Constructing an Office Domain Ontology using Knowledge Engineering Process

© 2018, AJCSE. All Rights Reserved 19
RESEARCH ARTICLE
Constructing an Office Domain Ontology using Knowledge Engineering Process
Sameia Suha, S. R. Rashmi, Akshaj Jain, Krishnan Rangarajan, D. R. Ramesh Babu
Department of CSE, Dayananda Sagar College of Engineering, Bengaluru, Karnataka, India
Received: 15/01/2018; Revised: 28/02/2018; Accepted: 16/03/2018
ABSTRACT
Knowledge-based identification of human activities in systems depends primarily on rich contextual
domain knowledge casing all of the information about the human, objects around human and also
relations among them. Knowledge engineering plays an important role in building knowledge-based
expert systems, to solve complex problems such as human activity recognition. This requires formal
representation of the knowledge which is based on the conceptualization of the domain. Ontology is a
widely chosen representational model that depicts knowledge as a set of concepts. In this work, we have
applied knowledge engineering process for constructing the domain ontology of the office environment
in agreement with the ontology development life cycle.
Key words: Domain knowledge, expert systems, ontology, reasoning
INTRODUCTION
An expert system is a software that uses experts’
knowledge which is stored in the knowledge base
to take decisions.[1]
There are various domains in
which expert systems can be applied. The first
expert system was named Dendral which was
used to analyze the chemical compounds. Expert
systems are widely applied in the medical domain
for improving the diagnosis. Other domains in
which experts systems can be used are monitoring
systems,processcontrolsystems,knowledgedomain,
and finance or commerce domain to name a few.
The facts stored in the knowledge base of an expert
system are the understanding of the domain expert
in divergence with the facts of a book.[1]
Rule-based intelligent systems with human-
curated rule sets have a huge disadvantage over
the ontological-based intelligent systems as the
rule-based intelligent systems have a tendency to
omit the implicit meaning in statements or given
rules. Whereas, the ontological-based intelligent
systems always contain the implicit meaning in
statements too, thus progressing way beyond the
rule-based intelligent systems toward achieving full
consciousness within the given system. Important
control information can get lost while using rule-
Address for correspondence:
Sameia Suha,
E-mail: sameia.suha@gmail.com
based intelligent systems while this is not the case
withontologicalintelligentsystems,thusontological
intelligent systems are far superior to rule-based
intelligent systems. For example, suppose we
have two meanings of the same sentence, but one
is explicit and another is implicit, the rule-based
intelligent system will most probably take into
account only the explicit meaning given by the
sentence, whereas the ontological system will take
into account the explicit meaning in the sentence.
Let’s take, for example, a statement which says
if x and y are 0, so will its product. The rule-
based system would just represent this statement
and it would not add the formulation that if the
product is 0, x 0, so will y. The ontological
type system would take into consideration both
the first statement and the second statement and
thus there is no loss of implicit information.[15]
Figure 1 shows the basic building blocks of an
expert system and are discussed further down. The
knowledge engineer consults the domain expert to
build the knowledge base. The knowledge base
and the inference engine are the cruxes of the
expert system, where inference engine applies
reasoning on the knowledge base to realize the
inferences and provides it to the user.
Domain expert
A trained person who has expertise in solving
problems of a specific domain. This person can
Available Online at www.ajcse.info
Asian Journal of Computer Science Engineering 2018;3(2):19-25
ISSN 2581 – 3781

Suha, et al.: Running title missing???
AQ1
AJCSE/Apr-Jun-2018/Vol 3/Issue 2 20
give his best results in the domain for the problem
stated. The knowledge engineer takes input
from the domain experts to build the system.[12]
Although the experts do not need to understand
the algorithms of the system, they play an
important role in the development process by
communicating their understanding and facts of
the subject matter to the engineer, hence, giving
an ample amount of time to the project. Domain
expert is also involved in testing the specifications
of the expert system.
Knowledge engineer
A knowledge engineer is someone who designs
and builds the expert system. The main tasks of
knowledge engineer are as follows:
• Discussion of the problem statement with the
domain expert as to how the problem can be
solved.
• To develop the reasoning methods, find
relation between the objects and represent the
facts in the knowledge base.
• Choose some development software or tool, for
implementation of the knowledge represented.
• Testing, maintenance, and integrating the
components of the expert system.
Knowledge base
Knowledgeisacollectionoffactsandunderstanding
cultured through experience and learning.[8]
The
knowledge base contains the knowledge of the
domain as given by the expert. The expert systems
are also known as knowledge-based system.
Knowledge is something which is known or may
be the result of understanding the subject through
experience. Knowledge base is one which stores
the knowledge of a particular domain or the
complete system. Facts and rules are constructed
into a knowledge base and used by expert systems
to extract the conclusions.
Therearevariousformsof knowledgerepresentation.
One such custom is to use ontology which provides
a common understanding of the domain knowledge.
Ontology is a way to explicitly represent the
knowledgeusingclassesandtherelationshipsamong
classes. In ontology, new facts can be inferred from
IF-THEN rules. If part lists the set of conditions. If
the IF part is satisfied, then part is concluded.
For example: “If you are hungry, then eat.”
Inference Engine
Inference engine fetches new facts from the
knowledge base. The conclusions can be drawn
through forward chaining which uses the rules or
backward chaining inference method that works
backward from the goals.
End user
End user is one who is in need of an expert system
and uses the system for the desired operation. For
example, a doctor may need an expert system
which can assist the doctor in diagnosis.
The rest of the paper is divided into different
sections as follows: Section 2 is about knowledge
engineering process and sections 3 and 4 explain
the steps involved in constructing the ontology
and an example of office ontology, respectively,
followed by the conclusion.
KNOWLEDGE ENGINEERING
APPROACH: ONTOLOGY
DEVELOPMENT LIFE CYCLE
Figure 2showstheknowledgeengineeringprocess
used for the construction of ontology. However,
there is no unique ontology development life
cycle for the construction of the ontology. The
subsequent iterative steps can be considered as
a conventional way of applying the knowledge
engineering process to develop an ontology
depending on the application.
Knowledge
Base
n e en e
eng ne
o a n
es
Knowledge
eng nee
se s
Figure 1: Block diagram of expert system

AQ1
Specification
The initialization steps involve identification
and specification of the problem statement for
the ontology to be constructed. It also involves
identifying the important characteristics of the
problems, the resources used to solve the problem
and the reasoning strategies to be used. This step is
carried out by the domain expert in collaboration
with the knowledge engineer.
Conceptualization
Concepts are the entity that exists in real world.
Ontology is an explicit way of specifying the
concepts. It structures the concepts and the
relationship between the concepts in the domain
of interest. Conceptualization is explicitly done by
the knowledge engineer with the consultation of
the domain expert.
Formalization
The knowledge engineer coverts the concepts
into formal language. Web ontology language
abbreviated as OWL is a language that provides
axiomstodefinetheconceptsasahierarchyofclass
and subclasses. This formal conceptualization
also includes properties as relationship between
the concepts and set of individuals belonging to
the class along with the property of individual.[13]
Implementation
Implementation part refers to the realization of the
actual specification listed by knowledge engineer
and domain expert. Actual implementation tools
are used here for the construction of ontology
where the formalization is fed into the expert
system shell by the engineer.
Testing and maintenance
The system is subjected to the process of testing
by thedomainexperttocheckforthespecifications.
The maintenance refers to the taking care of the
product after the delivery. It includes correction
of bugs and improves performance of the system
along with other attributes. This is done by
knowledge engineer.
PROCEDURE FOR IMPLEMENTATION
OF DOMAIN ONTOLOGY
Ontology construction is conversion of knowledge
into software representation. It provides a common
way of understanding the knowledge and domain
reusability. Once the concepts and the essential
specification are obtained from the expert of the
domain, analysis of domain knowledge is possible.
Underneath traditional conditions, ontology
creation itself is not the goal. Ontology can be
analogous to the delineation of knowledge and its
structure.
Simple ontology development method consists of:
1. Determine the aim, domain, and scope of the
ontology
2. Reuse prevailing ontologies if appropriate
3. Input the domain expert’s knowledge into the
system
4. Label important terms in the ontology
5. Define classes of ontology along with their
hierarchy
6. Define relationships or properties along with
its domain and range
7. Creating instances
Step 1
The first step consists of determining the desired
domainonwhichtheontologymustbeconstructed.
In the given example of the ontology below, the
aim, domain, and scope are to recognize activities
that take place in the office. The domain is selected
by the knowledge engineer who is interested in
creating the ontology. The necessary information
required to construct the ontology is collected
Specificaon
Conceptualizaon
Formalizaon
Implementaon
Tesng and maintenance
Figure 2: Ontology development life cycle

AQ1
from the domain expert. This phase must decide
the application that will use the ontology.
Step 2
One of the main advantages of ontology is
reusability.Any existing ontology of the interested
domain if possible can be reused. The existing
ontologies for specific domains can be utilized by
filtering and expanding the prevailing properties
and classes. This reduces the time required to
construct the ontology.
Step 3
Theknowledgeengineershouldtaketheinformation
from the trained expert and he should input this
same information into the ontology. The domain
expert here was the office manager and the
knowledge engineer took the information of day-
to-day activities from the manager and inserted
the information into the ontology.
Step 4
List all significant terminologies and statements
withthehelpofdomainexpert,oughtnottoconsider
the idea of overlapping relationships of concepts or
terms, and even concept of class or property.
For example, the important terms for office domain
include actions, events, person, and objects,
among others.
Step 5
The next step is to determine and define ontology
classesofthedomainfortheontologyconstruction.
In OWL, we can expressly declare the resources to
be a class by stating that it’s of RDF:type owl:class.
Syntactically, this amounts to exploitation of an
owl:class component. It is accustomed OWL
convention to name classes with singular nouns.
It is additionally customary for the names to start
out with uppercase and to use camel case for
multiword names. The employment of uniform
resource identifier to name classes and properties
is a vital facet of the Semantic Web. The classes in
the office ontology are as follows:
• Action
• Event
• Person
• Objects
Step 6
OWL outlines two varieties of properties:
i. Object property
ii. Data type property
Object properties express relationships between a
pairsofclass,whereasdatatypepropertiesdescribe
relationships between a class and a data value.
Similar to classes, we can describe properties
by including subcomponents. RDFs:subproperty
of tells that a resource or object related with one
property can also be related with another property.
The rdfs: domain and rdfs:range state the domain
and range of a property, which tells how a property
links a subject with an object. The rdfs: domain of
a property applies to the subject of any statement.
Whereas, the rdfs: range of a property applies
to the object of the statement. Although these
properties could appear simple, they will result
in variety of misunderstandings and should be
used fastidiously.
Step 7
The individuals of a class are created in the
instance tab by selecting the class and naming
the individuals. Defining an individual instance
of a class requires choosing a class, creating an
individual instance of that class, and naming the
instance.[14]
Once an ontology is created its consistency can
be checked and reasoning can be applied. It
is important to note that there can be different
ontologies for the same domain, the design of
ontology is dependent on the application it will be
used for.[14]
Consistency Check
The knowledge of a domain consisting of the
concepts and their roles covered in the ontology
can be checked for consistency. A consistent
ontology is the one which holds a set of conditions.
An ontology can be checked for:
1. Structural consistency
2. Logical consistency
3. User-defined consistency.
AQ2

AQ1
An ontology is said to be consistent if it adheres
to the syntax and the logics of the language used
for the construction of the ontology and is built
according to the user requirements.
Reasoning
A reasoner is the tool that infers logical
consequences from a group of declared facts.
One of the prominent ways to express the rules
is through Semantic Web Rule Language which
sits above the ontology web language. The JESS
is used as the inference engine in the forthcoming
example.
IMPLEMENTATION OF AN OFFICE
DOMAIN ONTOLOGY
This section shows construction of office ontology
and inferences of the events using the constructed
ontology and SWRL rules.
First, the problem statement is to recognize
the events carried out by person in an office
environment. Hence, the domain in which the
ontology is constructed is the office domain. New
ontology is built; therefore, the reusability step
mentioned above is omitted. After the domain is
decided, the important terms related to the ontology
are listed down and the classes and the subclasses
are decided. Four classes and their subclasses along
with an object property are shown in Figure 3.
The next phase in ontology construction is the
design and implementation phase which can be
carried out according to the following steps.
List the events which are the result of a sequence
of actions that are carried out by a person. The
events picked are as follows:
1. Enter.
2. Exit.
3. Print.
4. Sit.
The next step is to create the OWL classes defining
actions, events, objects, and person [Figure 4].
The next step consists of defining the property along
with domain and range. In this example, an object
property named action is assigned its domain as
person and events, its range as actions and objects.
The domain and range tells in what way the property
links the subject with the object [Figure 5].
Create instances for each class in individual tab.
The instances are the individual of the classes.
The class for which an instance must be created is
selected from the individual class and an instance
is added by naming it [Figure 6].
The constructed ontology can be viewed in various
forms including the nested view, tree view, nested
composite view, domain-range view, and class
tree view. Nested composite view of the office
ontology using the jambalaya tool as a plugin of
protégé is shown in Figure 7.
In Figure 8, the box represents the classes and the
pink line is the relationship defined through the
domain and range of the object property.
The consistency of the ontology is checked using
the pellet direct reasoner. The results show that the
ontology is consistent with respect to the concepts
and inferred hierarchy.
Once the ontology is consistent, inference can be
done to recognize the activities performed by the
personintheofficeenvironment.Thisisdoneusing
the SWRL rules. SWRL tab must be activated and
the rule is written in the multiline rule editor. The
rules are fired using the JESS inference engine
which converts the OWL and SWRL to JESS and
back to OWL after the inference [Figure 9].
Figure 3: Classes and property of office ontology
Figure 4: Creation of classes in protégé 3.4

AQ1
To see the working:
1. Create instances of single person that have
properties mentioned in rules.
2. Run the JESS tab.
3. New instances of the class single person will be
added in the events classes [Figures 10 and 11].
The next phase is the testing and maintenance
phase where the ontology is tested with various
inputs to check the correct inference, further
improvements to the ontology can be done
during the maintenance phase. More events and
activities along with the rules can be added as
desired.
CONCLUSION
In this paper, we discussed about the procedure
for modeling ontologies for domain knowledge
representation. We built the ontology for office
domain using the tool Protégé 3.4 and used JESS
engine for reasoning on the ontology. Knowledge
engineering approach is thoroughly followed
including the ontology life cycle for building this
ontology, and it was tested for consistency. This
ontology is reusable and expandable. This work
will be helpful for the beginners in the ontology
research.
Figure 5: Property along with domain and range
Figure 6: Creation of instance for class
Figure 7: Nested composite view
Figure 8: Consistency check results
Figure 9: Multiline rule editor

AQ1
REFERENCES
1. Rani S. Expert System of AI, 01 October 2014, Vol. 4,
No. 5.
2. Xi FE, Jihua S. A Study in Knowledge Ontology
Building in the Area of Knowledge Engineering. Third
International Conference on Semantics, Knowledge
and Grid; 2007.
3. Kureychik VM. Overview and Problem State of
Ontology Models Development. Russia: Southern
Federal University; 2017.
4. Ying-ying Y, Zong-yong L, Zhi-xue W. Domain
Knowledge Consistency Checking for Ontology-based
Requirement Engineering. China: ICCSSE; 2008.
5. Jin Z. Ontology-based requirements elicitation.
J Computers 2000;23:486-92.
6. Li Z, Wang Z, Yang Y. Towards a Multiple Ontology
Framework for Requirements Elicitation and Reuse. In:
Proceedings of 31st
IEEE International Computer Software
and Application Conference, Beijing; 2007. p. 189-95.
7. Vel´asquez JD, Jain LC, editor. Advanced Techniques
in Web Intelligence-1. Verlag: Springer; 2010.
8. Hyeon J, Oh K, Chung YJ, Kang BH, Choi HJ.
Constructing an Initial Knowledge Base for Medical
Domain Expert System using Induct RDR. Daejeon,
Republic of Korea: School of Computing, KAIST;
2016.
9. Xia FE, Jihua S. Specifying Ontologies: Linguistic
AspectsEngineering, Third International Conference on
Semantics, Knowledge and Grid; 2017.
10. Source: Portal.surrey.ac.uk.
11. Wei Q. Development and Application of Knowledge
EngineeringBasedonOntology,2010ThirdInternational
Conference on Knowledge Discovery and Data Mining;
01/2010.
12. Staab S, Studer R, Schnurr HP, Sure Y. Knowledge
processes and ontologies. IEEE Intell Syst 2001;16:26-34.
13. Andrea M, Francesco R, Maurizio T, Salvatore M.
Formalizing Knowledge by Ontologies: OWL and
KIF. Pisa, Italy: CNR, IIT Department, Via Moruzzi 1,
I-56124; 2015.
14. Noy NF, McGuinness DL. Ontology Development 101:
A Guide to Creating YourFirst Ontology. Stanford, CA:
Stanford University, 94305; 2001.
15. Nilsson NJ. Principles of Artificial Intelligence. Los
Altos, CA: Morgan Kaufmann; 1980.
AQ3
AQ4
Author Queries???
AQ1:Kindly provide running title
AQ2:Kindly review the sentence.
AQ3:Kindly cite references 2-7, 9,10 and 11 in the text part
AQ4:Kindly provide complete reference details
AQ5:Kindly cite table 1 in the text part
Figure 10: Input instances
Figure 11: Inferred instance in the events: Enter subclass
Table 1: Comparison of expert system and human expert
Resource
parameters
Human expert Expert system
Availability On working days/retire 24*7
Geographical
location
Specific location Across globe
Replacement Not replaceable Can be replaced
Performance Variable Constant
Speed Variable Constant
Updation Inconvenient Easy to update
AQ5

Constructing an Office Domain Ontology using Knowledge Engineering Process

More Related Content

Similar to Constructing an Office Domain Ontology using Knowledge Engineering Process (20)

More from BRNSSPublicationHubI (20)

Recently uploaded (20)

Constructing an Office Domain Ontology using Knowledge Engineering Process