You are on page 1of 6

A User Model for Personalization Services

YeSun Joung, Magda El Zarki, Ramesh Jain


Dept. of Computer Science
Universitv of California Irvine
Irvine, USA, CA 92697
{foungv, magda , fain}ics.uci.edu


Abstract

A user model is essential to support personali:ation
services. However, the user models that are used in most
svstems are designed in an ad-hoc manner and often
related to their application domains. It hinders service
interoperabilitv and increases the amount of work. In this
paper, we focus on defining a general user model in order
to represent user information and the user context. This
paper contributes two points. a survev on existing context
based svstems and a user model for personali:ation. The
survev analvses previous context based svstems and
provides different kinds of features used in their svstems.
Based on this survev, we propose a user model to capture
user information and contexts for personali:ation
application.

1. Introduction

We Iace a huge amount oI inIormation in everyday liIe.
Current search engines provide lists oI webpage links that
might have inIormation related to the user`s interest. This
approach has been successIul, but it requires the user`s
eIIorts to Iind the inIormation. How can we improve the
eIIiciency oI the current search process?
Personalization has become an important research area
to deal with this inIormation overload and the ineIIiciency
oI the current search process. The deIinition oI
personalization is the tailored services according to each
user`s characteristics or preIerences. From the deIinition,
personalization is using methods to adjust the diIIerences
among users and a user model is commonly adopted Ior
characterizing users. In early approaches to personalization
|1, 2|, they emphasize the three important research
challenges:

How to model the user and user`s situation in an
extensible and uniIied way?

How to collect user data Ior instantiating the user
model?

How to enable and manage instances oI the user
model in personalization systems eIIiciently?

In this paper, we want to Iocus on deIining a general
user model that represents a user and the user`s situations.
As mentioned in early works |1, 2|, they advocate context
as means oI capturing user`s situations such as user needs,
preIerences, interests, expertise, and so on. However, their
notion oI context is restricted. It represents a user in a
dynamic aspect because their user context like user needs
changes through diIIerent time Irames. We want to deIine
context in more general and systematic manner. As shown
in Figure 1, the environment oI personalization is
composed oI users, services and resource. Users have
personalization services which use related resources. The
context Ieatures are related to these components, time and
locations. Our user model uses these context Ieatures to
describe users and their situations. Thus, our notion oI
context is the set oI Iacts or circumstances that surround a
situation or event.


Figure 1 Components of Personalization

In this paper, we try to Iind what kinds oI Ieatures are
related to User, Service and Resource Contexts through
studying previous context based systems. Based on the
lessons Irom the survey, we propose a user model that
could be more Ilexible and useIul in many applications.
The rest oI the paper is organized as Iollows, Section 2
describes a survey on context based systems, Section 3
proposes a user context model, and Section 4 draws
conclusions based on the model.

2. A Survey on Context based Systems

2.1. The Analysis Framework
Much research has been done in deIining notions oI
context, proposing context models and implementing
context aware systems using the context models |5-23|. In
addition, interesting comparisons already exist |3, 4|. The
analysis Irameworks compare available context models
and characterize the models Irom various perspectives.
However, we Ielt the need to build a Iramework Ior
personalization. In the Iramework, we intend to veriIy
what kinds oI Ieatures are included in user, service and
resource contexts and the diIIerent applications they build
using context models.

Features aspects: The set oI Ieatures to represent each
component: User, Service and Resource.

User: is a subject oI personalization services, it is
important to capture the user`s inIormation such as
personal inIormation or preIerences. The
978-1-4244-4254-6/09/$25.00 2009 IEEE
inIormation can be provided by users directly or
derived Irom other inIormation like usage history oI
users. Thus, we analyze what kinds oI Ieatures were
considered as user contexts.

Service: an application on systems. In order to
provide a personalized service to users, service
contexts will be used to identiIy user contexts and
check capabilities oI resources according to user
contexts. Hence, we analyze what kinds oI Ieatures
are considered as service contexts.

Resources: are available sources to execute
personalized services. To execute services, what
kinds oI Ieatures are used to describe resource?

Space: does the context model capture spatial
Ieatures?

Time: does the context model capture temporal
Ieatures?

Usage History: does the context Iocus on the user`s
activities? This Ieature is important to provide
personalization services because usage history can
be used to derive the user's preIerences.

Usage aspects: The way the context is used.

Platform: does an application oI the context model
run on a mobile (M) or server (S) platIorm? In the
case oI the application that runs on a mobile
platIorm, it should provide a personalization service
and design the context model Ior the service.
'PlatIorm indicates what the core context Ieatures
Ior the personalization service.

Application Domain: what kinds oI applications are
implemented using the proposed context model?
Depending on the application domain that the
context model is designed Ior, each model
emphasizes on diIIerent Ieatures. We notice Iour
kinds oI application domains, which are an
application oI data-device-presentation (P), location
(L), user activity (U) and inIormation management
(I).

2.2. The summary of survey
Table 1 shows the result oI our studies using the
analysis Iramework. As we mentioned in the previous
chapter, we consider the environment as composed oI
users, services and resources. Each component has the
context inIormation such as user contexts, service contexts
and resource contexts. Except the three components, we
notice that space and time data are important Ieatures oI
the context models in most studied systems. Space and
time Ieatures are essential to describe users and resources.
In stead oI providing a description oI each system, we
Iollow each application domain and highlight the
characterization oI each context model. As shown in Table
1, we classiIy each system into Iour application domains,
but some oI systems Iall into several application domains
such as CoBrA |13|, MAIS |18|, MPEG-21 |19| and
5W1H |23|. The reason is that these systems build a
general context model which can apply to various
application domains at their systems design stage.
The application oI data-device-presentation (P) aims to
present data on devices adaptively. This application
domain is characterized by limited or absent location and
time Ieatures, Iocused on describing device and data
Ieatures in detail, and considering user inIormation and
preIerences. AMM |5|, CACP |7|, MobiLiIe |8|,
Hydrogen |9|, Experimental Documents |15|, SOCAM
|20| and WIS |22| are classiIied into this application
domain. In case oI SOCAM, it has Iull Ieatures oI location
and time because it is also designed to provide location-
aware service.
The application oI location (L) provided a location-
aware service. Models oI this application in general
support accurate location and time Ieatures and context
reasoning using agent, location status and user status
Ieatures. CASS |6| and SIMPLICITY |10| systems Iocus
on locations oI resources while MobiLiIe, ActiveMap |11|,
COMANTO |14| HIPS/HyperAudio |17|, SOCAM
consider locations oI users more. The systems concentrated
on users` locations include user status Ieatures.
The application oI user activity (U) is Iocuses on what
the user is doing. Usage history and user status are
considered as important Ieatures. Most oI context models
in this category have rich user Ieatures. Even though Table
1 shows that some oI systems have less Ieatures than other
systems, the granularity oI Ieatures are very Iine. For
example, the Experimental Documents system aims to
build an experiential system Ior providing the enriched
multimedia learning environment. In the user context, it
includes general inIormation and preIerence only oI the
user. However, those two Ieatures contain more attributes
to describe user context than others. COMANTO,
Experimental Documents, Graphical Notion |16| and
HIPS/HyperAudio are included in this application domain.
One interesting observation is that this application domain
is implemented with other application domains.
COMANTO system considers a user activity and uses the
activity inIormation to provide a location-aware
application. The reason is that the user activity data is used
as constraints oI adaptation in other applications. II a user
accesses his or her mobile device, the application oI data-
device-presentation should adjust its data Ior the mobile
device. Systems in this application category provide more
similar personalization service.
The application oI inIormation management (I) uses the
context inIormation to select or manage inIormation. Space,
Time and User Ieatures are well developed in this
application domain. The context model in this application
is more Ilexible than context models in other application
domains. Usage history and context reasoning are not
considered. ADDICT |12|, Graphical Notion and U-
Learning |21| are classiIied in this application domain. For
example, the goal oI ADDICT system is to select or tailor
inIormation relevant to the user`s interests. In their context
model, they Iocus on building the context model around
the user`s interests. Based on the context model, it Iilters
inIormation that meets the user`s demands.
Through this survey, we observe various context
models in various application domains. However, up to
this point, modeling a context is application-speciIic and
most oI them are designed Ior particular domain.

S
y
s
t
e
m
s

F
o
c
u
s

A
p
p
l
i
c
a
t
i
o
n


P
l
a
t
f
o
r
m

(
0
o
b
i
l
e

o
r

6
e
r
v
e
r
)

U
s
a
g
e

H
i
s
t
o
r
y

S
p
a
c
e

T
i
m
e

U
s
e
r

R
e
s
o
u
r
c
e

S
e
r
v
i
c
e

P
h
y
s
i
c
a
l

L
o
c
a
t
i
o
n

L
o
g
i
c
a
l

L
o
c
a
t
i
o
n

P
h
y
s
i
c
a
l

T
i
m
e

L
o
g
i
c
a
l

T
i
m
e

G
e
n
e
r
a
l

n
f
o

A
c
t
i
v
i
t
y

P
r
e
f
e
r
e
n
c
e

M
e
d
i
c
a
l

n
f
o

S
o
c
i
a
l

N
e
t
w
o
r
k

N
e
t
w
o
r
k

D
e
v
i
c
e
s

D
a
t
a

S
e
n
s
o
r

H
a
r
d
w
a
r
e


C
o
d
e
c

O
S

S
o
f
t
w
a
r
e

A
p
p
l
i
c
a
t
i
o
n

A
g
e
n
t

R
e
s
o
u
r
c
e

S
t
a
t
u
s

U
s
e
r

S
t
a
t
u
s

L
o
c
a
t
i
o
n

S
t
a
t
u
s

AMM P M V V V V V V V
CASS L M V V V V V V V
CSCP P M V V V V V V V V
MobiLife P,L M V V V V V V V V V V V V V V
Hydrogen P M V V V V V V
SMPLCTY L M V V V V V V V V V V V
Active MAP L S V V V V V V V V V V V V V V
ADDCT S V V V V V V V V V V V
CoBrA All S V V V V V V V V V V V V
COMANTO L,U S V V V V V V V V V V V V
Experimental
Documents
P,U S V V V V V V V V V V
Graphical
Notion
,U S V V V V V V V V V V V
HPS/
HyperAudio
L,U S V V V V V V V V V V V V V V V V V V V V V
MAS All S V V V V V V V V V V V V V V V
MPEG-21 All S V V V V V V V V V V V V V V V V V V
SOCAM P,L S V V V V V V V V V V V V V
U-Learning S V V V V V V V V V V V V V V V V V V
WS P S V V V V V V V V V
5W1H All S V V V V V V V V V V V
Table 1 Summary of Context Models in Studied Systems
In order to provide personalization in applications,
modeling a user is important, but most observed user
models are limited and tailored Ior speciIic application
domains as mentioned beIore.
In this work, we present a general and rich user model
and represent it using ontology.

3. A User Model
We use the result oI the survey to create a user model
and ontology to represent the user. Ontology have been
proven an eIIective means Ior modeling concepts and
relationships in an explicit and high level abstraction way.
In this chapter, we describe a user model and ontology
description oI the proposed model.

3.1. Modeling
Based on the survey on existing context systems, we
consider that general inIormation, activity, medical
inIormation, social aspect and preIerence as Ieatures that
must be attached to a user. We categorized these Ieature
groups into Iour major classes in the proposed context
model oI user: general inIormation, preIerences, events
and social networks as shown in Figure 2.
In the General InIo class, we need to capture general
inIormation about a person such as name, contact
inIormation, biologic inIormation, accessibility
inIormation, online inIormation and background. This
inIormation can be used Ior reasoning or inIerring
semantic inIormation about the person. For example, iI we
knew a person`s nationality, we could guess which
language was preIerred Ior content.
The Events class is used to capture activity and status
inIormation around the person. Event inIormation can be
organized in diIIerent Iormats such as Schedules,
Situations, Usage Histories and EventReI as shown in
Figure 2.
The PreIerences class is used to capture what the person
likes. The PreIerence Condition subclass describes the
validity oI each preIerence in time and location. The FOA
(Focus oI Attention) subclass is Ior recording preIerred
keywords or objects in media data.

The Content PreIerence subclass describes a preIerence


oI the content creation, content classiIication, content
dissemination and content summary inIormation. The
Resource PreIerence subclass is to capture preIerences
inIormation oI resources such as devices and networks.
The Location PreIerence subclass contains where the
preIerred locations are. The ProIile PreIerence subclass is
Ior indicating the privacy preIerence oI inIormation.
The Social Networks class explains social connections
around a person using Communication Networks, Groups
and SocialNetworksReI subclasses. The
CommunicationNetworks subclass contains who is in a
person`s contacts database and the kinds oI relationships
they have to the user. The Groups class contains
inIormation about what kind oI communication groups a
person belongs to. The SocialNetworksReI class consist oI
links to other ontologies and descriptions oI social
networks.

3.2. Ontology Description
This section presents a short description oI user
ontology. This ontology can be extended though
inheritance or the addition oI more classes. This ontology
presents user inIormation that is mostly static and
permanent. More dynamic inIormation like the current
position oI the user is considered as a service context. We
consider location or time inIormation oI the user such as
when the user entered the school or where the school is
located. This inIormation is static and will not be changed
over the time. Figure 2 presents an overview oI the
proposed ontology upper level classes. The 'Person class
is the central one in the ontology as it contains all the user
characteristics like general inIormation, preIerences,
events and social networks. This class has ID and
description as attributes. The 'ID is used to identiIy
people in systems and the 'description is Ior a short
comment.
General inIormation class contains several sub classes
like shown in Table 2. This class is capturing the basic
inIormation such as name, IDs, background, contact
inIormation, biological inIormation, online inIormation,
medical inIormation and Iinance inIormation. The
inIormation captured in this class can be used Ior
reasoning or inIerring the user`s preIerences. For example,
iI personalized service has a person`s nationality data, the
service could provide a personalized content service with a
proper language.

&ODVV1DPH &ODVV'HVFULSWLRQ
Name Names oI user
IDs Various IDs oI user such as SSID, Employee
ID and so on
Background Education or working experience inIormation
Contact InIo Basic contact inIormation such as addresses,
phone numbers, E-mails and so on
BIO InIo Biologic inIormation such as gender, weight,
height, size oI shoes and so on
Online InIo User inIormation in cyberspace such as web
page, blogs, online accounts, bookmarks and
so on
Medical InIo Medical background inIormation such as
insurance number, medical histories and so on
Finance InIo Finance background inIormation such as
account numbers, credit scores and so on
Table 2 Description of General Information Class

In the case oI preIerences, we provide PreIerences class
to represent each person`s interests on contents, resources,
and locations as shown in Table 3.
PreIerence Condition describes conditions oI
preIerence in time and location. For example, a person can
set the content preIerence at work and at home. While the
person is in the company during daytime, he or she preIer
to read contents which are related to his or her work, and
while the person is in the home, he or she preIer to watch
entertainment contents.
FOA (Focus oI Attention) subclass speciIies a person`s
interests using keywords or objects. The keywords can
contain a user`s interests in various Iields. II a user is
interested in traveling to Europe, user can speciIy keyword,
'Europe in FOA class. The object can be entity in media
data or a certain range oI media data. For example, a Iriend
on picture, a ball on soccer match or a speciIic part oI
music is the object. We can indicate a user is interested in
Figure 2 Overview of Proposed User Context Model

the objects using FOA class and application can provide


summarization services upon a person`s preIerred objects.
Content PreIerence class describes a person`s
preIerences on contents directly and this preIerence related
to the creation, classiIication, dissemination and summary
inIormation oI contents. When or where contents were
created and who created contents are included in creation
inIormation. People can express their contents preIerences
by providing preIerred classiIication oI contents. Mostly,
content can be classiIied using Iorm, genre, languages and
parent guide inIormation. Dissemination inIormation is
important Ior applications because presentation
applications depend on Iormats or copyrights oI data. In
the case oI a content player, they have to decide which
codec they have to use while system veriIies Iormats oI
data or copyright oI data. The Summary class is used to
navigate or browse multimedia contents. AIter searching
multimedia data using keywords or other methods,
eIIicient presentation oI search results has become
important in personalized service. To provide this
personalized service, we need to capture a person`s
preIerence using the Summary class. Resource PreIerence
class is a tool Ior describing preIerred environments
around resources such as devices and networks. MPEG-21
|19| deIines Device properties in their Usage Environment
Description Tools in details such as device type, power
capability, CPU, display capability, audio capability, codec
capability, storage and data IO. In the user model, we need
diIIerent aspects oI metadata upon resource such as what
kinds oI devices are Irequently used by users or which
networks are used with which devices.
Resource PreIerence class describes preIerred devices
using DeviceID, DeviceType and DeviceReI, and also
expected networks using NetworkType,
NetworkCapability and NetCondition.

Class Name Class Description
PreIerence
Condition
Conditions oI preIerence in time and location
Focus oI
Attention
Person`s preIerred keywords or objects
General
PreIerence
Person`s general preIerence, Ior example 'likes
white color, or 'dislikes carrots.
Content
PreIerence
Person`s preIerences on contents directly and
this preIerence related to the creation,
classiIication, dissemination and summary
inIormation oI contents
Resource
PreIerence
Person`s preIerred environments around
resources such as devices and networks
Location
PreIerence
Person`s Iavorite locations such as home, work
and other Iavorite location that the person
visits Irequently
Table 3 Description of Preferences Class

Location PreIerence class captures a user`s Iavorite
locations such as home, work and other Iavorite location
that the user visits Irequently.
Events class is used to capture experiences around the
user such as activity and status inIormation. Events
inIormation can be organized in the diIIerent Iormats such
as Schedules, Situations, Usage Histories and EventReI
classes. The Schedules class is a way to represent
sequences oI a user`s activities or status into calendar
Iorms such as what, when, where Iormats. Another way to
describe Events is to use the Situation class. Using the
Situation class, we can explain events around objects and
objects can be people, group oI people, and other entities.
For example, people want to describe a birthday party such
as who attend the party, when the party happened and
where the party took place. We added the Usage Histories
class into the Events class because we wanted to provide
tools to capture events on cyberspace such as when users
access resources, contents or applications on computers.
Usage Histories describe which items are accessed and
what applications are used. Usage Histories inIormation is
important to inIer user preIerences on contents or resources.
Also, we open a way to reIer other event descriptions using
EventReI class.
Social Networks explains social connections around
users using Communication Networks, Groups and Social
Networks ReI class. There are several well known
description languages Ior social networks such as FOAF
(Friends oI A Friend) and VCard. In the proposed user
ontology, we can reIer exist social networks descriptions
using SocialNetowrksReI class. Communication networks
keep who are in a user`s contacts and what kinds oI
relationships they have. The RelationshipType is based on
Relationship ontology. The Relationship type includes
relationships among Iamily members such as siblingOI and
spouseOI, relationships among colleagues such as
worksWith and colleagueOI, and so on. We extended this
Relationship ontology to adopt relationships on
cyberspaces such knowsOnline, IriendsOnline and so on.
The Groups class contains inIormation what kinds oI
communication groups a user belongs. When we analyze
Emails, we notice that people belongs to many kinds oI
social groups such as mailing lists or social gatherings. In
order to capture this inIormation, we deIine the Group
class.

3.3. Case Study/Example
In order to demonstrate the usage oI the proposed user
model, we present one case study which collects user
inIormation Irom Facebook |27|, builds a user inIormation
database based on the proposed user model, and provides a
personalized search. The case study is implemented using
Yahoo API |26| to show how the search results can be
personalized by the proposed user model.

Figure 3 Query example of the case study

Using the case study, we can get more personalized
search results based on users` inIormation Irom social
networks. Figure 3 shows a simple example oI the query
example. The user, John, is searching Iriends` inIormation
and he entered query on search interIace in a search system.
The query module veriIies user and meaningIul terms on
the query such as Iriends and school. User InIo Engine
module is asked to Iine who John`s Iriends are and what
school he is in. The module answers back to the query
module with the list oI Iriends and 'UCI. The search
system shows more personalized search results to the user.


Figure 4. A snapshot of the case study system

As shown in the Figure 4, the leIt side oI the user
interIace shows personalized search results using the
proposed user model, and the right side shows non
personalized search results Irom Yahoo search. In the
personalized search results, it contains the list oI users`
Iriends, links oI inIormation which are more related to user.
Through this case study, we want to show that the
proposed user model is rich enough to use Facebook API
and can be used in existing system, Yahoo search.

4. Conclusions
Modeling users is very important work in order to
provide personalized services. As we studied on various
context-aware systems, we Iailed to Iind a rich enough user
model to capture various aspects oI user inIormation. As
long as we have enough user inIormation, it will help to
accurate personalized services in many application
domains. We propose a rich user model to capture user
inIormation Ior personalization application. In Iuture work,
we will Iind more case studies to demonstrate the proposed
user model and it will trigger to revise the proposed
ontology. In addition, we will implement the mentioned
case study. We believe some oI the more interesting
research issues will be identiIied during implementing the
system.

5. References
|1| Goker A. and Myrhaug H.I. User Context and Personalization.
European ConIerence on Case Based Reasoning, Workshop on Case
Based Reasoning and Personalization, 4-7 September 2002.
|2| Kramer J., Noronha S., and Vergo J.. A User-Centered Design
Approach to Personalization. In ACM Computing Surveys, 08 2000.
|2| Cristiana Bolchini, etc, 'A Data-oriented Survey oI Context
models, SIGMOD Record, December 2007 (Vol. 36, No. 4).
|3| Mattias BaldauI, Schahram Dustdar and Florian Rosenberg, 'A
survey on context-aware systems, Int. J. Ad Hoc and Ubiquitous
Computing, Vol. 2, No. 4, 2007.
|4| Vallet, D., etc, "Personalized Content Retrieval in Context Using
Ontological Knowledge," Circuits and Svstems for Jideo
Technologv, IEEE Transactions on , vol.17, no.3, pp.336-346,
March 2007
|5| Thomas Springer and Kay Kadner, 'Middleware Support Ior
Context-Awareness in 4G Environments, Proceedings oI the 2006
International Symposium on a World oI Wireless, Mobile and
Multimedia Networks (WoWmoM`06), 2006.
|6| Fahy, P. and Clarke, S. CASS - Middleware Ior Mobile Context-
Aware Applications. In Workshop on Context Awareness,
MobiSys 2004.
|7| S. Buchholz, T. Hamann, and G. Hubsch, 'Comprehensive
structured context proIiles (CSCP): Desing and experiences, The
2nd IEEE ConI. on Pervasive Computing and Communications
Workshops (PerCom 2004 Workshops), IEEE Computer Society,
2004, pp. 43-47.
|8| R. Kernchen, David BonneIoy, A. Battestini, B. Mrohs, M.
Wagner, and M. Klemettinen: 'Context-awareness in MobiLiIe,
Proc. oI the 15th IST Mobile Summit, Mykonos, Greece, June
2006.
|9| HoIer, T., Schwinger, W., Pichler, M., Leonhartsberger, G.,
Altmann, J., and Retschitzegger, W., 'Context-Awareness on
Mobile Devices the Hydrogen Approach, In Proceedings oI the
36th HICSS`03, IEEE Computer Society, Washington, DC.
|10| Enrico Rukzio, etc, 'Context Ior Simplicity: A Basic Ior Context-
aware Systems Based on the 3GPP Generic User ProIile,
International ConIerence on Computational Intelligence 2004
(ICCI 2004), 17th-19th, December 2004 Istanbul, Turkey.
|11| J. Coutaz, J. L. Crowley, S. Dobson, and D. Garlan., 'Context is
Key, Communications oI the ACM, 48(3): 49-53, 2005.
|12| C. Bolchini, C. Curino, E. Quintarelli, F. A. Schreiber, and L.
Tanca. Context-ADDICT. Technical Report, 2006.044, Dip.
Elettronica e InIormazione, Politecnicodi Milano, 2006.
|13| Chen, Harry, Tim Finin, and Anupam Joshi. "An Intelligent
Broker Ior Context-Aware Systems." Adjunct Proceedings oI
Ubicomp 2003, Seattle, Washington, USA, October 12-15, 2003.
|14| Ioanna G. Roussaki, etc, 'Hybrid context modeling: A location-
based scheme using ontologies, Proceedings oI the Fourth Annual
IEEE International ConIerence on Pervasive Computing and
Communications Workshops (PERCOMW`06), 2006.
|15| H. Sridharan, H. Sundaram, and T.Rikakis, 'Computational
models Ior experiences in the arts, and multimedia, Proceedings
oI the 2003 ACM Workshop on Experiential TelePresence (ETP),
ACM, 2003, pp. 31-44.
|16| Karen Henricksen, Jadwiga Indulska and Andry Rakotonirainy,
'Modeling Context InIormation in Pervasive Computing Systems,
Pervasive 2002, LNCS 2414, pp. 167-180, 2002.
|17| D. Petrelli, E. Not, C. Strapparava, O. Stock, and M. Zancanaro.
Modeling context is like taking pictures. In Proc. oI the Workshop
The What, Who, Where, When, Why and How oI Context-
Awareness in CHI2000, 2000.
|18| MAIS, 'MAIS: Multi channel Adaptive InIormation System,
http://www.mais-project.it/.
|19| Ian S Burnett, Fernando Pereira, Rik Van de Walle and Rob
Koenen, The MPEG-21 Book, John Wiley & Sons Ltd. 2006.
|20| Tao Gu, Hung Keng Pung, Daqing Zhang. A Service-Oriented
Middleware Ior Building Context-Aware Services. Elsevier Journal
oI Network and Computer Applications (JNCA), Vol. 28, Issue 1,
pp. 1-18, January 2005
|21| Stephen J.H. Yang, etc, 'Context Model and Context Acquisition
Ior Ubiquitous Context Access in U-Learning Environments,
Proceedings oI the IEEE International ConIerence on Sensor
Networks, Ubiquitous, and Trustworthy Computing (SUTC`06),
2006.
|22| R. D. Virgilio and R. Torlone, 'A general methodology Ior
context-aware data access, Proceedings oI the International
Workshop on Data engineering Ior wireless and mobile access.
ACM, 2005, pp. 9-15.
|23| dongpyo Hong, hedda R. Schmidtke, Woontack Woo, 'Linking
Context Modeling and Contextual Reasoning, The 6th
International and interdisciplinary ConIerence on Modeling and
Using Context (Context`07), August 2007.
|24| N. F. Noy, D. L. McGuiness, 'Ontology Development 101: A
Guide to Creating Your First Ontology, StanIord Knowledge
Systems Laboratory Technical Report KSL-01-05, March 2001.
|25| W3C RDF, http://www.w3.org/RDF, 2000.
|26|Yahoo Search API, http://developer.yahoo.com/search/boss/
|27| Facebook Developers, http://wiki.developers.Iacebook.com/

You might also like