An Adaptive e-Advertising Delivery Model: The AEADS Approach

Alaa A. Qaffas and Alexandra I. Cristea

Department of Computer Science, The University of Warwick, Coventry, CV4 7AL, U.K.

Keywords: e-Advertising, e-Commerce, Personalisation, Adaptive Advertising, Delivery Model, Delivery Engines.

Abstract: e-Advertising adaptation plays a main role in delivering personalised advertisements to internet users. In this

time of the Internet revolution, many websites need to use the adaptation process to adapt their advertisements.

This paper focuses on a lightweight delivery model, easy to integrate into wide range of existing websites.

This model includes three engines, in order to deliver personlised advertisements to Internet users easily. It

also presents a study that assesses the effectiveness of a tool based on this model, called AEADS, via a trial

run of a model prototype with users.

1 INTRODUCTION

Providing suitable content and products for different

users meets the needs of both businesses and

customers. It increases business profit and allows

greater customer satisfaction. Recently, there has

been a rapid growth of e-commerce and web

applications (Abu-Taieh, 2009; Al Qudah et al., 2014;

Kazienko and Adamski, 2007), and thus the

improvement of the delivery systems is important, to

match such growth. E-commerce has given customers

the power to choose from a variety of options offered

by different companies, and thus competition has

emerged (Puntambekar, 2008). Adaptation attempts

to match content and products to profiles of targeted

customers. Delivering adaptive advertising will

support this process, by both maximising the profits

of businesses and increasing customer satisfaction.

Still, adaptation has not been applied consistently and

effectively in e-advertising. Moreover, whilst

businesses large and small may wish to add adaptive

advertisement to their sites, the process currently is

too cumbersome for an easy transition. Currently,

there is no solution that can be added in a lightweight

fashion to existing websites of businesses. Thus, our

research targets the following main research question:

How can we create a model for lightweight

adaptive advertising and design the

corresponding system that can be integrated with

most websites?

To answer this question, we recommend a collection

of tools, Adaptive e-Advertising Delivery System

(AEADS), which facilitate the creation of adaptive e-

advertising. This paper focuses on a vital component

in adaptive delivery systems, the Delivery Model

(DM). Here we propose a lightweight DM, with a set

of features that we consider essential to adaptive

advertising, and which can be easily added to wide

range of static commercial websites. This model is

implemented and evaluated with real Internet users

and customers.

2 RELATED RESEARCH

Many methods of modelling delivery specification

have been proposed. The following were selected

based on their similarity to AEADS.

ADE (Scotton et al., 2011) is written in Java,

using Servlets and JSP technology, and can be run on

a standard Tomcat server, to display any content

which can be described using standard web mark-up

languages. The delivery processes in ADE are located

in the adaptation and presentation layers. Based on

user model, domain model, and adaptation strategies,

ADE delivers the appropriate course contents for

users. ADE is able to adapt to the type of device being

used. In addition, ADE uses AJAX, to track the

network status and update the bandwidth variable in

the user profile, to tailor adaptation.

AdROSA (Kazienko and Adamski, 2007) extracts

knowledge stored in the web content page and the

historical user sessions, and recent behaviour of

online users, via data-mining techniques. Banners

visited by users are stored in the form of vectors of

124

Qaffas, A. and Cristea, A.

An Adaptive e-Advertising Delivery Model: The AEADS Approach.

DOI: 10.5220/0005996301240131

In Proceedings of the 13th International Joint Conference on e-Business and Telecommunications (ICETE 2016) - Volume 2: ICE-B, pages 124-131

ISBN: 978-989-758-196-0

user behaviour. The delivery part of AdRosa applies

advertising policy and priority features to

advertisements alongside user behaviour, to display

the most appropriate advertisements for each user.

Based on the LAOS framework (Cristea and de

Mooij, 2003), the MyAds system (Al Qudah et al.,

2014) encapsulates the delivery part in the adaptation

model - where the connection between the user model

and the appropriate advertisement is established - and

the presentation model - where the personalised

advertisement is displayed to the users. The

Personalisation and Decision Making Engine delivers

adaptive advertisements, matching the UM with the

appropriate product, to show adaptive advertisements

to each user.

Although ADE delivers adaptation content

efficiently, it mainly targets course adaptation,

meaning that there are certain limitations for its use

in the delivery of adaptive advertisements. The

parameters applied to introduce adaptive

advertisements and adaptive courses are different,

since, for example, the course adaptation depends

mainly on experience, as well as has a more narrative

structure. Moreover, it is a standalone system and

cannot be easily incorporated into existing websites.

The AdROSA and MyAds systems are designed to be

used in the portal model of advertising, since they

match the publishers' interests and many advertisers'

interests. The delivery tool in AEADS controls and

adapt advertisements located and owned by

businesses. Finally, the delivery engine in AEADS is

superior to AdROSA and MyAds, as it allows

businesses to control the number and location of

advertisements on each webpage automatically.

Additionally, it can be integrated easily into a wide

range of websites.

3 AUTHORING ADAPTIVE

E-ADVERTISING

The overall Authoring model of Adaptive e-

Advertising, informed by prior research, includes:

1. The Domain Model - used by businesses to

organise, label and categorise advertisements

(Qaffas and Cristea, 2014b).

2. The Adaptation Model (Qaffas and Cristea,

2014a) - enabling businesses to adapt the

advertisements they have organised, using the

domain model tool for their customers’ needs.

3. The User Model - representing the personal data

of an individual user, to base adaptive changes on

system behaviour (Qaffas and Cristea, 2015).

These tools are used to author personalised

advertisements via XML files, used by the delivery

model to deliver personalised advertisements.

4 DELIVERING ADAPTIVE

E-ADVERTISING

The delivery model (DM) (Figure 1) is resident on the

same website server, in order to deliver

advertisements to Internet users. This part parses the

contents of the XML files and uses adaptation

strategies to send appropriate advertisements to the

respective users, based on a user model. It consists of

three engines: inference, decision and modifier.

Figure 1: Delivery Engines of the AEADS System.

4.1 The Inference Engine

The inference engine gathers data from the domain

model, the adaptation model and the user model, to

infer multiple sequences of advertisements, to be sent

to the decision engine. First, it checks whether or not

the current user is logged in. If not, the inference

engine only applies the plan recognition process. This

will depend on the plan libraries, which the

businesses create in the authoring part. The inference

engine checks the clicked items and the plan libraries,

to acquire a sequence of advertisements to send to the

decision engine (Figure 2). An XML file contains the

library of plans. Using XML files should enhance the

portability, easy processing and generalisation of the

system, as discussed. Each node represents an

advertisement, and inside this node, an edge will be

inserted with the advertisement ID referring to the

linked advertisement. The simple structure of the

XML file allows authors to easily add plans.

An Adaptive e-Advertising Delivery Model: The AEADS Approach

125

Figure 2: Plan Recognition in the Inference Engine.

f the current user is logged in, then general rules

will be applied by the inference engine, to assign a

group of advertisements to the current user, according

to features, e.g., gender and age - based on stereotypes

created. This data is sent to the modifier engine, to

update the user model. Next, behaviour rules,

representing adaptation strategies, are next applied. A

sequence of advertisements is also retrieved and

passed to the decision engine, based on these rules.

The inference engine also applies the plan recognition

process and passes it to the decision engine. Finally,

all of these advertisements must apply the general

rules from the first step (Figure 3).

Figure 3: Inference Engine Process (User Logged In).

4.2 The Decision Engine

The decision engine is responsible for displaying

advertisements to the current user. Firstly, a flexible

method that allows businesses to put any number of

advertisements anywhere they want, is used by the

decision engine. The businesses are only assigned the

ID of the html element that contains the

advertisement image with a fixed name

"Image_Universal_AdLocation". As shown in Figure

4, the ID of the link that represents this advertisement

will be assigned the name

"A_Universal_AdLocation", and this code is to be

repeated on all webpages. This allows businesses to

add any number of advertisements in any location on

the webpage (Figure 5). Furthermore, the number and

location of advertisements can vary from page to

page, based on businesses views.

Figure 4: Advertisements Location Determination Code.

When a user loads a webpage, the decision engine

searches for the IDs, which represent the

advertisements, and changes their names, by giving

them a number in increasing order. The decision

engine then determines the number of advertisements,

which will appear on the current webpage. This

process is aimed at allowing the system flexibility and

usability for businesses to insert advertisements, since

the business owners have the ability to control the

number of advertisements and the location of each

advertisement on the webpage (Figure 5).

Figure 5: Advertisements on the Webpage.

If the current user is not logged in (Figure 6), then

higher priority advertisements will be displayed first.

The decision engine arranges the available

advertisements, as per following algorithm.

1. Display the advertisements from the plan

recognition, firstly.

2. Randomly display advertisements from the entire

domain, if the plan recognition advertisements is

finished.

On the other hand, if the current user is logged in,

then a sequence of advertisements from the inference

engine, which meet the behaviour rules, will be

retrieved and sent to the decision engine. A sequence

of advertisements based on plan recognition from the

inference engine will be given to the decision engine

ICE-B 2016 - International Conference on e-Business

126

as in the following algorithm:

1. The fourth behaviour rule, “show after” explained

in (Qaffas and Cristea, 2014a), has first priority, if

it exists.

2. If there are advertisements from the plan

recognition, display them. If they are exhausted,

display advertisements, which meet the other

behavioural rules.

Figure 6: Decision Engine Process (User not Logged In).

4.3 The Modifier Engine

The modifier engine acquires information from the

inference and decision engines, to update the user

model. The user model is updated based on certain

events; for example, during the user’s login, the

modifier engine detects whether or not the device

type and bandwidth have changed and it updates the

user model. When the decision engine delivers

advertisements to be shown, the modifier engine also

updates the user model.

5 CASE STUDY

To test the AEADS system and obtain feedback with

regards to its effectiveness (usefulness) and efficiency

(ease of use), the AEADS system was integrated with

an online bookstore. To evaluate the AEADS system,

samples of Internet users were asked to use the

system. The user modelling profile attributes of the

AEADS system was integrated into the online

bookstores user profiles (Figure 7). In the figure, the

‘name’, ‘user name’, ‘password’ and ‘email’

attributes form the online bookstores user profile

attributes, while the attributes ‘age’, ‘gender’,

‘bandwidth’, ‘education level’, ‘education type’ and

‘hobbies’ are the AEADS user modelling profile

attributes. The user modelling tool in AEADS has

been designed to be simple—i.e., to possess only a

few user model features and have an XML data

structure — the latter so that it is lightweight and can

be integrated with any potential website user model

(Qaffas and Cristea, 2015). AEADS includes two

methods of login: registering (explicit data retrieval)

and Facebook login (implicit data retrieval), as

discussed in (Qaffas and Cristea, 2015).

Figure 7: Book Store Registration.

The main aim of this survey was to determine

whether Internet users responded favourably to the

new lightweight advertising delivery system.

Currently, there are around three billion worldwide

Internet users (InternetWorldStats, 2012). Thus, a

suitable sample group requires 267 participants at a

confidence level of 90; alternatively, we can use a

sample group of 377 participants, at a confidence

level of 95 (raosoft). Aiming at international

applicability (confidence level 90-95), 450 different

Internet users were sent the user questionnaire.

5.1 Hypotheses

Hypotheses have been defined to evaluate the

AEADS system, from Internet users’ perspective:

H0a: The AEADS system and its functions is useful

for adaptive advertising.

H0b: The AEADS system and its functions is easy to

use for adaptive advertising.

H0x are the basic hypotheses. Specific hypotheses

were also tested via the questionnaire method:

H1: The various functions in the AEADS system are

well integrated.

H2: AEADS has a shallow learning curve.

H3: AEADS overcomes the privacy concerns.

H4: Users prefer to login via Facebook account

rather than register.

H5: The collected data is enough and acceptable for

users.

H6: The AEADS system interface is user-friendly.

H7: The AEADS system performance is adequate.

An Adaptive e-Advertising Delivery Model: The AEADS Approach

127

H8: The AEADS system reliability is achieved.

H9: The AEADS system increases the clicking

behaviour on advertisements.

5.2 Case Study Setup

The AEADS system was tested by a number of

students who were studying a variety of disciplines at

the King Abdul-Aziz University in Saudi Arabia.

They were studying Principles of Marketing,

Introduction to Business, Management Information

Systems and e-Marketing. These students were

chosen as suitable system testers, as they were

frequent Internet users and often made online

purchases. In effect, these students were familiar with

online platforms and had first-hand knowledge of

existing online providers. Additionally, students were

representing a diverse range of subjects, to obtain

generalisable results and avoid focusing only on

computer science students. However, evaluating with

students presents drawbacks, as, whilst they represent

the young population knowledgeable of the Internet

and its tools, especially e-business tools, they do not

represent the population at large.

All users were required to use, assess and evaluate

AEADS. This process involved a number of different

stages, as outlined below.

Table 1: AEADS System Features.

1 Registration process

2 Logging in using Facebook account

3 Managing the user profile

Automatic extraction of device information (location,

device type, device software, bandwidth)

5 The advertisements that are appropriate for users

6 The personalised advertisements is acceptable for users

7 I notice that the advertisements were personalised

8 The system collects enough information from you

Your behaviour on the website is tracked to give you

suitable advertisements

The participants were first given a general

overview of the AEADS system and the concept of

adaptive advertising. They were then asked to use the

system and evaluate its functionality. At this stage, a

five-part survey was distributed, to facilitate the

assessment process. The opening section of the

questionnaire asked participants to provide personal

demographic details, e.g., age, gender, level of

education, etc. The following section asked

participants to answer system usability scale (SUS)

questions. The next step required users to general

questions, while the fifth section required them to

offer more in-depth responses. This section utilised a

Likert scale for responses, as participants were

required to analyse and evaluate the effectiveness and

applicability of the system. Numerical data was used

to represent feelings or opinions: for instance, 1 = ‘not

at all useful’ / ‘very difficult’; whereas 5 = ‘very

useful’ / ‘very easy to use’. The last section then

asked a number of qualitative questions.

5.3 Results

A total of 381 questionnaires were completed

accurately and returned to the researcher. The number

of completed surveys is impressive, considering that

students were assured that participation was

completely voluntary. From respondents, almost two

thirds were aged between 18 and 24 while a further

22.8% were aged between 25 and 34. In terms of

gender, over two thirds of those who took part in the

survey were male, while only 27% were female.

Finally, in terms of education level, the majority of

participants held a Bachelor’s degree, while only

14.2% were pursuing a post-graduate qualification.

This indicates that the data may be skewed towards

younger, well-educated males. Nonetheless, this

demographic is crucial for web providers, as they are

the most prolific Internet users, likely to maintain a

high rate of Internet usage in the future. It is therefore

imperative that web providers meet the needs of this

niche social group.

For SUS results, the majority agreed that the

system is simple to understand and use by Internet

users, without specialised training or advanced

knowledge, which support hypothesis H2. They also

considered the system well integrated, and stated that

they would like to use the system on a frequent basis,

which support hypothesis H1. They strongly agreed

that AEADS is easy to use, with 96.9%, and 95.6%

stating that they felt very confident using it. Most

users understood how to use the system from the

presentation given at the beginning of the evaluation.

They were confident when they used the system.

Additionally, they further backed up these statements

via open-ended question (section 5.4). Furthermore,

the overall SUS score for AEADS is 87.70 out of 100.

Cronbach’s Alpha for SUS scores is 0.93  [≥ 0.9],

meaning the results of the SUS questionnaire were at

an ‘excellent’ level of reliability. These findings

support hypothesis H0b, which posits that AEADS is

easy to use.

The second section of the survey posed a series of

general questions about the functionality of AEADS

and its overall effectiveness. This section focused

primarily on the influence of AEADS in encouraging

users to click on sponsored links or make purchases

on the basis of personalised advertisements. It also

focused on the degree to which participants were

ICE-B 2016 - International Conference on e-Business

128

concerned about their online security and the safety

of their personal information. Of those questioned,

93.5% stated that the system would encourage them

to click on more links and make more purchases,

while 90.9% claimed that they were largely

unconcerned about their privacy and online security.

This supports hypothesis H9, which posits that the

AEADS system increases the clicking behaviour on

advertisements. Furthermore, these findings also

substantiate hypothesis H5, as 90.2% of participants

felt that the system was justified in collecting private

information and were willing to offer such data in

exchange for a more effective adaptive advertising

mechanism, as the AEADS system collects only the

data that is needed to personalise the advertisement.

In addition, 85.7% of the participants stated that they

would login via Facebook, if they were to use this

system regularly, which substantiates hypothesis H4,

on preferring to login into the system using their

Facebook account.

A significantly large proportion of participants

(95.9%) strongly agreed that the information

requested by the system overcomes privacy concerns,

by collecting only data needed for personalisation of

advertisements, which left users feeling confident

with the AEADS system. These findings support

hypothesis H3. Generally, the majority of users were

extremely satisfied with the effectiveness of the

system and believed that it performs exceptionally

well. In addition, the majority of those questioned had

faith in the reliability of the system. These findings

support hypothesis H8. The Cronbach’s Alpha score

was 0.96 [≥ 0.9], meaning that the reliability of the

psychometric test is excellent.

A comparatively low score was obtained in

relation to the user interface of the system, as only

79.5% of those questioned considered the system

interface to be user-friendly. This relatively low level

of satisfaction could be attributable to the interface of

the website on which the assessment was performed.

Though the design of the website was beyond our

control, the system nonetheless scored highly in terms

of usability and ease of use, supporting hypothesis

H6, positing that the user interface of the AEADS

system is user-friendly.

Participants were next asked to evaluate the

various features and functions of the AEADS system

on a Likert scale. The main functions of the system

were generally well-received by users, with more

than 84.8% of participants stating that they found the

various features extremely useful. The standard

deviation values in this instance were between .46-.54

and a mean value of 4.24-4.69. Thus, the system can

be considered ‘useful’. The Cronbach’s Alpha score

is 0.90  [≥ 0.9], meaning that the reliability of the

psychometric test is excellent.

In terms of which features proved the most

popular, the majority of those questioned agreed that

the advertisements shown were suitable, given their

interests and preferences. In addition, the majority

found the advertisements shown to be acceptable, and

were satisfied that their behaviour on the website was

monitored, in order to generate the most relevant

advertisements. The participants clearly enjoyed the

advertisements they were shown during the

evaluation processes because the advertisements had

been personally adapted, through methods based on

the personal data found within the user profiles, along

with the participants’ behaviour, which had been

monitored by the system. These findings substantiate

hypothesis H0a, as the AEADS system and its

functions is useful for adaptive advertising.

The least-liked features included ‘automatic

extraction of device information (location, device

type, device software, bandwidth) is useful’ and

‘logging in using a Facebook account is useful’.

Nonetheless, as these features still scored above 4,

they cannot be considered as disliked features. In fact,

the lower score obtained by these features could be

attributable to the user’s lack of understanding of the

purpose of each feature. Another interpretation is that

they might have been worried about the system

extracting information without their knowledge (as in

the extraction of the device information).

Additionally, they might have been worried about the

information that the system would have access to, if

they were to login via their Facebook accounts. In the

open-ended question section, one user questioned

whether the system would continue to track their

online activities once they had closed the webpage, as

is further discussed in section 5.4. Nevertheless, as

both rules achieved a minimum rate of 4, they can still

be deemed useful (Figure 8).

Figure 8: Usefulness (Ox axis detailed in Table 1).

These findings substantiate hypothesis H0a,

which posits that the AEADS system and its functions

is useful for adaptive advertising.

An Adaptive e-Advertising Delivery Model: The AEADS Approach

129

The usability of the distinct features was

separately evaluated through questionnaire questions

(1-9, defined in Table 1). In terms of usability and

ease of use, the mean values fell between 4.17-4.74.

In addition, the standard deviation values for usability

fell between .45-.51. These results indicate that

AEADS can be considered usable, as it can be easily

operated by any user, without the requirement for

formal training, or an existing knowledge of online

platforms. In addition, the Cronbach’s Alpha score is

0.91 [≥ 0.9], meaning that the reliability of the

psychometric test is excellent. These findings were

then subject to analysis and it was discovered that the

most popular elements in terms of usability were

‘Your behaviour on the website is tracked to give you

suitable advertisements’ and ‘login via Facebook is

easy to use’. Obviously, users preferred to receive

personalised advertisements based on their

characteristic and preferences, as the personalised

advertisements were presented to them during the

evaluation processes based on their data contained

within the user profiles, along with their behaviour,

which was monitored by the system. Moreover, in

2005, 80% of Internet users were interested in

receiving personalised content on sites that they

visited (ChoiceStream, 2005) and the percentage has

only increased since then.

Conversely, the least popular features were

‘Registration is easy process’ and ‘I can manage my

profile easily’. However, although these features

received the lowest scores, they still obtained a

minimum rate of 4, which means that they can still be

considered usable; however, they simply may not be

as easy to use in comparison to the other more highly-

rated features. Broadly speaking, these findings imply

that the system as a whole is easy to use. Obviously,

the participants preferred to login into the system

using their Facebook account. These findings also

substantiate hypothesis H0b, which posits that the

AEADS system and its functions is easy to use for

adaptive advertising.

5.4 Qualitative Answers and Discussion

One user made the commented that it was clear how

each of the displayed advertisements were linked. In

other words, they understood how each advertisement

related to one another as well as related to the interests

or preferences of the users. Basically, the users

acknowledged the effectiveness of the system in

customising the selection of advertisements based on

the unique details of each user. Another user also

highlighted how the advertisements that were

displayed reflected aspects of the user’s profile,

which again indicates that the system worked

effectively for the majority of participants. In fact,

many of those questioned expressed their

appreciation of personalised advertisements and were

impressed with how the system tailored the

advertisements displayed, based on their profile, user

preferences and online behaviour. The system also

allows the user to accept or reject the use of cookies,

which was highlighted by one user as a useful feature.

However, another user stated that the system did not

include their personal hobbies in their list of common

interests. This fell in line with the quantitative data,

as they considered the registration and managing of

their profiles as their least popular features. It should

be noted that the attributes are a changeable list that

can be modified, based on the business owner's view.

More details about attributes are discussed in (Qaffas

and Cristea, 2015).

Another issue highlighted by the users within the

qualitative section of the questionnaire concerns the

security of private data and the system’s monitoring

of online activity. For instance, one user wondered

whether the system would continue to track their

online activities once they had closed the webpage.

This implies that some users might be concerned

about the possibility of the system monitoring all of

their online behaviour. Thus, measures should be

taken to ensure that the system’s users are fully aware

of how the system operates and when the system is

tracking activity, in order to deliver the most relevant

and user-specific advertisements. Another user

commented that the user interface of the website

needs to be more attractive. Again, this relatively low

level of satisfaction could be attributed to the

interface inherited from the website, upon which the

assessment was performed. Though the original

design of the website was beyond our control and the

AEADS extensions were applied in a manner that was

true to the principles of our research, in a lightweight

manner, without changing the look&feel of the

original website, the system nonetheless scored

highly overall in terms of usability and efficiency.

In terms of usefulness and usability, one user

simply stated that they ‘liked the system’, which

indicates their full overall satisfaction with the

system’s features and functionality. Within the

analysis of the quantitative data process, users

revealed the belief that AEADS had aided them in

receiving personalised advertisements much more

than any normal e-business system would have. The

users stated that they had been confused by Google

advertisements when attempting to find certain

content and most especially when trying to download

specific software. One user also stated that they liked

ICE-B 2016 - International Conference on e-Business

130

the location of the advertisements, which indicates

that the AEADS system displays advertisements in an

eye-catching, yet unobtrusive manner. Another user

claimed that the frequent display of different

advertisements was both convenient and effective. In

addition, another user stated that the system pushed

them to think about developing their own online

business, as the features and functions of the system

facilitated the marketing and advertising required for

their company.

These insights into the system reflect the

effectiveness and functionality of the current system

from the perspective of Internet users, while

highlighting possible areas in which future versions

of the system could be modified.

6 CONCLUSIONS

The delivery model is introduced in this paper, its

design and internal processes are described in detail.

It consists of three engines: inference, decision, and

modifier engines. The system, its features and

usability have been evaluated by real users, and the

overall outcome has been positive. Based on this

outcome, it can be seen that the delivery model in

AEADS is necessary and introduces flexible

adaptation.

REFERENCES

Abu-Taieh, E. M. 2009. Utilizing Information Technology

Systems Across Disciplines: Advancements In The

Application of Computer Science: Advancements in the

Application of Computer Science, IGI Global.

Al Qudah, D., Cristea, A. I., Shi, L., Al-Sayyed, R. M. &

Obeidah, A. 2014. Myads: A Social Adaptive System

for Online Advertisement from Hypotheses to

Implementation.

Choicestream. 2005. Choicestream Personalization Survey:

Consumer Trends and Perceptions. Available:

Http://Www.Choicestream.Com/Pdf/Choicestream_Pe

rsonalizationsurveyresults2005.Pdf [Accessed 10

March 2014].

Cristea, A. I. & De Mooij, A. 2003. Laos: Layered Www

AHS Authoring Model and their Corresponding

Algebraic Operators. WWW03 (The Twelfth

International World Wide Web Conference), Alternate

Track on Education, Budapest, Hungary.

Internetworldstats. 2012. World Internet Usage and

Population Statistics. Available: Http://Www.

Internetworldstats.Com/Stats.Htm [Accessed 25

September 2013].

Kazienko, P. & Adamski, M. 2007. Adrosa—Adaptive

Personalization of Web Advertising. Information

Sciences, 177, 2269-2295.

Puntambekar, A. A. 2008. Data Structures and Files,

Technical Publications.

Qaffas, A. & Cristea, A. 2014a. How to Create An E-

Advertising Adaptation Strategy: The Aeads Approach.

In: Hepp, M. & Hoffner, Y. (Eds.) E-Commerce and

Web Technologies. Springer International Publishing.

Qaffas, A. & Cristea, A. 2014b. How to Create an E-

Advertising Domain Model: The Aeads Approach. The

2014 International Conference on e-Learning, E-

Business, Enterprise Information Systems, and E-

Government (EEE'14). Las Vegas, United State.

Qaffas, A. A. & Cristea, A. I. 2015. An Adaptive E-

Advertising User Model the Aeads Approach. The

International Conference on E-Business (Ice-B).

Colmar, Paris.

Raosoft. Sample Size Calculator [Online]. Available:

Http://Www.Raosoft.Com/Samplesize.Html [Accessed

25 September 2013].

Scotton, J., Stewart, C. & Cristea, A. I. 2011. Ade: The

Adaptive Display Environment for Adaptive

Hypermedia. Proceedings of the ACM Hypertext 2011

International Conference, Eindhoven, The

Netherlands.

An Adaptive e-Advertising Delivery Model: The AEADS Approach

131