An Exploratory Study to Design an Adaptive Hypermedia System

for Online-advertisement

Dana A. Al-Qudah, Alexandra I. Cristea and Shi Lei

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

Keywords: Adaptive Hypermedia, e-Advertising, e-Business, Social Networks, Requirements Gathering.

Abstract: The revolutionary world of the World Wide Web has created an open space for a multitude of fields to

develop and propagate. One of these major fields is advertisement. Online advertisement has become one of

the main activities conducted on the web, heavily supported by the industry. Importantly, it is one of the

main contributors to any businesses’ income. However, consumers usually ignore the great majority of

adverts online. This research paper studies the field of online advertisement, by conducting an exploratory

study to understand end users’ needs for targeted online advertisement using adaptive hypermedia

techniques. Additionally, we explore social networks, one of the booming phenomena of the web, to

enhance the appropriateness of the advertising to the users. The main current outcome of this research is that

end users are interested in personalised advertisement that tackles their needs and that they believe that the

use of social networks and social actions help in the contextualisation of advertisement.

1 INTRODUCTION

Adaptive hypermedia aims at tailoring the content

presented based on users’ knowledge, capabilities

and interests, amongst others (Brusilovsky, 2007).

Its main application area is e-learning. This paper

however focusses on another application for

adaptive hypermedia, which is much less explored:

that of online advertising, defined as the process of

delivering a marketing message using the World

Wide Web, to attract more consumers (Goy et al.,

2007). It is a well-known fact that consumers usually

ignore adverts (Nielsen, 2003). The overall research

aim is to find a way in which online advertising can

be provided, so that it is not intrusive to users and is

smoothly integrated into the general purpose of the

website visited, so that users should be drawn to the

advert. Thus the overarching research question is:

Can e-advertising be designed in such a way that

it is non-intrusive, smoothly integrated, aligned with

user expectations, and attractive to users?

Our main hypothesis is therefore:

H0: Personalisation, based on customisation and

adaptive hypermedia techniques, as well as social

networking data provide the means to create non-

intrusive, smoothly integrated, attractive adverts,

aligned with user expectations.

Here, we test the validity of this hypothesis by

dividing it into sub-hypotheses, evaluated with end-

users. Concretely, this paper is a result of an

exploratory design experiment (section 3) conducted

to extract the main user needs and properties

required of an adaptive hypermedia system for

online advertisement, based on social networks.

The experiment uses a user centred methodology

that guides the participants through the stages of

exploring and designing the new system (Adams,

2004). The additional advantage of this approach is

that it not only allows us to test the main hypothesis,

but it also provides a starting point for any system

designer who would wish to create viable

implementations for this relatively unexplored area

(see sections 2, related work, and 5 on discussion).

2 RELATED WORK

Online advertisement is one of the main incomes for

many online companies, as well as an innovative

marketing tool providing a sense of loyalty and trust

in customers. As a part of e-commerce, e-advertising

shares its problems, including the elusive ‘one

application fits all’ (Holden et al., 2009). Whilst it is

great that companies can address customers around

the globe, clients from different countries or

locations can have very different needs, or be subject

368

A. Al-Qudah D., I. Cristea A. and Lei S..

An Exploratory Study to Design an Adaptive Hypermedia System for Online-advertisement.

DOI: 10.5220/0004369903680374

In Proceedings of the 9th International Conference on Web Information Systems and Technologies (WEBIST-2013), pages 368-374

ISBN: 978-989-8565-54-9

 2013 SCITEPRESS (Science and Technology Publications, Lda.)

to completely different legal frameworks. A major

challenge of e-commerce is to produce a more

personalized commercial as well as advertising

experience (Goy et al., 2007). In 2005, statistics

showed that 80% of internet users were interested in

getting personalized content for the sites they visit

(ChoiceStream, 2005). Adaptive advertisement

looks at each user as a standalone case and provides

personalized content based on user-modelling

approaches (Kazienko and Adamski, 2007). User

modelling is one of the key aspects in user-adaptive

systems (Kobsa, 2007). Their foremost objective is

to collect data about the user to respond to the users’

needs. The correct definition and maintenance of

user models is predicted to be central to the

application of adaptive advertising as well, with one

of the main challenges the proper selection of the

user model variables, and their relationships.

Another important issue is the data collection

source. Part of the research is aimed at data

collection from social networks, to gather some (or

all of) the relevant user model information. These

platforms provide a rich source of user-information

for within-platform as well as outside applications.

Famous platforms like Facebook and Twitter have

large and growing numbers of users (more than

908,000,000 users on Facebook and 500,000,000

users on Twitter in 2004), and increasing wealth of

personal information about these users

(Socialbakers, 2012). Hence, social platforms have

become the main target of online advertisement,

with more than 20% of the ads already promoted via

these platforms (Dunay and Krueger, 2009).

What is clear is that, due to the huge availability

of information about products, and the loss of trust

in traditional advertisement, businesses need to

rethink their advertisement strategies (Qiao, 2008).

One strategy is to look at social network as a source

of user data, where personalisation can be provided

based on users’ profiling (Qiao, 2008). Our research

aims to explore this fast growing area and find a

balance between parameters to be modelled and user

response. Thus, the research described here starts

with the users from the very beginning, which can

improve the chances of success of a system (Preece

et al., 2002). The aim was to understand different

customers’ perceptions, which are crucial in

designing a system that fulfils their needs (Sanders,

2002). Thus, the methodology applied in this

experiment was a user-centred design process.

3 EXPERIMENT

In order to implement the user-centred experimental

design process (Vredenburg et al., 2002) and the

participatory design (Schuler and Namioka 1993),

we needed to enrol the help of real users.

Fortuitously, when it comes to online advertising,

any web user qualifies as online adverts user.

Certainly in the Western world, with a close second

in Eastern Europe, the great majority of the

population is a web user, with more than

2,405,520,175 users in the world and 518.6 million

users in Europe as per a recent survey conducted on

June 2012 (internetworldstats, 2012).

To perform a controlled experiment, it was

decided that the experiment was to be conducted

with the help of a class of 3

year students enrolled

in the Computer Science degree, Faculty of

Engineering Sciences in Foreign Languages, at the

University “Politehnica” of Bucharest, Romania,

studying a course entitled ‘Web Application and

Development’. Out of an overall student population

of 35, 12 volunteered to take part of the experiment.

The positive effect of this process was that these

students were actively engaged and determined to

help, instead of being coerced in any way. Also, the

relatively small sample size ensured that the whole

experiment was relatively easy to coordinate, that all

opinions could be properly listened to, discussed and

recorded, and that the overall atmosphere could be

kept quite informal, and thus conducive of honest

and straightforward discussions. The experiment

lasted slightly over two hours, based on the natural

flow of the interactions and (monitored) discussion.

In these two hours, the methodology of the user-

centred design process was applied, based on two

important thinking techniques: the brainstorming

technique and the six hats thinking technique.

The brainstorming technique, a very popular

supervised thinking approach (Osborn, 1963), is

used to collect as much as data as possible on the

problem, then classify it into main points for further

investigation, producing so called “spider diagrams”

(Howse et al., 2005). Due to its popularity, ease of

use, fast results, and its dealing well with ill-defined

search spaces, we have selected it for our

experiment. The six thinking hats technique (De

Bono, 1985) proposes that each person in the group

actively and purposefully thinks differently (thus

dons another hat), so a full analysis from all

perspectives can be covered. This technique is useful

with small number of participants, guaranteeing that

important aspects of a design process are not

omitted, and ensuring that users really consider all

AnExploratoryStudytoDesignanAdaptiveHypermediaSystemforOnline-advertisement

369

angles, after having discussed their preferred one

with the previous technique. Still, in some cases,

users may not be capable to completely explore

some aspects. For instance, a system administrator’s

perspective may not be as clear to the average user.

As in our study we used Computer Science students,

they were able to embrace relatively easily most

aspects of system design. The experiment was

conducted over three main phases, as follows.

In the first phase, a questionnaire was conducted

to examine the current knowledge on topics related

to e-commerce, e-advertisement and personalisation.

This phase lasted for 20 minutes.

The second phase was a short seminar, for

around half an hour, introducing participants to the

experiment and the framework

The third phase was the most labour-intensive

for the students, comprising the system requirements

gathering stage, as well as the participants’

involvement in the design process. This phase lasted

the longest, for over an hour, as participants were

encouraged to discuss their ideas in full.

These stages were conducted to follow the user

centred methodology. Thus, the participants

expressed what they thought by filling in the

questionnaire. Then, they moved to the doing stage,

where they used the existing tools. That was covered

in the short seminar, where participants could get

familiar with the platforms that use related principles

to the current system’s goals. The final stage is to let

the participants make what they want to have,

through getting them to actually design a

preliminary version of the proposed system.

3.1 The Questionnaire

The first step of the experiment was to help

participants express what they think, as per user-

centred methodology (Abras et al 2004). At the

beginning of the experiment, the participants were

not sure about what they were expected to deliver.

So the questionnaire was the tool to make them

express their ideas, by examining their current level

of their understanding in relation to the expected

outcome of the experiment. The questionnaire was

divided into three main sections. After collecting the

needed demographic data, their current level of e-

commerce knowledge was examined. This was

through a set of questions about popular e-commerce

websites (such as Amazon and eBay). It also

covered concepts related to e-advertisement, in

terms of concept definition, importance, reaction to

online ads and important related social networks and

websites that are famous for online ads. The final

section was about their future expectations from e-

commerce and e-advertisement websites,

considering that e-advertisement is a model of e-

commerce. In all the sections, questions related to

personalised advertisement were asked. Also, the

knowledge or opinions about the type of social

interaction related to these ads, the targeted websites

that the users may find most beneficial in terms of

online ads was elicited.

The overarching hypothesis introduced at the

beginning of the paper is further broken down into

sub-hypotheses, which were the underpinning to the

questionnaire (Table 1):

H1: Advertising is considered to be a major

activity, performed on the web by individuals – not

only by companies or businesses.

H2: Social networks are useful for personalised

e-commerce and e-advertisement applications.

H3: There are many factors affecting online

advertising and commerce, such as scalability,

reliability, privacy and security.

H4: Personalization of e-commerce and e-

advertisement tools will increase their usage.

The questions were aimed to be simple, direct

and to the point, to not confuse the students. The

whole process took place in English, as the degree

the students were following taught all classes in

English. Nevertheless, the questionnaire design took

into account that English was only the second

language for these students. Questions were

designed in such a way as to be neutral. For

instance, instead of asking a positively loaded

question, such as ‘Do you like the online

advertisement that you see?’ the question used was

phrased as: ‘What do you think of the advertisement

you see online?’ (see Q1 in Table 1 below, showing

questions and related hypotheses). Moreover,

participants were asked to comment, where

necessary, on their answers, to contextualize them.

Table 1: Questions, Answer range, and their related (sub-)

Hypotheses.

Questions and Answer range /

/ Related Hypothesis

Q1: What do you think of the advertisement you see

online? (Useless; Useful; It doesn’t make any

difference; Other; Please specify: )

Q2: When you come across an online advertisement

what do you do? (Ignore it; Look at it; It depends on

the advertisement; Please Explain: )

Q3: If you have used online advertising before, from

where did you click to reach the product website?

(Facebook; Twitter; Google; Google+; Linkedin;

Other - Please specify: )

WEBIST2013-9thInternationalConferenceonWebInformationSystemsandTechnologies

370

Table 1: Questions, Answer range, and their related (sub-)

Hypotheses(cont.).

Questions and Answer range /

/ Related Hypothesis

Q4: Do you prefer an e-advertisement that is based

on social interaction? (Yes/No)

If yes, choose the social properties you would like to

have (Chat; Comment; Rank; Rate;

Recommendations; Others – please specify: )

If you had the opportunity to design an e-advertising

tool, which social network would you use?

(Facebook; Twitter; Google+; Linkedin; Other -

Please specify: )

Q5: What factors affect you when using an e-

commerce website? (Popularity; Reliability; Privacy;

Security; Other - Please specify: )

Q6: If you had the opportunity to design an e-

advertising tool, would you consider using adaptive

techniques? (e.g,, showing specific content

customized to users, change content based on user

change of preference ) (Yes/No/Please explain: )

Q7:If you use an e-advertisement tool, what would

you prefer it to do? (Change content ‘by itself’ –

based on system metrics and parameters; Change

content based on parameters set by the user)

3.2 The Seminar

The questionnaire established the baseline, the

beliefs and needs of the participants based on their

own prior knowledge. Phase 2 was dedicated to

expanding this knowledge, via a seminar. This

corresponded to the second phase of the user-centred

methodology, where the participants needed to

become familiar with related systems, via a lecture-

like process, as well as via hands-on experiences

(Abras et al., 2004). During the session, participants

interacted via discussions, as they familiarized

themselves with the examples displayed, building on

their own experiences and knowledge. The seminar

discussed e-commerce platforms, their importance;

the models derived from these platforms - such as

online stores, online-auctioning and online

advertisements. It also exposed upon social networks

and popular examples, such as Facebook, Twitter

and Google+ (http://plus.google.com). A part of the

allocated time was spent on the topic of adaptive

hypermedia systems, their application and some case

studies of online advertisements in adaptive

hypermedia. The presentation also mentioned the

well-known commercial system AdSence, by

Google (google.com/ads, 2012) as well as a

research-based system called AdRosa, created by

Kazienko and Adamski (Kazienko and Adamski,

2007). The final part of the seminar introduced the

thinking techniques that were going to be used in the

next phase, since some of the students were not

familiar with brainstorming and the six-hat

technique.

3.3 The Design Phase

In this phase, the third step of the user-centred

methodology, participants (Abras et al., 2004) were

encouraged to design their own version of the

system, by setting up a list of requirements that they

wanted to see fulfilled. The twelve students worked

in two equal sized groups. They were allowed to

choose in which group to work, as some felt more

comfortable working with certain peers. The

participants were supervised by two facilitators: (1)

an expert on e-commerce systems, ensuring that

participants were deploying the appropriate

knowledge, without directly intervening with any

design ideas; (2) an expert on experiments

monitoring, providing feedback on the experiment

atmosphere and timeframe.

The participants started the process by firstly

using the brainstorming thinking technique. Here,

they created a spider diagram (Howse et al., 2005),

and they suggested functionality-oriented solutions.

Next, the participants used the six-thinking hats

technique, where they expressed stronger beliefs, by

giving voice to their emotions, and attempting to

think outside the box. As a result, in this phase, they

produced a list of usability-related problems, and

discussed possible solutions. By the end of this

session, students had a clearer understanding and a

set of expectations of an adaptive online system for

e-commerce. They presented their work to each

other and to the facilitators in a semi-formal

presentation, and received feedback from both the

other participants and the facilitators. They created,

beside the set of requirements based on their

expectations, also visual representations of the

design modules required in their ideal systems. The

feedback resulted in a set of recommendations

presented in the result section, and then discussed.

4 RESULTS

Our two sets of results are described below.

4.1 Questionnaire Results

As their starting level of understanding of e-

commerce and e-advertising, students showed that

they were mostly familiar with websites such as

AnExploratoryStudytoDesignanAdaptiveHypermediaSystemforOnline-advertisement

371

Amazon and eBay. H1 was confirmed to some

degree by the number of students agreeing that

online advertising is considered as a major activity

on the web. It should be noted that, due to the

limited number of participants in the experiment, the

results here cannot claim quantitative statistical

significance. Instead, they represent an exploratory,

idea-generating study, aiming at qualitative results.

The results from the questionnaire showed that

individuals – student in this case – use advertisement

as one of the major activities that they perform on

the web. The great majority (69%), when asked

about their behaviour when finding online adverts

(Q2, Figure 1), feels that ‘it depends on the

advertisement’. This shows that students don’t find

adverts a negative experience per se (only 16%

ignore them), nor positive (15% response). Thus,

contextualisation (and potentially adaptation,

customisation) of the advert may be crucial in

ensuring that the advert appears at the right time in

the right place. Similarly, H2 is supported, as

students considered that social networks have an

impact on the usage of any e-advertisement or e-

commerce activity. The second image in Figure 3

indicates the exposure and popularity of social

networks. Interestingly, 45% of the users get their

advertisement information from Facebook.

Figure 1: Reaction to online advertisements (Q2) and

Source of online advertisement use (Q3).

Figure 2: Factors affecting online advertisement.

The third hypothesis suggested that there are

many factors affecting online advertising and e-

commerce. The response to the question Q5 related

to this hypothesis was as shown in Figure 2.

The students correctly noticed that social

applications offer little in terms of privacy, but make

up for it in terms of personal recommendations and

reliability, thus explaining their popularity.

The security of the applications is related

somewhat to the privacy provided, and thus is of

lesser importance. Hence, a preliminary pre-

selection and order of relevance of the factors that

are expected to influence online advertising in e-

commerce based on social networks has been

obtained. The students were very interested in

getting personal recommendations and personalised

services.

The final, fourth hypothesis suggested that

personalisation will play a role in the usage of the

system. The students responded highly positively to

questions related to this hypothesis. They answered

the questions about getting personalised

advertisements with a percentage of 83% of them

desiring a personalised experience, while only 17%

being against it. Students have also shown interest in

different types of advertisements, listed below

according to the importance students associated to

them (percentages in brackets), as follows:

 Recommendation for products (47%),

 products that satisfy their needs (41%), and

 Customised profiles (12%).

4.2 Requirement List

As students reached the stage of designing their own

system, the final product was to create a set of

requirements they considered necessary (and, in

most cases, sufficient) in their ideal systems. These

requirements were further grouped by the facilitator

based on their experience, as well as the diagrams

and requirements lists and feedback from the users,

into two main categories, the input and output of the

desired personalised advertisement system:

1. It should be based on (input):

 User modelling techniques.

 Browsing and purchasing history.

2. It should provide (output):

 Live notifications about the advertisement in terms

of what has been clicked on or viewed by other

users.

 Targeted advertisements using social networks.

 Extended advertisements to cover mobile

applications.

WEBIST2013-9thInternationalConferenceonWebInformationSystemsandTechnologies

372

 Social capability to interact, chat, comments about

the advertisement.

5 DISCUSSION & CONCLUSIONS

The main aim of the experiment was to understand

the user needs regarding personalised advertisement

systems based on social networks. Many systems on

the web offer different types of targeted

advertisements. We explored ideas from the end-

users’ perspective regarding their perceptions of an

adapted system taking in consideration their needs.

The main important outcome showed the students’

interest in personalised advertisement using social

interaction, supporting the overarching hypothesis,

H0. The social aspect was crucial in their view, as

they saw the need of both using social networks, as

well as additional social interaction. Thus, they

discussed social networks as a platform of delivery

as well as data collection. At the same time, they

requested social capabilities to be performed on the

advertisement itself, such as allowing functionality

of commenting, rating and chatting about the

advertisement. However, the results also highlighted

that the students, like many users in fact, didn’t have

a clear understanding of what adaptation means and

how content can be automatically adapted. They

looked at adaptation as an extra feature a system can

provide, and not as the main approach of system

delivery. Therefore, their level of understanding was

not reliable enough to detail the system design of the

adaptive process and give a wider perception and a

detailed description of the system requirements.

Overall, the main outcomes are:

1. Advertising is an important activity performed

on the web and needs further research.

2. Social networks play an important role in today’s

businesses and advertising industries. They can be

used both as a data collection tool and delivery tool

for targeted advertisements.

3. The main factors affecting online advertising are

the correct targeting of the users, and reliable,

popular content. People won’t be interested in

unrelated material.

4. Any form of personalisation will have an impact

on system usage. It could increase the system

exportability and performance. However, the exact

way in which the personalisation has to be designed

needs further research. A possible user group that

can help with this aspect are the researchers of the

adaptive hypermedia community.

ACKNOWLEDGEMENTS

This work is done with partial support from the FP7

BlogForever project.

REFERENCES

Abras, C., Maloney-Krichmar, D., Preece, J. (2004) User-

Centered Design. In Bainbridge, W. Encyclopaedia of

HCI. Thousand Oaks: Sage Publications.

Adams, R., (2004) “Intelligent advertising”, AI & Society

18 (1) 68–81.

Brusilovsky, P. (1996), “Methods and Techniques of

Adaptive Hypermedia”, UMUAI Vol. 6, 87-129

Brusilovsky, P. (2007). “Adaptive Navigation Support”

Adaptive Hypermedia, 263-290.

Brusilovsky, P., Millán, E.(2007) “ User models for

adaptive hypermedia and adaptive educational

systems”. In: Brusilovsky, P., Kobsa, A., Neidl, W.

(eds.) LNCS, Vol. 4321, 3–53. Springer, Heidelberg

ChoiceStream, Choicestream personalization survey:

consumer trends and perceptions (2005) http://

www.choicestream.com/pdf/ChoiceStream_Personaliz

ationSurveyResults2005.pdf, last viewed 15

August

2012

Dunay P. and Krueger R. (2009) Facebook Marketing for

Dummies, ForDummies Publishing, NY.

De Bono, E. (1985). Six Thinking Hats: An Essential

Approach to Business Management. Little, Brown& Co.

Goy, A., et al (2007). “Personalization in e-commerce

applications”. In Brusilovsky, Kobsa, Nejdl,.(Eds).:

LNCS 4321, 485-520.

Holden G., Belew S., Elad J. and Rich J. (2009), E-

Business, Wiley.

Howse, J. et al (2005) “Spider Diagrams” London Math.

Soc. J. of Computation & Mathematics, Vol 8, 145–194

Kobsa, A. (2007) “Generic User Modeling Systems”, The

Adaptive Web, Vol. 4321 SpringerVerlag Berlin

Heidelberg 136-154.

Kazienko, P., & Adamski, M. (2007).“AdROSA-Adaptive

personalization of web advertising”.Information

Sciences, 177 (11), 2269-2295.

Nielsen, J. (2003) "Usability 101: Introduction to

usability," Jakob Nielsen's Alertbox, Retrieved

October 2012 from http://www.useit.com/

alertbox/20030825.html, 2003.

Osborn, A.F. (1963) Applied imagination: Principles and

procedures of creative problemsolving(Third Revised

Edition). New York, NY: Charles Scribner’s Sons.

Preece, J., Rogers, Y., & Sharp, H. (2002). Interaction

design: Beyond human-computer interaction. New

York, NY: John Wiley & Sons.

Qiao, Y. (2008). Social Networks and E-commerce,

Helsinki University of Technology, http://

www.cse.tkk.fi/en/publications/B/1/papers/Qiao_final.

pdf Retrieved: 16

of June 2012.

AnExploratoryStudytoDesignanAdaptiveHypermediaSystemforOnline-advertisement

373

Schuler, D., Namioka,A., (1993) Participatory Design:

Principles and Practices, Lawrence Erlbaum

Associates, Inc., Mahwah, NJ.

Socialbakers (2012) Facebook Statistics by Country

retrieved February 2013 from http://

www.socialbakers.com/facebook-statistics/?interval

=last-week#chart-intervals

Sanders, Elizabeth B. N. (2002), "From User-Centred to

Participatory Design Approaches," in Design and

social Sciences , J Frascara, ed. Taylor & Francis.

Vredenburg, K., Mao, J.-Y., Smith, P. W., & Carey, T.

(2002). “A survey of user-centered design practice”,

Proc. of SIGCHI CHI 2,(1),471.

www.internetworldstats.com/stats.htm viewed on 31

October 2012

WEBIST2013-9thInternationalConferenceonWebInformationSystemsandTechnologies

374