An Approach to Manage the Web Knowledge

Filippo Eros Pani, Maria Ilaria Lunesu, Giulio Concas and Gavina Baralla

Department of Electrics and Electronics Engineering, University of Cagliari, Piazza d'Armi, Cagliari, Italy

Keywords: Knowledge Management, Multimedia Content, Semantic Web, Knowledge Base, Taxonomy.

Abstract: The spread of the Social Web is influencing the evolution of Semantic Web: the way of producing and

consulting information changes, as well as the way people relate themselves with the Internet and the

services it gives. Users will participate at first hand to the developing of the Web which therefore becomes

interactive. This study considers this feature, trying to link the worlds of Social Media and Semantic Web,

with the aim of proposing a semantic classification of the information coming from the Web, which do not

always follow a well-defined order and organization. Starting from a precise analysis of the information of

the Web through an accurate and meticulous study on how these are presented and used, in order to give a

sorted and easily usable data structure, this approach wants to define a taxonomy able to represent

knowledge through an iterative combined approach, where top-down and bottom-up analyses are applied on

the knowledge domain we want to represent.

1 INTRODUCTION AND

RELATED WORK

Over the last decade a broader knowledge of the

Web has strengthened and fostered the developing of

new applications: the Web has turned into a

multifunctional platform where users no longer get

the information passively; in fact, they become

authors and makers. This has been mainly possible

thanks to the developing of new applications which

allow users to add contents without knowing any

programming code. The social value which the Web

has acquired recently is therefore unquestionable;

the Web's structure grows and changes depending on

the user's needs, becoming every day more complex.

The new frontier for the Internet is represented by

the Web 3.0 (Berners-Lee et al., 2001): with the

evolution of the Web into its semantic version, a

transition to a more efficient representation of

knowledge is a necessary step. Particularly, data are

no longer represented just by the description of their

structure (syntax) but also by the definition of their

meaning (semantics). In fact, a data can have a

different meaning depending on the contexts; the use

of tools like ontologies and taxonomies helps the

classification of information, as shown also in

(Decker et al., 2000; Maedche and Staab, 2001;

Jacob, 2003; Davies et al., 2003; Strintzis et al.,

2004; Jewell et al., 2005; Hepp, 2007; Gruber, 2008)

and (Simperi, 2009).

The Web becomes clever and is conceived as a

big database in which data are orderly classified.

“Information”, therefore, is one of the keywords at

the base of the success of both search engines

(Google, Yahoo, Bing, …), which become more

refined in data retrieval and presentation, and Social

Networks (Youtube, Facebook, Twitter, Flickr, …),

which allow exchange and sharing, creating an

interconnection among users and content makers.

However, such data, despite being formally

available, are often unreachable as for their semantic

meaning and cannot be used as real knowledge.

Various proposals to solve these problems can be

found in literature, also to overcome the semantic

heterogeneity problem (Euzenat and Shvaiko, 2007)

and to facilitate knowledge sharing and reuse

(Fensel et al., 2001; Gómez-Pérez and Corcho,

2002). In (Schreiber et al., 2001) an approach based

on the use of an ontology to make annotating photos

and searching for specific images more intelligent is

described; and in (Jaimes and Smith, 2003) a data-

driven approach to investigate semi-automatic

construction of multimedia ontologies is used. With

the emergence of the Semantic Web, a shared

vocabulary is necessary to annotate the vast

collection of heterogeneous media: in (Jewell et al.,

2005) an ontology is proposed to provide a

Eros Pani F., Ilaria Lunesu M., Concas G. and Baralla G..

An Approach to Manage the Web Knowledge.

DOI: 10.5220/0004547900810088

In Proceedings of the International Conference on Knowledge Engineering and Ontology Development (KEOD-2013), pages 81-88

ISBN: 978-989-8565-81-5

 2013 SCITEPRESS (Science and Technology Publications, Lda.)

meaningful set of relationships which may enable

this process.

Particularly, in (Lunesu et al., 2011) the problem

of representing and managing the knowledge which

can be found on the Internet is discussed as for the

User Generated Content (UGC), classifying this

knowledge through a top-down (TD) and bottom-up

(BU) combined approach. To reach such target, an

ontology was built as a base to define a repository of

multimedia contents, putting a special focus on the

georeferencing of multimedia objects. As for the TD

approach, the standards used for the multimedia

objects (XMP, Exif, etc.) have been defined by

selecting data of interest to represent them on the

ontology, which in turn was defined through rules of

correspondence. As for the BU approach, UGCs of

two particularly exemplifying platforms (Flickr and

Youtube) have been analyzed, in order to extrapolate

some structured tags, folksonomies and attributes of

multimedia objects (characteristically Exif, as for

Flickr, and proprietary tags, as for Youtube). In

conclusion, this ontology allowed for the

construction of a repository to store the information

extracted from UGC systems, where all the

information related to multimedia objects are shown

(compatibly with the XMP standard) as well as other

tags of general interest, apart from representing also

the information which can be found on the as-is

folksonomies.

This study aims at defining a new approach for

the problem of the contents on the Internet,

especially semi-structured contents coming from

heterogeneous sources referring to a common

knowledge domain. Through a combined TD and

BU approach, knowledge of a specific domain was

extracted defining a common structure through a

taxonomy, in order to classify and make the majority

of such knowledge available.

With the TD approach the knowledge of interest on

the domain was defined, following the specifications

and the analysis of the ontologies and other

classifications, in order to define a reference

taxonomy.

On the other hand, the BU approach started from

the selection of some websites concerning the

domain of interest, to pinpoint the knowledge in

them. Then, these contents were classified with the

taxonomy previously defined and the mapping rules

between contents and taxonomy.

This taxonomy allowed for the definition of a

reference knowledge which may later be managed in

terms of really usable and interesting knowledge,

fostered by the whole knowledge of all the selected

websites.

We chose to test this approach on the knowledge

domain of Italian wines reviews. As for the

validation, we verified how this KMS allowed such

knowledge to become available on systems that were

compliant with the Wines ontology as defined as an

example of Semantic Web by W3C; then we

checked other websites of Italian wines reviews,

verifying how their contents of interest could be

represented and managed on the KMS through some

simple mapping rules.

The paper is structured as follows: in the second

section of this paper we present our proposed

approach for knowledge management and in the

fourth we explain the case study. The next section

includes the analysis of results and verification.

Finally, the fifth section includes the conclusion and

reasoning about the future evolution of the work.

2 THE APPROACH TO MANAGE

THE KNOWLEDGE

The proposed approach aims at defining a taxonomy

able to represent knowledge through a mixed-

iterative approach, where TD and BU analyses of the

knowledge domain which has to be represented are

applied: these are typical approaches for this kind of

problems. In this case, they are applied following an

iterative approach which allows, through further

refinements, for the efficient definition of the

taxonomy able to represent the domain's knowledge

of interest.

The knowledge to be represented is the most

popular among users of a certain domain. To

determine which is the users' knowledge of real

interest we chose to select the most used websites by

users, the most important and looked up ones. For

this definition, websites with a higher ranking on

Google among the domain of interest are typically

chosen.

2.1 Top-down Phase

When our knowledge or our expectations are

influenced by perception, we refer to schema-driven

or TD elaboration. A schema is a model formerly

created by our experience. More general or abstract

contents are indicated as higher level, while concrete

details (senses input) are indicated as lower level.

The TD elaboration happens whenever a higher level

concept influences the interpretation of lower level

sensory data. Generally, the TD process is an

information process based on former knowledge or

KEOD2013-InternationalConferenceonKnowledgeEngineeringandOntologyDevelopment

acquired mental schemes; it allows us to make

inferences: to “perceive” or “know” more than what

can be found in data. TD methodology starts,

therefore, by identifying a target to reach, and then

pinpoints the strategy to use in order to reach the

established goal.

Our aim is therefore to begin by a formalization

of the reference knowledge (ontology, taxonomy or

others) to start classifying the information on the

reference domain.

The model could be, for instance, a formalization

of one or more classifications of the same domain,

formerly made in a logic of metadata. Therefore, the

output of this phase will be a table with all the

elements of knowledge formalized through the

definition of the reference metadata.

2.2 Bottom-up Phase

With this phase the knowledge to be represented is

analyzed by pinpointing, among the present

information, the ones which are to be represented

together with a reference terminology for data

description.

When an interpretation emerges from data, it is

called data-driven or BU elaboration. Perception is

mainly data-driven, as it must precisely reflect what

happens in the external world. Generally, it is better

if the interpretation coming from a system of

information is determined by what is effectively

transmitted at sensory level rather than what is

perceived as an expectation. Applying this concept,

we analyzed a set of websites containing the

information of the domain of interest; from these

websites, both information whose structure needed

to be extrapolated and the information in them were

pinpointed. Typically, reference websites for that

information domain are selected, namely the ones

which users mainly use to find information of their

interest over the domain itself.

Primary information, important ones, already

emerge during the phase of websites analysis and

gathering: during a first skimming phase, the

minimum, basic information necessary to well

describe our domain can be noticed. Then, important

information are extrapolated by choosing fields or

keywords which best represent the knowledge, in

order to create a knowledge base (KB). In this phase,

one of the limits could be the creation of the KB

itself, because each website is likely to show a

different structure and a different way of presenting

the same information. Therefore, it will be necessary

to pinpoint the present information of interest,

defining and outlining them. After this analysis of

gathering of information, a classification is made

and it has to reflect, in the most faithful way, the

structure of the knowledge proposed by every single

website, respecting both its contents and hierarchy.

To analyse data, we chose to build a tabular

system for each website coming from a precise

identification of each information area existing in

every website taken as a knowledge base.

For each website we created a table which

accurately gathers and describes the information that

can be found in it, with a detailed field of

descriptions. With this stage we obtained a complete

representation of the knowledge which can be found

on the chosen websites, but not a usable one because

it had not been classified yet.

2.3 Integration Phase

In these phases we will try to reconcile these two

representations of knowledge of the domain, as

represented in the former phases.

Thus, we want to pinpoint, for each single TD's

metadata, where the information can be found in the

table's fields representing the knowledge of each

website (which, for us, represents the knowledge we

want to represent, considering the semantic concept

and not the way to represent it, absolutely subjective

for every website).

At this point we check if, in every table, the

information of our representation of knowledge

coming from the TD can be found in the tables

coming from the BU, verifying if it exists as a field

or can be found in a field or is missing.

Then, we will create a mapping macro-table of

knowledge containing, for each item of the

taxonomy, the correspondence if and where that

information exists in the various websites and also

the information of the websites which are not

represented by the taxonomy.

From the macrosystem a KB originates, which is

able to represent both the formalization of

knowledge and the present knowledge.

2.4 Formalization of Knowledge

Starting from this KB, further iterative refining can

be made by re-analyzing the information in different

phases: 1) with a TD approach, checking if the

information which are not represented by the chosen

formalization can be formalized; 2) with a BU

approach, analyzing if some information of the

websites can be connected to formalized items; 3)

with the mixed phase by which these concepts are

AnApproachtoManagetheWebKnowledge

reconciled. This is obviously made only for the

information to be represented.

The knowledge we want to represent is the one

considered of interest by the users for the domain:

for this reason, the most important and looked up

websites are chosen. At the end of this analysis we

will define a taxonomy able to represent the

knowledge of interest for this domain, which may

also not have items from the taxonomy (or ontology

from which we started in the TD analysis), but may

have items which did not exist in it, emerged from

the BU analysis. The final result of this phase will be

a reference taxonomy, where, for each item, there is

a linked information about where the knowledge of

interest can be found on each website.

3 CASE STUDY ABOUT WINES

In this study we chose as a case study the domain of

wines and, particularly, the one belonging to the

technical files and/or descriptions of “Italian wines”:

the choice was not made randomly as the world of

wines is rich in contents and complete enough to

give a good starting point for our study. In fact, there

are thousands of contents which can be found on the

Internet; also, there are different studies on the

classification of wines from which we can draw on.

3.1 Knowledge Base of Interest

Contents on wine available on the Web are

thousands, offering a significant KB.

Our study takes into consideration a subdomain

of wine, represented by all the most important

reviews which can be found on the Internet. From

the analyses of the domain on the web and the

Google Ranking of these websites, we chose a list of

suitable and representative websites, having

considered the popularity and the reliability given by

the Web. The websites we took into consideration

are the following: Decanter.com; DiWineTaste.com;

Lavinium.com; GamberoRosso.it; introspective.com

Snooth.com; Vinix.com. These websites are

considered as representative for our study also

because of their own information structures,

particularly various and differentiated. Each website

has its own structure and a different representation

of the information. To correctly define our domain it

was therefore necessary to precisely analyse the

contents in each of them and the layouts. The

structure of the page showing the review is useful to

understand if the same website always uses the same

structure and the same items for every review.

Unfortunately we saw that some of them show the

same information differently depending on the

review, using, for instance, different tags for the

same data. This, obviously, is a limit in the process

of classification of contents. It is thus necessary to

align the different items for the same website, used

to represent the same data.

3.2 Top-down Phase

In this phase we analyzed the existing formalizations

for the representation of knowledge of this domain.

A very interesting formalization which we

pinpointed was the one by the Associazione Italiana

Sommelier (AIS), The Italian Sommelier

Association, providing a detailed description of all

the terms associated with wine. Another important

formalization was the one by the European law

defining the reference features of a certain wine,

such as type, colour, grape variety, etc. From these

two, a reference taxonomy for those features was

created. As an additional formalization, we chose a

reference scheme, represented by an ontology

already existing on the Web and made by W3C:

wine ontology [http://www.w3.org/TR/owl-

guide/wine.rdf]. An ontology is more complex than

a taxonomy. It has, apart from class hierarchies,

property hierarchies with cardinalities for the

assignable values. It offers a general view of the

world of wines, with a less detailed description for

certain fields as stated on the reviews found on the

Web. Moreover, from this ontology we took into

consideration only the areas of interest existing in

our classification, omitting those ones representing

elements not of interest (such as, for instance, each

winemaker’s property).

Starting from these reference formalizations, a

first taxonomy was built in which we pinpointed the

items to create the reference table. After choosing

the items of interest in the reference ontology, we

analyzed the direct correspondence among tags of

the two representations, directly extracting the

ontology ones from the OWL code. To standardise

our taxonomy we decided to take into consideration

the RDF standard indicating, just for the items with

a correspondence, its URI. The RDF standard allows

to associate a URI also to the properties. website,

used to represent the same data.

3.3 Bottom-up Phase

The BU analysis required a detailed analysis of the

contents of these websites, trying to pinpoint the

information we considered as important; then we

KEOD2013-InternationalConferenceonKnowledgeEngineeringandOntologyDevelopment

studied the structure of each single source, useful to

see the existing data and their position in the layout

of the page.

Once the KB for the domain of interest

composed by the websites was defined, the next step

was classifying all the chosen information. Such

classification is made by creating a classification of

the BU contents because it was built from the

bottom: information on the websites are thus

accurately analyzed.

We start from the analysis of the specific to

reach a general classification of data. One of the

initial steps of our project contemplated the study of

the structure of each source, useful to see the

existing data and their position on the page's layout.

This procedure happens to be important also at

this point of the study, because allows for the

evaluation of the classification of information. Both

the item “maturazione”, but also the organoleptic

analysis (visual, olfactory and gustatory test) if

existing, are systematically shown on the websites

taken into consideration, into the area which we

identified as “tasting notes”. For this reason, to build

the hierarchy we tried to respect the original, already

existing one.

The type of classification was also revealed

during the data analysis phase, during the study of

semantics and uniformation.

With the creation of the tables we tried to

represent the knowledge in the shape of fields as

faithful as possible to those ones already existing in

the samples taken into consideration.

The evaluation of this phase is subjective and left

to the intuition of the analyst, which freely interprets

the information at their disposal, intuitively

obtaining the taxonomic tree. This step happens to

be very tricky, because is susceptible to accidental

mistakes. However, we could say that the various

structures found in the domain which we considered,

apart from the caption used to define each field, are

not so different, thus the classification did not raise

any big doubt, as for the representation.

Thus, the macrosystem made 7 tables, one for

each website. Every table has the list of information

of the website it represents.

3.4 Mixed Phase

During this phase, the items of the fields existing in

the taxonomy defined in the TD phase were

compared to the fields of the tables created in the

BU phase. To do this, we built a mapping macro-

table of knowledge containing, for each item of the

taxonomy, the correspondence if and where that

information exists on the various websites and also

the information existing on the websites which were

not represented by the taxonomy.

To each item we thus assigned a numerical value

to represent this mapping: 1) existing and extractable

information; 2) existing but not extractable

information; 3) sometimes existing and extractable

information; 4) sometimes existing but not

extractable information; 5) always missing

information.

For the fields with values 1 and 3, the

corresponding field and the mapping rule to

extrapolate the information are also indicated.

The information with value 2 and 4 is embedded

(hidden in the text) and, therefore, should be

specifically looked for with tools of semantic

analysis. Anyway, the field in which it exists is

indicated.

With this analysis and classification of every

single data we managed to solve the inhomogeneity

of the information existing in the Web, as for the

domain of interest. This allowed to study both its

structure and the type of information existing, giving

us the chance to examine how data are presented and

the classification given for each website.

When creating the taxonomy, which wants to be

a semantic classification, we also tried to represent

the structure of data and the existing hierarchies of

the sample websites.

3.5 Formalization of Knowledge

This activity was iteratively repeated to best

represent the knowledge and its connections

described in the macro-system mentioned above. As

expected, not all the fields were taken into

consideration, neither among those existing in the

initial taxonomy nor among the extrapolated ones,

and those ones which appear just once in the whole

macro-system were rejected (evaluation made

considering the field value = 5), such as, for

instance, “Bicchiere consigliato” or “Temperatura di

servizio consigliata”.

The inhomogeneity among the information

existing in the different websites was analyzed by

looking for the semantic correspondences

represented in the macrosystem with the column

‘field details’. The same principle was used to

uniform fields with numerical values. The final

range takes into account the classification used by

the majority of websites.

A simplifying table summarizing the procedure

of classification described above is shown below.

The result of these phases was the knowledge

AnApproachtoManagetheWebKnowledge

Table 1: Classification.

Macrosystem Items Final Tags

Wine's identification name Wine

Winery: address, telephone,

fax, e-mail, web, map, other

wine, other info winery

Classification: Vino da

tavola, IGT, DOC, DOCG

<type>

Colour: white, rose, red

<type>

specification

Qualification: embedded

Qualification: classic,

reserve, superior

<uve>

grape

<alcol>

Alcohol

Label Label

<zona>

State/Region

Tasting notes

<$>

Price

Food pairing suggestion

Author

Date

<decanter rating>: max 5

stelle

<rated>: max 5 bicchieri

<valutazione>: max 5

chiocciole

<punteggio>: max 5

diamanti

<voto>: max 5 chiocciole

Punteggio: max 3 bicchieri

Rate: 60-70; 71-75; 76-80;

81-85; 86-90;

91-100

base formalized through the taxonomy. The table

shows some items of it, with a field of value 1 or 3

and expressed in textual form (for instance, those

ones directly extractable through tags or metadata).

Other fields, represented by an icon, were rejected,

though their presence was considered.

4 ANALYSIS OF RESULTS AND

VERIFICATION

During the validation phase we verified how our

KMS made the acquired knowledge usable for the

systems compliant with other ontology of wines and

for other websites on Italian wines reviews. We went

on verifying how the contents of interest of these

websites could be represented and managed on the

KMS through some simple mapping rules.

Then, we tried to solve the clear inhomogeneity

by paying more attention to the semantic meaning

and not to the notation used to represent those

contents. In fact, the purpose of the study was not to

describe the whole world of wines, but just the part

of it represented by the information which can be

found on the Web.

After matching the two systems, Ontology and

Taxonomy, the information were generalized and

made coherent. This allowed us to verify that our

system is able to represent and combine specific

information, and at the same time understands the

main variances between the two systems, namely the

difference of some considered information.

This kind of study can also be used to enrich an

already existing ontology with fields coming from a

general classification, evaluating a possible

integration of such information without damaging

the existing hierarchy, so that we can have a broader

and more accurate view over the analyzed domain.

4.1 Choice of Samples

To continue with the phase of verification of the

created taxonomy, we decided to take into

consideration another set of samples – again, wine

reviews which can be found on the Web.

The choice of the websites for the testing phase

followed the same criteria used during the analysis

of the domain. The main obstacle we found was due

to the popularity of the product and the large amount

of followers who have a very subjective way of

representing the information about wine and the

acquired knowledge. Here comes the need of

pinpointing sources with clear, easily extractable and

objective information.

One of the main features which these sources

needed to have was the presence of differentiated

fields with a single notation rather than a broad

textual field. So, also in this case, all the websites

gathering a large quantity of information in just a

macro-textual area were rejected. In fact, these kind

of websites, though full of contents, were not

suitable for the testing phase. The embedded

KEOD2013-InternationalConferenceonKnowledgeEngineeringandOntologyDevelopment

information, though fostering the acquisition of a

general knowledge, do not facilitate its own

structured classification. Similarly, some apparently

suitable sources happened to have very few contents,

with a database so poor that it did not mention the

most appreciated wines.

After these considerations, the websites we

decided to take into consideration for the tests were

the following: guida-vino.com; vinogusto.com;

kenswineguide.com; buyingguide.winemag.com.

4.2 Testing Phase

For each sample website, in this testing phase we

verified whether the information in them could be

found in the classification proposed by us, and

whether our taxonomy could be able to represent

them. For each website, therefore, the following

table was built, representing the specific fields of

information which was the same for every review

that we analyzed.

Table 2: Testing phase.

Existing

Information

Field Details Taxonomy Item

Label Label's image Wine.label

Producer

About the

producer

Wine.winery

Classification

IGT, DOC,

DOCG

Wine.classification

Grape variety Grape variety Wine.grape

Range of prices Price Wine.prices

Others years Others years

Wine.winery.info

Winery.otherWines

Presentation/

comments

Wine tasting Wine.tastingNotes

Rate: max 5 stars Rate Wine.rate

In the light of the results obtained in this testing

phase, we are satisfied with the taxonomy which we

created. In fact, with this testing phase, we saw that

the classification defined in our study reflects the

type of contents needed. Such classification,

therefore, is usable, re-usable and possibly

extendible to the domain of interest of wine.

5 CONCLUSIONS

The spread of the Social Web is significantly

influencing the evolution of Semantic Web: users

themselves are creating rules for the representation

of information. The structure of the Web grows and

changes giving the user the chance to actively

participate in the developing of the Web. For this

reason, our study took into consideration this feature

with the uniformation of UGCs, trying to link the

two worlds: Social Media and Semantic Web. Also

the main search engines (Google, Yahoo, Bing, …)

and the main Social Network (Youtube, Facebook,

Twitter, Flickr, …) are evolving, specializing and

interconnecting themselves on data retrieval,

presentation, exchange and sharing.

That being so, the basic idea of our study was to

propose a solution to the problem of the different

contents of the Web, coming from different sources

but belonging to the same domain of knowledge.

Our proposal is to define a taxonomy able to

represent knowledge through a mixed iterative

approach, articulated in a top-down analysis and a

bottom-up one of the domain of knowledge which is

to be represented. Thus, first we tried to define the

knowledge of interest on the domain, depending on

the specifications, and through the analysis of the

existing ontologies, in order to define a reference

taxonomy. Then, the knowledge we considered as

important (and as an element of common interest)

was extracted from a selection of websites belonging

to the domain of interest. These contents are to be

classified in the taxonomy mentioned before, also

using mapping rules made ad-hoc. The taxonomy

created allowed for a definition of the reference

knowledge which could then be managed as an

actual usable knowledge, fostered by all the

information existing on the selected websites. Due to

the large amount of the information available, we

chose as domain of knowledge a sub-domain of

wine, represented by the reviews which can be found

on the Web.

From the analysis of the domain on the Web and

the Google Ranking of many websites, we chose a

list of some suitable and representative ones after

considering popularity and reliability given from the

Web.

We chose to validate the proposed approach by

verifying how the KMS allowed to make the

acquired knowledge usable and accessible to the

systems compliant with the Ontology of Wines taken

into consideration along with other websites of

Italian wine reviews, underlining how, also in this

case, the collected information could be represented

and managed on the KMS through some simple

mapping rules. Such a system could be enriched by

deducing an ontology of information existing on the

Web to be compared with another ontology

representing the same domain. A similar comparison

AnApproachtoManagetheWebKnowledge

has the advantage to be simple, less disorganized

and surely less susceptible to mistakes than the one

proposed in our project.

A further, interesting development could be the

creation of repositories able to collect the

information previously classified and, through a

system made ad-hoc, they would be presented to the

final user in a structured and customized way,

depending on the requests, and possibly developing

a graphic interface which could be able to draw the

curiosity and the interest of the user.

REFERENCES

Barton, J., Currier, S., Hey, J. M. N. (2003). Building

Quality Assurance into Metadata Creation: an

Analysis based on the Learning Objects and e-Prints

Communities of Practice. In Sutton, S., Greenberg, J.

and Tennis, J., Eds. Proceedings 2003 Dublin Core

Conference: Supporting Communities of Discourse

and Practice - Metadata Research and Applications,

Seattle, Washington (USA).

Berners-Lee, T., Hendler, J., Lassila, O., 2001. In The

Semantic Web, Scientific American, pp. 28-37.

http://www.scientificamerican.com/2001/0501issue/05

01berners-lee.html

Davies, J., Fensel, D., van Harmelen, F., 2003. Towards

the Semantic Web: Ontology-driven Knowledge

Management, John Wiley & Sons, ISBN:

9780470848678.

Decker, S., Melnik, S., Van Harmelen, F., Fensel, D.,

Klein, M., Broekstra, J., Erdmann, M., Horrocks, I.,

2000. The Semantic Web: the roles of XML and RDF.

In: Internet Computing, IEEE, Vol. 4, Issue 5, pp. 63-

73.

Euzenat, J., Shvaiko, P., 2007. Ontology matching,

Springer-Verlag, Berlin Heidelberg (DE), ISBN: 978-

3-540-49612-0.

Fensel, D., Van Harmelen, F., Horrocks, I., McGuinness,

D. L., Patel-Schneider, P. F., 2001 OIL: an ontology

infrastructure for the Semantic Web. In Intelligent

Systems, IEEE, Vol. 16, Issue 2, pp. 38-45.

Gómez-Pérez, A., Corcho, O., 2002. Ontology languages

for the Semantic Web. In Intelligent Systems, Vol. 17,

Issue 1, pp. 54-60.

Gruber, T., 1993. A Translation Approach to Portable

Ontology Specification. In Knowledge Acquisition,

Vol. 5, pp. 199-220.

Gruber, T., 2008. Ontology. In Liu L., Özsu, M. T. (Eds.),

Encyclopedia of Database Systems, Springer-Verlag.

Hepp, M., 2007. Ontologies: State of the Art, Business

Potential, and Grand Challenges. In Hepp, M., De

Leenheer, P., de Moor, A., Sure, Y. (Eds.), Ontology

Management: Semantic Web, Semantic Web Services,

and Business Applications, ISBN 978-0-387-69899-1,

Springer, pp. 3-22.

Horrocks, I., 2008. Ontologies and the semantic web. In

Communications of the ACM, 51(12):58-67.

Jacob, E. K., 2003. Ontologies and the Semantic Web. In

Bulletin of the American Society for Information

Science and Technology, Wiley Periodicals, Vol. 29,

Issue 4, pp. 19–22.

Jaimes, A., Smith, J., 2003. Semi-automatic, data-driven

construction of multimedia ontologies. In Proceedings

of IEEE International Conference on Multimedia and

Expo (ICME), Vol. 2.

Jewell, M. O., Lawrence, K. F., Tuffield, M. M., Prugel-

Bennett, A., Millard, D. E., Nixon, M. S., Schraefel,

M. C., Shadbolt N. R., 2005. OntoMedia: An

Ontology for the Representation of Heterogeneous

Media. In Multimedia Information Retrieval

Workshop, ACM SIGIR.

Lassila, O., Swick, R., 1999. Resource Description

Framework (RDF): Model and Syntax Specification,

Recommendation W3C. http://www.w3.org/TR/REC-

rdf-syntax

Lunesu, M. I., Pani, F. E., Concas, G., 2011. An approach

to manage semantic informations from UGC. In

International Conference on Knowledge Engineering

and Ontology Development , Paris, France.

Lunesu, M. I., Pani, F. E., Concas, G., 2011. Using a

standards-based approach for a multimedia

knowledge-base. In International Conference on

Knowledge Management and Information Sharing

Paris, France.

Maedche, A., Staab, S., 2001. Ontology learning for the

Semantic Web. In Intelligent Systems, IEEE, Vol. 16,

Issue 2, pp. 72 – 79.

Noy, N. F., McGuinness, D. L., 2001. Ontology

Development 101: A Guide to Creating Your First

Ontology, Stanford Knowledge Systems, Laboratory

Technical Report KSL-01-05, 2001.

Schreiber, A. Th., Dubbeldam, B., Wielemaker, J.,

Wielinga, B., 2001. Ontology-Based Photo

Annotation. In IEEE Intelligent Systems, Vol. 16, pp.

66-74.

Simperi, E., 2009. Reusing ontologies on the Semantic

Web: A feasibility study. In Data & Knowledge

Engineering, Vol. 68, Issue 10, pp. 905-925.

Strintzis, J., Bloehdom, S., Handschuh, S., Staab, S.,

Simou, N., Tzouvatras, V., Petridis, K., Kompatsiaris,

I., Avrithis, Y., 2004. Knowledge representation for

semantic multimedia content analysis and reasoning.

In Proceedings of the European Workshop on the

Integration of Knowledge, Semantics and Digital

Media technology.

W3C, Wine ontology. http://www.w3.org/TR/owl-

guide/wine.rdf

KEOD2013-InternationalConferenceonKnowledgeEngineeringandOntologyDevelopment