=Paper=
{{Paper
|id=Vol-441/paper-4
|storemode=property
|title=Delivery Context Descriptions - A Comparison and Mapping Model
|pdfUrl=https://ceur-ws.org/Vol-441/p03.pdf
|volume=Vol-441
}}
==Delivery Context Descriptions - A Comparison and Mapping Model==
https://ceur-ws.org/Vol-441/p03.pdf
Delivery Context Descriptions – A Comparison and
Mapping Model
Christian Timmerer, Johannes Jabornig, and Hermann Hellwagner
Department of Information Technology (ITEC)
Klagenfurt University
Universitätsstrasse 65-67
A-9020 Klagenfurt
{christian.timmerer, hermann.hellwagner}@itec.uni-klu.ac.at
j.jabornig@jabjo.net
Abstract: Nowadays, mobile devices have implemented several transmission
technologies which enable access to the Internet and increase the bit rate for data
exchange. Despite modern mobile processors and high-resolution displays, mobile
devices will never reach the stage of a powerful notebook or desktop system (for
example, due to the fact of battery powered CPUs or just concerning the small-
sized displays). Due to these limitations, the deliverable content for these devices
should be adapted based on their capabilities including a variety of aspects (e.g.,
from terminal to network characteristics). These capabilities should be described in
an interoperable way. In practice, however, there are many standards available and
a common mapping model between these standards is not in place. Therefore, in
this paper we describe such a mapping model and its implementation aspects. In
particular, we focus on the whole delivery context (i.e., terminal capabilities,
network characteristics, user preferences, etc.) and investigated the two most
prominent state-of-the-art description schemes, namely User Agent Profile
(UAProf) and Usage Environment Description (UED).
1 Introduction and Motivation
Today, the access to the Internet via mobile phones and other devices which are limited
in power capacity and/or rendering functionality increases. Additionally, manufacturers
equip their devices with technologies to access various networks, and mobile providers
offer services for connecting these devices to the Internet. Thus, using small, mobile
devices enables the access to the Internet but often the available content of Web pages is
not suitable according to their capabilities. In order to solve this issue, some projects and
standards have been released which deal with the description of capabilities and
characteristics of all kind of devices. Terminal capabilities and network conditions as
well as user characteristics may allow the adaptation of the content for certain purposes.
Amongst others, there are two standards which were designed to meet the requirements
of device and user descriptions and are often compared to each other. The first one was
released by the WAP Forum (now the Open Mobile Alliance) and is named User Agent
Profile (UAProf) [OM06] and the second one is the Usage Environment Description
(UED) standard which was standardized within MPEG-21 Digital Item Adaptation
(DIA) [VT05]. The aim of this paper is to build a mapping model for these description
formats enabling context-aware content delivery independent of the actual description
format used. Recently, W3C has started a new work item with the aim to define a
delivery context ontology [LF08] but, still, the mapping issue remains.
While the User Agent Profile standard is very popular and has been implemented in a
wide range of mobile devices the Usage Environment Descriptions are only limited
available and tools which would ease the creation of such descriptions rarely exist.
However, a high availability of Usage Environment Descriptions is desired by research
projects [Da06][Ax08][En08] and, thus, obtaining information about terminals, networks
and users from projects with similar aims to that of Usage Environment Descriptions
should be enabled. For example, an implementation compliant to a standardized delivery
context description format A requires a mapping module if it receives descriptions
compliant to another standard B and vice versa. In order to keep the mapping effort
minimal and scalable the proposed method in this paper enable the implementation of a
service that performs such kind of mapping.
This paper is organized as follows. Section 2 highlights the main requirements on
standards for delivery context descriptions by classifying terminals and their properties.
An analysis and comparison of delivery context description formats is presented in
Section 3 while Section 4 describes the actual mapping model. Finally, the
implementation details are described in Section 5 and the paper is concluded with
Section 6 which contains also future work items.
2 Requirements on Standards for Context Descriptions
In this section we have identified different classes of terminals including their hardware
and software capabilities. This kind of information should provide us a rough estimation
on the requirements for delivery context description standards from the terminal’s point
of view which also includes the access networks.
When the Internet became popular, the only way to access the Web was through a
personal computer (PC) or a workstation. In general, these computers had large color
displays with full graphic capabilities, sufficient computational power without battery
issues, and a decent network connection [GLS06]. Nowadays, people tend to access the
Web using smaller and mobile devices with various constraints on display capabilities,
user input/output facilities, computational power, electric power, and access networks
ranging from high-speed Wireless Local Area Network (WLAN) to low-speed General
Packet Radio Service (GPRS). In the following we will provide a classification of the
various terminals based on their hardware (HW) and software (SW) characteristics.
Table 2 in the Appendix provides an overview of HW/SW characteristics of different
end user devices (i.e., desktop PC/workstation, notebook/tablet PC, sub-
notebook/netbook, handheld, smart phone, and mobile phone) with respect to
performance (i.e., CPU), display, permanent storage, memory, network connectivity,
electric power, user input/output, extensibility, operating system support, and software in
general.
A summary and comparison of terminal’s display and memory properties is depicted in
Figure 1. As one can see there is still a huge gap between classical mobile devices (i.e.,
phones) and devices that may have full power supply. Thus, a comprehensive delivery
context standard needs to accommodate all these HW/SW properties in an easy-to-
understand/use, extensible, and manageable way.
Figure 1. Summary and comparison of terminal’s display and memory properties.
However, not only HW properties like screen size, color capabilities, or user
input/output facilities are important, also SW properties like supported operating
systems, audio/video/image codecs, etc. become more and more important as diversity of
devices increases. In particular, the number of different coding formats a terminal is
capable to support – both for encoding and decoding – is of interest for delivery context
description formats. As there are so many coding formats available, some may have
certain profile/level definitions, and even a class of terminals may define its own
constraints, there exists a strong requirement to describe these properties effectively. A
key functionality is the possibility to add new coding formats – e.g., through a
registration authority – in a convenient and relatively unbureaucratic way as they arise
rather rapidly on the market.
3 Analysis and Comparison of Context Description Formats
3.1 Composite Capabilities/Preference Profiles (CC/PP)
The Composite Capabilities/Preference Profiles (CC/PP) [Kl04] comprises descriptions
based on the Resource Description Framework (RDF) which cover device capabilities
and user preferences by introducing a two-level hierarchy consisting of components and
attributes. Components are groups of attributes with related meaning such as the
software or the hardware properties of a terminal. A CC/PP document shall contain at
least one component each identified by an rdf:type attribute which indicates the type
of the component. Attribute values may be simple, i.e., string and integer or rational
numbers, or complex, i.e., set (rdf:bag) or sequence (rdf:seq) of simple values.
However, CC/PP does not define a vocabulary of terms but provides a common structure
for holding any arbitrary vocabulary. Thus, another description format is required which
specifies the actual vocabulary, e.g., the User Agent Profile as described next.
3.2 User Agent Profile (UAProf)
The Open Mobile Alliance (OMA) defines the User Agent Profile (UAProf) [OM06]
which is based on CC/PP and defines a vocabulary for describing characteristics and
capabilities of mainly WAP-enabled mobile devices. The components can be clustered
into
— HardwarePlatform defines manufacture, CPU type, display/audio output
capabilities, device interaction possibilities, and keyboard layout;
— SoftwarePlatform comprises supported media types, preferred language,
operating system, audio/video codecs, etc.;
— BrowserUA includes whether the browser supports certain (X)HTML features,
frames, tables, and its JavaScript capabilities;
— NetworkCharacteristics holds information about supported and current bearers,
security options, and some details about Bluetooth support; and
— various WapCharacteristics and PushCharacteristics.
3.3 Usage Environment Description (UED)
The Usage Environment Description (UED) is defined in Part 7 of MPEG-21, i.e.,
Digital Item Adaptation (DIA) [VT05]. The UED is a very comprehensive vocabulary
organized in so-called properties. It is based on XML Schema and its properties can be
divided into four categories:
— User characteristics provide information pertaining to the user plus his/her usage
preferences/history, presentation preferences, accessibility characteristics, and
location information including the user’s movement.
— Terminal capabilities comprise codec capabilities, input/output characteristics
including display/audio output capabilities, and device properties such as device
class, power/storage characteristics, data input/output facilities, and CPU capabilities.
— Network characteristics include static and dynamic properties pertaining to the
capabilities (e.g., max. bandwidth) and conditions (e.g., available bandwidth) of a
network.
— Natural environment characteristics provide means for describing lightning
conditions, noise level, time, and location.
3.4 Delivery Context Ontology (DCO)
The Delivery Context Ontology (DCO) [LF08] is based the Web Ontology Language
(OWL) [Mv04] and is divided into four entities, namely:
— Environment including information about the location and network.
— Hardware provides information about various hardware capabilities including
display, input, memory, camera, Bluetooth, CPU, etc.
— Measure defines terms related to units with respect to physical electric charges and
length as well as unit conversions.
— Software describes whether the delivery context supports certain APIs, formats (i.e.,
audio, video, image, text, binary), operating systems, application-layer protocols,
Java/Web browser specifics, etc.
3.5 Analysis/Comparison
The previous sections provided an overview of existing standards for delivery context
description formats. In this section we will analyse differences and commonalities of
these formats. First of all, and most importantly, all standards make use of XML that
provides extensibility but UED is based on XML Schema whereas CC/PP and,
consequently, UAProf are based on RDF. The most recent standard in this series is DCO
which adopted already OWL which is based on RDF. Hence, we observe an
incompatibility at the level of technology used for these description formats, mainly
between XML Schema and RDF. Although it is possible to provide a high-level mapping
between these two technologies, the mapping of concrete schemas/instances itself is a
difficult and cumbersome task [HL01].
The second observation we made was that there are only a few characteristics or
capabilities that are common across all delivery context description formats in question,
e.g., display capabilities and file/coding formats. However, there is sometimes a huge
difference in the actual syntax used. For example, display resolution described as
horizontal=1024 and vertical=768 versus 1024x768 or using MIME types for
file/coding formats versus classification schemes (i.e., URNs).
Finally, CC/PP defines only a basic structure (i.e., components and attributes) without
specifying a particular vocabulary of terms. UAProf adopts CC/PP and provides a
concrete vocabulary mainly targeting WAP-enabled mobile devices. A repository of
some specific device profiles is available [W308]. Other industry adoptions of CC/PP
are not known but some are envisaged and documented in Annex E of [Ki07]. The UED
defines both the structure (i.e., properties) and a comprehensive vocabulary while DCO
defines an ontology including not only a vocabulary of delivery context terms but also
basic measure units.
In conclusion, there is a need to describe the relationship between commonalities of the
different delivery context description formats, i.e., a mapping model which will be
described in the next section. To the best of our knowledge, such a mapping model has
not been published yet.
4 Mapping Model
UAProf and UED are based on different data models as the former is based on RDF
whereas the latter is based on XML Schema with having their pros and cons [HL01].
That is, RDF provides support for rich semantic descriptions but provides limited
support for the specification of local usage constraints, e.g., cardinality and datatype
constraints. On the other hand, XML Schema provides support for explicit structural,
cardinality and datatype constraints but provides little support for the semantic
knowledge necessary to enable a flexible mapping between metadata domains.
The main issue is to find a suitable technology for the mapping process which includes
the advantages of both standards. Basically, the mapping can be performed by two
approaches as discussed in the following and depicted in Figure 2:
— Direct mapping model: creating mapping functions that perform direct mapping
from one standard to another.
— Integration model: integrating both models into a new one with functions to convert
between this new model and the initial model.
Figure 2. Direct Mapping Model vs. Integration Model.
As the direct mapping model provides an explicit mapping from one format to another
format it lacks of flexibility with respect to the integration of other formats. Thus, it can
be only applied for specific solutions whereby the number of explicit mappings increases
exponentially with the number of formats between which mappings should be provided.
The integration model defines a common interface that allows the provisioning of the
individual description formats. For new formats to be added, only the methods for
converting to/from this model needs to be implemented without taking into account the
existence of the other formats. Thus, the number of mappings increases linearly with the
number of formats. However, the integration model needs to be implemented with a
certain technology and those in question are XML Schema, RDF, or even OWL:
— One could define an XML Schema that covers all components of each standard and
well-established XPath/XML processors could be used to extract the required data
for a certain standard. Unfortunately, datatype or value range incompatibilities (e.g.,
UED: colorCapable={true,false} vs. UAProf: ColorCapable={Yes,No})
cannot be represented with XML Schema which requires external knowledge to be
provided. Thus, it would be better to use tools which are able to express the relations
between, e.g., datatypes or attribute values.
— OWL is based on RDF and provides means to describe the relationship between
classes and properties, e.g., classes can be declared distinct or equal, restrictions on
properties can be defined as transitive or functional, or the use of cardinality restricts
the number of values which can be associated to properties.
4.1 Mapping Levels
The relationships between different delivery context description formats can be found at
different levels within the entities of their respective schemas (i.e., XML Schema of
OWL). In this paper we introduce four different mapping levels, namely component,
datatype, element, and value. An example thereof is shown in Table 1.
Table 1. Example of Mapping Levels for Network Characteristics.
Level UAProf Example UED Example
Component prf:NetworkCharacteristics dia:NetworkType
Element prf:InputCharSet dia:CharacterSetCode
Datatype prf-dt:Boolean xsd:Boolean
Value Yes true
The component mapping level tries to map a predefined group of elements/attributes
(e.g. prf:NetworkCharacteristics) to similar group of the other description format
(e.g. dia:NetworkType). Difficulties may arise in case the structure is different, e.g.,
one has only attributes defined whereas the other includes also elements within a nested
structure.
Thus, one needs to dig a level deeper and the element mapping level tries to map
attributes/elements with equal semantics but possibly different syntax, i.e., different tag
names. Note that a mapping at the component level is not always sufficient as indicated
above which requires describing relationships between individual elements/attributes or
even beyond.
The datatype mapping level tries to map datatypes with equal domains but different
syntax whereas the value mapping level tries to map datatypes with different domains
but equal semantics.
An example of mapping attributes with the equal semantics and Boolean values is
described in Listing 1 which maps the prf:ColorCapable to the map:colorCapable
(assuming map:colorCapable is the RDF/XML representation of the colorCapable
attribute of the dia:DisplayCapabilityType).
Listing 1. Mapping prf:ColorCapable to map:colorCapable.
1 ...
2
3
4
5 ...
A problem arises as both datatypes are Boolean types but with different syntax: while
xsd:boolean (as used within UED) accepts true and false, prf-dt:Boolean (as
used within UAProf) accepts only Yes and No which requires an appropriate mapping.
Listing 2 shows one possibility for such a mapping of these Boolean datatypes. Lines 2
to 6 and lines 8 to 12 describe properties to create a relation between a new defined
resource for a Boolean value (prefixed by btd) and the Boolean values used by UAProf
and UED. The rest provides the mapping between the values and the bdt Boolean
datatypes.
Listing 2. Mapping the values of prf-dt:Boolean and xsd:Boolean.
1 ...
2
3
4
5
6
7
8
9
10
11
12
13
14
15 true
16 Yes
17
18
19
20 false
21 No
22
23 ...
Another example is shown in Listing 3 which maps a Uniform Resource Name (URN)
identifying a certain key input type to the equivalent string representation of UAProf.
The usage of URNs to uniquely identify predefined terms is heavily used within UED.
Listing 3. Mapping of key input types: DIA-KeyInputCS-NS:1 and Querty.
1 ...
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16 urn:mpeg:mpeg21:2003:01-DIA-KeyInputCS-NS:1
17
18 Querty
19
20 ...
Datatypes such as prf-dt:Number and xsd:nonNegativeInteger can be mapped
directly to each other because both cover the same range of values. However, integer
values also raise problems when a mapping from dia:bitsPerPixel to
prf:BitsPerPixel is provided because the UED standard defines an xsd:integer
datatype and the UAProf standard uses the prf-dt:Number datatype which is equal to
xsd:nonNegativeInteger. Of course, it is unlikely to describe the number of bits per
pixel or the horizontal and vertical resolution with a negative number but there is the
possibility to do that. OWL lacks to offer appropriate tools to describe what to do if such
problems arise.
Another issue is raised with the datatype prf-dt:Dimension which is used to describe
the resolution of a terminal’s display within one string of the form
“HorizontalResolution” x “VerticalResolution” (e.g., 1024x768). This means
that two values of the UED (i.e., horizontal and vertical attributes of the
Resolution element) are combined to represent one value within UAProf. Thus,
external tools are needed to enable such mappings.
4.2 Mapping Classes
In practice, the tag names and datatypes of different description formats can be clustered
into four classes which are described in the following.
Direct. Elements falling into this class have equal semantics and compatible datatypes
with equal domains but may differ in their syntax (i.e., tag name).
For example, dia:bitsPerPixel of type xsd:integer and prf:BitsPerPixel of
type prf-dt:Number where these datatypes are compatible. Another example is the
dia:CharacterSetCode of type mpeg7:characterSetCode (equal to
xsd:string) and prf:CcppAccept-Charset of type prf-dt:Literal (equal to
xsd:anySimpleType) which are used to store string representations of the supported
character sets.
Advance. The class advance comprises elements describing the same concept (i.e., equal
semantics) but with different, non-compatible datatypes and/or domains. Thus, the actual
format is that much different and requires major changes if mapped from one format to
the other. For example, the dia:Resolution includes two attributes (horizontal and
vertical) for describing the resolution of a screen whereas prf:SreenSize uses only
one value (e.g., 480x320). Another example is the usage of classification schemes
versus MIME types as detailed in Section 4.3. Thus, an advanced mapping mechanism is
required.
Derive. This class includes mappings where element values can be derived from one or
more elements of the respective other description format. The difference to the advance
class is that for the derive class the semantic equality is not necessarily a requirement.
For example, prf:SoundOutputCapable indicates whether a terminal is able to output
sound which could be derived from only the presence of a
dia:AudioOutputCapability element.
Extend. Elements that cannot be mapped directly, in an advanced way through
additional mapping rules, or derived from other elements require proprietary extensions
of the respective other description format. For example, properties defined within
UAProf but not defined in UED require an extension of the UED schema by adding
additional elements and datatypes representing these UAProf properties.
In our example, the UAProf standard defines six components and 77 elements which
have been mapped – with respect to UED – to the classes described above (quantities in
brackets): direct (4), advance (7), derive (4), and extend (62). The specific support for
mobile phones within UAProf causes the high number within the class extend whereas
UED does not provide means for describing WAP or push characteristics. One could
now argue that such a mapping is not required. Please note that for most of the
application scenarios – in particular, multimedia content adaptation – the required
elements/attributes/tags fall into direct, advance, and derive classes, e.g., adaptation to
screen size, codec, bitrate which are covered in all delivery context description formats.
Therefore, the class ‘extend’ can be ignored for this kind of applications.
4.3 Additional Mapping Rules for Coding Formats
This section specifically discusses means for describing supported coding formats as this
seem to be an inherent part of each delivery context description standard. Unfortunately,
the standards in question adopt different technologies for describing this property. In
particular, the CC/PP and, thus, UAProf adopts an approach which is based on MIME
media types [FB96] whereas MPEG-21 UED relies on classification schemes introduced
within MPEG-7 [MSS02].
MIME media types are well known within the Internet – thanks to its adoption for
HTTP, etc. – and comprises five discrete top-level media types, i.e., text, image, audio,
video, and application, as well as two composite top-level media types, i.e., multipart
and message. These top-level media types are referred to as content types and the actual
coding format is identified through the content sub-type (e.g, video/mp4). It is also
possible to associate an arbitrary number of parameters in form of key-value pairs to
media types which could be used to describe specifics usually defined within
profiles/levels. However, most of the audio/video/image MIME type definition does not
make use of this possibility. Thus, it is up to the application to identify the exact data
format by other means. For example, video/mp4 may contain a bitstream compliant to
MPEG-4 Part 2 (Visual) or MPEG-4 Part 10 (Advanced Video Coding), not mentioning
all the available profile/level combinations.
An MPEG classification scheme is an XML document that may contain terms –
identifiable by URN – and corresponding definitions of arbitrary semantics in a
hierarchically fashion. Thus, it is also possible to include profile/levels of a certain
coding format as shown in Listing 4. Although classification schemes are extensible as
they are based on XML there is a lack of an approved registration authority to
accommodate future coding formats. However, the European Broadcasting Union (EBU)
maintains a set of classification schemes used within their specifications (including TV-
Anytime) [EBU09].
Listing 4. Excerpt of Visual Coding Format Classification Scheme MPEG-4 Visual Simple Profile.
MPEG-4 Visual
MPEG-4 Visual Coding Format
MPEG-4 Visual Simple Profile
MPEG-4 Visual Simple Profile @ Level 0
MPEG-4 Visual Simple Profile @ Level 1
MPEG-4 Visual Simple Profile @ Level 2
MPEG-4 Visual Simple Profile @ Level 3
Listing 5 describes the mapping from the MIME type image/jpeg to an equivalent URN
representation which is used in UED. Line 1 of Listing 5 defines a resource for the
MIME type image/jpeg. The other prefixes csm, mit and owl are shortcuts for the
resources where the used vocabularies are defined (e.g., owl as shortcut for
http://www.w3.org/2002/07/owl#). Line 2 defines the mapping from the resource
&uic;jpeg to the resource urn:mpeg:mpeg7:cs:VisualCodingFormatCS:2001:4
which represents JPEG as a reference to a classification scheme term. Line 3 defines
which string representation for JPEG should be used in UAProf descriptionss. Lines 4
and 5 define all representations for the image/jpeg MIME media type. Line 6 uses
standard OWL syntax to define that the resource &uic;jpeg is different from the
resource &uic;bitmap.
Listing 5. Mapping MIME media type image/jpeg to
urn:mpeg:mpeg7:cs:VisualCodingFormatCS:2001:4.
1
2
3 image/jpeg
4 image/jpeg
5 image/jpg
6
7
8
5 Implementation Details
The aim of this section is to provide an overview of our implementation that currently
performs a mapping between UAProf descriptions and UEDs (and vice-versa). The high-
level architecture is depicted in Figure 3 and comprises three components:
— Validator.
— Transformer.
— Profile Creator.
Figure 3. High-Level Architecture of the UAProf/UED Mapping Implementation.
The Validator is responsible for validating incoming and outgoing UAProfs and UEDs.
If the received profile is a UED profile, a transformation to an RDF/XML document is
needed for further processing which the Transformer accomplishes. UAProfs need not
be translated because they are already written in RDF/XML syntax which is a
requirement of the profile creator. The Profile Creator queries data from the profile data
which is available in a consistent syntax and creates the desired profile as output which
is again checked by the validator before it is delivered. These three components are
further detailed in the following.
Validator (cf. Figure 4). The purpose of this component is to validate instances against
its specification. This is performed both for inputs and outputs of our implementation.
For UAProf we have integrated the DElivery context LIbrary (DELI ) [Bu08] which is
one of some rarely available tools that are able to validate UAProfs and to extract data
from these documents. As the UED schema is an XML schema we have used standard
XML schema validation tools such as the Java built-in XML validation Application
Programming Interface (API).
Figure 4. Architecture of the Validator.
Transformer (cf. Figure 5). The transformer is responsible for translating the input
instances into an integration model based on RDF as already introduced in Section 4.
Therefore, we have implemented style sheets based on the Extensible Stylesheet
Language Transformation (XSLT) [Cl99], i.e., only one style sheet is required for each
delivery context description language keeping the overall approach scalable.
Figure 5. Architecture of the Transformer.
Profile Creator (cf. Figure 6). Finally, this component generates the designated target
delivery context description based on the integration model. In order to query the RDF-
based integration model we have used SPARQL Protocol And RDF Query Language
(SPARQL) [PS08] and A SPARQL Processor for Jena (ARQ) [HP08] as the actual
query engine. The implementation adopts predefined templates and queries to generate
desired output format based on the integration model.
Figure 6. Architecture of the Profile Creator.
6 Conclusions and Future Work
In this paper we have presented a model that allows one to map context delivery
descriptions between different formats (e.g., OMA UAProf and MPEG-21 UED) that are
based on different technologies (i.e., XML Schema and RDF/OWL). For this model we
have investigated state-of-the-art terminals in terms hardware and software capabilities
as well as analyzed and compared existing delivery context description formats. Based
on this analysis and comparison we concluded that there is a need for describing the
commonalities and relationships between these description formats using a common
model, i.e., following the integration model approach introduced earlier. The mapping
model clusters the properties of the individual description formats based on their levels
into four classes, namely direct, advance, derive, and extend. Based on these classes we
have defined the integration model and formulated templates (i.e., using SPARQL and
OWL) to query information from the integration model in order to generate the target
context delivery format. The feasibility of the approach has been validated through a
prototype and implementation details have been described in this paper.
The major findings can be summarized as follows:
— The overlap between different context delivery description formats is not that
huge as expected but is clustered around those properties which are considered
by the majority of applications areas (e.g., screen size, coding formats, etc.).
— Hence, the classes direct, advance, and derive are sufficient for most of the
application areas.
— The relationship between different delivery context description formats needs to
be described manually with respect to an integration model (i.e., the mapping
function) and requires a thorough analysis of these formats which is sometimes
cumbersome (cf. also [HL01]). For each format the mapping functions need to
be defined only once with respect to an integration model.
— However, in this paper we have demonstrated that it is feasible – in principle –
but requires the integration of many XML-based technologies ranging from
XML Schema and RDF to SPARQL and OWL.
The following items are to be considered for future work. The integration of an OWL
reasoner may be used to automatically recognize related data and extract specific
information by using inference (e.g., mapping between different versions or slight syntax
variations). Another future work item is a more detailed investigation of W3C’s Delivery
Context Ontology (DCO) and whether it can be used as the basis for the integration
model for both UED and UAProf. Finally, as the newest description format under
development (i.e, DCO) is based on OWL it confirms our decision to use OWL as
underlying technology for the integration model.
7 References
[Ax08] IST-FP6: AXMEDIS (Automating Production of Cross Media Content for Multi-channel
Distribution), http://www.axmedis.org/ (last accessed: December 2008).
[Bu08] Butler, M.: DELI: A Delivery Context Library For CC/PP and UAProf, October 2006.
http://delicon.sourceforge.net/ (last accessed: December 2008).
[Cl99] Clark, J.; (ed.): XSL Transformations (XSLT) Version 1.0, W3C Recommendation,
November 1999. http://www.w3.org/TR/1999/REC-xslt-19991116 (last accessed:
December 2008).
[Da06] IST-FP6: DANAE (Dynamic and Distributed Adaptation of scalable multimedia coNtent
in a context-Aware Environment), http://danae.rd.francetelecom.com/ (last accessed:
December 2008).
[EBU09] EBU Metadata Classification Scheme Mainenance Form,
http://www.ebu.ch/metadata/maintenanceforms/index_cs.html (last accessed 2009).
[En08] IST-FP6: ENTHRONE (End-to-End QoS through Integrated Management of Content,
Networks and Terminals), http://www.ist-enthrone.org/ (last accessed 2008).
[FB96] Freed, N.; Borenstein, N; (eds.): Multipurpose Internet Mail Extensions (MIME) Part
Two: Media Types, RFC 2064, November 1996.
[GLS06] Gimson, R.; Lewis, R.; Sathish, S.; (eds.): Delivery Context Overview for Device
Independence, W3C Working Group Note, March 2006. http://www.w3.org/TR/di-dco/
(Last accessed: December 2008).
[H01] Hunter, J.: Adding Multimedia to the Semantic Web - Building an MPEG-7 Ontology,
Proceedings of the International Semantic Web Working Symposium (SWWS),
Stanford, August 2001; pp. 261-281.
[HL01] Hunter, J.; Lagoze, C.: Combining RDF and XML Schemas to Enhance Interoperability
Between Metadata Application Profiles, Proceedings of the 10th World Wide Web
Conference, Hong Kong, May 2001.
[HP08] HP Laboratories: ARQ - A SPARQL Processor for Jena,
http://jena.sourceforge.net/ARQ/ (last accessed: December 2008).
[Ki07] Kiss, C.; (ed.): Composite Capability/Preference Profiles (CC/PP): Structure and
Vocabularies 2.0, W3C Working Draft, April 2007. http://www.w3.org/TR/CCPP-struct-
vocab2/ (last accessed: December 2008).
[Kl04] Klyne G.; et.al. (eds.): Composite Capability/Preference Profiles (CC/PP): Structure and
Vocabularies 1.0, W3C Recommendation, January 2004. http://www.w3.org/TR/CCPP-
struct-vocab/ (last accessed: December 2008).
[LF08] Lewis, R.; Fonseca, J.M.C.; (eds.): Delivery Context Ontology, W3C Working Draft,
April 2008. http://www.w3.org/TR/dcontology/ (last accessed: December 2008).
[MSS02] Manjunath, B.S.; Salembier, P.; Sikora, T.; (eds.): Introduction to MPEG-7: Multimedia
Content Description Interface, Wiley & Sons, April 2002.
[Mv04] McGuinness, D.L.; van Harmelen, F.; (eds.): OWL Web Ontology Language Overview,
W3C Recommendation, February 2004. http://www.w3.org/TR/owl-features/ (last
accessed: December 2008).
[OM06] Open Mobile Alliance Ltd: User Agent Profile, Approved Version 2.0. Technical report,
OMA, February 2006.
[PS08] Prud'hommeaux, E.; Seaborne, A.: SPARQL Query Language for RDF, W3C
Recommendation, January 2008. http://www.w3.org/TR/rdf-sparql-query/ (last accessed:
December 2008).
[VT05] Vetro, A.; Timmerer, C.: Digital Item Adaptation: Overview of Standardization and
Research Activities, IEEE Transactions on Multimedia, vol. 7, no. 3, June 2005; pp. 418-
426.
[W308] w3development.de: UAProf profile repository. http://w3development.de/rdf/uaprof
repository/ (last accessed: December 2008).
8 Appendix
Table 2. HW/SW Characteristics of Desktop PC/Workstation, Notebook/Tablet PC, Sub-Notebook/Netbook, Handheld, Smartphone, and Mobile Phone
Desktop Notebook/Tablet Sub-Notebook/Netbook Handheld Smart Phone Mobile Phone
PC/Workstation PC
Performance High performance Power saving dual Medium performance Medium and low performance with various processors Low performance with various
with dual, triple or core processor special power saving processors
quad core processors processors
Display 17" - 30", multiple 12" - 20" 7" - 12" 4" (or smaller), 1,8"-3,5", color (and monochrome) 1,8"-3,5", color and monochrome
displays color and
monochrome
Storage Up to 2 TB Up to 1 TB 2GB to 160GB 8MB-256MB (also Up to 16GB (expandable, e.g., Up to 64MB
16GB) microSD-Cards)
Memory Up to 8 GB Up to 4 GB Up to 1 GB Similar to Often 128MB Similar to permanent storage
permanent storage
Network Gigabit Ethernet, Gigabit Ethernet, WLAN, Modem, IrDA, Bluetooth, IrDA, Bluetooth, USB, WLAN, GSM, IrDA, Bluetooth, GSM, GPRS,
WLAN, cable/xDSL, Bluetooth, (UMTS, HSDPA) WLAN, GSM, GPRS, EDGE, UMTS, HSDPA, UMTS
etc. GPRS, UMTS, HSUPA, GPS, DVB-T, etc.
HSDPA, GPS
Power AC AC, DC, battery 2-6h (typically 2,5h) DC, battery (3-15h) Battery 100h with normal use (up to 400h Standby)
User I/O Keyboard, Mouse, Keyboard, Touchpad, LCD, Touchscreen Touchscreen and Touchscreen, QUERTY-Keyboards Phone Keyboard, (joystick or
Monitor, Loudspeaker, and Stylus, Loudspeaker, Microphone, Stylus, (often very small), limited Keyboards, navigation buttons)
Microphone, Webcam, Webcam Loudspeaker, Loudspeaker, Microphone, Cameras
etc. Microphone
Extensibility USB, Firewire, PCIe, USB, Firewire, USB CF and SD card USB, TV-out Usually not supported
PCI, etc. eSATA, slot for devices and
ExpressCard, memory, USB
PCMCIA
OS Windows Vista, Windows XP, Mac OS, Windows XP, special Windows Mobile, Symbian OS, Windows Mobile, RIM Symbian OS, other proprietary
Linux, etc. Linux versions, Palm OS Blackberrry, Apple OS X (special systems
Windows Mobile edition for iPhone), Google Android
(Linux and Java based)
Software Almost unlimited Limited to Operating Limited to Operating System and hardware capabilities Preinstalled, (sometimes
System and hardware expandable through Java-Midlets)
capabilities