Exploiting Tagging in Ontology-based e-Learning
Nicola Capuano
1
, Angelo Gaeta
1
, Francesco Orciuoli
1
and Stefano Paolozzi
2
1
Dipartimento di Ingegneria dell’Informazione e Matematica Applicata, University of Salerno
Via Ponte Don Melillo, 84084 Fisciano, Salerno, Italy
2
Dipartimento di Informatica e Automazione, University of Roma Tre
Via della Vasca Navale, 79, I, 00146 Roma, Italy
Abstract. The ontologies are used to state the meaning of the terms used in data
produced, shared and consumed within the context of Semantic Web applica-
tions. The folksonomies instead are an emergent phenomenon of the Social Web
and represent the result of free tagging of information and objects in a social en-
vironment. Both ontologies and folksonomies are considered useful mechanisms
to manage the information and are pretty always exploited, independently, in sev-
eral areas of interest in order to cope with different problems related to searching,
filtering, categorization and organization of content within some applications for
e-commerce, e-learning, e-science, etc. In our opinion the two mechanisms are
not in opposition but could be synergically used. In this paper we propose an ap-
proach based on the convergence between ontologies and folksonomies in order
to improve personalised e-learning processes.
1 Introduction
In a famous presentation [6], T. Berners-Lee explains his vision of the Semantic Web.
He believes that the Web of the future foresees machines able to understand the content
of web pages. The core of his Semantic Web architecture is composed of three main
layers. At the bottom there are the Markup Languages for the resources description.
On top of this layer he places the ontologies used to define terms and their relation-
ships between other terms. Ontologies enable the next layer, namely the Logic Layer,
where we can deduce new information by analyzing assertions from the Web. Hence,
ontologies are considered an important building block for the evolutions of the Web.
Gruber, in [7], states that an ontology is an explicit specification of a conceptualization
and, more pragmatically, he explains that a common ontology defines the vocabulary by
which queries and assertions are exchanged among agents (both human and software
agents).
Let us to illustrate two important aspects related to the use of ontologies: (i) the
fields of application in which ontologies should be used and (ii) the approaches to on-
tologies definition on the Web.
With respect to the first issue, in [8] it is well described that the ontologies can
provide guidance on how to correctly relate pieces of information in a specific domain,
can provide a more effective basis for information extraction or content clustering, may
Capuano N., Gaeta A., Orciuoli F. and Paolozzi S. (2009).
Exploiting Tagging in Ontology-based e-Learning.
In Proceedings of the 1st International Workshop on Ontology for e-Technologies OET 2009, pages 3-12
DOI: 10.5220/0002222000030012
Copyright
c
 SciTePress

be a source of structure and controlled vocabularies helpful for disambiguating context,
can provide guiding structure for browsing or discovery within a domain, can provide
the basis to reason or infer over its domain and can be organized hierarchically from the
most specific to the most abstract one.
Moreover, concerning the approaches to ontologies on the Web, both structure and
formalism are dimensions for classifying them, which combined are often referred to
as an ontology expressiveness. Some interesting ontology approaches are:
– Hierarchical Faceted Metadata. A faceted classification [9] allows the assign-
ment of more than one classification to an object, enabling the classifications to be
organized in multiple ways, rather than in a unique taxonomic order. For instance,
a collection of photos might be classified using an author facet, a subject facet, a
date facet, etc. Faceted classification supports the navigation of information along
multiple paths corresponding to different orderings of the facets.
– Folksonomies. A folksonomy [10] is the result of free tagging of content (anything
with a URL) for the aims of simplifying succesive searching operations. When the
tagging is done in a social environment (e.g. Flickr, Del.icio.us, Furl, etc. ), a folk-
sonomy is created from the act of tagging by the users consuming the information.
The value in this external tagging is derived from people using their own vocabu-
lary and adding explicit meaning, which may come from inferred understanding of
the tagged information/content.
– Topic Maps. A Topic Map
3
represents information using topics (i.e. any concept
or subject coming from a specific domain), associations (relationships between top-
ics) and occurrences (relationships between topics and resources/content relevant
to them).
– OWL Lite, DL, Full. The Web Ontology Language (OWL)
4
is a language (based
on XML and RDF/RDF-S) used for defining ontologies for the Web. An OWL
ontology includes descriptions of classes, properties and instances and is designed
for use by applications that need to process the information content.
– Higher-order Formal and Upper-level Ontologies. SUMO
5
, DOLCE
6
and Open-
Cyc
7
are examples of these ontologies that have the goal of enabling AI applications
to perform human-like reasoning.
More formal ontologies (e.g. OWL-Full, OWL-DL, Higher-order formal and upper-
level ontologies, etc.) have greater expressiveness, structure and inferential power than
the less formal ones (e.g. folksonomies, Topic Maps, Hierarchical Faceted Metadata,
etc.). The higher the formality, the higher the effort and rigor required, resulting in
an approach that is more powerful but also more rigid and less flexible. Furthermore,
formal ontologies require strong knowledge representation skills that are not simply
findable. These are all aspects that should be taken into consideration when choos-
ing an ontology approach. In particular, approaches like folksonomy are fine when on-
tologies producers are typical Web 2.0 users and the expected results are to categorize
3
http://www.ontopia.net/topicmaps/materials/tao.html
4
http://www.w3.org/2004/OWL/
5
http://www.ontologyportal.org/
6
http://www.loa-cnr.it/DOLCE.html
7
http://www.cyc.com/cyc/opencyc/overview
4

content and support search operations. Otherwise, the adoption of Topic Maps is prefer-
able when we aim to model educational domains (e.g. Physics, Mathematics, etc.), the
ontologies producers are domain experts (e.g. professors) having no Knowledge Engi-
neering skills and there are domain-specific inference rules to be executed. In the end,
we could choose OWL if we need a standard reasoning like, for instance, checking if a
concept C subsumes the concept D or if an individual a is an instance of concept F . Is
there a way to combine some of them in order to make the ontologies construction pro-
cess more rapid and simpler, and the ontologies management processes more effective
and efficient?
The aim of this work is to answer the previous question illustrating the existing
works and proposing an approach to face and solve the problem in the e-learning area,
making specific reference to an existing e-Learning Platform. In the section 2 we will
illustrate some approaches for the convergence of more formal ontologies and less for-
mal ones and some specifications useful to manage the ontological structures. In section
3 we will present the Intelligent Web Teacher (IWT) [13], an e-Learning Platform en-
abling the definition and execution of personalized learning experiences based on the
explicit representation of educational domains knowledge. In the section 4 we will pro-
pose an approach to improve the ontology-related processes in IWT. In the end, in sec-
tion 5 we will provide considerations about this work and some ideas for future research
activities related to it.
2 Background and Related Works
Semantic Web researchers have become increasingly interested in studying different
ways to use the technologies of the Semantic Web to organize data coming from the
Social Web. Some groups have studied possible means to improve the way the social
tagging system is organized. Laniado [17] suggests using WordNet
8
to improve the
performance of the related tags search of Del.icio.us. A promising field of research has
been the use of folksonomies as corpora to extract domain specific ontologies.
VanDamme [19] proposes an approach (named FolksOntology) that integrates mul-
tiple resources and techniques to achieve this goal. Some examples of techniques: com-
pute the similarity degree between two tags by counting the number of times they are
used to describe the same object, cluster the actors and the groups that share the same
tags or the same objects. WordNet, Wikipedia
9
and Leo Dictionary
10
are used as re-
sources to identify spelling errors in tags, to substitute tags with their hyponyms and to
translate tags from one language into another.
Torniai [22] analyses this problem in the context of e-Learning: it is difficult to
create and maintain the ontologies that describe different study courses. Torniai’s group
extended LOCO-Analyst, an e-learning ontology based learning tool, in order to exploit
the students annotations to update domain ontologies. Students can associate tags to
learning objects using OATS (Open Annotation and Tagging System). Teachers can see
both the ontology graph and the tag cloud of the students annotations: when a teacher
8
http://wordnet.princeton.edu/
9
http://www.wikipedia.org/
10
http://dict.leo.org/
5

selects a lesson, the related tag cloud is visualized (the size of each tag depends on its
popularity) and the related concepts are highlighted in the graph. If a single concept in
the graph is selected, related tags are colored with a darker shade of blue.
The folksonomies, anyway, have not been considered only as a corpus of infor-
mation for data mining processes. In recent times the Semantic Web community has
noticed that social web sites behave like data silos and that information cannot be ex-
changed between them in an easy manner. There is a need to improve sharing and inter-
operability. Gruber was one of the firsts to propose the creation of an ontology of folk-
sonomies [16]. He sees them as a substrate to be used to interface with different social
tagging systems and share tags between them improving the classification and helping
research work. According to Gruber, the tagging relationship should be expressed as a
tuple (object, tag, tagger, source, + or -), where the source is the social tagging system
from which the tag is taken and the fifth field is a polarity argument (plus for positive
tagging, minus to signal the tag as spam). Other researchers tried to use variations of
his tagging relation to construct a tag ontology, sometimes using existing ontologies as
building blocks. According to Knerr [20] the folksonomy system architecture should be
changed in order to record in one place a users personal information and relations, cou-
pled with the tags they entered in all the folksonomies they are part of. Knerr proposes
a Tag Ontology (known as TagOnt), in which Tagging, the central element of the on-
tology, is expressed as a tuple (time, user, domain, visibility, tag, resource, type).
This definition is similar to that of [16] (with domain corresponding to Grubers source)
but it adds the parameter time that keeps track of the tagging date.
The existing ontologies are used to map individual parts of an ontology: Dublin
Core
11
for time, FOAF
12
for users, SKOS
13
for tags, RDFS
14
for resources. The visi-
bility parameter can take the values public, private, or protected. Newman has created
another Tag Ontology, which is widely used in the Semantic Web field [18]. The tagging
relation is expressed by the tuple (User, Resource, T ag). Every single tag is identified
with an URI, allowing easier research and clustering. The tag relations are modeled us-
ing SKOS; even the Tag class inherits from the SKOS Concept Class. Even if it is widely
used, the Tag Ontology has suffered some criticism because of the lack of semantics in
tag relationships: we cant specify if two words are related because they are synonyms or
because they are variations of the same word. One of the most important ontologies used
to exchange data taken from social communities is SIOC (Semantically-Interlinked On-
line Communities), described in detail in [1]. With SIOC is possible to keep track of the
main argument of a discussion, the number of page views, the author and the relation-
ships between users.
Moreover, the MOAT project [21] is a framework created by Alexandre Passant
and composed of an ontology, a client and a server. The MOAT ontology differs from
previous tagging ontologies because the tagging relationship includes a new concept:
meaning. The Tag Ontology Class is extended: each tag has a set of meanings, described
with an URI (usually taken from lexical resources like DBpedia) and the set of users
11
http://dublincore.org/documents/dces/
12
http://www.foaf-project.org/
13
http://www.w3.org/2004/02/skos/
14
http://www.w3.org/TR/rdf-schema/
6

that associated a certain meaning to a given tag. The MOAT client can be added to every
blog application and offers an interface to associate meanings (in URI format) to tags,
while users who subscribe to the MOAT server can share tags and meanings with other
people.
In the end, SCOT (Social Semantic Cloud of Tags) is an ontology created with the
purpose of enhancing tag sharing and interoperability among different social commu-
nities [18]. The tagging activity is represented as a ternary relation between users, tags
and resources. The SCOT model is based on three existing ontologies: like TagOnt, it
reuses FOAF and SKOS for mapping users and tags, but uses SIOC to represent re-
sources. Int.ere.st
15
is a social tagging, bookmarking and sharing service based on the
SCOT ontology [18]. A more complete coverage of these needs can be achieved by
federating these specifications and exploit the advantages of each one of them.
3 Ontological Structures in the Intelligent Web Teacher
The Intelligent Web Teacher (IWT) is primarily an e-Learning Platform that enables the
definition and the execution of personalized learning experiences packaged in a Unit of
Learning (i.e. a course, a module or a lesson structured as a sequence of Learning Ac-
tivities represented by Learning Objects and/or Learning Services). The foundation for
the Unit of Learning (UoL) building process is the Learning Model described in [11].
The Learning Model allows to automatically generate a UoL and to dynamically adapt
it during the learning process according to the learners’ preferences and cognitive state
(personalization process). In order to achieve the expected adaptation capability, the
Learning Model uses three specific sub-models: the Knowledge Model, the Learner
Model and the Didactic Model, which are exploited by a specific process used to define
personalized e-learning experiences.
The Knowledge Model describes, in a machine-understandable way, the subset of
the educational domain that is relevant for the e-learning experience we want to define,
concretize and broadcast. The Knowledge Model exploits the ontologies. The used on-
tology approach is really similar to that of Topic Maps intoduced in section 1 and is
described in [14]. In our approach the vocabularies are composed of terms representing
subjects that are relevant for the educational domain we want to model. Subjects are
associated to other subjects through a set of several conceptual relations. The most im-
portant relations are: HasPart (in brief HP) that is a part-of relation and IsRequiredBy
(in brief IRB) that is an order relation. The ontologies constructed following the quite a
few aforementioned informal rules are called e-learning ontologies. Observe that when
we refer to concepts in the e-learning ontologies we are referring to the subjects of the
educational domain we are modeling. Let us now consider how to build an e-learning
ontology. Suppose we have to model the educational domain D, so we try to conceptual-
ize the knowledge underlying D and find a set of terms representing relevant concepts in
it. The result of the previous step is the list of terms T = C, C
1
, C
2
, C
3
where T is one
of the plausible conceptualizations of D. The existence of relations HasP art(C, C
1
),
HasP ar t(C, C
2
) and HasP art(C, C
3
) means that in order to learn a subject C the
15
http://int.ere.st/
7

learners have to learn subjects C
1
, C
2
and C
3
without considering a specific order. If
we add the relations IsRequiredBy(C
1
, C
2
) and IsRequiredBy(C
3
, C
2
) to the pre-
vious set of relations we can state that C
1
has to be necessarily learned before C
2
and
C
3
has to be necessarily learned before C
2
.
Now, we would like to introduce new elements called Learning Objects. You can
interpret the connection between a concept and a Learning Object, for instance C
1
and
LO
1
, as a HasResource (in brief HR) relation. The relation HasResource(C
x
, LO
y
)
means that the educational content packaged in Learning Object LO
y
explains con-
cept C
x
. So, if we assume that HasResource(C
1
, LO
1
), HasResource(C
2
, LO
2
),
HasResource(C
3
, LO
3
) and that our Learning Objective is C
1
then the corresponding
assembled e-learning experience is composed only by [LO
1
], otherwise if the Learn-
ing Objective is C then the assembled e-learning experience will be composed as
[LO
1
, LO
3
, LO
2
].
For sake of simplicity, we disregard the Didactic Model (used to model learning
methods to be applied to specific learning experiences) and the Learner Model (used
to model cognitive states and learning preferences of single learners in order to sup-
port the personalization process) given that it is not fundamental for the focus of this
work. Excluding the selection of the Learning Objective over an ontology and some
other customization parameters, the UoL building process is fully automatic and re-
alized through the execution of several algorithms. The most important are: Learning
Path Generation Algorithm and Presentation Generation Algorithm [12]. Using these
algorithms it is possible to generate courses tailored to a class, to a specific group and
even to single learners.
According to the IWT approach, a Learning Object is a learning content (or a pack-
aged aggregation of learning content) that can be delivered through a Web Browser, that
is annotated with an instance of a metadata schema interoperable with IEEE LOM [15]
and that is stored and indexed into a Learning Object Repository. The binding between
Learning Objects and subjects of ontologies is realized by storing subjects references in
a specific attribute (Classification.Taxonpath.Taxon [15]) within the metadata instances
associated with Learning Objects. The binding operation is performed by users (with
resource manager permissions) selecting the object, the metadata attribute, the ontology
of interest and one or more subjects upon the selected ontology.
In IWT, the Learning Objects, and in general all content managed by the Platform,
can be tagged both in a personal and in a shared area. The tagging process is free (no
guidance for tags definition). In the context of the e-Learning Platform, the tagging
process is used in order to improve the searching functionality, by using the Tag Cloud
facility, and to support collaborative learning activities where the teacher asks for the
students to collaboratively categorize a set of resources as a part of a whole learning
scenario. The assigned tags are stored into the General.Keyword [15] Learning Object
metadata attribute and used to create indexes for performance aims. The result of the
tagging process is a folksonomy that grows when new learning content are added and
represents a rich source of knowledge that could be better exploited with respect to the
optimization of the e-Learning Platform processes.
8

The mechanisms concerned with the e-learning ontologies and the content tagging
are independently managed and controlled by the IWT Platform services. This indepen-
dence leads to:
– Possible redundancy of labels applied to content. The same Learning Object could
be annotated at the same time with the tag ”SOA” and with a reference to a subject
”SOA” within the ”Software Architectures” ontology.
– Possible disorientation for users. The users could be confused if involved in dif-
ferent, but similar, tagging operations to perform in different contexts and with
different goals.
– Difficulty in sharing, within the users community, the intended meaning of tags,
keywords, labels, etc. From the point of view of handling unambiguous meanings,
a unique space of shared terms is easier to manage than multiple spaces .
– Unnecessary complexity of the search engine. Having more than one indexing
mechanism triggers the growth in complexity of the search engine that has to query
several indexes and harmonize the result sets.
– Incapability to exploit tags within the Unit of Learning building process. The afore-
mentioned Presentation Generation Algorithm retrieves Learning Objects using
only subjects previousely extracted from the reference ontology and cannot ex-
ploit other semantic enrichments provided by users in order to improve the filtering
action of the algorithm.
4 A Proposed Approach: The Tagging System
In order to face the problems illustrated in section 3 we propose an approach based
on the unification of the e-learning ontologies management and the content tagging
mechanisms within the IWT Platform.
The identified solution is focused on the action of providing a unique tag space
populated by several content and knowledge managing processes. The conceptual ar-
chitecture of the Tagging System is reported in figure 1(a).
So, with respect to the creation of e-learning ontologies, the subjects (nodes in the
ontologies) will be retrieved from tags stored in the tag space. The link between ontolo-
gies and Learning Objects (defined as the HasResource in section 3) is realized only
using the General.Keyword attribute of the Learning Object Metadata schema, that is
also used to store simple tagging information. The Classification.Taxonpath.Taxon will
no longer be used. In figure 1(b) we show the use of Tag Ontology, SCOT and MOAT
to implement respectively the tag modeling, the e-learning ontologies structure repre-
sentation and the synonyms handling with respect to the proposed approach.
Numerous are the interaction flows involving users, tag space and other IWT mod-
ules. We can deal with the aforementioned flows by decomposing them into classes:
Insert, Import, Search, Organize, Suggest and Export flows.
The Insert flows concern with tag space population. The new subjects defined by
the authors of ontologies will become tags in the tag space, everytime the users add a
keyword to a given content (e.g. Learning Objects), this keyword is inserted into the tag
space as a new tag. Furthermore, new tags used in the Blog, Wiki, Forum, etc. will be
inserted as tags in the tag space.
9

Fig. 1. Tagging System.
The Import and the Export flows concern the interoperability with external tag
spaces. With respect to the interoperability issues, the SCOT specifications represents a
good solution.
The Search flows regard the way users can exploit the tags associated with the con-
tent. Two are the main search mechanisms taking advantage of the tag space. The first
one is the keyword-based search and the second one is the Tag Clouds [5], i.e. visual
representations of social tags, displayed in paragraph-style layout, usually in alphabet-
ical order, where the relative size and weight of the font for each tag corresponds to the
relative frequency of its use.
The Suggest flows are involved in exploiting specific elaboration performed on the
tag space in order to support the users when they use tags. Suggestions about tags
already existing in the tag space can be useful to avoid the occurrence of different tags
with the same implicit meaning. This suggestion could be provided by the system using
approaches like [4] where a technique is proposed to compute semantic relatedness
between two terms using Wikipedia or like [3] where the authors define the Google
distance as a measure of semantic relatedness computed by counting the number of hits
that two keywords have if used together in the Google search engine.
The Organize flows concern a way of supporting users who build e-Learning On-
tologies by providing draft structures (to be manually refined) automatically extracted
from the tag space. Works like [2] and [19] are interesting for this goal. In particular
in [2], the authors aim at adding meaning to tags, using generic ontologies (Word-
Net) and domain specific ontologies. Domain specific ontologies could be created by
analysing set of Flickr pictures related to the same subject, the most common tags are
extracted and used to individualise the basic concepts of the ontology.
5 Conclusions and Future Works
In the present work we have proposed (section 4) an approach with the objective to face
and solve the problems identified at the end of the section 3. In particular:
10

– The redundancy of labels applied to the content is loosened by the adoption of a
unique tag space.
– The disorientation for users is loosened by the unique way to tag content and by the
use of tags as ontologies subjects and as keywords.
– The difficulty in sharing, within the users community, the intended meaning of tags,
keywords, labels, etc is loosened by the adoption of specifications like MOAT and
SCOT and by the model provided by the Tag Ontology.
– The complexity of the search engine is loosened by the adoption of a model in
which content are described only by tags. Advanced searches are provided by
means of similarity algorithms and other tag space elaborations.
– The incapability to exploit tags within the Unit of Learning building process is
loosened by modifying the Presentation Generation Algorithm in order to retrieves
Learning Objects exploiting tags stored in General.Keyword field of metadata schema
and used as subjects within the e-Learning Ontologies.
Future works will concern the analysis, design and prototyping of a system based
on the proposed approach and integrated with the IWT Platform in order to perform
experiments and evaluate our approach.
References
1. Breslin, J.G., Harth, A., Bojars, U., Decker, S.: Towards semantically-interlinked online
communities. ESWC 2005. Heraklion, Crete, Greece, May 29 - June 1, 2005, Proceedings.
(2005).
2. Lee, S.S. , Yong, H.S.: OntoSonomy: Ontology-based Extension of Folksonomy
doi.ieeecomputersociety.org (2008)
3. Cilibrasi, R., Vitanyi, P.: The Google Similarity Distance. IEEE Transactions on knowledge
and data engineering 19(3) (2007) 370383.
4. Strube, M., Ponzetto, S.P.: WikiRelate! Computing semantic relatedness using Wikipedia. In
Proc. of AAAI-06, pp. 14191424. (2006).
5. Hearst, M. A., Rosner, D.K.: Tag Clouds: Data Analysis Tool or Social Signaller? HICSS
2008. ¡http://people.ischool.berkeley.edu/ hearst/papers/tagclouds.pdf¿.
6. Berners-Lee, T.: Semantic Web on XML, Keynote presentation for XML 2000. Slides
available at: http://www.w3.org/2000/Talks/1206-xml2k-tbl/slide1-0.html. Reporting avail-
able at: http://www.xml.com/pub/a/2000/12/xml2000/timbl.html .
7. Gruber, T. R.: Towards Principles for the Design of Ontologies used for Knowledge Sharing.
International Journal of Human-Computer Studies, 43:907-928, 1995.
8. Bergman, M. K.: An Intrepid Guide to Ontologies. Theres an Endless Variety of World
Views, and Almost as Many Ways to Organize and Describe Them. AI3:::Adaptive 2007.
Retrieved from: http://www.mkbergman.com/?p=374
9. Denton, William.: How to Make a Faceted Classification and Put It On the Web. Nov. 2003.
http://www.miskatonic.org/library/facet-web.
10. Vanderwal, T.: Folksonomy Coinage and Definition.
http://www.vanderwal.net/folksonomy.html, 2004.
11. Albano, G., Gaeta, M., Salerno, S.: E-learning: a model and process proposal. IJKL, vol. 2,
no.1, 2006, pp. 73-88.
12. Capuano, N. , Gaeta, M., Miranda, S. , Orciuoli, F. and Ritrovato, P.: LIA: An Intelligent
Advisor for e-Learning. 1st world summit on The Knowledge Society, 2008.
11

13. Albano, G., Gaeta, M., Ritrovato, P.: IWT: an innovative solution for AGS e-Learning model.
IJKL, vol. 3, n. 2-3, 2007.
14. Gaeta, M., Orciuoli, F. , Ritrovato, P.: Advanced ontology management system for person-
alised e-Learning, Knowl. Based Syst. (2009), doi: 10.1016/j.knosys.2009.01.006
15. Learning Technology Standards Comittee of the IEEE: Draft Standard for Learning Objects
Metadata, IEEE Draft P1484.12.1/D6.412, 2002 http://ltsc.ieee.org/doc/wg12/LOM 1484 12
1 v1 Final Draft.pdf.
16. Gruber, T. R.: Ontology of Folksonomy: A Mash-up of Apples and Oranges. Invited keynote,
November 2005.
17. Laniado, D., Eynard, D. and M. Colombetti: Using WordNet to turn a folksonomy into a
hierarchy of concepts. In Proc.of Semantic Web Application and Perspectives. 4th Italian
Semantic Web Workshop, pages 192201, Bari, Italy, 2007.
18. Kim, H.L., Breslin, G.J., Yang, S.K., and Kim, H.G.: Int.ere.st: Building a Tag Sharing Ser-
vice with the SCOT Ontology. In Proceedings of the AAAI 2008 Spring Symposium on
Social Information Processing, Stanford University, California, 2008.
19. Van Damme, C., Hepp, M. and Siorpaes, K.: FolksOntology: An Integrated Approach for
Turning Folksonomies into Ontologies. Bridging the Gap between Semantic Web and Web
2.0 Workshop at the ESWC 2007. 2007, Innsbruck, Austria.
20. Knerr, T.: Tagging ontology- towards a common ontology for folksonomies. (2006) Re-
trieved June 14, 2008, from http://tagont.googlecode.com/files/TagOntPaper.pdf.
21. Passant, A. and Laublet, P.: Meaning Of A Tag: A collaborative approach to bridge the gap
between tagging and Linked Data. In Proceedings of the WWW 2008 Workshop Linked Data
on the Web (LDOW2008) Beijing, China, Apr 2008.
22. Torniai, C., Jovanovic, J., Bateman, S., Gaevic, D., Marek, H.: Leveraging Folksonomies
for Ontology Evolution in E-learning Environments. The IEEE International Conference on
Semantic Computing, 2008.
12