workshop on Ontologies and
Problem-Solving Methods (KRR5), Stockholm,
Sweden.
D. L. Goetsch, and S. B. Davis, 1998Understanding and
Implementing ISO 9000 and ISO Standards, Prentice-
Hall, USA.
C. S. Hou, N. F. Noy and M. A. Musen, 2002, “ A
Template Based Approach Toward Acquisition of
Logical Sentences”, Proceedings of the Conference on
Intelligent Information Processing (IIP-2002), pp. 77-
89, Canada
ICEIS 2004 - ARTIFICIAL INTELLIGENCE AND DECISION SUPPORT SYSTEMS
268