Fuzzy-Ontology-Enrichment-based Framework for Semantic Search

Hajer Baazaoui-Zghal and Henda Ben Ghezala

Riadi-GDL Laboratory, ENSI, Manouba University, Tunis, Tunisia

Keywords:

Fuzzy Ontology, Web Search, Query Reformulation.

Abstract:

The dominance of information retrieval on the Web makes integrating and designing ontologies for the on-

line Information Retrieval Systems (IRS) an attractive research area. In addition to domain ontology, some

attempts have been recently made to integrate fuzzy set theory with ontology, to provide a solution to vague

and uncertain information. This paper presents a framework for semantic search based on ontology enrich-

ment and fuzziness (FuzzOntoEnrichIR). FuzzOntoEnrichIR main components are: (1) a fuzzy information

retrieval component, (2) an incremental ontology enrichment component and (3) an ontology repository com-

ponent. The framework aims on the one hand to capitalize and formulate extraction-ontology rules based on

a meta-ontology. On the other hand, it aims to integrate the domain ontology enrichment and the fuzzy ontol-

ogy building in the IR process. The framework has been implemented and experimented to demonstrate the

effectiveness and validity of the proposal.

1 INTRODUCTION

Ontologies are deﬁned as an explicit formal speciﬁca-

tion of a shared conceptualization. They can be classi-

ﬁed as lightweight ontologies gathering concepts and

relations’ hierarchies which can be enriched by classi-

cal properties called axiom schemata (algebraic prop-

erties and signatures of relations, abstraction of con-

cepts, etc.) and heavyweight ontologies which add

properties to the semantics of the conceptual primi-

tives and are only expressible in axiom domain form.

The axioms schemata describe the classical proper-

ties of concepts and relations (subsumption, disjunc-

tion of concepts, algebraic properties and cardinalities

of the relations, etc.). The domain axioms character-

ize domain properties expressible only in an axiom

form. They specify the formal semantics constraining

the conceptual primitive interpretation.

Currently, ontologies are playing a fundamental

role in knowledge management and semantic Web.

Building ontology manually is a long and tedious

task. Thus, many approaches have been proposed dur-

ing the last decade to make this task easier. Infor-

mation Retrieval (IR) deals with models and systems

aiming to facilitate accessibility to sets of documents

and provide to users the corresponding ones to their

needs, by using queries. Generally, Information Re-

trieval System (IRS) integrates techniques allowing

selection of relevant information. The ﬁrst research

on ontologies for the IRS dates back to the late 90s

(McGuinness, 1998), and aims to remedy limits of

the traditional IRS based on the keywords. This re-

search topic presents one of the main actual axes of

the semantic Web.

In addition to domain ontology, the integration

of the fuzzy logic shows that it presents an interest-

ing way to solve uncertain information problems. In

fact, fuzzy logic is used in IR to solve the ambigu-

ity issues by deﬁning ﬂexible queries or fuzzy indexes

(e.g., (Baazaoui-Zghal et al., 2008). A fuzzy ontology

can be considered as an extension of domain ontology

by embedding a set of membership degrees in each

concept of the domain ontology and adding fuzzy re-

lationships among these fuzzy concepts (Zhou et al.,

2006).

In this paper, we present a framework for seman-

tic search based on fuzzy ontologies. It includes an

ontology repository (meta-ontology generating a do-

main ontology and ontology of domain services), in-

cremental approach of domain ontology learning and

fuzzy ontology enrichment method. The framework

has been implemented to demonstrate the proposal ef-

fectiveness and to evaluate it.

The remaining of this paper is organized as fol-

lows. Section 2 presents related works to information

retrieval and fuzzy ontologies. Section 3 describes

our proposal. Section 4 presents and discusses some

experimental results of our framework. Finally, sec-

tion 5 concludes and discusses directions for future

research.

123

Baazaoui-Zghal H. and Ben Ghezala H..

Fuzzy-Ontology-Enrichment-based Framework for Semantic Search.

DOI: 10.5220/0004923801230130

In Proceedings of the 10th International Conference on Web Information Systems and Technologies (WEBIST-2014), pages 123-130

ISBN: 978-989-758-024-6

 2014 SCITEPRESS (Science and Technology Publications, Lda.)

Figure 1: An example of a fuzzy ontology.

2 RELATED WORKS AND

MOTIVATIONS

Several studies were presented showing how the

fuzzy logic could be integrated to IRS, to solve un-

certain information problems (Zhou et al., 2006). We

precise here, that in a fuzzy ontology, each index term

or object is related to every term (or object) in the on-

tology, with a degree of membership assigned to the

relationship and based on fuzzy logic.

The fuzzy membership value µ is used for the re-

lationship between terms or objects, where 0 < µ < 1,

and µ corresponds to a fuzzy membership relation

such as ”strongly”, ”partially”, ”somewhat”,”slightly”

etc, where for each term:

i=n

∑

i=1

= 1 ;

n is the number of relations that a particular object

has, n = (N − 1), with N representing the total num-

ber of objects in the ontology (Lee et al., 2005).

The insertion of the fuzzy logic and the ontol-

ogy in the process of information retrieval has im-

proved the quality and the precision of the returned

results. Thus, integration of the fuzzy ontology into

the IR process is an interesting area of research and

can lead to more relevant results than in the case

where ontology and fuzzy logic are used separately

(Chien et al., 2010; Bordogna et al., 2009; Calegari

and Ciucci, 2006). Several existing IRSs (Zhou et al.,

2006; Chien et al., 2010; Calegari and Ciucci, 2006)

generally use semi-automatic or automatic methods,

which allow the fuzziﬁcation only of the ”IS-A” rela-

tion. In addition, from the state of the art (Chien et al.,

2010; Colleoni et al., 2009), it is noticed that there is a

lack of information retrieval system integrating fuzzy

ontology allowing a document classiﬁcation and as-

sisting users in their searches.

First, in (Sayed et al., 2007), document classiﬁca-

tion is not based on fuzzy ontologies.

In fact, classiﬁcation based only on domain ontology

could not take into account dynamic aspect of fuzzy

ontology, mainly when the aim is to improve query

reformulation and information retrieval results. Both

in (Parry, 2006; Akinribido et al., 2011), only ”IS-

A” relations are taken into account. Nevertheless, all

relations are important mainly in case of query refor-

mulation.

From the conducted survey made on methods for

fuzzy ontology construction, we have noticed that au-

tomatic methods can take as input a database (Lee

et al., ; Quan et al., 2006), a documentary corpus

(Widyantoro and Yen, 2001) or an existing ontol-

ogy (fuzziﬁcation) (Parry, 2006),(Sayed et al., 2007),

(Chien et al., 2010), (Calegari and Ciucci, 2006).

Figure 1 illustrates an example of a fuzzy ontol-

ogy. The number related to relations represents the

membership value of the relationship between the

concept ”Hotel” and other concepts (room, Suite).

The related value to a concept describes the impor-

tance of this concept into the ontology. These differ-

ent relations and concepts will have different mem-

bership values depending on the context of the query,

and particularly the user’s view of the world.

In this paper, our main objective is to develop a

WEBIST2014-InternationalConferenceonWebInformationSystemsandTechnologies

124

framework allowing ontology’s building for the se-

mantic Web. The proposed framework includes an

ontology repository (meta-ontology generating a do-

main ontology and ontology of domain services, and

fuzzy ontology), incremental approach of domain on-

tology learning, fuzzy ontology building method and

ontology based retrieval process.

So, the aims of this paper can be summarized in:

• The integration of the domain ontology enrich-

ment and the fuzzy ontology building in the IR

process.

• The capitalization and formulation of extraction

ontology rules based on a meta-ontology aiming

to explicitly specify knowledge about the con-

cepts, relationships, instances and axioms extrac-

tion, the learned patterns and frames, and the se-

mantic distance

3 FUZZY-ONTOLOGY-

ENRICHMENT-BASED

FRAMEWORK FOR SEMANTIC

The general structure of the framework (FuzzyOn-

toEnrichIR) is given by Figure 2. FuzzOntoEnrichIR

framework is based on a fuzzy ontology building, an

incremental ontology learning and an ontology repos-

itory.

The main FuzzOntoEnrichIR components are: a

fuzzy ontology building component, an incremen-

tal ontology learning component and an ontological

repository component.

The ﬁrst component is composed of:

• Two methods of information retrieval based on a

domain ontology and an individual fuzzy ontol-

ogy.

• An automatic method of fuzzy ontology building:

allowing the fuzziﬁcation of all existing relations

in the initial domain ontology initial, and assur-

ing the updating of membership values at the end

of every information retrieval, which is made dy-

namically by the user

• A classiﬁcation of documents by service using a

domain ontology

This ﬁrst component aims to automate the collection

of the relevant documents which will be used as entry

of the second component

The second component is based on a meta-

ontology (which is a high-level ontology of abstrac-

tion (Baazaoui-Zghal et al., 2007a)) and an incremen-

tal ontology learning which may require enrichment

phase. It allows incremental construction of ontolo-

gies from the Web documents. Thus, this component

proposes a composite ontological architecture of three

interdependent ontologies: a generic ontology of web

sites structures, a domain ontology and a service on-

tology.

These offer a representation of the domain and ser-

vices behind Web content, which could be exploited

by the semantic search engines. This latter is instanti-

ated with the contextual information of concepts and

relations of the ontology extracted incrementally from

texts. The semi-automatic construction of the domain

ontology is the main objective of this component.

The third component is composed of a meta-

ontology generating a domain ontology and ontology

of domain services, and fuzzy ontology.

The details related to each component will be

given in the next subsections.

3.1 Fuzzy Ontology Building

Component

FuzzyOntoEnrichIR integrates an automatic fuzzy

ontology building method, with an automatic fuzziﬁ-

cation of all the existing relations in the domain ontol-

ogy, not restricted to ”Is-a” relations. Indeed, in con-

ventional ontologies, particular objects may occur in

multiple locations. So, a simple expansion that does

not understand the intended location of the query term

may lead to many irrelevant results being returned. A

fuzzy ontology membership value could therefore be

used to identify the most likely location in the ontol-

ogy of a particular term. Each user would have own

values for the membership assigned to terms in the

ontology, reﬂecting their likely information need and

worldview.

Then, the use of an individual fuzzy ontology

approach allows the convenient representation of the

relationships in a domain according to a particular

view, without sacriﬁcing commonality with other

views; the ontology framework is common, just the

membership values are different. An individual fuzzy

ontology using an automatic fuzziﬁcation based is

built.

Initialization of Membership Values

We propose to build an individual fuzzy ontology

using an automatic fuzziﬁcation based on Jiang-

Conrath similarity measure (Jiang and Conrath,

1997). To calculate IC (Information Content), we use

the formula presented by (Seco et al., 2004) which

is based on the structure of the ontology hierarchy.

In fact, this frequency has the advantage of bringing

the occurrence frequency of the concept itself and the

Fuzzy-Ontology-Enrichment-basedFrameworkforSemanticSearch

125

Figure 2: FuzzyOntoEnrichIR famework architecture.

concepts it subsumes, which allows supporting all

relations’ types.

Updating the Membership Value of Concepts and

Relations

We suppose that a deﬁned fuzzy ontology is not avail-

able in any context. Thus, it is necessary to deﬁne an

update process of fuzzy values, taking into account

the user’s needs. The membership value should con-

sider the previous values, the retrieved documents and

the query. In the literature, there are researchers that

have presented similar ways of updating membership

value (Calegari and Ciucci, 2006; Parry, 2006). In-

spired by these methods, we have integrated in our

case two updating membership values respectively for

concepts and relations.

new

= µ

old

µ − µ

old

Q + 1

(1)

where µ

old

is the current membership value, Q is the

number of update performed to this value and µ

new

is the new value. µ is a value that evaluates the new

change added to the relation or the concept. µ must

take into account the query and the returned docu-

ments content that have been selected by the user.

The fuzzy ontology is used for query reformula-

tion and for documents and query indexing. A fuzzy

ontology is an individual an ontology owned by each

user.

To show the purpose of the given formulas, we

take as example the query sent by the user contain-

ing the concept ”Hotel”. The framework computes

themeasures of this concept and all its related con-

cepts (like: Rate, Restaurant, Motel. . . ) using the re-

turned documents selected by the user. Finally, the

membership value of relations using the formula 5 are

also updated. In this example the membership value

of the relation ”has-restaurant” between ”Hotel” and

”restaurant”’ will be updated (cf. Figure 3).

3.2 Ontological Repository Component

A dedicated architecture is proposed based on two in-

terdependent ontologies to build a knowledge-base of

a particular domain, constituted by a set of Web doc-

uments, and associated services. Thus, two ontolo-

gies are distinguished: domain ontology and ontology

of domain services, which are in interaction. These

two ontologies are built in accordance with the meta-

ontology. Domain ontology is a set of concepts, rela-

tions and axioms that specify shared knowledge con-

cerning a target domain. Ontology of domain services

speciﬁes for each service, its provider, its interested

users, possible process of its unrolling, main activities

and tasks that constitute this service (Baazaoui-Zghal

et al., 2007b).

This ontology contains axioms specifying the re-

lations between domain services and precise main do-

main concepts which identify each service. These on-

tologies are semantically linked and relationships be-

tween them are deﬁned. The meta-ontology is a spec-

iﬁcation of meta-models of domain ontology and on-

tology of domain services. Besides, knowledge con-

cerning the semi-automatic construction of domain

WEBIST2014-InternationalConferenceonWebInformationSystemsandTechnologies

126

Figure 3: Relation between domain ontology and service ontology.

ontology is also speciﬁed by this meta-ontology.

The proposed architecture is composed of three

ontologies, namely a generic ontology of web sites

structures, domain ontology and service ontology.

The generic ontology or meta-ontology contains

knowledge representation related to each ontology,

which required knowledge for ontology construc-

tion, knowledge representation speciﬁed by the meta-

model related to these ontologies. It is mostly

based on generic concepts: ”meta-concept”, ”meta-

relation”, and ”meta-arxiom”. The class of ”meta-

concept” is divided into subclasses which repre-

sent respectively the domain meta-concept, the meta-

concept of domain services and the meta-concept of

element. Besides, the class ”meta-relation” and the

class ”meta-axiom” are designed in the same way.

The meta-ontology is, consequently, made up of three

homogeneous knowledge ﬁelds. The ﬁrst ﬁeld is a

conceptualization of knowledge related to learning

concepts, relations and axioms related to a target do-

main.

Besides, for each instance of the class Do-

main Concept and the class Domain Relation, the

technique leading to its discovery is speciﬁed.

The service ontology speciﬁes the common ser-

vices that can be solicited by web users and can be

attached to several ontologies deﬁned on subparts of

the domain (cf. Figure 3 showing the relation between

domain ontology and service ontology)

3.3 Incremental Ontology Learning

Component

The incremental ontology learning component is

based on a process of ontology learning from

Web content according to LEO By LEMO (LEarning

Ontology BY Learning Meta Ontology) approach

(Baazaoui-Zghal et al., 2007b). Our approach is

based on learning rules of ontology extraction from

texts in order to enrich ontologies in three main

phases:

• Initialization phase

• Incremental phase of learning ontology

• Result analysis phase.

The initialization phase is dedicated to data source

cleaning. The input of this phase is constituted by

a minimal ontology, the meta-ontology, the termino-

logical resource ”Wordnet” (Miller”, 1995) and a set

of Web sites delivered by a search engine and classi-

ﬁed by domain services. A minimal ontology is de-

signed and built to be enriched in the second phase.

It is called ”minimal domain ontology” as the number

of concepts and relations are reduced. Consequently,

data source preparation consists of: searching Web

documents related to the domain corresponding to a

query based on concepts describing a target service

(these concepts are obtained from the projection of

the corresponding service speciﬁed in the ontology of

domain services), selecting Web sites provided by a

search engine tool (the number of chosen sites is lim-

ited because analyzing an important number of Web

pages requires very important execution time), classi-

fying Web pages according to domain services.

Finally, cleaning Web pages by eliminating

markup elements and images, text segmentation and

tagging in order to obtain a tagged textual corpus.

One hypothesis is that we deal with Web documents

written in a target language. The meta-ontology ad-

justment is thus done according to linguistic knowl-

edge related to the target language. The second phase

is a learning iterative process. Each one of the iter-

ations is made up of two main steps. The ﬁrst one

is the meta-ontology instantiation and the second one

enables us to apply the meta-ontology axioms related

to the learning of ontology. An iteration is processed

in two steps. In the ﬁrst step, techniques are applied

to the corpus. In this context, we have adapted the

construction of a word space (Baazaoui-Zghal et al.,

2008) by applying the N-Gram analysis instead of a 4-

Fuzzy-Ontology-Enrichment-basedFrameworkforSemanticSearch

127

Figure 4: Modular architecture framework.

gram analysis. We have also proposed a disambigua-

tion algorithm (Baazaoui-Zghal et al., 2007b). It aims

to determine the right sense of a lexical unit. This al-

gorithm is based on the study of term co-occurrence

in the text and the selection of the adequate sense. Be-

sides, we propose to use many similarity measures to

build the similarity matrix which describes the con-

textual similarity between concepts.

The last phase is useful to verify the coherence of

enriched ontology by analyzing learning results. We

admit that the maintenance of the meta-ontology al-

lows the readjustment of the rules according to the

results obtained in order to improve the ontology

construction during a further execution of the sec-

ond phase of the process. Moreover the correction

of meta-ontology generates a more valid ontology

scheme and richer.

4 EXPERIMENTATION AND

RESULTS ANALYSIS

The implementation and experimentations of the pro-

posed framework, have been done in order to eval-

uate the proposed architecture. Figure 4 gives an

idea about the developped modules and the applica-

tion structure. FuzzEnrichIR is composed of the con-

struction and updating module which allows manip-

ulation of the fuzzy ontology. The processing mod-

ule regroups classes assuring the different treatments

done on the request and on the documents (as index-

ing and downloading). The class module regroups the

different useful classes to the document classiﬁcation

by service. A pre-processing module of data sources

which pre-process textual corpus, its POS (Part-of-

Speech) tagging and importation of terminological

and conceptual resources (minimal ontology, Ontol-

ogy of domain services and terminological resource

”Wordnet”). An editing module of the meta-ontology

which allows concepts and ontology axioms update

by integrating the Plug-in of Protege-OWL tool. A

module for domain ontology generation. An alimen-

tation module of the meta-ontology which consists

of conceptual elements in the meta-ontology from

text and implements the incremental process of on-

tology domain construction proposed by the ”LEO-

By-LEMO” approach. Finally, a module of domain

ontology learning which is the result of the associa-

tion and the development of learning techniques set

(concepts and relations).

The framework was implemented in Java, provid-

ing an online service and using the Jena Api to handle

ontologies and Google Api to search through the Web.

Several experiments were conducted to investigate

the performance of our proposal and to evaluate:

• The impact of ontology enrichement on informa-

tion retrieval relevance

• The impact of fuzzy ontology enrichement on in-

formation retrieval relevance with and without up-

date

The adopted protocol is centered on users, and the

used data for the experimentation and the evaluation

was composed of a domain ontology and users’ re-

quests. Fifteeen queries in the tourism’s domain and

WEBIST2014-InternationalConferenceonWebInformationSystemsandTechnologies

128

Figure 5: Scenarios’ results.

ten users were considered. Users evaluate the results

obtained by using the domain ontology, the individ-

ual fuzzy ontology and the updated individual fuzzy

ontology.

Three scenarios were designed to evaluate the pro-

posed framework:

• Scenario based on domain ontology (Sc1)

• Scenario based on individual fuzzy ontology,

without update (Sc2)

• Scenario based individual fuzzy ontology, with 4

updates (Sc3)

Figure 5 shows the results in terms of precision (P)

and recall (R), for Top 5, 10, 20, 30 and 50 retrieved

documents. To evaluate the recall values, we consider

the same queries used for the precision. We anal-

ysed the ﬁrst 50 relevant URL’s returned by every sce-

nario. The obtained precision and recall related with

FuzzOntoEnrichIR for scenario Sc3 are clearly higher

than the ones obtained by Sc1 and Sc2. Indeed, re-

sults related to the comparison of exact precision ob-

tained with FuzzOntoEnrichIR based on domain on-

tology only, FuzzOntoEnrichIR based on fuzzy on-

tology with and without update, present a precision

of 0.95 which is superior to the two other scenarios.

These results show that the use of fuzzy ontology sup-

porting update process increases the precision more

than the use of simple domain ontology.

To complete these results we computed the im-

provements given by Table 1, which show that indi-

vidual fuzzy ontology updated four times is reported

to achieve 18,75% precision and 37,50% recall. In-

deed, the results show that the use of fuzzy ontology

increased the precision more than the use of simple

domain ontology.

In the experimentations, the initial used ontology

and fuzzy ontology are composed of 8 concepts, en-

riched with 20 concepts after the ﬁrst iteration, with

Table 1: The improvement of the average recall and the av-

erage precision of the FuzzEnrichOntoIR framework.

Precision/ Precision Recall

Recall improvement improvement

TS3 vs.TS1 TS3 vs.TS1

(in %) (in %)

P/R 05 18,75 37,50

P/R 10 16,25 35,29

P/R 20 14,28 25,00

P/R 30 12,00 22,44

P/R 50 07,50 17,50

20 concepts, enriched with 27 concepts after the sec-

ond iteration, enriched with 40 concepts after the third

iteration, and enriched with 100 concepts after the

fourth iteration. From the obtained results after the

experimentations, we note that the incremental en-

richment of the domain ontology improves the rele-

vance.

However, relevance becomes stable after the third

iteration, when the size of the ontology is enough

great to cover a unique complex query. Indeed, the

100 concepts don’t cover the ﬁeld of the same request,

but they serve to other composed requests in the same

domain. For this reason, the variance of the relevance

of the ﬁrst iteration of the enrichment has a remark-

able impact on the relevance of improvement.

5 CONCLUSIONS AND FUTURE

WORKS

In this paper, we proposed a fuzzy-ontology-

enrichment framework based on fuzzy ontology,

namely FuzzOntoEnrichIR. Since ontologies have

proven their capacity to improve IR, fuzzy ontology-

based IR is becoming an increasing research area.

FuzzOntoEnrichIR’s framework takes place in four

Fuzzy-Ontology-Enrichment-basedFrameworkforSemanticSearch

129

main phases:

• Initialization of membership values,

• Updating the membership value of concepts and

relations,

• Updating the membership value of the existing

concepts in the user’s query

• Updating the membership value of relations re-

lated to the existing concepts in the user’s query.

Fuzzy ontologies buiding method is integrated to

IR process, and returned results are classiﬁed taking

into account fuzziﬁed relations.

So, in this work, our ﬁrst contribution concerns the

fuzzy ontoly’s building process. Our method consid-

ers automatic fuzziﬁcation of a domain ontology tak-

ing into account both taxonomic and non taxonomic

relations, however, all relations are important mainly

in case of query reformulation.

Second contribution concerns the integration of

our fuzzy ontology method into the IR process. In-

deed, query reformulation is based on the weights as-

sociated to all the relations existing in the fuzzy ontol-

ogy, and this fuzzy ontology is used to classify docu-

ments by services.

Finally, the obtained results establish the great

interest and FuzzOntoEnrichIR’s contribution to im-

prove the performance of the retrieval task. Experi-

ments and evaluations have been carried out, which

highlight that overall achieved improvement are ob-

tained thanks to the integration of fuzzy ontologies

into IR process, integration of update and classiﬁca-

tion. These components contribute to signiﬁcantly in-

crease the relevance of search results, by enhancing

documents ranking as shown by the obtained results.

As an evolution of this work, integration of mod-

ular ontologies in order to facilitate the updates is in

progress. Otherwise, the ontology will be extended

to different domains so that architecture will support

a multi-domain use of the ontology. A multi-domain

retrieval based on modular and fuzzy ontologies will

be possible.

REFERENCES

Akinribido, C. T., Afolabi, B. S., Akhigbe, B. I., and Udo,

I. J. (2011). A fuzzy-ontology based information re-

trieval system for relevant feedback. In International

Journal of Computer Science Issues.

Baazaoui-Zghal, H., Aufaure, M.-A., and Mustapha, N. B.

(2007a). Extraction of ontologies from web pages:

Conceptual modelling and tourism application. Jour-

nal of Internet Technology (JIT), Special Issue on On-

tology Technology and Its Applications, 8:410–421.

Baazaoui-Zghal, H., Aufaure, M.-A., and Mustapha, N. B.

(2007b). A model-driven approach of ontological

components for on- line semantic web information re-

trieval. Journal of Web Engineering, 6(4):309–336.

Baazaoui-Zghal, H., Aufaure, M.-A., and Soussi, R. (2008).

Towards an on-line semantic information retrieval sys-

tem based on fuzzy ontologies. JDIM, 6(5):375–385.

Bordogna, G., Pagani, M., Pasi, G., and Psaila, G.

(2009). Managing uncertainty in location-based

queries. Fuzzy Sets and Systems, 160(15):2241–2252.

Calegari, S. and Ciucci, D. (2006). Towards a fuzzy ontol-

ogy deﬁnition and a fuzzy extension of an ontology

editor. In ICEIS (Selected Papers), pages 147–158.

Chien, B.-C., Hu, C.-H., and Ju, M.-Y. (2010). Ontology-

based information retrieval using fuzzy concept docu-

mentation. Cybernetics and Systems, 41(1):4–16.

Colleoni, F., Calegari, S., Ciucci, D., and Dominoni, M.

(2009). Ocean project a prototype of aiwbes based on

fuzzy ontology. In ISDA, pages 944–949.

Jiang, J. J. and Conrath, D. W. (1997). Semantic similar-

ity based on corpus statistics and lexical taxonomy.

CoRR, cmp-lg/9709008.

Lee, C.-S., Jian, Z.-W., and Huang, L.-K. (2005). A fuzzy

ontology and its application to news summarization.

IEEE Transactions on Systems, Man, and Cybernet-

ics, Part B, 35(5):859–880.

Lee, C.-W., Shih, C.-W., Day, M.-Y., Tsai, T.-H., Jiang, T.-

J., Wu, C.-W., Sung, C.-L., Chen, Y.-R., Wu, S.-H.,

Hsu, and Wen-Lian. Asqa: Academia sinica question

answering system for ntcir-5 clqa.

McGuinness, D. L. (1998). Ontological issues for

knowledge-enhanced search. In Proceedings of For-

mal Ontology in Information Systems.

Miller”, G. A. (1995). ”wordnet: A lexical database for

english”. Commun. ACM, 38(11):39–41.

Parry, D. (2006). Chapter 2 fuzzy ontologies for informa-

tion retrieval on the {WWW}. In Sanchez, E., editor,

Fuzzy Logic and the Semantic Web, volume 1 of Cap-

turing Intelligence, pages 21 – 48. Elsevier.

Quan, T. T., Hui, S. C., Fong, A. C. M., and Cao,

T. H. (2006). Automatic fuzzy ontology generation

for semantic web. IEEE Trans. Knowl. Data Eng.,

18(6):842–856.

Sayed, A. E., Hacid, H., and Zighed, D. A. (2007). Using

semantic distance in a content-based heterogeneous

information retrieval system. In MCD, pages 224–

237.

Seco, N., Veale, T., and Hayes, J. (2004). An intrinsic infor-

mation content metric for semantic similarity in word-

net. In ECAI, pages 1089–1090.

Widyantoro, D. and Yen, J. (2001). A fuzzy ontology-based

abstract search engine and its user studies. In Fuzzy

Systems, 2001. The 10th IEEE International Confer-

ence on, volume 3, pages 1291–1294.

Zhou, L., Zhang, L., Chen, J., Xie, Q., Ding, Q., and Sun,

Z. X. (2006). The application of fuzzy ontology in

design management. In IC-AI, pages 278–282.

WEBIST2014-InternationalConferenceonWebInformationSystemsandTechnologies

130