USING A STANDARDS-BASED APPROACH FOR A
MULTIMEDIA KNOWLEDGE-BASE
Maria Ilaria Lunesu, Filippo Eros Pani and Giulio Concas
Department of Electrics and Electronics Engineering, University of Cagliari, Piazza d'Armi, Cagliari, Italy
Keywords: Multimedia content, Ontology, Mapping, Knowledge-base.
Abstract: In recent years, we witnessed the diffusion and rise in popularity of software platforms for User Generated
Content management, especially multimedia objects. These platforms handle a big quantity of unclassified
information. UGC sites (i.e. YouTube and Flickr) do not force the users to perform classification operations
and metadata definitions, leaving space to a logic of free-tags (Folksonomies). In the context of an industrial
project financed by the Autonomous Region of Sardinia, the idea of producing a Geolocalized Guide based
on a Knowledge-base came forth. Such Guide would be able to share georeferenced content with their users,
originated from UGC sources as well as from users themselves. For this purpose, we defined an ontology
that can represent the semantics of multimedia content, especially its metadata, which in turn can be given
an unambiguous meaning. The innovation in this work is represented by the use of the Adobe XMP,
DUBLIN CORE, EXIF, IPTC standards as a starting point. In order to unify metadata coming from
different sources we defined all laws of mapping toward a structure defined by sources like YouTube and
Flickr.
1 INTRODUCTION
It has been observed how, in recent years, many
software platforms managing big quantities of
multimedia content have risen in popularity within
the Web-2.0. UGCs in particular, the most famous of
which are Youtube, Flickr, Del.icio.us, Zooomr,
Picasa, own their great success to a spread of digital
technology accessible by a mass, paralleled by the
quantity and quality of the services offered. The
prominent features of such platforms are their ease
of use, the possibility for users to create and manage
their own spaces (personal channels or pages),
carrying and sharing any kind of multimedia content
from various sources, the implementation of
efficient content research and localization methods,
the definition of access and usage types for them,
and storage of information about legal restrictions
and rights management. When the first problems
about interoperability of applications and
management of shared means arose inside those
platforms, moving on to a more effective
representation of knowledge became a necessity,
along with the evolution of the Web in its semantic
form named Web 3.0.
In particular, we wanted data and resources to be
conceived and represented not through a description
of their structure (syntax), but by a definition of their
meaning (semantics), and for this to be shared within
the community using those same data.
In the context of an industrial project financed by
the Autonomous Region of Sardinia, we fostered the
idea of developing a Guide based on a Knowledge-
base which could manage content from UGC
sources like Flickr and YouTube available for their
users, in addition to contents the users add into the
system, with georeferenced information.
The purpose of this study is to give an
unambiguous meaning to those information so that
they can be managed with a single Knowledge-base.
In order to achieve this goal we chose to define
an ontology that is able to represent the semantics of
these multimedia contents and their metadata, with
the most used standards to define metadata in this
domain (Adobe XMP, DUBLIN CORE, EXIF,
IPTC) as starting points as proven by Gruber (2008).
The innovation of this work is that we chose to
base our approach on the use of the Adobe XMP
standard as a description of the domain of
multimedia metadata, which gives the opportunity to
exhaustively represent multimedia content and its
metadata as well as the other mentioned standards
(DC, EXIF e IPTC).
87
Lunesu M., Pani F. and Concas G..
USING A STANDARDS-BASED APPROACH FOR A MULTIMEDIA KNOWLEDGE-BASE.
DOI: 10.5220/0003663300870095
In Proceedings of the International Conference on Knowledge Management and Information Sharing (KMIS-2011), pages 87-95
ISBN: 978-989-8425-81-2
Copyright
c
2011 SCITEPRESS (Science and Technology Publications, Lda.)
In the second section of this paper we recall
some aspects about the state of the art concerning
ontologies and standards used to represent metadata
related to multimedia contents.
In the third, we present our methodology and
address the analysis of the metadata typical of our
two UGC sources, Youtube and Flickr, and
Folksonomies. We also describe the testing done,
with a description of the different phases of ontology
building and mapping technology we developed.
The fourth section includes the conclusion and
reasonings about the future evolution of the
proposed project.
2 STATE OF THE ART
2.1 Standard Description
Ontologies are an increasingly popular tool thanks to
the advantages they offer when sharing information.
For a few years they have played a leading role in
the representation and utilization of knowledge
processes. In the past, the study of ontology focused
mainly on its philosophical context, but recently it
has assumed an important role in a different research
and industrial section. Ontologies are in fact able to
isolate, retrieve, organize and integrate information
according to their core feature, their semantic
context. Multimedia Ontologies have many
application fields, including Content Visualization,
Knowledge Sharing and Learning as proven by
Jaimes and Smith (2003) and by Kameas and
Seremeti (2007). Their structure and semantics are
accurately modelled to be broadly consistent with
existing multimedia description standards like
MPEG-7 as shown by Martínez, Koenen and Pereira
(2002).
There are two different approach to this kind of
ontologies. Media-Specic Ontologies (that use
different kinds of multimedia taxonomies and
describe different kinds of object properties) and
Content-Specic Ontologies (that describe the
content of resources, e.g. the scenarios or the
participant) as proven in Alesso and Smith (2008).
The construction of multimedia ontologies is
rather complex, it is an iterative process whereby a
phase for the selection of concepts to include in the
ontology, a phase to create properties and relations
linking them together, and a phase for maintenance
of the ontology. We could mention many attemps to
create a multimedia content ontology. In (Jaimes and
Smith, 2003) and (Benitez and Chang, 2003) new
methods for extracting semantic knowledge from
data are presented.
In (Strintzis et al., 2004), we used a Visual
Descriptors Ontology and a Multimedia Structure
Ontology with a domain ontology that aims to
support the note of content. There are other works as
shown in (Bertini et al., 2005), (Bertini et al., 2006),
(Petridis et al., 2006) that are not applicable in this
contest.
As regards our work (that we will describe in the
next sections) we referred to standards as domain
reference, which fit with the management and
categorization of different types of content and
georeferenced data. These standards will be
described below.
2.1.1 XMP Standard
The Adobe Extensible Metadata Platform (XMP) is
a standard, created by Adobe Systems Inc., for
processing and storing standardized and proprietary
information relating to the contents of a file.
XMP standardizes the definition, creation, and
processing of extensible metadata. Serialized XMP
can be embedded into a significant number of
popular file formats, without breaking their
readability by non-XMP-aware applications.
Embedding metadata avoids many problems that
occur when metadata is stored separately. XMP is
used in PDF, photography and photo editing
applications.
XMP encapsulates metadata inside the file, using
RDF (Resource Description Framework), a basic
tool proposed by W3C for encoding, exchange and
reuse of the structured metadata as proven by W3C.
The standard allows, in addition to other things,
interoperability between the different applications
that interact on the web. The reason for its use is that
it is a common standard for a wide range of
applications, which allows us to work efficiently and
effectively on metadata. These properties have
encouraged the rapid spread at many companies
operating in the digital media, which integrate their
applications with this technology. XMP has been
designed and thought also to define, create and
elaborate user-defined metadata which are compliant
with the standards.
2.1.2 Dublin Core (DC) Standard
Dublin Core is a standard for metadata that consists
of a core of essential elements for the description of
any digital material accessible via a computer
network as proven by (Becker et al., 1977). They
proposed a set of 15 basic elements extended also to
KMIS 2011 - International Conference on Knowledge Management and Information Sharing
88
sub-elements or qualifiers: each element is defined
using a set of 10 properties obtained by a standard
ISO 11179. The main features of DC are follows:
1) ease of use the standard is aimed at specialized
cataloguers which are not experts in cataloguing, as
users;
2) semantic interoperability, which gives rise to a
complex and precise data system the meaning of
which has been agreed in advance, along with a
value that allows the DC to be a standard for quality
research in Internet; flexibility, as it allows you to
integrate and develop the data structure with
different semantic meanings and a congenial
application environment.
2.1.3 EXIF Standard
The EXIF standard was created by Japan Electronics
and Information Technology Industries Association
for metadata of digital images. This is a structure
supported by the main producers of cameras and is
studied to give users the opportunity to supply
photos with interchangeable information between
imaging devices to improve processing and printing.
The rapid spread of digital cameras and related
tools increased the need to exchange images directly
from cameras or other instruments, or to display an
image taken with a camera through either another, or
a different device altogether.
EXIF offers a set of specific tags in itself,
concerning shooting parameters and settings of the
device at the time of capture (JEITA, 2002).
2.1.4 International Press
Telecommunication Council (IPTC)
Standard
IPTC is a standard that offers an advantage to
relations and exchanges among entities devoted to
information creation and distribution. It is sponsored
and defined by a consortium based in London that
encompasses the leading news companies in the
information world, such as Reuters, Associated
Press and France Press. IPTC does not hold, among
its metadata, fields related to technical information
on a digital object; in fact, the metadata of the digital
object itself, present in other standards like EXIF,
are not defined. The focus of IPTC in defining the
standard is on analyzing what surrounds the many
situations of telecommunications, and on studying
their production process. Therefore, a range of
metadata were defined, which are useful to define
and certificate all digital object production activities
for print or editing (IPTC, 2008).
2.2 Support Tools
The work needed a preliminary and careful study of
ontologies and the most suitable tools for ontology
editing. We chose Protégé (Stanford Center for
Biomedical Informatics Research), a knowledge-
base and open source framework developed at
Stanford University.
Protégé is able to implement a rich set of
structures and knowledge-modelling actions to
support creation, viewing and manipulation of
ontologies in various representation formats.
Below is an overview of the software platforms
analyzed and taken as an example for mapping
strategy development.
2.3 YouTube
YouTube is an Internet site for video sharing.
Founded in February 2005 by Chad Hurley, Steve
Chen and Jawed Karim (all former PayPal
employees), it is now property of Google Inc. It is
the third most visited site in the world next to
Google itself and Facebook.
YouTube uses the Adobe Flash technology to
play its contents, and it aims to host only videos
created directly by the uploader. This platform
makes an intensive usage of feeds containing
objects, such as web link to content sources. The
interaction between YouTube and clients is managed
through a protocol named YouTube Data API
Protocol, a program communication interface
application.
The data existing in the API are shown by the
protocol as views or projections, with the ability to
modify the form a feed is to be presented with.
However, content is preserved as it is. In this way,
two different projections from the same feed will
identify the same objects, but using different XML
tag sets in (Bray et al., 1998) and (W3C, 2007).
2.4 Flickr
Flickr, developed by Ludicorp (a Canadian company
in Vancouver founded in 2002), is a multilingual
web site that allows users to share personal pictures
with whoever has access to the Internet, in a Web
2.0 environment. The site, owned by the Yahoo!
group, has an ever growing library and was one of
the first to implement tag clouds, visual
representations of user-generated tags. Tag clouds
allow access to images tagged with the most popular
keywords. Thanks to this support for tags, Flickr
was mentioned as the first example of actual
USING A STANDARDS-BASED APPROACH FOR A MULTIMEDIA KNOWLEDGE-BASE
89
folksonomy use, although Thomas Vander Wal
suggested Flickr is not the best example (Vander
Wal, 2007).
Flickr supports standard metadata sets (it shows
the entire Exif metadata set for every picture),
keywords for searches and a group of tags belonging
to folksonomies.
As far as the georeferencing is concerned, Flickr
allows users to organize their pictures in 'sets', that is
groups of images sharing the same gallery. Sets are
more flexible than the traditional folder organization
method for files: a picture can belong to one or more
sets, or to none. Those sets represent a form of
category metadata, instead of a physical hierarchy.
The pictures in a set can be geotagged, and every
set of geotagged pictures can be put in relation with
a map using ImapFlickr2. Such a map can then be
embedded in Web sites.
2.5 Folksonomies
Folksonomies are a set of terms used by a group of
users to tag its own contents (Vander Wal, 2007).
An important aspect of folksonomies is that they
are made up of terms belonging to an open
namespace, i.e. there is neither a hierarchy nor a
parent-child relation among the terms. The only
standard they conform to is the one the tags are
formalized with: namespace:property=value (with
no rules on the other three parts that complete the
assertion) (Vander Wal, 2005); (Mathes, 2004).
3 ONTOLOGY BUILDING
Considering the current literature, the approach we
followed in building a multimedia content ontology
assumes the XMP, Dublin Core, EXIF and IPCT
standards, as well as the related XML schemas and
the integration with the semantics through RDF, as a
basic Knowledge-base for the starting domain as
proven by Lassila and Swick (1999) and by Brickley
and Guha (2000). Thus a complete modelling of the
domain of multimedia content properties, coming
from different sources, is guaranteed together with
an uniform representation of the variety of
associated metadata.
The reason for choosing this approach lies in the
fact that such standards allow for cataloguing
different aspects of multimedia content and natively
possess the specification tools for georeferenced
information. The ontology was then modelled on
those standards, selecting the relevant elements.
Once the basic ontology was decided, an analysis
on cataloguing and classification of metadata in
contents from the main software platforms of the
Net followed. Thanks to this we could acknowledge
alternative standards and proprietary formats used.
Given the great number of available platforms on
the Net and since analyzing all of them would have
been too onerous for us, it was necessary to narrow
the scope and choose which ones should be
considered.
After a study of all the features related to
metadata on the chosen UGC sites, we worked on a
mapping mechanism that allows such data and
associated metadata to be represented within the
ontology.
3.1 Ontology Modelling on Standards
The first step in creating an ontology implies
acquiring knowledge about the domain to be
modelled (Noy and McGuiness, 2001). In order to
do this, we started from the assumption that the
reference domain is the one that includes every kind
of multimedia content, both currently available in
the web or through modern digital technologies,
equipped with sets of metadata belonging to the
above mentioned standards.
The purpose of the ontology is to model the
semantics of metadata from various multimedia
contents, providing for georeferencing and mapping
of the different standards related to metadata. Their
representation can thus comply with the
reconciliation standard provided by the MWG and
with Adobe XMP.
The ontology must be able to receive a content
coming from social networks or software platforms
for content management without information loss or
alteration.
The resulting ontology can be used as a
Knowledge-base supporting the Geolocalized Guide.
As per the definition of the ontology, we chose to
categorize the concepts according to the expression
sublanguage named OWL DL (Description Logic),
due to its computational completeness, its
decidability and the fact it guarantees maximum
expressivity (Carroll and De Roo, 2004);
(McGuinness and Van Harmelen, 2004); (Heflin,
2004). The structure of the ontology was created by
modelling the concepts mentioned above as classes
or properties, following a middle-out approach. First
of all we proceeded with the definition of relations
and main entities which were progressively
generalized and specialized.
These structures were integrated with RDF
schemas. In particular, the entire set of metadata
KMIS 2011 - International Conference on Knowledge Management and Information Sharing
90
required by the EXIF standard, together with the
entire Dublin Core set (complete with its refiniment
terms), was imported. Both schemas allow the
ontology to exploit their metadata, making them
available as particular properties, datatype
properties and object properties at the same time,
probably so as to satisfy every kind of usage needs.
In this specific case they were used solely as
object properties, i.e. to link class instances with
other class instances. The main classes involved in
ontology building are basically four:
1) MultimediaContent: this class models the
concept of multimedia content. It is a simple class,
without subclasses, which formalizes its link with
the class representing file formats
(MultimediaFormat).
2) MultimediaFormat: represents the most common
file formats currently available in the Net. This class
has a two-level hierarchy. The first level represents
format file categorizations depending on the content
type they express. The second level is represented
within each categorization, where classes,
representing the actual formats, are located. Each
format is identified by its own extension.
3) Metadata: its subclasses represent every type of
metadata considered in the study of reference
standards and reference application context.
4) XMPtype: all properties concerning the Metadata
class and its subclasses have the XMPtype class as
codomain. It includes a number of classes which
represent the different data type the XMP standard
uses to describe information inside its tags. One
problem is that some applications avoid the complex
operation that stores information inside files. They
opt instead for executing it in external files or
databases, although that operation could lead to the
loss of metadata as well, when the same file is used
in different applications. XMP, for example, is one
of the standards that requires writing of its own
metadata set inside the file, but it is not the only
standard that enables this action. Every file format
often has its own blocks, different from the ones
XMP uses, to store certain metadata schemas. For
example, a JPEG image has some containers for
storage of the EXIF, IPTC-IIM, and Photoshop
standards.
Metadata are stored in different semantic groups
inside each block. For example, the following
groups can be found inside the XMP APP1 block:
Dublin Core, IPTC-Core, EXIF/TIFF; inside the
Photoshop APP13 block is the IPTC-IIM group.
This problem required a data reconciliation
which was performed through the mapping
technique. The creation of the mapping meant the
execution, where feasible, of a set of non-
automatable, strongly subjective operations.
The search for XMP tags that could map the ones
used on the analyzed platforms was an integral part
of our work.
We searched for tags with the same semantics as
the ones we needed, among those available in the
standards within XMP. This search was performed
with particular care so as to avoid mistakes due to
unclear or poor descriptions and consequent
semantic association mistakes.
Figure 1: XMP block's Schema.
3.2 Mapping of UGC Metadata
YouTube. A significant part of the metadata used by
YouTube was used in order to define the Ontology.
One of the problems we faced during the
mapping was reducing the semantics of certain tags
to a single representation.
Each tag was represented with all of its attributes
and subtags within the ontology. The significance of
the information YouTube associates with some tags
depends on the feed where they appear.
The set we chose to represent was the one related
to the video feed, i.e.:
id,published?,updated,category*,title,c
ontent?,link*,author?,gd:comments?,
media:group?,yt:statistics?,gd:rating?,
yt:location?,yt:recorded?,yt:accessCont
rol+ georss:where?, app:control?
This subset describes the information related to
videos, which are the main content of YouTube, and
has pieces of information that are to be taken into
account when complying to specifications, such as
comments and georeferencing information.
However, due to the importance of YouTube,
this subset was represented whole in the Ontology.
Non-mappable tags were created ex novo as well as
implication relations that could tie together
mappable tags with 'mapping' tags.
USING A STANDARDS-BASED APPROACH FOR A MULTIMEDIA KNOWLEDGE-BASE
91
Table 1: XML Schemas and related namespaces.
Schema Namespace
Prefix
Schema URL
Atom
Syndication
Format
Default
namespace
www.w3.org/2005/Atom
Open Search
Schema
openSearch
A9.com/-
/spec/opensearch/1.1
Media RSS media Search.yahoo.com/mrss
YouTube XML
Schema
yt
Gdata.youtube.com/schemas
/2007
Google Data
Schema
gd schemas.google.com/g/2005
GeoRSS georss www.georss.org/georss
Geography
Markup
Language
gml www.opengis.net/gml
Atom Publishing
Protocol
app www.w3.org/2007/app
Google Data API
Batch Processing
batch
schemas.google.com/gdata/b
atch
The mapping happened in two different ways,
depending on whether the semantic correspondence
was direct (same meaning of information, and same
format, same data type as well) or indirect, that is to
say there was discordance in its form (same meaning
but different representation. In the direct case, we
exploited the feature by which it is possible to create
property hierarchies: each property can have its own
subproperties, which specialize their superproperties
just like a subclass specializes a superclass. This
means that an implication relation among nested
properties is in place: if the superproperties have a
domain and a codomain, those will be necessarily
inherited by their subproperties. Even on a visual
level, mapped tags will appear under the mapping
ones. Therefore direct mappings were performed by
assigning the mapped tag its mapping tag as
superproperty.
To make such operation clearer and the ontology
more readily accessible by users, every direct
mapping came together with an annotation of the
rdf:comment type with information related to the
'mapping' tag.
On the other hand, in the indirect case the
implication relation cannot be used, because the
information must be broken down in its elementary
parts first, and then those parts must be traced back
to direct mode. These steps are described inside the
rdf:comment associated to the mapped tag.
In particular it explains how to split and convert
the information, and where to store it.
Flickr. Our approach to Flickr was quite different
compared to YouTube. It was firstly because of how
metatada related to available content were managed,
and secondly due to the lack of documentation about
them. Initially Flickr used to equip its content with a
simple set of pure Exif data, so natively mappable in
the ontology because they strictly complied with the
standard.
Folksonomies. In the more recent years a reversal
took place, and metadata in Flickr started becoming
part of the so-called Folksonomies. As previously
discussed on section two, the usage of folksonomies
causes metadata not to belong to any kind of
hierarchy. They become unidentifiable in a
namespace and make recognition and mapping
impractical. For this reason the tags of Flickr not
included in the Exif standards were represented with
a class called 'FlickrFolkosomies' inside the
ontology. This class has, as a property, a set of tags
that allow to generate a Feed Atom, that in turn
includes all such information in bulk, non-
standardized.
Metadata belonging to standards but non-mapped
and lacking on the ontology Knowledge-base were
mapped and presented in the same way.
In particular, the set of unknown metadata or
metadata belonging to folksonomies must be stored
inside the Atom Syndication Format tag
atom:content. “This specification describes Atom's
XML markup vocabulary. Markup from other
vocabularies (foreign markup) can be used in an
Atom Document. Note that the atom:content element
is designed to support the inclusion of arbitrary
foreign markup.” (The Internet Society, 2005).
3.3 Mapping Example
We shall now consider the mapping applied to the
metadata related to the 'Hammamet' picture on
Flickr. The metadata associated to this picture by
Flickr are listed in Table 2.
These metadata are partly complying with the
Exif standard (and mapped with the typical rules of
the standard as such) and partly belong to
Folksonomies.
The above mentioned mapping rules were
applied, and part of the data were inserted, while the
rest was inserted in the FlickrFolksonomies class.
KMIS 2011 - International Conference on Knowledge Management and Information Sharing
92
Table 2: Tag EXIF.
Fotocamera Panasonic DMC-FS5
Esposizione 0,002 sec (1/640)
Aperture f/5.6
Lente 5.8 mm
ISO 100
Exposure Bias 0 EV
Flash Auto, Did not fire
Orientation Horizontal (normal)
X-Resolution 180 dpi
Y-Resolution 180 dpi
Software f-spot version 0.5.0.3
Date and Time
(Modified)
2009:08:21 19:02:51
YCbCr Positioning Co-sited
Exposure Program Landscape
Date and Time
(Original)
2009:08:19 08:07:14
Date and Time
(Digitized)
2009:08:19 10:07:14
Compressed Bits
Per Pixel
4
Max Aperture
Value
3.3
Metering Mode Multi-segment
Light Source Unknown
Color Space sRGB
Sensing Method One-chip color area
Custom Rendered Normal
Exposure Mode Auto
White Balance Auto
Digital Zoom Ratio 0
Focal Length
In35mm Format
33 mm
Scene Capture
Type
Standard
Gain Control None
Contrast Normal
Saturation Normal
Sharpness Normal
Image Quality High
Firmware Version 0.1.1.2
White Balance Auto
Focus Mode Auto
AFMode 3-area (auto)?
Image Stabilization On, Mode 2
Table 2: Tag EXIF (cont.).
Macro Mode Off
Shooting Mode Scenery
Audio No
Data Dump
(Binary data 8200 bytes, use -
b
option to
extract)
White Balance Bia
s
0
Flash Bias 0
Internal Serial
Number
AAAAAAAAAAAAAAAAAAAAAA
Panasonic Exif
Version
0260
Color Effect Off
Time Since Power
On
00:00:49.25
Burst Mode Off
Sequence Number 0
Contrast High
Noise Reduction Standard
Self Timer Off
Rotation Horizontal (normal)
Color Mode Normal
Optical Zoom
Mode
Standard
Conversion Lens Off
Travel Day n/a
World Time
Location
Home
Text Stamp Off
Program ISO n/a
Maker Note
Version
0121
Scene Mode Scenery
WBRed Level 1833
WBGreen Level 1054
WBBlue Level 1964
Flash Fired No
Compression JPEG (old-style)
Orientation Horizontal (normal)
As for the mapping, it was necessary to manually
enter what was not provided for by the scheme of
the ontology. We inserted the information related to
all properties and created the link amongst them and
between them and the various metadata so that they
could be represented univocally and no information
could be lost. In our example the first thing to be
created was, with the aid of the tool, the
USING A STANDARDS-BASED APPROACH FOR A MULTIMEDIA KNOWLEDGE-BASE
93
MultimediaContent class; the name 'Hammamet'
was then associated to it, exploiting the 'instance
browser. It could be noticed that, for the properties
previously created, the hasMetadataLocation and
doesExpress fields appear already compiled.
On the other hand, we had to define the elements
to insert in the hasMetadataDescription field and the
ExifSchema,UnknownMetadata,ExifSchemaHamma
met and UnknownmetadataHammamet instances.
The latter belongs to the class devoted to the
representation of unknown metadata belonging to a
standard.
At this stage, the ExifSchemaHammamet
instance could be filled out with all the fields
returned by the Flickr tool. In this way an univocal
correspondence between information and metadata
related to it was created.
The entire Exif schema must be checked in order
to know which tags of the picture are present or not.
We entered the missing data manually.
Once the values were ready to be entered into the
tags, we created a different data-type instance for
each data. Afterwards a Date_1-type instance was
created for the tiff:dateTime tag.
Since the data type belongs to the EXIF schema,
it requires some additional attributes for temporal
information
(
exif:subSecTimeDigitized,exif:subSecTi
meOriginalexif:subSecTime
); thanks to the
existing relations, the fields related to such attributes
were displayed as well.
4 CONCLUSIONS
The purpose of this work was to study, design and
create an ontology that could formalize the
multimedia content semantics and geocoded data,
starting from the most used standards in representing
that domain, especially the Adobe XMP standard.
The aim was to offer a structure enhanced with
semantics, that could serve as base support for the
creation of a software platform for web content
management.
Another powerful tool is represented by the same
shared standards: in fact, they guarantee
interoperability, i.e. the ability for various
technological systems and services to communicate
and exchange information among themselves and
with other systems, which is a feature of utmost
importance. Given the way the ontology was
organized and structured, we might think of its
future application as a support to a software platform
which would allow different subjects to develop
high-value services based on the input of multimedia
content in a context of semantic organization,
integrated by localization services.
The system could be accessible through mobile
devices such as PNAs (Personal Navigator
Assistant), that would use a geolocalization system
to know their location.
The platform, thanks to the modelled concepts,
could give users the chance to collect and add
contents originated from varied sources (websites,
web portals, local files) and to influence the value of
the contents though ratings, comments and
preferences. Thus contents could be gathered,
aggregated and geocoded, and then distributed to
each user.
Such a platform should clearly be provided a
powerful tool capable to “conform” every piece of
information about the added contents to the form
designated as representation standard within itself.
In other words, it must be able to map any kind of
metadata present in contents.
Once again the ontology we created would be an
impressive tool for fulfilling that requirement.
REFERENCES
Adobe Systems Incorporated, (2010). Adobe XMP
Specifications, additional properties. Retrieved from:
http://www.adobe.com/content/dam/Adobe/en/devnet/
xmp/pdfs/XMPSpecificationPart2.pdf/
Adobe XMP, http://www.adobe.com/products/xmp
Alesso, H. P., Smith, C. F. (2008). Thinking on the Web:
Berners-Lee, Gödel and Turing. Wiley.
Becker, H., Chapman, A., Daviel, A., Kaye, K., Larsgaard,
M., Miller, Nebert, D., Prout, A., Wolf, M. P. (1977).
Dublin Core element: Coverage. Retrieved from:
http://www.alexandria.ucsb.edu/public-documents/met
adata/dc_coverage.html
Benitez, A., Chang S. (2003). Automatic multimedia
knowledge discovery, summarization and evaluation.
IEEE Transactions on Multimedia.
Bertini, M., Cucchiara, R., Del Bimbo, A., Torniai, C.
(2005). Video annotation with pictiorally enriched
ontologies. In Proceedings of IEEE Int'l Conference
on Multimedia and Expo (ICME 2005).
Bertini, M., Del Bimbo, A., Torniai, C., Cucchiara, R.,
Grana, C. (2006). MOM: Multimedia Ontology
Manager. A Framework for Automatic Annotation and
Semantic Retrieval of Video Sequences. ACM MM
2006, Santa Barbara, California, USA.
Bray, T., Paoli, J., Sperberg-McQueen, C. M. (1998).
Extensible Markup Language (XML) 1.0, W3C
Recommendation. Retrieved from: http://www.w3.
org/TR/1998/REC-xml-19980210
Brickley, D., Guha, R.V. (2000). Resource Description
Framework (RDF) Schema Specification 1.0. W3C
KMIS 2011 - International Conference on Knowledge Management and Information Sharing
94
Proposed Recommendation. Retrieved from: http://
www.w3.org/TR/PR-rdf-schema
Carroll, J. J., De Roo, J. (2004). OWL Web Ontology
Language Test Cases. World Wide Web Cobsortium
(W3C) Recommendation. Retrieved from: http://
www.w3.org/TR/2004/REC-owl-test-20040210
Del.icio.us, http://www.delicious.com
Flickr, http://www.flickr.com
Groza, T., Handschuh, S. (2009). A Hybrid Approach
Towards Information Expansion based on Shallow and
Deep Metadata. In Proceedings of KEOD 2009 (pp.
109-116).
Gruber, T. (2008). Ontology. In Liu L., Özsu, M. T.
(Eds.), Encyclopedia of Database Systems. Springer-
Verlag.
Hammamet picture. Retrieved from: http://www.flickr.
com/photos/35513418@N04/38429 94991/in/photo
stream
Heflin, J. (2004). OWL Web Ontology Language Use
Cases and Requirements. World Wide Web
Consortium (W3C) Recommentation. Retrieved from:
http://www.w3.org/TR/webont-req
Hillmann, D. (2005). Using Dublin Core. Retrieved from:
http://dublincore.org/documents/usageguide
IPTC, Information Technology for news (2008). Standard
Photo Metadata 2008 IPTC Core Specification 1.1
and IPTC Extension Specification 1.0. Retrieved from:
http://www.iptc.org/std/photometadata/2008/specificat
ion/IPTC-PhotoMetadata-2008.pdf
Jaimes, A., Smith, J. (2003). Semi-automatic, data-driven
construction of multimedia ontologies. Proceedings of
IEEE International Conference on Multimedia and
Expo (ICME 2003, Vol. 2).
Kameas, A., Seremeti, L. (2007). Multimedia ontologies.
Proceedings of the 3rd international conference on
Mobile multimedia communications, Nafpaktos,
Greece.
Lassila, O. and Swick, R. (1999). Resource Description
Framework (RDF): Model and Syntax Specification.
Recommendation W3C. Retrieved from: http://
www.w3.org/TR/REC-rdf-syntax
Martínez, J. M., Koenen, R., Pereira, F. (2002). MPEG-7:
the generic Multimedia Content Description Standard,
part 1. In IEEE Multimedia (Vol. 9, pp. 78-87).
Mathes, A. (2004). Folksonomies - Cooperative
Classification and Communication Through Shared
Metadata, Computer Mediated Communication,
University of Illinois Urbana-Champaign.
McGuinness, D. L., Van Harmelen F. (2004). OWL Web
Ontology Language Overview. World Wide Web
Consortium (W3C) Recommendation. Retrieved from:
http://www.w3.org/TR/owl-features
Noy, N. F., McGuiness, D. L. (2001). Ontology
Development 101: A Guide to Creating Your First
Ontology. Knowledge Systems, AI Laboratory,
Stanford University.
Petridis, K., Anastasopoulos, D., Saatho, C.,
Timmermann, N., Kompatsiaris, I., Staab, S. (2006).
M-OntoMat-Annotizer: Image Annotation. Linking
Ontologies and Multimedia Low-Level Features.
Engineering Applications of Semantic Web Session
(SWEA) at the 10th International Conference on
Knowledge-Based and Intelligent Information and
Engineering Systems (KES 2006), Bournemouth, U.K.
Picasa, picasaweb.google.com
Picca, D. (2010). Building Multilingual Lexical Resources
On Semiotic Principles. In Proceedings of KEOD
2010 (pp. 412-415).
RDF Working Group, The World Wide Web Consortium
(2004). Resource Description Framework (RDF).
Retrieved from: http://www.w3.org/RDF
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 (EWIMT 2004).
Technical Standardization Committee on AV & IT
Storage Systems and Equipment (2002). Exchangeable
image file format for digital still cameras: Exif version
2.2. Published by: Standard of JEITA (Japan
Electronics and Information Technology Industries
Association). Retrieved from: http://www.
exif.org/Exif2-2.pdf
The Internet Society (2005). The Atom Syndication
Format. Retrieved from: http://www.ietf.org/rfc/
rfc4287.txt
The Protégé, Ontology Editor and Knowledge Acquisition
System (Stanford Center for Biomedical Informatics
Research). Retrieved from: http://protege.stanford.edu
Vander Wal, T. (2005). Explaining and Showing Broad
and Narrow Folksonomies. Retrieved from:
http://www.personalinfocloud.com/2005/02/explainin
g_and_.html
Vander Wal, T. (2007). Folksonomy Coinage and
Definition. Retrieved from: http://www.vanderwal.net/
folksonomy.html
W3C (2007). XML Schema. Retrieved from:
http://www.w3.org/XML/Schema
Youtube, http://www.youtube.com
YouTube APIs and Tools (2010). Reference Guide: Data
API Protocol. Retrieved from: http://code.google.com/
intl/it/apis/youtube/2.0/reference.html
Zooomr, www.zooomr.com
USING A STANDARDS-BASED APPROACH FOR A MULTIMEDIA KNOWLEDGE-BASE
95