RESTful User Model API for the Exchange of User’s Preferences among

Adaptive Systems

Martin Bal´ık and Ivan Jel´ınek

Department of Computer Science and Engineering, Faculty of Electrical Engineering,

Czech Technical University, Karlovo n´amˇest´ı 13, 121 35 Prague, Czech Republic

Keywords:

Adaptive Hypermedia, Personalization, User Model, Data Exchange, REST API.

Abstract:

Adaptive Hypermedia Systems observe users’ behavior and provide personalized hypermedia. Users interact

with many systems on the Web, and each user-adaptive system builds its own model of user’s preferences and

characteristics. There is a need to share the personal information, and the current research is exploring ways to

share user models efﬁciently. In this paper, we present our solution for personal data exchange among multiple

hypermedia applications. First, we designed a communication interface based on the REST architectural style,

and then, we deﬁned data structures appropriate for the data exchange. Our user model is ontology-based and

therefore, the data from multiple providers can be aligned to achieve interoperability.

1 INTRODUCTION

Hypermedia applications on the World Wide Web

(WWW) are used by people on a daily basis. Peo-

ple use applications intended for work, entertainment,

communication, learning, etc. To improve usability

and facilitate the orientation within the large amount

of information, many of the applications provide auto-

mated personalization and adaptation features. Such

applications are called AdaptiveHypermedia Systems

(AHS) (Brusilovsky, 2001; Knutov et al., 2009) and

belong to the category of user-adaptive systems.

User-adaptive systems monitor users’ behavior

and keep track of characteristics, behavior, prefer-

ences, etc. of each individual user. The collection of

personal data associated with a speciﬁc user is called

the User Model (UM). The User Model is typically

built individually within each user-adaptive applica-

tion. This can cause a lot of issues, including the

cold start problem (Salton and McGill, 1983), infor-

mation inconsistency (Vassileva et al., 2003), both-

ering users with initial setup and asking for similar

data input, narrow personal information domain, etc.

In our work, we focus speciﬁcally on Adapitve Hy-

permedia Systems (AHS). Many different AHSs are

used by a single person on a daily basis. The effort of

extracting user’s characteristics is commonly repeated

in multiple self-contained applications. Moreover, in-

dividual user models can include complementary in-

formation. Therefore, it would be very beneﬁcial to

share the personal information and keep it synchro-

nized between all applications used by the person.

Sharing a User Model is a required, but a very

challenging task. Fortunately, the users are willing

to share personal information (Kobsa and Teltzrow,

2006; Gross and Acquisti, 2005). It is important

to ﬁnd a balance between information revelation and

personal privacy. The convenient choice of informa-

tion providers and proper integration of data sources

will bring many beneﬁts to the user.

The user usually works with diverse services on

the Web. Sometime, even the domain is quite similar.

In other cases, there can be connections with other

people with corresponding interests and new links to

reusable data. On the web, everything is intercon-

nected. Though, adding semantics to the raw data

brings new possibilities how the data can be used.

Through semantic annotations, the vision of Linked

Data (Bizer et al., 2009) is becoming true.

The main problem of personal information shar-

ing is the heterogeneity of User Models. To deal with

the interoperability problems, especially the semantic

heterogeneity of User Models needs to be addressed

(Carmagnola, 2009). Users’ data are stored in AHSs

in different formats and use different syntactic and

conceptual structures. Solving the AHS interoperabil-

ity problems by utilizing an Application Program In-

terface (API) is an appropriate research direction as

discussed in the context of adaptive educational sys-

tems in (Aroyo et al., 2006).

627

Balík M. and Jelínek I..

RESTful User Model API for the Exchange of User’s Preferences among Adaptive Systems.

DOI: 10.5220/0004896806270634

In Proceedings of the 16th International Conference on Enterprise Information Systems (ICEIS-2014), pages 627-634

ISBN: 978-989-758-028-4

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

The paper proposes a new adaptive hypermedia

system architecture and a method of personal data ex-

change. RESTful web service is used to expose both

user’s preferences and through user modeling process

obtained characteristics.

The paper is structured as follows. Section 1 deals

with the description of AHS and the tasks to solve. In

Section 2, a current state of the art of the discussed

topic is being reviewed. In Section 3, both the pro-

posed personal data storage structure and the RESTful

user model integration interface are described in de-

tail. In Section 4, a use case scenario utilizing the ben-

eﬁts of the RESTful user model interface is demon-

strated. Finally, the paper concludes by summarizing

results of the research and indicates the directions of

the future work; see Section 5.

2 RELATED WORK

User-adaptive applications are fundamentally based

on a process of observing user’s behavior and stor-

ing relevant user-speciﬁc information. This process

is called user modeling, and data is stored into a

user model. The user modeling process is a marginal

topic for the presented research, and an overview of

existing user modeling techniques can be found in

(Sosnovsky and Dicheva, 2010; Aroyo and Houben,

2010). In next subsections, we will ﬁrst review exist-

ing approaches used to share user-speciﬁc informa-

tion stored in the user model, and second, we will

explain fundamentals of the Web architecture, Web

services and Representational State Transfer.

2.1 User Model Sharing Approaches

Recently, motivated by the expansion of mobile and

ubiquitous devices, the researchers noticed the need

of sharing personal proﬁles of users, to enhance the

adaptation abilities of user-adaptiveapplications. Var-

ious techniques of sharing personal data have been

proposed (Carmagnola, 2009; Sosnovsky et al., 2009;

Cena and Furnari, 2009). The main obstacle is usu-

ally to solve the data interoperability problems. There

are currently two main approaches. The ﬁrst possibil-

ity is the way of standardization. The systems need

to adhere to a ﬁxed representation that needs to be re-

spected by all service providers. The second approach

is using mediation techniques to transfer the data from

one representation to another.

Standardization-based approach of AHS integra-

tion assumes a common semantic representation of

user models within all participating systems, usually

expressed by a shared ontology (Heckmann et al.,

2005). Implementing domain models of adaptive sys-

tems as ontologies is the ﬁrst step toward interoper-

ability. A standardized user-modeling ontology is a

possible solution to make the information exchange

possible. However, the fundamental requirement of

this approach is that all participants agree upon the

standardized ontology, which may pose an issue for

some of them (Berkovsky et al., 2009).

As a generalization of a standardization-based ap-

proach, the central user modeling server can be as-

sumed. A solution, where adaptive systems do not

need to support user modeling was used in (Kay

et al., 2002).The networked adaptive applications act

as clients, they simultaneously update the central user

model on the server, and they can request back per-

sonal information when needed. Another solution

based on a central server is presented in (van der

Sluijs and Houben, 2005), where the exchange of user

data between applications is supported by Semantic

Web technologies. The authors call the component

providing user model storage the Generic User Mod-

eling Component.

More complex solutions than standardization are

utilized by mediation (Berkovsky et al., 2007). With-

out any standard vocabulary, it is necessary to solve

syntactic and semantic heterogeneity issues. Ontol-

ogy mediation is the process of reconciling differ-

ences between heterogeneous ontologies in order to

achieve inter-operation between data sources anno-

tated with and applications using these ontologies

(de Bruijn et al., 2005). To overcome the hetero-

geneity of user modeling data, two steps are required.

First, the reasoning and inference mechanisms for

converting data between various representations, ap-

plications and domains need to be developed and ap-

plied. Second, the semantically enhanced knowledge

bases are exploit, facilitating the above reasoning and

inference (Berkovsky et al., 2009).

Although, there have been projects like (Heck-

mann et al., 2005) primarily focusing on the standard,

widely accepted user modeling ontology, (Martinez-

Villase˜nor et al., 2012) claims that the standardization

approach is not a feasible solution. Such statement

suggests to follow the second direction, the media-

tion of different domains based on natural language

processing and artiﬁcial intelligence. However, there

is also a possibility of a hybrid approach combining

both standardization and mediation approaches. Such

uniﬁcation is the aim of our approach.

2.2 The Web Architecture

Representational State Transfer (REST) (Fielding and

Taylor, 2002) architectural style was proposed by

ICEIS2014-16thInternationalConferenceonEnterpriseInformationSystems

628

Table 1: Vocabulary namespaces used in the REST API.

Preﬁx Namespace URI Description

dcterms: http://purl.org/dc/terms/ Dublin Core vocabulary

foaf: http://xmlns.com/foaf/0.1/ Friend of a Friend (FOAF) vocabulary

sioc: http://rdfs.org/sioc/ns# Semantically-Interlinked Online Communities (SIOC) vocabulary

owl: http://www.w3.org/2002/07/owl# Web Ontology Language (OWL) vocabulary

rdf: http://www.w3.org/1999/02/22-rdf-syntax-ns# RDF vocabulary

rdfs: http://www.w3.org/2000/01/rdf-schema# RDF Schema vocabulary

xsd: http://www.w3.org/2001/XMLSchema# XML Schema (XSD) vocabulary

um: http://intelleo.eu/ontologies/user-model/ns IntelLEO User Model Ontology

gomawe: http://fel.cvut.cz/gomawe/0.1/ GOMAWE Architecture Ontology

Roy Fielding and is based on a set of principles

for designing network-based software architectures.

The set of principles was deﬁned by four interface

constraints: identiﬁcation of resources, manipulation

of resources through representations, self-descriptive

messages, and hypermedia as the engine of applica-

tion state.

Web services based on the principles of REST

are called RESTful and can be considered as an al-

ternative to the SOAP-based web services. In the

past, RESTful services were used only for simple ad-

hoc services, and the area of enterprise systems was

scoped to the WS-* standards, e.g., SOAP, WSDL,

WS-Addressing, WS-Security. This is no longer the

case and RESTful services have been successfully ap-

plied in many enterprise applications. The challenge

is to use them correctly and to be able to align them

to solve the real problems (Adamczyk et al., 2011).

The comprehensive comparison of both technologies

was presented in (Pautasso et al., 2008; Pautasso and

Wilde, 2009). Conclusions of the comparison give

an advantage to the RESTful services in Web inte-

gration scenarios and prefer WS-* Web services for

enterprise application integration, where advanced se-

curity and Quality of Service (QoS) is required. The

practical comparison in (Guinard et al., 2011) con-

cludes that REST is lightweight, scalable, very easy

to understand, learn, and implement.

There are two important terms when speaking

about REST methods – safety and idempotence

(Fredrich, 2012). Safety means that calling the

method does not cause side effects and does not

change the state of the server. For example, the

GET

method of an API must adhere to the safety deﬁnition,

otherwise it can cause problems for other services and

result in unintended changes on the server. Idempo-

tence refers to a method that will produce the same

results if executed once or multiple times. The

PUT

and

DELETE

methods are deﬁned to be idempotent.

Therefore, it should be ensured that making multi-

ple requests produce the same result on the server.

Safe methods are idempotent at the same time, be-

cause they do not cause any changes on the server.

3 GENERIC AHS

ARCHITECTURE

In our previous work, we have formalized the Generic

Ontology-based Model for Adaptive Web Environ-

ments (GOMAWE) (Bal´ık and Jel´ınek, 2008). This

model deﬁnes the fundamental components of an

adaptive hypermedia application. One of the compo-

nents is the integration interface intended for expos-

ing user-speciﬁc data in the local storage to external

applications, and, at the same time, for obtaining user-

speciﬁc data from external providers.

Based on the GOMAWE formal speciﬁcation,

the Adaptive System Framework (ASF) (Bal´ık and

Jel´ınek, 2013) was built to support AHS development.

ASF provides the most typical AHS components serv-

ing as building blocks for further development. The

most recent extension of the framework is the inte-

gration component based on the principles described

in this article.

3.1 User Model

In our design, the user model has a special architec-

ture that was devised from the requirements, includ-

ing types of stored information and methods of in-

formation retrieval. We divided the user data into two

parts – the User Proﬁle and User Model. This division

corresponds to explicit and implicit personalization

styles. Implicit personalization is performed by the

adaptivesystem. On the other hand, explicit personal-

ization is performed by the user using special features

of the system. A system with such personalization

features is called adaptable system. Our architecture

can be perceived as a hybrid solution combining both

personalization methods.

The User Proﬁle contains explicit user’s prefer-

ences, i.e., preferences explicitly ﬁlled by the users.

This data is stored as key-valuepairs, see Deﬁnition 1.

RESTfulUserModelAPIfortheExchangeofUser'sPreferencesamongAdaptiveSystems

629

UserModelAttribute

xsd:string

UserProleAttribute

foaf:Person

xsd:string xsd:stringxsd:string

xsd:DateTime

associatedUser

owl:Thing

xsd:DateTime

dateUpdated

foaf:name

hasValue

associatedDomainInstance

foaf:name

hasValue

dateUpdated

Figure 1: Selected parts of the GOMAWE Architecture Ontology.

The key is usually a constant string deﬁned by the de-

veloper of the application. Typical origin of the data

is the “settings page” of the application. However,

when we assume integration with other personal data

providers, the User Proﬁle data can be ﬁlled utiliz-

ing social networks’ user proﬁles available in services

like Facebook, Twitter or LinkedIn.

Deﬁnition 1 (User Proﬁle). The User Proﬁle of a user

u is a tuple UP

(u)

= (A, V, r)

r : A → V|∀a ∈ A : r

(a)

∈ V

, (1)

where A is a set of attributes deﬁned as the vocabu-

lary of user’s characteristics, V =

a∈A

is a set of

attribute values and V

is the range of attribute a.

The User Model denotes in our terminology

a model containing implicit user’s characteristics.

While the explicit characteristics are set by the user

oneself, the implicit characteristics are devised by the

adaptive system. The Domain Model captures the

most important types of objects in the application

context. It deﬁnes the conceptual framework and se-

mantics of hypermedia content. The system collects

various values related to a speciﬁc object of the do-

main and stores the values into the User Model, see

Deﬁnition 2. Moreover, the characteristics are cate-

gorized to a set of dimensions. The assignment of

attributes and dimensions is utilized in the user mod-

eling process, however, it will not be discussed here

as it is not substantial for the web services and the

application integration interface.

Deﬁnition 2 (User Model). The User Model of a user

u is a tuple UM

(u)

= (D, A, V, r)

r : D× A → V|∀a ∈ D× A : r

(a)

∈ V

, (2)

where D is a set of domain instances, A is a set of

attributes deﬁned as the vocabulary of user’s implicit

characteristics, V =

a∈A

is a set of attribute values

and V

is the range of attribute a.

As examples of User Model attributes we can

mention knowledge of a lesson in an educational ap-

plication, or bought item indication in a web com-

merce application. The User Model is able to capture

even relationships among users. Based on the fact that

a user entity is part of the Domain Model, the User

Model can include attributes friend, follower, etc. re-

lating the user to other users. This information can be

acquired from social networking applications.

3.2 A RESTful User-speciﬁc Data

Interface

Based on the comparison in Section 2, we decided to

use RESTful web services to design the user-adaptive

application’s personal data interface. First, the inter-

face does not require advanced security, and second,

the most important requirement is ﬂexibility supple-

mented with easy and intuitive development.

The exchange of information is based on standard

metadata vocabularies and ontologies. Utilizing the

proposed uniﬁed user model data structure, ontolo-

gies can be aligned with relative ease and transla-

tion between two domains can be achieved. At the

same time, the design does not force participating sys-

tems to agree fully on a ﬁxed domain model ontology,

and advanced mediation techniques can be used when

needed.

3.2.1 RDF Vocabularies

Our user-data-integration interface uses three types

of vocabularies – ﬁrst, the standard vocabularies,

second, user-model-speciﬁc vocabularies and ﬁnally,

the domain-speciﬁc vocabularies. RDF resources

and their attributes reuse existing, widely adopted

vocabularies such as the Dublin Core (Dub, ),

Friend of a Friend (Brickley and Miller, 2010), and

Semantically-Interlinked Online Communities (Sio, )

ICEIS2014-16thInternationalConferenceonEnterpriseInformationSystems

630

Client REST

REST UM

Integration service

«rdfsResource»

User Model

«rdfsResource»

User Profile

Adaptation

Manager

GET /api/users/lookup?

email={emailAddress}

:gomawe:User

GET /api/users/{user_id}/profile

getUserProfileAttributes(user_id)

getAl lAttributes(user_id)

:List<UserProfil eAttribute>

GET /api/users/{user_id}/model?

modifiedSince={dateTime}

getUserModelAttributes(

user_id, dateT ime)

getAl lAttributes(

user_id, dateT ime)

:List<UserModel Attribute>

Figure 2: Example of request sequence to retrieve users’s personal data.

vocabularies. To ensure good interoperability, ASF

maps as many attributes as possible to these standard

vocabularies. User-model-speciﬁc vocabularies used

in our design include IntelLEO User Model Ontol-

ogy (Jovanovic et al., 2012) and GOMAWE Archi-

tecture Ontology, see Fig. 1. Domain-speciﬁc vocab-

ularies differ application from application and refer to

domain-speciﬁc concepts to which the user’s charac-

teristics are related.

The Table 1 lists the vocabulary namespaces used

in the ASF-based application REST API.

3.2.2 ASF-based Application API Resources

The fundamental resources of the RESTful API are

listed in Table 2.

Table 2: Primary resources of the REST API.

Type Example

gomawe:UserProﬁleAttribute /api/users/{user id}/proﬁle

gomawe:UserModelAttribute /api/users/{user id}/model

foaf:Person

1) /api/users/{user id}

2) /api/users/lookup?email={emailAddress}

In Table 2,

gomawe:UserProfileAttribute

refers to the list of User Proﬁle attributes as-

signed to the user with the matching ID.

gomawe:UserModelAttribute

refers to the list

of User Model data of the user with the matching

ID.

foaf:Person

refers to the owner of the user

account corresponding to the speciﬁed id (1) or email

address (2).

3.2.3 REST Operations

This subsection summarizes the REST operations

supported by the API. The API supports three HTTP

operations:

GET, PUT, POST

. The

DELETE

operation

is not supported, as the user data can only be extended

or updated, and no attributes can be removed.

Only the RDF/XML format is supported at this

time, therfore the HTTP Accept header value should

be set to:

GET /api/users HTTP/1.1

Accept: application/rdf+xml

The following parameters can be used to reﬁne the

requests:

• Modiﬁed since – Parameter

modifiedSince

can

be used to limit the listed attributes to only those

that were modiﬁed after the entered time value.

RESTfulUserModelAPIfortheExchangeofUser'sPreferencesamongAdaptiveSystems

631

The parameter value is of type

xsd:date

xsd:dateTime

• Resource paging – Parameter

limit

can be used

to limit the number of listed attributes to avoid

large data transfers. Parameter

offset

can be

used to request additional pages starting with the

speciﬁed item. For both parameters integer type

value is allowed.

The following are the typical uses of HTTP meth-

ods with explanatory examples:

• HTTP GET – The

HTTP GET

method is supported

by all API resources. It is used to retrieve the

user’s account and his or her user modeling data.

First, the application needs to negotiate the cor-

rect resource IDs and after matching user’s ac-

count, the user model attributes can be requested,

see Fig. 2. The following request can be used to

retrieve user proﬁle attributes of a user:

GET http://example.org/api/users/

{user_id}/profile HTTP/1.1

Accept: application/rdf+xml

• HTTP PUT – The

HTTP PUT

method is supported

only by UserProﬁle and UserModel resources. It

is used to update the models by an external ap-

plication. The

PUT

method operation needs to be

idempotent, therefore, using this method means

creating new attributes or replacing values of ex-

isting attributes.

PUT http://example.org/api/users/

{user_id}/profile HTTP/1.1

• HTTP POST – The

HTTP POST

method is sup-

ported only by UserProﬁle and UserModel re-

sources. The operation of this method is reserved

for the cases of incremental updates of the models.

It will be more extensivelyutilized in future exten-

sions of the REST API. The more precise value

updates can be based on arithmetic mean or the

time of last update. This operation is not idempo-

tent and should be called only once to avoid inap-

propriate user model changes.

3.2.4 Authentication

To avoid misuse of both the personal and the domain

data provided by the RESTful application interface,

authentication of requests is required. We use HTTP

basic authentication. Using HTTPS protocol is rec-

ommended. Otherwise, the username and password

would be sent without encryption.

4 APPLICATION USE-CASE

The proposed approach of AHS integration will be

demonstrated in an adaptive learning scenario. There

are currently three separate systems used by students

in a programming course, each with a different pur-

pose. Although each of the systems stores different

information, they can beneﬁt from each other, ex-

change user’s data and extend understanding of the

user’s knowledge and preferences.

The integration module implementation is a work-

in-progress project, and the results will be evaluated

in the future. Currently, we do not focus on ontol-

ogy mapping issues. Our aim is to verify the web

service interface and the method of exposing user-

speciﬁc personal data.

There are three participating information systems

in our scenario. The ﬁrst system is an adaptive learn-

ing system containing learning materials and simple

test questions at the end of each lesson. The sec-

ond system is intended for selection of a program-

ming project topic and for submission of the com-

pleted works. The third system performs an auto-

matic evaluation of several programming tasks as-

signed to students in the course of a year. The adap-

tation features of the learning system can beneﬁt to

a considerable extend by additional information about

students’s progress on solving the assigned tasks and

their achieved results. The adaptive guidance within

the learning course can be also really well tailored

based on the student’s project topic, and the subtopics

related to his or her project.

In the current setup, see Fig. 3, each of the sys-

tems is equipped with the RESTful data interface, and

all systems are interconnected through a central medi-

ation service. Each of the systems conforms to a sim-

ilar domain, and their domain models overlap with

some mutually related concepts.

The User Model Attributes exchange is subject to

a certain level of mutual understanding of the domain

semantics by the communicating counterparts. In our

experimental scenario, for the ease of personal-data

interface demonstration, all three systems use the do-

main ontology of the adaptive learning system. The

data structure deﬁned by the ontology is mapped lo-

cally. Furthermore, the RESTful API of the learn-

ing systems is extended by the domain resources and

systems can negotiate the correct domain instance

references. All domain instances are represented as

RDFS-resources identiﬁed by a unique Uniform Re-

source Identiﬁer (URI). The URIs are used as a link

between the overlay User Model and the domain layer

of the application.

The User Proﬁle Attributes are not domain-

ICEIS2014-16thInternationalConferenceonEnterpriseInformationSystems

632

Service Mediator

ASF-based

Adaptive Learning System

Project assignment

application

Programming tasks

evaluator

REST API

ASF Extension

REST API

ASF Extension

REST API

Figure 3: Applications participating in the use-case integra-

tion scenario.

dependent. The meaning of the included prefer-

ences needs to be well represented using common

and widely used ontologies. In later implementations,

public services like Facebook, Twitter or LinkedIn

can be used as user’s personal information providers,

and therefore, it is important to be able to match the

preferences adequately.

5 CONCLUSION

The paper has proposed our novel solution to deal

with the syntactic and semantic heterogeneity of per-

sonal information in adaptive hypermedia systems.

To make the integration of users’ preferences and

characteristics possible, a special architecture of User

Proﬁle and User Model was deﬁned, and a RESTful

web service application interface was introduced.

Compared to other solutions presented in Sec-

tion 2, our proposal combines both shared format and

conversion approaches resulting in a hybrid solution

and utilizing advantages of both approaches.

The strength of our approach is the generic ar-

chitecture, and the fact that realization can be sup-

ported by the Adaptive System Framework. Even

non-adaptive systems can be extended by the ASF-

based integration module and operate as providers

of users’ personal data. An integration use-case of

that kind was presented using the case of multiple

learning-support applications. As a result, the adap-

tive learning application can beneﬁt from the integra-

tion and promptly react on achievements of students

recorded by related learning-support systems.

In our future work, we will thoroughly evaluate

the integration impact on a deeper understanding of

user’s knowledge, preferences and needs.

The special user modeling data storage and the in-

troduction of the web-service-based application inter-

face of a user-adaptive application is one of the im-

portant steps towards the generic AHS architecture

formalization.

ACKNOWLEDGEMENTS

The results of our research form a part of the scientiﬁc

work of a special research group WEBING.

REFERENCES

Dublin Core Metadata Initiative. http://dublincore.org/.

[Accessed: August 7, 2013].

SIOC Project. http://sioc-project.org/. [Accessed: August

7, 2013].

Adamczyk, P., Smith, P. H., Johnson, R. E., and Haﬁz, M.

(2011). REST and Web Services: In Theory and in

Practice. In Wilde, E. and Pautasso, C., editors, REST:

From Research to Practice, chapter 2, pages 35–57.

Springer New York, New York, NY.

Aroyo, L., Dolog, P., Houben, G.-J., Kravcik, M., Naeve,

A., Nilsson, M., and Wild, F. (2006). Interoperability

in personalized adaptive learning. Educational Tech-

nology & Society, 9(2):4–18.

Aroyo, L. and Houben, G.-J. (2010). User modeling and

adaptive Semantic Web. Semantic Web, 1(1–2):105–

110.

Bal´ık, M. and Jel´ınek, I. (2008). Towards Semantic Web-

based Adaptive Hypermedia Model. In ESWC Ph.D.

Symposium, pages 1–5, Tenerife, Spain.

Bal´ık, M. and Jel´ınek, I. (2013). Adaptive System Frame-

work: A Way to a Simple Development of Adaptive

Hypermedia Systems. In The Fifth International Con-

ference on Adaptive and Self-Adaptive Systems and

Applications (ADAPTIVE2013), pages 20–25, Valen-

cia, Spain. IARIA.

Berkovsky, S., Heckmann, D., and Kuﬂik, T. (2009). Ad-

dressing challenges of ubiquitous user modeling: Be-

tween mediation and semantic integration. In Ad-

vances in Ubiquitous User Modelling, pages 1–19.

Berkovsky, S., Kuﬂik, T., and Ricci, F. (2007). Mediation

of user models for enhanced personalization in recom-

mender systems. User Modeling and User-Adapted

Interaction, 18(3):245–286.

Bizer, C., Heath, T., and Berners-Lee, T. (2009). Linked

data-the story so far. International Journal on Seman-

tic Web and Information Systems, 5(3):1–22.

Brickley, D. and Miller, L. (2010). FOAF vocabulary spec-

iﬁcation 0.98. http://xmlns.com/foaf/0.1. [Accessed:

August 7, 2013].

Webing research group – http://webing.felk.cvut.cz.

RESTfulUserModelAPIfortheExchangeofUser'sPreferencesamongAdaptiveSystems

633

Brusilovsky, P. (2001). Adaptive Hypermedia. pages 87–

110.

Carmagnola, F. (2009). Handling semantic heterogeneity in

interoperable distributed user models. In Advances in

Ubiquitous User Modelling, pages 20–36.

Cena, F. and Furnari, R. (2009). A model for feature-based

user model interoperability on the web. In Advances

in Ubiquitous User Modelling, pages 37–54.

de Bruijn, J., Mart´ın-Recuerda, F., Ehrig, M., Polleres, A.,

and Predoiu, L. (2005). Ontology Mediation Manage-

ment v1. Deliverable D4.4.1. Technical report, SEKT.

Fielding, R. T. and Taylor, R. N. (2002). Principled design

of the modern Web architecture. ACM Transactions

on Internet Technology, 2(2):115–150.

Fredrich, T. (2012). RESTful Service Best Practices.

Gross, R. and Acquisti, A. (2005). Information revelation

and privacy in online social networks. In ACM Work-

shop on Privacy in the Electronic Society (WPES).

Guinard, D., Ion, I., and Mayer, S. (2011). In Search of

an Internet of Things Service Architecture: REST

or WS-*? A Developers’ Perspective. In Puiatti,

A. and Gu, T., editors, Mobile and Ubiquitous Sys-

tems: Computing, Networking, and Services (MobiQ-

uitous 2011), pages 326–337, Copenhagen, Denmark.

Springer.

Heckmann, D., Schwartz, T., Brandherm, B., and Schmitz,

M. (2005). GUMO The General User Model Ontol-

ogy. pages 428–432.

Jovanovic, J., Gasevic, D., and Siadaty, M.

(2012). IntelLEO User Model Ontology.

http://intelleo.eu/ontologies/user-model/spec/. [Ac-

cessed: August 7, 2013].

Kay, J., Kummerfeld, B., and Lauder, P. (2002). Personis:

A server for user models. In De Bra, P., Brusilovsky,

P., and Conejo, R., editors, Adaptive Hypermedia and

Adaptive Web-Based Systems (AH2002), pages 203–

212, Malaga, Spain. Springer.

Knutov, E., De Bra, P., and Pechenizkiy, M. (2009). AH

12 years later: a comprehensive survey of adaptive

hypermedia methods and techniques. New Review of

Hypermedia and Multimedia, 15(1):5–38.

Kobsa, A. and Teltzrow, M. (2006). Convincing Users to

Disclose Personal Data. Privacy-Enhanced Personal-

ization, pages 19–21.

Martinez-Villase˜nor, M. D. L., Gonzalez-Mendoza, M., and

Hernandez-Gress, N. (2012). Towards a ubiquitous

user model for proﬁle sharing and reuse. Sensors

(Basel, Switzerland), 12(10):13249–83.

Pautasso, C. and Wilde, E. (2009). Why is the web

loosely coupled?: a multi-faceted metric for service

design. In Proc. of the 18th international confer-

ence on World Wide Web (WWW ’09), pages 911–920,

Madrid, Spain. ACM.

Pautasso, C., Zimmermann, O., and Leymann, F. (2008).

RESTful web services vs. ”big” web services: mak-

ing the right architectural decision. In WWW 08: Pro-

ceeding of the 17th international conference on World

Wide Web, pages 805–814, New York, NY, USA.

ACM.

Salton, G. and McGill, M. J. (1983). Introduction to modern

information retrieval. McGraw Hill Book Co., New

York.

Sosnovsky, S., Brusilovsky, P., Yudelson, M., Mitrovic, A.,

Mathews, M., and Kumar, A. (2009). Semantic In-

tegration of Adaptive Educational Systems. In Ku-

ﬂik, T., Berkovsky, S., Carmagnola, F., Heckmann,

D., and Kr¨uger, A., editors, Advances in Ubiquitous

User Modelling, pages 134–158. Springer.

Sosnovsky, S. and Dicheva, D. (2010). Ontological tech-

nologies for user modelling. International Journal of

Metadata, Semantics and Ontologies, 5(1):32–71.

van der Sluijs, K. and Houben, G. (2005). Towards a generic

user model component. In Proc. of PerSWeb05, Work-

shop on Personalisation on the Semantic Web, pages

47–56, Edinburgh, UK.

Vassileva, J., McCalla, G., and Greer, J. (2003). Multi-agent

multi-user modeling in I-Help. User Modelling and

User Adapted Interaction, 13(1-2):179–210.

ICEIS2014-16thInternationalConferenceonEnterpriseInformationSystems

634