Ontology based Knowledge Transferability and Complexity
Measurement for Knowledge Sharing
Pornpit Wongthongtham and Behrang Zadjabbari
Curtin University, GPO Box U1987, Perth, WA 6845, Australia
Keywords: Ontology based Knowledge Transferability Measurement, Ontology based Knowledge Complexity
Measurement, Knowledge Sharing.
Abstract: Importance of knowledge sharing raises the issue of how organizations can effectively encourage individual
knowledge sharing behaviour and what factors enable promote or hinder sharing of knowledge. It is
important to explore the factors affecting knowledge sharing and remove barriers to participation in
knowledge sharing. Willingness and ability to share knowledge and willingness and ability of receiver to
achieve knowledge are one of key issues in knowledge sharing. Knowledge sharing also depends on
knowledge context including the nature, definition, and properties of knowledge which influence the ease
with which knowledge can be shared. In this research the context of knowledge is defined by two key
variables i.e. transferability and complexity which are subject of this paper. Ontologies are used mainly to
provide a shared semantically domain knowledge in a declarative formalism. Ontology specifies consensual
knowledge. In this paper, ontology is applied to explore knowledge context. It is then used to measure
transferability of knowledge between individuals from different backgrounds by comparing the similarity of
their ontologies. Then the difference of the ontologies is measured its complexity in order to determine how
complicated of new knowledge being shared.
1 INTRODUCTION
Knowledge sharing is one of the most critical
elements of effective knowledge processing and
organizations often face difficulties when trying to
encourage knowledge sharing behaviour (Saraydar
et al., 2002). It has been estimated that at least $31.5
billion are lost per year by Fortune 500 companies as
a result of failing to share knowledge (Babcock,
2004). Knowledge sharing refers to the provision of
task information and know-how to help and
collaborate with others to solve problems, share
ideas, or implement policies or procedures
(Cummings, 2004). Davenport and Prusak define
knowledge sharing as equivalent to knowledge
transfer and sharing amongst members of the
organization (Davenport and Prusak, 2003).
Knowledge sharing can occur in different forms
such as written correspondence, face-to-face
communications or through networking with other
experts, documenting, organizing and capturing
knowledge for others (Cummings, 2004).
Knowledge sharing is important for companies to be
able to develop skills and competence, increase
value, and sustain competitive advantages due to
innovation that occurs when people share and
combine their personal knowledge with others
(Matzler et al., 2008). The importance of knowledge
sharing raises the issue of how organizations can
effectively encourage individual knowledge sharing
behaviour and what factors enable, promote or
hinder sharing of knowledge. It is important to
explore the factors affecting knowledge sharing and
remove barriers to participation in knowledge
sharing within and between communities.
Researchers have found that organizational culture
affects knowledge sharing and the benefits of a new
technology were limited if long-standing
organizational values and practice were not
supportive of knowledge sharing across units
(DeLong and Fahey, 2000). Among the many
cultural dimensions that influence knowledge
sharing, trust is the important dimension and a
culture that emphasizes trust can help to alleviate the
negative effect of perceived cost on sharing
(Kankanhalli et al., 2005). Trust provides conduits
for the knowledge exchange and learning needed to
solve problems and achieve shared goals (Preece,
5
Wongthongtham P. and Zadjabbari B..
Ontology based Knowledge Transferability and Complexity Measurement for Knowledge Sharing.
DOI: 10.5220/0004103600050014
In Proceedings of the International Conference on Knowledge Management and Information Sharing (KMIS-2012), pages 5-14
ISBN: 978-989-8565-31-0
Copyright
c
2012 SCITEPRESS (Science and Technology Publications, Lda.)
2004). Trust has been recognized as the gateway to
successful relationships (Wilson and Jantrania,
1993). High levels of trust are the key to effective
communications as trust improves the quality of
dialogue and discussions (Dodgson, 1993). The
willingness to share knowledge is a key issue in
knowledge sharing (Connelly and Kelloway, 2003).
Willingness trust is considered as one of the key
variables in knowledge sharing measurement. Some
of the researches show that management support
affects both the level and quality of knowledge
sharing through influencing employee willingness to
make a commitment. Moreover, in an organizational
context, willingness to share knowledge can be
improved by management support, rewards and
incentives and organizational structure (Wang and
Noe, 2009). In interpersonal and team contexts,
willingness to share knowledge depends more on the
level of team cohesiveness (Bakker et al., 2006) and
the diversity of team members (Ojha, 2005). It is
understood by different researchers that the ability
and competency to share knowledge and to send or
receive knowledge is the most critical issue in
knowledge sharing (Jap, 2001). Competency trust is
considered as the next key variable in knowledge
sharing measurement and it is one of the key issues.
The reason is that competency trust refers to how the
partner is expected to perform, or does perform, in
relation to the underlining functions of the
relationship (Heffernan, 2004). Competency trust is
defined as whether a partner has the capability and
expertise to undertake the purpose of relationship
and meet the obligations of the relationship (Doney
and Cannon, 1997). In overall, willingness and
ability to share knowledge and willingness and
ability of receiver to achieve knowledge are key
issues in knowledge sharing.
Knowledge sharing also depends on knowledge
context including the nature, definition, and
properties of knowledge which influence the ease
with which knowledge can be shared and
accumulated (Argote et al., 2003). The context of
knowledge has been recognized by a number of
knowledge management researchers as being crucial
to improving the understanding and sharing of
knowledge. In this research the context of
knowledge is defined by two key variables i.e.
transferability and complexity which are subject of
this paper. Firstly, transferability of knowledge is
used to measure the nature of knowledge. It is based
on the fact that, in most cases, knowledge senders
and receivers are from different backgrounds such as
engineering, business, medicine etc. and when
individuals from different backgrounds start to share
knowledge, the meaning of this knowledge for each
party may differ. Complexity of knowledge is the
next variable used to measure the ease with which
particular knowledge can be shared. It is obvious
that explicit knowledge and routine or day-to-day
knowledge that people share in their daily
conversation is less complex, while technical
knowledge is more complex.
Ontology is an explicit specification of a
conceptualisation (Gruber, 1993) enabling
underlying knowledge representation. Ontologies are
used in widespread application areas e.g. semantic
web, medical informatics, e-commerce, etc. Mainly
ontologies are used to provide a shared semantically
domain knowledge in a declarative formalism.
Ontology specifies consensual knowledge accepted
by a community. In this paper, ontology is applied to
explore knowledge context. It is then used to
measure transferability of knowledge between
individuals from different backgrounds by
comparing the similarity of their ontologies. Then
the difference of the ontologies is measured its
complexity in order to determine how complicated
of new knowledge being shared. The rest of paper is
organized as follows. We discuss the related works
about ontology comparison and complexity in the
next section. Then we discuss ontology
transferability and its metrics in section 3. Ontology
complexity and its metric are presented in section 4.
Experiment is given in section 5 followed by
discussion in section 6. We conclude our work in
section 7.
2 EXISTING APPROACHES
There are many studies in semantic web applications
emphasizing on measuring ontology similarity and
difference know as ontology matching and mapping.
A number of approaches have been proposed to deal
with the heterogeneity of ontologies (Wang and Ali,
2005). Ontology integration approach maps different
ontologies into a generic ontology using vocabulary
heterogeneity resolution on various ontologies
(Kashyap and Sheth, 1998); (Weinstein and
Birmingham, 1999); (Mena et al., 2000);
(Stuckenschmidt and Timm, 2002). In this method,
the semantic transferability has not been measured
before merging into the generic ontology. Measuring
the semantic transferability is important in the
integrated ontology whether the ontologies should
be merged. Suggested Upper Merged Ontology
(SUMO) is developed to merge ontologies by
sharing ideas from all available ontologies and
KMIS2012-InternationalConferenceonKnowledgeManagementandInformationSharing
6
mapping the entries of merged ontologies with
WordNet entries (Pease et al., 2002). However this
approach does not address the requirement of
transferability of two ontologies. One approach
creates a computational model to assess semantic
similarity among entity classes from different and
independent ontologies without having to form a
shared ontology (Rodriguez and Egenhofer, 2003).
This approach is not practical to measure semantic
transferability of two ontologies due to the
complexity of matching process. Another approach
proposes ontology-based information retrieval
model by using domain ontology to extend the
original keywords input by users and calculates the
concept similarity. Yet this approach does not
address the requirement of transferability of two
ontologies.
In regard to existing work on ontology
complexity, there are existing metrics for analysing
ontology quality but only few of them focus on
complexity of ontology. Burton-Jones et al. (Andrew
et al. 2005) measure elements of quality i.e.
syntactic quality, semantic quality, pragmatic
quality, and social quality using a number of
attributes. Dazhou et al. (Dazhou et al., 2004)
present complexity measurement for ontology based
on UML. However UML cannot entirely represent
semantic richness like ontology does. UML is not a
suitable modeling language to represent an ontology,
thus, the method cannot measure the structure
complexity of ontology objectively. Chris Mungall
(Mungall, 2005) researched the increased
complexity of Gene Ontology which is similar to
Dalu et al. method (Zhang et al., 2006). Anthony et
al. (Anthony et al., 2007) also proposed a metric
suite to measure the increased complexity of tourism
ontologies throughout ontology evolution. However,
the metrics in (Mungall, 2005), (Zhang et al., 2006),
and (Anthony et al., 2007) are evaluating ontology
in ontology evolution. Idris (His, 2004) proposed
conceptual coherence and conceptual complexity
metrics based on graph theory. Orme et al.
(Anthony, Haining et al. 2006) examined coupling
between ontologies. Nevertheless, in (Mungall,
2005); (Zhang et al., 2006); (Anthony et al., 2007);
(His, 2004); and (Anthony et al., 2006), complexity
is analysed by the concept structure and does not
consider the number of restrictions.
In this paper we address the ontology
transferability and complexity as two key variables
for knowledge sharing.
3 ONTOLOGY BASED
KNOWLEDGE
TRANSFERABILITY AND ITS
MEASUREMENT
Knowledge is a combination of the data and
information being produced by human thought
processes. Knowledge can be distinguished into
general knowledge and specific knowledge. General
knowledge is explicit and is easily understood by
locals and neighbours since both their ontologies are
similar. Specific knowledge is more technical and
difficult to understand and depends on an
individual’s background and knowledge level
(ontologies are different). It is necessary to
understand the nature of knowledge in order to
analyse the process of knowledge sharing between
and within organizations or individuals. The
characteristics of knowledge influence the outcome
of knowledge sharing (Nonaka and Takeuchi, 1995).
The impact of the nature of knowledge on
knowledge sharing is part of this research’s
objective. The nature of the knowledge also affects
the importance of trust in knowledge sharing. When
the knowledge seems simple, competence-based
trust is not necessarily important and in this case,
people care more about benevolence-based trust. On
the other hand, when the knowledge is complex and
professional, people care more about competency-
based trust.
We divide knowledge type into easy or hard
transferable knowledge (transferability). Metrics to
measure the complexity of knowledge by using
ontology are presented. We develop a proposed
model and measure the transferability of knowledge
by comparing the two ontologies (sender and
receiver of the knowledge) and ascertaining whether
or not there are similarities.
Transferability of the knowledge is more related
to the members’ backgrounds and their domain
ontology. We use the similarity of ontologies to
measure the level of transformability between two
members. Transferability of the knowledge for both
transmitter and receiver will be given a value
between 0 and 1.
To measure the transformability of two
knowledge backgrounds, ontology similarity is
considered and calculated. In the means of obtaining
the senses and hyponyms of the each concept in the
ontologies and based on the structure of the
ontologies, the similarity of two ontologies can be
calculated. Precisely said knowledge transferability
is signified by ontology similarity. Nevertheless,
OntologybasedKnowledgeTransferabilityandComplexityMeasurementforKnowledgeSharing
7
there may be more than one sense for each concept.
The senses of subclasses of ontology can be
determined by their ancestors. To which sense from
the root of the ontology it is determined by users.
In this paper, our formulas give a numeric
measurement of ontology transferability. Assume we
measure transferability of two ontologies which can
be calculated by using ontology similarity formulas.
Wang and Ali (Wang and Ali, 2005)defined the
difference of set of concepts, S1, captured in
ontology 1, O1, from set of concepts, S2, captured in
ontology 2, O2 as shown in equation (1).
S1 – S2 = {x|xS1 xS2}
(1)
The semantic difference between O1 and O2 can be
defined by function Dif(S1, S2) in equation (2)
(Wang and Ali, 2005).
Dif(S1, S2) =
|S1S2|
|S1|
(2)
Based on the above formula, if the two ontologies
are totally different, the difference value is given 1
or the similarity value is given 0. On the contrary, if
the two ontologies are the same, the difference value
is given 0 or the similarity value is given 1.
Therefore, the similarity of set S1 from set S2 is
defined as {x|xS1 xS2}
The semantic similarity between O1 and O2 or
the transferability can be defined by function
Trans(S1, S2) in equation (3).
Trans(S1, S2) = 1 -
|S1S2|
|S1|
(3)
We compare in both directions i.e. Trans(S1, S2)
and Trans(S2, S1) which may be given different
value.
In domain ontology that two individuals
(receiver and sender) are sharing their knowledge (a
class in ontology), they first need to agree on a sense
of shared knowledge. Sense sets will be provided to
summarize the semantics of the shared knowledge
(the class in ontology). Basically the sense set is a
set of synonym words denoting the concept of the
class in ontology. A sense set is extracted from the
electronic lexical database WordNet which is
available online as Java WordNet Library (JWNL).
JWNL is used to obtain the semantic meanings of
concepts confined in ontologies.
Figure 1: Chair concept in two different ontologies.
The simple ontology transferability algorithm is
shown below:
OntologySenseSet(O)
begin
R = resultSet;
for all node n in Ontology O
p = parent node of n;
senseSetP = all senses of p;
senseSet = all WordNet senses of n;
if n = root
select related sense used in Ontology O;
else
relateFlag = false;
for each sense S in senseSet
hyperSet = hypernyms of each sense S of n;
for each h in hyperSet
if h is in senseSetP
relateFlag = true;
for each s in S
if s == n
R.add(s + “_is-a_” + p);
else
R.add(s);
endif
endfor
endif
endfor
endfor
if relateFlag == false
R.add(n);
endif
endif
endfor
return R;
end
OntologyTransferability(O1, O2)
begin
difference = 0;
for each r1 in OntoSenseSet(O1)
if r1 is not in OntoSenseSet(O2)
difference ++;
endif
endfor
Trans = 1-difference/size of
OntoSenseSet(O1);
return Trans;
end
Quantifying the transferability of knowledge is
intersection between two different ontologies and for
this purpose it is important to assess the semantic
similarity of difference between two ontologies. To
demonstrate the above algorithm we use simple
ontologies and show its transferability as example.
KMIS2012-InternationalConferenceonKnowledgeManagementandInformationSharing
8
Assume we have two ontologies i.e. Furniture
Ontology and Position Ontology as shown in Figure
1.
Furniture Ontology represents concepts of chair
and table as furniture while Position Ontology
represents concepts of secretary and chair as
position. We assess transferability between these
two ontologies. To assess transferability from
Furniture Ontology to Position Ontology, we need to
get the sense set of the two ontologies. In other
words, we get the concepts and their senses with
hypernyms for both Furniture Ontology and Position
Ontology. In process of getting sense set, users
initially choose which sense s/he means at the root
concept if there is more than one sense. Senses and
its hypernyms are obtained from WordNet. Among
those retrieve from WordNet, we also include is-a
relationship to differentiate concept from others if
there is more than one senses in that particular
concept.
Tables below show senses and hypernyms from
WordNet for Furniture Ontology and Position
Ontology. The highlighted senses are ones in sense
set or are ones that have meaning within the meant
content.
Table 1: Senses and hypernyms retrieved from WordNet
for Furniture Ontology.
Concept Senses Hypernyms
Furniture
furniture, piece of
furniture, article of
furniture
furnishing
Chair
chair seat
Professorship, chair
position, post, berth,
office, spot, billet,
place, situation
president, chairman,
chairwoman, chair,
chairperson
presiding officer
electric chair, chair,
death chair, hot seat
instrument of
execution
Table
table, tabular array array
table
furniture, piece of
furniture, article of
furniture
table
furniture, piece of
furniture, article of
furniture
mesa, table tableland, plateau
table gathering, assemblage
board, table fare
As can be seen in Table 2, there are 16 senses for
Position concept. Since Position concept is the root
concept, it need user to initially select which
sense(s) s/he means. In this example sixth sense
(position, post, berth, office, spot, billet, place,
situation) is what the user chosen and is what s/he
means by Position concept. The sixth sense will be
included in sense set for the Position Ontology.
There are 4 senses for Chair concept in Position
Ontology, shown in Table 2, the second sense
(professorship, chair) are selected and to be included
in the sense set because its hypernyms are matched
with selected root sense. We also need to
differentiate ‘chair’ from other ‘chair’ in other
senses by incorporating is-a relationship. To identify
the is-a relationship, we add ‘_is-a_’ follow with
parent concept to ‘chair’ becoming ‘chair_is-
a_position’. For Secretary concept, there is no
matched sense with parent (root) sense, we simply
include it into sense set.
Table 2: Senses and hypernyms retrieved from WordNet
for Position Ontology.
Concept Senses Hypernyms
Position
position, place point
military position,
position
point
position, view,
perspective
orientation
position, posture,
attitude
bodily property
status, position state
position, post, berth,
office, spot, billet,
place, situation
occupation, business,
job, line of work, line
position, spatial
relation
relation
position point
position role
placement, location,
locating, position,
positioning,
emplacement
activity
situation, position condition ,status
place, position Item, point
stance, posture attitude, mental attitude
side, position opinion, view
stead, position,
place, lieu
function, office, part,
role
position assumption
Secretary
secretary head, chief, top dog
secretary, secretarial
assistant
assistant, helper, help,
supporter
repository, secretary confidant, intimate
secretary ,writing
table, escritoire,
secretaire
desk
Chair
chair seat
professorship, chair
position, post, berth,
office, spot, billet, place,
situation
president, chairman,
chairwoman, chair,
chairperson
presiding officer
electric chair, chair,
death chair, hot seat
instrument of execution
From Table 1, the senses set for Furniture
OntologybasedKnowledgeTransferabilityandComplexityMeasurementforKnowledgeSharing
9
Ontology is {furniture, piece of furniture, article of
furniture, chair, table_is-a_furniture, table_is-
a_furniture}. From Table 2, the senses set for
Position Ontology is {position, post, berth, office,
spot, billet, place, situation, secretary, professorship,
chair_is-a_position}. To find transferability value
from Furniture Ontology to Position Ontology,
firstly we need to find sense(s) that appear in the
Furniture sense set but do not appear in the Position
sense set as follow:
Furniture sense set – Position sense set = {x|x
Furniture sense set
x
Position sense set} = 6
The transferability can be defined by function
Trans(Furniture Ontology, Position Ontology) as
follow:
Trans(Furniture Ontology, Position Ontology) = 1 -
= 0
The value of transferability 0 means that knowledge
is not transferable. Concept chair is used in both
ontologies but means differently.
4 ONTOLOGY BASED
KNOWLEDGE COMPLEXITY
AND ITS MEASUREMENT
Ontology complexity is related to the complexity of
conceptualization of the domain of interest. It is
measured to reflect how easy any ontology is to
understand. Definition of ontology complexity is
clarified in features that characterize complexity of
ontology i.e. (i) usability and usefulness and (ii)
maintainability. For example, a more complicated
ontology indicates a more specified knowledge.
However, it is difficult to comprehend and requires a
high value of competence-based trust. Usability and
usefulness of the knowledge may be then decreasing
which implies a major impact on knowledge sharing.
Additionally complicated ontology is hard to
maintain.
In order to measure the complexity of ontology,
number of ontology classes, number of datatype
properties, object properties, constraints, and
hierarchical paths are considered. Number of
Ontology Classes (NoOC) is needed to obtain
average value. Number of Datatype Properties
(NoDP) illustrates how well concepts are being
defined. In OWL the datatype properties are
indicated as owl:dataTypeProperty. Number of
Object Properties (NoOP) illustrates how well
spread of concepts within the ontology. In OWL the
object properties are indicated as
owl:objectProperty. Number of Constraints (NoC)
illustrates how well relations being restricted. In
OWL the constraints are indicated as
owl:allValuesFrom, owl:someValueFrom,
owl:hasValue, owl:cardinality, owl:minCardinality,
and owl:maxCardinality. Lastly Number of
Hierarchical Paths (NoHP) illustrates how fine
concepts being presented. In OWL the hierarchical
paths are represented as owl:subClassOf.
To calculate complexity of an ontology O, a
numeric measurement is defined by function
Complex(O) using above parameters in following
formula:
Complex(O) =
∑
NoOC
⁄
Where Max(NoDP) is maximum number of datatype
property, Max(NoOP) is maximum number of object
property, Max(NoC) is maximum number of
constraint, and Max(NoHP) is maximum number of
hierarchical path. The complexity value is ranged
between 0 and 1 which 0 means the ontology is not
very complicated while 1 means the ontology is very
complicated.
5 EXPERIMENT
We experiment pizza domain. We take Pizza
Ontology developed by CO-ODE team at
Manchester University (Drummond et al., 2007). We
have modified the Pizza Ontology and created
another 2 different Pizza ontologies namely
Vegetable Pizza and Meat Pizza for experiment
studies. The prototype is implemented using JAVA.
We use OWL2.0 API to load and manipulate
ontologies which are related to the domains of
people who are going to share the knowledge.
JWNL is the main API which is used to obtain the
semantic meanings of each concept captured in
ontologies.
Assuming people want to share knowledge about
pizza. Ones who are vegetarian have idea of
vegetable pizza which will be different from ones
who have idea of meat pizza and from others who
have idea of pizza in general. In other words, when
people start to share pizza knowledge, vegetarian
people will be thinking of vegetable pizza, meat
lover people will be thinking of meat pizza, and
other people will be thinking of pizza in general. We
assess how well they share the pizza knowledge. We
have modified Pizza Ontology and create Vegetable
Pizza Ontology and Meat Pizza Ontology. In
experimental studies, we firstly measure the
KMIS2012-InternationalConferenceonKnowledgeManagementandInformationSharing
10
Figure 2: Relation hierarchy of different ontologies.
transferability of pizza knowledge in different
ontologies. Figure 2 shows relation hierarchy of
Pizza Ontology, Meat Pizza Ontology, and
Vegetable Pizza Ontology.
Considering transferability between Vegetable
Pizza Ontology and Meat Pizza Ontology, it is as
follow. From Word Net 2.1, the number of senses
(|S1|) found in Vegetable Pizza Ontology is 56. The
number of senses (|S2|) for Meat Pizza Ontology is
72. After comparing,
S1 – S2 = {x|x
S1
x
S2} = 2
There are 2 distinct senses existing in Vegetable
Pizza sense set and are not in Meat Pizza sense set.
The two senses are “rosa and soho”. The
transferability from Vegetable Pizza Ontology to
Meat Pizza Ontology is as follow:
Trans(S1, S2) = 1 -
||
||
= 1 -
= 0.9642858
In opposite direction, the transferability from Meat
Pizza Ontology to Vegetable Pizza Ontology is as
follow:
Trans(S2, S1) = 1 -
||
||
= 1 -
= 0.75
There are 18 distinct senses existing in Meat Pizza
sense set and are not in Vegetable Pizza sense set.
The 18 senses are “american, cajun, fish_is-
a_topping, anchovy_is-a_fish, anchovy_is-a_fish,
prawn_is-a_fish, shrimp, seafood_is-a_fish, meat_is-
a_topping, beef_is-a_meat, boeuf, chicken_is-
a_meat, poulet, volaille, ham_is-a_meat, jambon,
gammon, and sausage_is-a_meat”.
Table 3 shows other results of different
transferability in different ontologies.
Next we calculate complexity of new knowledge
or complexity of the different part of ontology. If
one who has Vegetable Pizza ontology shares his/her
OntologybasedKnowledgeTransferabilityandComplexityMeasurementforKnowledgeSharing
11
knowledge to one who has Meat Pizza ontology,
complexity of new knowledge of one who has
Vegetable Pizza has to give to one who has Meat
Pizza ontology is measured. Figure 3 shows
properties and restrictions of classes rosa and soho
which are different parts in Vegetable Pizza
ontology.
Table 3: Transferability of different ontologies.
Ontology target Ontology source Transferability
Pizza Meat Pizza
1 -
= 0.9113925
Pizza Vegetable Pizza
1 -
= 0.9113925
Meat Pizza Vegetable Pizza
1 -
= 0.75
Meat Pizza Pizza
1 -
= 1
Vegetable Pizza Pizza
1 -
= 1
Vegetable Pizza Meat Pizza
1 -
= 0.9642858
Figure 3: Properties and restrictions of Rosa class and
Soho class in Vegetable Pizza ontology.
In order to measure complexity value of different
path in the Vegetable Pizza ontology, we need to
find number of classes, datatype properties, object
properties, constraints, and hierarchical paths that
have in Vegetable Pizza ontology but not appear in
Meat Pizza ontology. There are 2 classes i.e. Rosa
and Soho. As in Figure 3, class Rosa has 2 object
properties (i.e. hasTopping and hasBase) and has 5
constraints. As in Figure 3, class Soho has 2 object
properties (i.e. hasTopping and hasBase) and has 8
constraints. There is no hierarchical path in classes
Rosa and Soho. Therefore complexity value of the
different path in the Vegetable Pizza ontology is as
follow:
Complex(O) =
∑
⁄
=
2
⁄
= 0.85
Table 4 shows other results of different complexity
in different ontologies.
Table 4: Complexity of different ontologies.
Ontology target Ontology source Complexity
Pizza Meat Pizza
⁄
= 0.6
Pizza Vegetable Pizza
⁄
= 0.71875
Meat Pizza Vegetable Pizza
⁄
=
0.3232323
Meat Pizza Pizza
Vegetable Pizza Pizza
Vegetable Pizza Meat Pizza
⁄
= 0.85
Value of the new knowledge complexity is 1
which means the new knowledge is more
complicated. In contrarily, value of the new
knowledge complexity is 0 which means the new
knowledge is less complicated. Meat Pizza and
Vegetable Pizza are subset of Pizza so there is no
new knowledge to share between Meat Pizza to
Pizza or Vegetable Pizza to Pizza. Therefore the
complexity value is 0.
6 DISCUSSION
In this study we define two key variables for
knowledge sharing measurement i.e. knowledge
transferability and knowledge complexity. Since we
utilise ontology as knowledge representation in this
paper we propose procedure of measurement of
ontology transferability and ontology complexity. In
the experiment we numerically measure how well
ones share the particular knowledge given that they
have different background or have different
information domains. The process is simple by
measuring their knowledge background similarity
and then finding the difference of knowledge
background. Below is some of result summary from
the experiment:
People have same background knowledge
resulting in best knowledge sharing.
People have similar background knowledge and
the new knowledge is not complicated resulting
some value of knowledge sharing.
People have similar background knowledge and
the new knowledge is complicated. It results some
value of knowledge sharing.
People have different background knowledge
and the new knowledge is very complicated
KMIS2012-InternationalConferenceonKnowledgeManagementandInformationSharing
12
resulting low vale of knowledge sharing. This can
result people will not be able to share knowledge.
The value of knowledge transferability and
knowledge complexity can be put in fuzzy logic
system to explain in high, medium, or low levels.
There are some limitations in our prototype as
follow. In the process of finding transferability and
complexity value we implement sense set which is
extracted from the electronic lexical database
WordNet. By using WordNet, we can only define
ontology concept as a single word which mean it can
only be noun and cannot be adjective, verb, or
adverb. In transferability measurement process, we
only in this paper consider is-a relationship omitting
properties (i.e. object property and datatype
property), constraints, and concept relations e.g.
siblings. Nevertheless, we assess the new knowledge
complexity after finding its transferability
considered above mentioned ontology attributes.
7 CONCLUSIONS AND FUTURE
WORK
We have addressed knowledge complexity and
knowledge transferability as key variables for
knowledge sharing. We then proposed the ontology
based approach which measures ontology
complexity and transferability to correspond to
knowledge complexity and knowledge
transferability respectively. Experimental studies
were given taking Pizza domain and a prototype has
been developed for proof of concept.
For future work a key variable of trust especially
in form of competency trust and benevolence trust
will be incorporated to measure knowledge sharing
in business intelligent applications. Our approach
can be applied to other domains for example e-
commerce and health domains. Future work also
includes a better complexity measurement which
will incorporate depth of concepts i.e. properties
(object property and datatype property) and
constraints, and breadth of concepts i.e. concept
relations e.g. siblings. Comparative evaluation will
also be needed in future work in order to compare
result with other researches in areas of knowledge
sharing measurement and alike.
REFERENCES
Andrew, B.-J., C. S. Veda, et al. (2005). "A semiotic
metrics suite for assessing the quality of ontologies."
Data Knowl. Eng. 55(1): 84-102.
Anthony, M. O., Y. Haining, et al. (2006). "Coupling
Metrics for Ontology-Based Systems." IEEE Softw.
23(2): 102-108.
Anthony, M. O., Y. Haining, et al. (2007). "Indicating
ontology data quality, stability, and completeness
throughout ontology evolution: Research Articles." J.
Softw. Maint. Evol. 19(1): 49-75.
Argote, L., B. Mcevily, et al. (2003). "Managing
knowledge in organizations:An integrative framework
and review of emerging themes." Management Science
49(4): 571-582.
Babcock, P. (2004). Shedding light on knowledge
management. HR Magazine 49: 46-50.
Bakker, M., R. T. A. J. Leenders, et al. (2006). "Is trust
really social capital? Knowledge sharing in product
development projects." The Learning Organization 13:
594−605.
Connelly, C. E. and E. K. Kelloway (2003). "Predictors of
employees' perceptions of knowledge sharing
cultures." Leadership & Organization Development
Journal 24: 294−301.
Cummings, J. N. (2004). "Work groups, structural
diversity, and knowledge sharing in a global
organization." Management Science 50(3): 352−364.
Davenport, T. H. and L. Prusak (2003). Working
knowledge: How organization manage what they
know, Harvard Business School Press.
Dazhou, K., X. Baowen, et al. (2004). A complexity
measure for ontology based on UML. Distributed
Computing Systems, 2004. FTDCS 2004.
Proceedings. 10th IEEE International Workshop on
Future Trends of.
DeLong, D. W. and L. Fahey (2000). "Diagnosing cultural
barriers to knowledge management." Academy of
Management Executive 14(4): 113−127.
Dodgson, M. (1993). "Learning, trust and technological
collaboration." Human Relations 46(1): 77-95.
Doney, P. M. and J. P. Cannon (1997). "An examination
of the nature of trust in Buyer-Seller relationships."
Journal of Marketing 61: 35-51.
Drummond, N., M. Horridge, et al. (2007). "Pizza
Ontologyv1.5" Retrieved May 31, 2010, from
http://www.co-ode.org/ontologies/pizza/2007/02/12/.
Gruber, T. R. (1993). Toward principles for the design of
ontologies used for knowledge sharing. International
Workshop on Formal Ontology in Conceptual
Analysis and Knowledge Representation, Padova,
Italy, Kluwer Academic Publishers, Deventer, The
Netherlands.
Heffernan, T. (2004). "Trust formation in cross-cultural
business to business relationships." Qualitative
research-An international Journal 7: 114-125.
His, I. (2004). Analyzing the Conceptual Coherence of
Computing Applications through Ontological
Excavation PhD Thesis Proposal.
Jap, S. D. (2001). "The strategic role of the saleforce in
developing customer satisfaction across the
relationship." The Journal of Personal Selling and
Sales Management 21: 95-109.
OntologybasedKnowledgeTransferabilityandComplexityMeasurementforKnowledgeSharing
13
Kankanhalli, A., B. C. Y. Tan, et al. (2005). "Contributing
knowledge to electronic knowledge repositories: An
empirical investigation." MIS Quarterly 29(1):
113−143.
Kashyap, V. and A. Sheth (1998). Semantic Heterogeneity
in Global Information Systems: The Role of Metadata,
Context, and Ontologies. London, UK, Academic
Press.
Matzler, K., B. Renzl, et al. (2008). "Personality traits and
knowledge sharing." Journal of economic pcychology
29: 301-313.
Mena, E., A. Illarramendi, et al. (2000). "OBSERVER: An
Approach for Query Processing in Global Information
Systems Based on Interoperation across Preexisting
Ontologies." Distributed and Parallel Databases 8(2):
223-271.
Mungall, C. (2005). Increased complexity in the GO,
BDGP / GO Consortium.
Nonaka, I. and H. Takeuchi (1995). The knowledge
creation company: How Japanese companies create
the dynamics of innovation, New York: Oxford
University Press.
Ojha, A. K. (2005). "Impact of team demography on
knowledge sharing in software project teams." South
Asian Journal of Management 12: 67−78.
Pease, A., I. Niles, et al. (2002). The Suggested Upper
Merged Ontology: A Large Ontology for the Semantic
Web and its Applications. the AAAI-2002 Workshop
on Ontologies and the Semantic Web, Edmonton,
Canada.
Preece, J. (2004). "Etiquette, Empathy and Trust in
Communities of Practice: Stepping-Stones to Social
Capital." Journal of Universal Computer Science
10(3): 294-302.
Rodriguez, M. A. and M. J. Egenhofer (2003).
"Determining semantic similarity among entity classes
from different ontologies." IEEE Transactions on
Knowledge and Data Engineering.
Saraydar, C., N. B. Mandayam, et al. (2002). "Efficient
power control via pricing in wireless data networks."
IEEE Trans. Commun. 50(2): 291-303.
Stuckenschmidt, H. and I. J. Timm (2002). Adapting
communication vocabularies using shared ontologies.
the Second International Workshop on Ontologies in
Agent Systems, Bologna, Italy.
Wang, J. Z. and F. Ali (2005). An efficient ontology
comparison tool for semantic web applications. the
2005 IEEE/WIC/ACM International Conference on
Web Intelligence (WI'05), IEEE.
Wang, S. and R. A. Noe (2009). "Knowledge sharing: A
review and directions for future research." Human
Resource Management Review.
Weinstein, P. and P. Birmingham, (1999). Comparing
Concepts in Differentiated Ontologies. 12th Workshop
on Knowledge Acquisition, Modeling, and
Management., Banff, Canada.
Wilson, T. and S. A. Jantrania (1993). Measuring value in
relationship development. 9th IMP conference, Bath.
Zhang, D., C. Ye, et al. (2006). An evaluation method for
ontology complexity analysis in ontology evolution.
Managing Knowledge in a World of Networks, 15th
International Conference, EKAW 2006, Podebrady,
Czech Republic.
KMIS2012-InternationalConferenceonKnowledgeManagementandInformationSharing
14
