A FRAMEWORK FOR DELIVERING PERSONALIZED

E-GOVERNMENT TOURISM SERVICES

Malak Al-hassan, Helen Lu and Jie Lu

Quantum Computation and Intelligent Systems Centre, Faculty of Engineering and Information Technology

University of Technology, Sydney, P.O. Box 123 Broadway, NSW, Australia

Keywords:

e-Government, Personalization, Online tourism services, Ontology, Recommendation systems, Framework.

Abstract:

E-government (e-Gov) has become one of the most important parts of government strategies. Signiﬁcant

efforts have been devoted to e-Gov tourism services in many countries because tourism is one of the major

proﬁtable industries. However, the current e-Gov tourism services are limited to simple online presentation

of tourism information. Intelligent e-Gov tourism services, such as the personalized e-Gov (Pe-Gov) tourism

services, are highly desirable for helping users decide ”where to go, and what to do/see” amongst massive

number of destinations and enormous attractiveness and activities. This paper proposes a framework of Pe-

Gov tourism services using recommender system techniques and semantic ontology. This framework has

the potential to enable tourism information seekers to locate the most interesting destinations with the most

suitable activities with the least search efforts. Its workﬂow and some outstanding features are depicted with

an example.

1 INTRODUCTION

E-government uses innovative systems made possi-

ble by information and communication technology to

achieve better services to citizens or businesses, as

well as to enhance process and management of pub-

lic sector. In the last few years, e-Gov initiatives have

been launched by most of the governments around the

globe (Millard et al., 2004). E-Gov initiatives cover

three main activities, e-services, e-democracy, and e-

administration. This study focuses on e-government

services, particularly e-Gov tourism services.

It has been shown in the literature that intelligent

personalization is a clear direction in the development

of e-Gov services. The current e-Gov development,

however, is mainly at the stage of the implementa-

tion of transaction services with a few exceptions,

such as the development in Singapore and Canada

(Pieterson et al., 2007; Wauters et al., 2007). Both

of these countries have offered their citizens sim-

ple personalized services through their ofﬁcial por-

tal websites (Accenture, 2004). More advanced and

intelligent e-Gov systems are highly desirable. We

have recently proposed a new conceptual framework

of Pe-Gov services from citizen-centric approach (Al-

Hassan et al., 2009). Using this framework, Pe-Gov

services within a speciﬁc organization can be deliv-

ered to citizens based on their interests, preferences,

characteristics and personal needs without excessive

input from users.

Within tourism domain, many governments

around the world seek to promote tourism industry

as non-proﬁt services by providing information about

attractions, activities and events that could be visited

at speciﬁc destination, because tourism is one of the

major proﬁtable industries (International tourism re-

ceipts grew to 944 billion USD in 2008). E-Gov

tourism services have the potential for delivering bet-

ter governmental tourism services to citizens and

overseas tourists, improving the quality of the pro-

vided services, improving the access to information

(24 hours a day, 7 days a week). The current e-Gov

tourism service is mainly about presenting tourism in-

formation online, but not personalized, e.g., everyone

who uses the system will be exposed with the same

set of information.

In order to make e-Gov tourism services more at-

tractive to users, it has been realized that the e-Gov

tourism services must be delivered in a user-centric

manner, in which e-Gov could improve the delivery

of its services to users on a personalized basis to en-

sure that heterogeneous users’ needs and interests are

met without excessive data input from users (Und-

heim and Blakemore, 2007).

263

Al-hassan M., Lu H. and Lu J.

A FRAMEWORK FOR DELIVERING PERSONALIZED E-GOVERNMENT TOURISM SERVICES.

DOI: 10.5220/0002803802630270

In Proceedings of the 6th International Conference on Web Information Systems and Technology (WEBIST 2010), page

ISBN: 978-989-674-025-2

Based on our previous work (Al-Hassan et al.,

2009); this paper proposes a speciﬁc framework

for delivering personalized e-Gov tourism (Pe-Gov

tourism) services from user-centric approach. This

framework employs advanced recommender system

(RS) techniques and semantic ontology to enable

users to ﬁnd the most interesting destinations with the

most favourable attractions and the most suitable ac-

tivities with least effort.

The rest of this paper is organized as follows: Sec-

tion 2 reviews some of existing online tourism service

systems. Section 3 presents the proposed framework

in details and discusses its components. Section 4 il-

lustrates how the framework works using an exam-

ple. Finally, the conclusion and future work are high-

lighted in Section 5.

2 ONLINE TOURISM SERVICES

Tourism industry has undertaken signiﬁcant evolution

along the development of web-based technology. Nu-

merous travel-related websites government and non-

government agencies have been established to offer

travel-related information and services. Nowadays,

online presence of tourism information is one of the

important services offered by e-Gov. Nearly, all the

leading countries that adopted e-Gov initiatives offer

online tourism services (Accenture, 2004). For ex-

ample, Tourism NSW Government Service website

(http://corporate.tourism.nsw.gov.au/) offers tourism

services and information, which includes attractions,

activities and events that could be visited at a speciﬁc

destination. In addition, it provides some links to pri-

vate travel and tourism agencies for other tourism and

travel services.

The current commercial tourism service systems

(i.e. websites) offer tourism services to users through

using either search tools or content browsing. Several

popular online travel websites have emerged, such as

Expedia, Travelocity and ebookers. The offered ser-

vices through tourism and travel websites could in-

clude destinations to visit, airlines and hotels reserva-

tion, tour operators, travel planning and many more

(Rabanser and Ricci, 2005; Ricci et al., 2006).

Even though the existing travel-related websites

can help users ﬁnd interesting tourism information

and services, it seems that these websites are rela-

tively poor in aiding users to plan their trips, which

includes the determination of where, when and how

to go for a trip and what to do at a certain destina-

tion from the huge amount of information available.

Users may spend signiﬁcant amount of time and ef-

fort in ﬁnding what they really want. This could lead

users to waste much more time or effort until they

ﬁnd their request (Fesenmaier, 2006). Additionally,

tourists demand tourism products tailored to their ac-

tual preferences. On the other hand, they often have

complex and multi-interests and their travel desires

change rapidly and usually require the tourism prod-

ucts with shorter life cycles (Berka and Plnig, 2004).

All these make the recommendation of the most ap-

propriate tourism products to a speciﬁc tourist be-

come a challenging tasks.

Commercial travel and tourism applications have

started adopting intelligent techniques, such as RS

techniques to assist users in choosing their preferred

tourism services (Berka and Plnig, 2004; Fesen-

maier, 2006). TripleHop’s TripMatcher (used by

www.skieurope.com) and VacationCoach’s expert ad-

vice platform (used by travelocity.com) have been de-

veloped to offer recommendations to users by sug-

gesting a list of destinations to visit (Rabanser and

Ricci, 2005; Staab et al., 2002). Both systems use the

content based recommendation techniques. Dietorecs

RS (dietorecs.itc.it), on the other hand, has devel-

oped to provide users with personalized recommen-

dations of complex tourist products including destina-

tions, accommodations and activities. Dietorecs uses

the case-based reasoning technique to generate rec-

ommendations (Ricci et al., 2006).

Most of the existed online travel and tourism rec-

ommendation systems only use single recommenda-

tion technique, and therefore, lack of intelligent and

effective support to users, as described in the above

examples (Ricci et al., 2006). Recently, an expert

software agent, named ”Traveller” was proposed to

offer tourism services that able to help users (tourists)

in planning their travels. A hybrid technique was used

to build this agent, which includes two RS techniques,

the collaborative ﬁltering and the content-based rec-

ommendation techniques (Schiafﬁno and Amandi,

2009). Another intelligent personalized RS is pro-

posed to recommend tourists attractions in a speciﬁc

city. This system is built using applying ontology

theory, bayesian network, analytic hierarchy process

and spatial web service technology (Huang and Bian,

2009).

There are growing interests in building intelligent

systems for e-tourism and travel applications, not only

from commercial travel service providers’ view point,

but also from governments’ view point. Unlike non-

government agencies, where they attempt to promote

tourism services as a business application, e-Gov fo-

cuses on promoting tourism services as non-proﬁt ser-

vices by introducing attractions, activities and events

that could be visited at speciﬁc destination.

E-Gov tourism services should be delivered in a

WEBIST 2010 - 6th International Conference on Web Information Systems and Technologies

264

user-centric manner, in which e-Gov could improve

the delivery of its services to users on a personal-

ized basis to ensure that the actual users’ needs and

interests are met with least time and efforts. In-

telligent system techniques, particularly RS, can be

used for achieving personalization in e-Gov tourism

services. However, the commonly used RS tech-

niques in e-commence applications cannot be used di-

rectly for e-Gov services due to the special feature

of e-Gov tourism services, such as enormous vari-

ety in tourism items, heterogeneous interests among

the users, multi-interests for individual user, and dy-

namic nature of these interests. Therefore, new RS

techniques are highly desirable for developing RS for

e-Gov, particularly for online tourism services.

3 FRAMEWORK OF Pe-Gov

TOURISM SERVICES

Aiming to serve citizens better and deliver the right

information to the right citizen with less time and ef-

fort, we propose a conceptual framework of Pe-Gov

tourism services based on our proposed framework

in (Al-Hassan et al., 2009), as depicted in Figure

1. This framework is comprised of four main com-

ponents: Customized User Interface (CUI), Knowl-

edge Base Repository (KBR), Intelligent Recommen-

dation Engine (IRE), and User Data Collector (UDC).

These components are jointly work together to offer

users recommendations that meet their interests and

requirements. These recommendations include the

most favourable attractions with the most suitable ac-

tivities at a speciﬁc destination. The following sub-

sections will give a general description of each com-

ponent with focus on conceptual functionality rather

than technical details.

3.1 Customized User Interface

A user needs to register with the system before he/she

can login to the system to get recommendations on the

tourism e-Gov services. The user interface of the sys-

tem consists of two units: Registration or login unit,

and the customized user interface (CUI) unit. The

CUI unit is the interactive interface between the sys-

tem and the active user. It contains the tourism items

presented in a tree-like catalogue and ”MyInterstAl-

boum”. The CUI is generated based on the user’s in-

formation, including user’s demographic information,

personal interests and preferences within the tourism

e-Gov domain. Each registered user has a ”MyIntere-

stAlbum” in his/her own customized page.

”MyInterestAlbum” is used to facilitate users in-

teractive with the system efﬁciently and friendly. It

contains all selected tourism items that the user plan-

ning to visit or already visited, rating to the tourism

items that have already been visited and a feedback to

the selected tourism items. Each time a user enters the

system, he or she will be asked to give a feedback and

rating for any outstanding tourism item in the ”My-

InterestAlbum”, which has been visited or conducted

by the user but not rated yet. The new feedbacks and

ratings that are provided by the user will be collected

and stored in the user proﬁle DB by the UDC com-

ponent. The feedbacks would be used for further im-

provements of the system whereas the ratings would

be used by IRE component for further processing.

3.2 User Data Collector

The user data collector component is responsible for

collecting user related information to the tourism e-

Gov services, including the demographic informa-

tion, personal interests and preferences and the user’s

tourism e-Gov service usage history.

This component performs two tasks. Firstly, it

collects user related data from the system through

user registration, rating experienced items and feed-

backs. Implicit and explicit acquisitions can be used

for collecting user related data for web personal-

ization (Eirinaki and Vazirgiannis, 2003; Markellou

et al., 2005). The implicit acquisition will be used

to collect data about citizens’ preferences or interests

implicitly. It can be collected via tracing a user’s nav-

igation sequences when he/she navigates the hierar-

chical tourism e-Gov service catalogue. While the ex-

plicit acquisition will be used during registration and

interaction between the user and the system to capture

users personal information, including users’ demo-

graphic data, their interests and preferences through

pre-deﬁned forms, as well as to collect users’ rat-

ings of tourism items and feedbacks in ”MyIntere-

stAlboum”. Secondly, UDC updates the user proﬁle

DB regularly by recording the captured data from the

UI component into the KBR component.

3.3 Knowledge Base Repository

This component contains relevant knowledge of the

tourism e-Gov domain. It contains tourism e-Gov on-

tology and the users proﬁle database (DB).

3.3.1 Tourism e-Gov Ontology

Tourism ontology (O) describes the fundamental con-

cepts/classes (C) that compose the tourism e-Gov do-

A FRAMEWORK FOR DELIVERING PERSONALIZED E-GOVERNMENT TOURISM SERVICES

265

Figure 1: Framework of Pe-Gov Tourism Service System.

main; relationships among these concepts, the prop-

erties a concept can have or share. The tourism ontol-

ogy under government context could include attrac-

tions, activities and events that are available at a spe-

ciﬁc destination. For this study, the Australian online

information of tourism items, including attractions,

activities and events, is the source to create concepts

and associated relationships in the tourism e-Gov on-

tology. The related Australian government tourism

websites have been visited to collect the required con-

cepts to create the ontology. Typically, developing do-

main ontology starts with the design of the ontology

schema and ﬁnishes with the populating of the ontol-

ogy data (Noy and McGuinness, 2001). The ontology

schema contains the deﬁnition of the various classes,

properties and relationships that represent a speciﬁc

domain world, whereas in the populating ontology,

the instances for each class in the ontology schema

can be created.

Figure 2 depicts a diagram of tourism e-Gov on-

tology schema. It includes the destination class,

which has ”attraction”, ”event”, and ”activity” sub-

classes. Each subclass can also have its subclasses.

All classes are characterised by a set of speciﬁc prop-

erties. For example, the ”Attraction” class can have

properties ”Time of visit”={summer, autumn, win-

ter, spring}, ”location”, and ”Attraction visiting pur-

pose”={relax, adventure, treatment, exploring}. Ad-

ditionally, classes are linked using relationship types.

For instance, the ”has-events” relationship links the

”Attraction” class and the ”Event” class.

Ontology schema can show the semantic rele-

vance among tourism e-Gov concepts at different

level of hierarchy (as taxonomies), therefore, it can

guide the hierarchical representation of available e-

Gov tourism items in the CUI component and can be

used as a means of locating and accessing services

(Vassilakis and Lepouras, 2006). It can also be used

by the IRE component to facilitate the semantic rule

reasoning to ﬁnd the matching (similarity) between

the users’ requests and the available tourism items,

which could enhance the retrieval of the appropriate

content for each user.

3.3.2 User Proﬁle Database

The user proﬁle DB contains all the data that related

to users, including users’ demographic data and the

online tourism items that used by the users. The user-

related data is stored in the DB as two parts: static

and dynamic parts. The static data part is about users’

demographic data, including name, age, residential

status, favourite type of attractions and general inter-

ests related to tourism. This data can be collected ex-

plicitly through the tourist account registration when

a user visits the system for a ﬁrst time. In contrast,

dynamic data includes the preferences and interests

of users that can be extracted from the interaction be-

tween users and tourism websites. In our framework,

dynamic data could be the visited attractions and the

activities/events that are preferred by users and their

opinions (ratings) about the tourism items. This data

can be attained implicitly or explicitly. Implicit data

can be attained via exploring a user’s web-navigation

patterns (i.e. click streams) to infer the user’s prefer-

ences and interests while the explicit data can be at-

tained from ”MyInterestAlbum”, in which the tourism

items that a user prefers and user’s rating for these

items are available. The user proﬁle is a crucial part

for generating recommendations. It forms one of

the main sources of information that will be used to

WEBIST 2010 - 6th International Conference on Web Information Systems and Technologies

266

Figure 2: Tourism e-Gov Ontology Diagram.

accomplish the recommendation process in the IRE

component.

3.4 Intelligent Recommendation Engine

This component is responsible of recommending for

users the most interesting attractions and the most

suitable activities at a speciﬁc destination.

IRE component uses a new hybrid methodology

in order to overcome the limitations of the existing

RS techniques and to enhance the recommendations.

The typical RS techniques, such as the Content Based

(CB) and the collaborative ﬁltering (CF), suffer from

a set of limitations which may hinder them from be-

ing directly applied to e-Gov tourism domain. For

instance, the CB technique would recommend highly

similar tourism items that belong to the same group

or category of the given item. Hence, they ignore the

items that belong to other groups but might have in-

terest to the target user. Such technique would not be

suitable for e-Gov tourism services as most of users

would like to see new tourism items (Heterogeneous

needs) which may be in different categories. Further-

more, the rating method used by the CF techniques

seems not directly applicable to the e-Gov tourism

domain, because it may be hard to use a single rat-

ing value to infer a person’s taste as people can give

the same rating to an item for vastly different reasons

(Schafer et al., 2007). The CF techniques also suffers

from data sparsity problem which occurs when the at-

tained ratings are few compared to the number of the

available items, and the cold-start problem which oc-

curs when a new user or a new item enters to the sys-

tem.

A combination of recommendation techniques,

which called hybrid approach, is used in the literature

to overcome the drawbacks of the existing techniques

and to improve the recommendation performance for

a particular application. Different hybrid approaches

have been developed in the literature which combine

CF with CB and some other techniques in different

ways (Adomavicius and Tuzhilin, 2005). The new

hybrid methodology, in our framework, combines the

probability technique, item-based CF, and semantic

matching based-ontology to tackle the main problems

in tourism services, such as the variety in the e-Gov

tourism items, the multi-interest of users, and data

saving and retrieving of e-Gov tourism data. This

methodology consists of the following four steps:

Step 1. Develop a probability-based technique to

ﬁnd implicitly the rating of items.

A probability-based technique will be developed

to calculate the rating of items for each user based on

users’ proﬁles. This technique consists of the follow-

ing two stages:

• Cluster users into a number of groups, called com-

munities. Each community includes a group of

users who share common interests. User commu-

nities will help IRE to solve the new user problem.

The system will recommend proper tourism items

to a new user based on his or her user community

proﬁle.

• Apply a probability technique to compute implic-

itly the rating values of items for the users in each

community. The rating of a speciﬁc item will

be generated based on the following three crite-

ria: (1) previous user behaviour, (2) the prefer-

ences of user’s neighbours to a speciﬁc item, and

(3) the semantic relevance among concepts at dif-

ferent level of hierarchy in the domain ontology.

The third criterion is useful for inferring whether

a given item is related to the concepts that are pre-

ferred by a speciﬁc user.

A FRAMEWORK FOR DELIVERING PERSONALIZED E-GOVERNMENT TOURISM SERVICES

267

In tourism domain, it is difﬁcult to attain explicit rat-

ing for all tourism items, such as destinations, attrac-

tion and activities. This technique can ﬁnd implicit

ratings for unrated items for a particular user. Hence,

the sparsity problem associated with the item-based

CF can be alleviated.

Step 2. Find similarity among e-Gov tourism items

using CF item-based similarity.

In this step, the obtained implicit ratings of items

from step 1 and the explicit ratings of items that rated

by users will be used to compute similarities among

all instances of tourism e-Gov ontology concepts.

One of the popular CF algorithms, the item-based CF

algorithm will be utilized for the similarity calcula-

tions (Adomavicius and Tuzhilin, 2005). Item-based

CF algorithm has shown better results than user-based

CF one in terms of enhancing the performance and the

quality of recommendation (Sarwar et al., 2001).

Step 3. Develop a semantic matching based-

ontology approach to compute similarity among the

available instances of concepts (tourism items) in the

tourism e-Gov ontology.

Ontology can support ﬁnd the matching (or simi-

larity) of objects that are conceptually close but not

identical. A few methods have been proposed to

assess similarities among instances within ontology.

Ehrig and his colleagues (2005) presented a compre-

hensive framework for measuring semantic similar-

ity within and between ontologies. The framework

identiﬁed three main levels on which the similarity

between two entities (concepts or instances) can be

measured: data layer, ontology layer, and context

layer, which deal with the data representation, ontol-

ogy meaning, and the usage of these entities, respec-

tively (Ehrig et al., 2005).

To ﬁnd similarities among instances of tourism

e-Gov ontology, semantic matching will take into

account the structural comparison between two in-

stances in terms of their classes, and the compari-

son between their attributes and their relations. These

comparisons can be inferred using semantic analysis

of the inference rules which interconnect instances in

the ontology. The involved attributes and relations in

comparison of different instances will be determined

based on the context of the tourism e-Gov ontology.

Step 4. Generate recommendations by integrating

the similarities.

The obtained similarities from steps 2 and 3 will

be combined together for each tourism item and the

combined similarity will be stored in KBR compo-

nent as a matrix. This combined similarity will be

used to build a discovery model. This model can

be used to predict the most relevant item among the

available tourism items that a particular user might

be interested in. Weighted sum model and regression

model are examples used in recommendation systems

for prediction (Sarwar et al., 2001).

Combination of the ontology based semantic

matching and the similarity from the item-based CF

would improve the quality of recommendations and

the prediction of new tourism items from different

categories that could be preferred by a speciﬁc user.

The main reason for this combination is that the gen-

erated recommendations do not merely depend on the

opinions of users to the preferred items, but also on

the semantic matching of these items based on their

underneath hierarchal structure and their attributes

and relationships that could affect the application con-

text. Additionally, considering semantic matching in

this integration could eliminate the sparsity problem

if the item-based CF similarity algorithm.

4 WORKFLOW OF Pe-Gov

TOURISM SERVICE SYSTEM

Under the proposed framework, the e-Gov tourism

service systems allow users to get speciﬁc informa-

tion about ”where to go” or ”what to do/see” at a spe-

ciﬁc destination. Figure 3 illustrates a workﬂow ex-

ample of the system.

Example of a scenario: Jack is a local user in Aus-

tralia; he wants to visit a Museum (as an attraction)

at a speciﬁc destination (e.g. Sydney). How the pro-

posed framework can help?

Jack is a registered member in the system, so he

needs to login to the system using his own user name

and password, as depicted in Figure 3. After login to

the system, he can see his own page. A catalogue of

all available tourism items (destinations and its attrac-

tions and associated activities) presented in a style of

directory tree as categories. He can browse the cata-

logue tree staring from select ”Sydney” under the des-

tination category. Then, browse tourism attractions

related to Sydney until he ﬁnds the Museum attrac-

tion node, and select this node. This selection will

automatically trigger the IRE component to generate

the most interested museums attraction items to Jack.

The discovery model in IRE component is re-

sponsible of generating recommendations. To accom-

plish that, both of Jack’s proﬁle (including his rat-

ing record) and the combined similarity matrix (that

obtained from the item-based CF similarity and the

semantic-based similarity, as described in section 3.4,

particularly, the one related to the museum category)

WEBIST 2010 - 6th International Conference on Web Information Systems and Technologies

268

Figure 3: A Description Scenario of the workﬂow of Pe-Gov Tourism Service System.

will be uploaded to the memory. Then the discovery

model will use, for example, a weighted sum model

to predict the most interested museum items in Syd-

ney. To perform the prediction, the weighted sum

model works as follow: Firstly, it ﬁnds for each un-

rated museum item by Jack the most similar muse-

ums using the combined similarity matrix; Secondly,

it computes for each given unrated item the intersec-

tion set between the most similar museums items to

the given one and the already rated museums items

by Jack; Finally, the gained intersection set will be

used to compute the prediction value for this unrated

item. The prediction value for each unrated museum

item of Jack can be computed as a ratio of summation

of the ratings given by Jack on the museums items

similar to the given unrated one. Each rating will be

weighted by the corresponding similarity between the

given unrated item and those ones in the intersection

set. Based on the computed prediction values for un-

rated museums items of Jack, the system will gener-

ate the top-N most interested items (higher predicted

items) as an ordered list of recommendation in the

GUI component.

After that Jack can browse the retrieved museum

items list until he chooses a speciﬁc museum attrac-

tion item. The system, subsequently, will present for

each selected museum item the most similar muse-

ums. That can be carried out according to the com-

bined similarity matrix that explained in step 4, in

section 3.4. Beside that, the system also will ask him

whether he is planning to visit this selected Museum.

If the answer is positive, the system will ask Jack

to add this attraction to his own ”MyInterestAlbum”.

Generating a few most relevant museums items (top-

N) for Jack will save his signiﬁcant amount of time

and effort, and will bring his pleasant searching expe-

rience.

The recommendation that generated upon Jack’s

request is called on-demand recommendation. The

on-demand recommendation could be homogenous or

heterogeneous recommendation. Homogenous rec-

ommendation includes recommendations to tourism

items that belong to the same category of the re-

quested item, as in the above case, where an ordered

list of Museums is generated upon Jack’s request.

While heterogeneous recommendation involves rec-

ommendations to tourism items from different cate-

gories that are similar to the requested item. Hetero-

geneous recommendation can be generated based on

the given user’s similar neighbours (with similar com-

munity or behaviour) and semantic similarity among

tourism items. The recommendation will be presented

to users in UI component as an ordered list of recom-

mendation by category. Figure 3 depicts that Jack can

receive also recommendations from other categories

that are similar to the Museum, e.g., items from Gal-

leries Studio category.

Furthermore, the intelligent Pe-Gov tourism ser-

vice system can present new recommendations to

users each time they login to the system. For instance,

the system will recognize Jack each time he visits

the system and regularly will update the content of

his page automatically, a new suggestions for tourism

items can be presented. These auto recommendations

can be generated dynamically ofﬂine to users depend-

ing on users proﬁle, their community and the changes

that might be occurred to any tourism items. For

example, new events of a speciﬁc attraction can be

A FRAMEWORK FOR DELIVERING PERSONALIZED E-GOVERNMENT TOURISM SERVICES

269

recommended for users who are interested in or vis-

ited that attraction. Interestingly, the system would

be more effective for regular users who visit the sys-

tem frequently. Irregular users would also be offered

recommendations but limited to their available infor-

mation.

5 CONCLUSIONS AND FUTURE

WORK

Intelligent personalization has become a clear direc-

tion in the development of delivering e-Gov services

by different government agencies. This paper pro-

poses a new conceptual framework for delivering Pe-

Gov tourism services using RS techniques and se-

mantic ontology. The proposed framework can help

users ﬁnd, efﬁciently and friendly, the most interest-

ing tourism attractions with the most appropriate ac-

tivities/events according to their interests, needs and

the behaviour/experience of other similar users. The

main components of this framework were discussed.

The potential of the proposed framework of offer-

ing better tourism services to users has been illus-

trated by a scenario example. The future direction,

of our research, would be to develop a working sys-

tem/prototype to deliver Pe-Gov tourism services to

users.

REFERENCES

Accenture (2004). egovernment leadership: high perfor-

mance, maximum value. ﬁfth annual accenture egov-

ernment study.

Adomavicius, G. and Tuzhilin, A. (2005). Toward the next

generation of recommender systems: A survey of the

state-of-the-art and possible extensions. IEEE Tran.

on Know. and Data Eng., 17(6):734–749.

Al-Hassan, M., Lu, H., and Lu, J. (2009). A framework for

delivering personalized e-government services from a

citizen-centric approach. In Proceed. of the 11th Inter.

Conf. on Inf. Integration and Web-based Applications

and Services.

Berka, T. and Plnig, M. (2004). Designing recommender

systems for tourism. In Proceedings of ENTER.

Ehrig, M., Haase, P., Stojanovic, N., and Hefke, M. (2005).

Similarity for ontologies-a comprehensive framework.

In 13th European Conf. on Inf. Systems.

Eirinaki, M. and Vazirgiannis, M. (2003). Web mining for

web personalization. ACM Transactions on Internet

Technology (TOIT), 3(1):1–27.

Fesenmaier, D. (2006). Introduction: Recommendation sys-

tems in tourism. In Fesenmaier, D., Wober, K., and

Werthner, H., editors, Destination Recommendation

Systems: Behavioral Foundations and Applications.

Huang, Y. and Bian, L. (2009). A bayesian network and

analytic hierarchy process based personalized recom-

mendations for tourist attractions over the internet.

Expert Systems with Applications, 36(1):933–943.

Markellou, P., Mousourouli, I., Sirmakessis, S., and Tsaka-

lidis, A. (2005). Personalized e-commerce recom-

mendations. In Proceedings of the 2005 IEEE In-

ter. Conf. on e-Business Engineering, pages 245–252.

IEEE Computer Society.

Millard, J., Havliek, J., Ticha, I., and Hron, J. (2004).

Strategies for the future egovernment. In The Inter.

Conf. Agrarian Perspectives XII. CAB Abstracts.

Noy, N. and McGuinness, D. (2001). Ontology devel-

opment 101: A guide to creating your ﬁrst ontol-

ogy. Technical report, Stanford Knowledge Systems

Laboratory Technical Report KSL-01-05 and Stanford

Medical Informatics Technical Report.

Pieterson, W., Ebbers, W., and Dijk, J. (2007). Personal-

ization in the public sector: An inventory of organi-

zational and user obstacles towards personalization of

e-services in the public sector. Government Informa-

tion Quarterly, 24(1):148–164.

Rabanser, U. and Ricci, F. (2005). Recommender systems:

Do they have a viable business model in e-tourism?

In Frew, A., editor, Information and Communication

Technologies in Tourism, pages 160–171. Springer.

Ricci, F., Fesenmaier, D., Mirzadeh, N., Rumetshofer, H.,

Schaumlechner, E., Venturini, A., Wober, K., and

Zins, A. (2006). Dietorecs: A case-based travel ad-

visory system. In Fesenmaier, D., Wober, K., and

Werthner, H., editors, Destination Recommendation

Systems: Behavioral Foundations and Applications.

Sarwar, B., Karypis, G., Konstan, J., and Reidl, J. (2001).

Item-based collaborative ﬁltering recommendation al-

gorithms. In Proceed. of the 10th Inter. Conf. on

WWW. ACM.

Schafer, J. B., Frankowski, D., Herlocker, J., and Sen, S.

(2007). Collaborative ﬁltering recommender systems.

In Brusilovsky, P., Kobsa, A., and Nejdl, W., editors,

The Adaptive Web, pages 291–324. Springer.

Schiafﬁno, S. and Amandi, A. (2009). Building an expert

travel agent as a software agent. Expert Systems with

Applications, 36(2):1291–1299.

Staab, S., Werthner, H., Ricci, F., Zipf, A., Gretzel, U., Fes-

enmaier, D., Paris, C., and Knoblock, C. (2002). Intel-

ligent systems for tourism. IEEE Intelligent Systems,

17(6):53–64.

Undheim, T. A. and Blakemore, M. (2007). A handbook for

citizen-centric egovernment.

Vassilakis, C. and Lepouras, G. (2006). An ontology for

e-government public services. Encyclopedia of E-

Commerce, E-Government and Mobile Commerce.

Wauters, P., Nijskens, M., and Tiebout, J. (2007). The user

challenge, benchmarking the supply of online public

services.

WEBIST 2010 - 6th International Conference on Web Information Systems and Technologies

270