Emotion-oriented eCommerce

Simone Leon Alexander Nikov

MPhil. student Senior lecturer and Head of Usability Lab
simoneleon2008@[Link] [Link]@[Link]

ABSTRACT intelligent behaviour in complex and changing

Emotion-oriented eCommerce is a new and fascinating environments” 4. Computational models are used to
research field that poses many opportunities to facilitate studies on the modelling, interpretation and
understanding the purchasing behaviour of online analysis of human behaviour. CI possesses the ability to
consumers. Computational models can be easily applied understand the beliefs, motives, intentions, perceptions
to analyze and assess predictive behaviour in an and inference involved in human perception and
eCommerce environment. Intelligent emotion recognition behaviour. Appropriate techniques for emotion
techniques using computational intelligence are presented. recognition are Fuzzy Systems, Genetic Algorithms,
The purpose of this particular study is to develop a model Neural Networks, and Swarm Intelligence.
for emotion-oriented eCommerce systems. Developments
within this field of study were noted.
EMOTION RECOGNITION
Electroencephalography has been identified as the
Researchers are using a number of techniques which
preferred method of emotion technique.
captures the emotions of the user. Emotion recognition
Author Keywords involves assigning computers with the ability to observe,
Emotion-oriented eCommerce, computational interpret and generate affect features 8. In an eCommerce
intelligence, electroencephalography (EEG), simulation. environment, the intent of emotion recognition is to
improve the quality of communication between the
customer and eCommerce system.
INTRODUCTION Many theorists have discussed what emotions are and the
Electronic commerce is becoming an increasingly popular role it plays in our daily lives. Humans do experience
tool used in the marketing and selling of products and various dimensions of emotions on a daily basis. When
services. In the online business world, electronic designing an eCommerce website it is important to
commerce is an integral part of the online shopping recognise emotions of consumers in order to build a good
experience 1. eCommerce is capturing the attention of interface for human to computer interaction.
entrepreneurs and consumers globally. Due to the success A user would use a computer in order to achieve a
achieved by many of the well known eCommerce particular result - typing a document, accessing email, or
companies like eBay and Amazon, eCommerce is seen as purchasing a product. The extent to which an interface
an essential ingredient in the commercial market 2. encourages or limits the user’s satisfaction, will in due
Emotions influence consumers’ processes and eShopping course affect their emotional state 4.
behaviour 3. Understanding human moods and emotions Zhang et. al. 9 believed that Maslow’s basic hierarchy of
is an important factor in predicting the purchasing needs should be revisited in order to determine what
behaviour of the consumer. Designing an interface humans want or what they need in their lives and then
involves more than simply trying to make the website employ technologies to support humans’ higher needs in
look good. It is not only about presenting what product or the needs hierarchy. When the needs of the consumer are
service you have to offer but it is also about how you satisfied, greater is the chance of a positive emotional
want the consumer to feel if they are to purchase this reaction. The factors that interfere with satisfying these
product or service. The main concept behind creating an needs would cause negative emotions. In emotion-
emotion-oriented eCommerce website is eliciting an oriented eCommerce, the design of the interface requires
emotional reaction that would lead to a purchasing good understanding of the consumers’ needs.
decision. Every aspect of the website should appeal to the
emotions of the consumer. A user’s previous emotional state can also affect the
experience of subsequent emotions 10. There are a
Computational Intelligence (CI) techniques are more number of internal and external disturbances that may
suitable for emotion recognition. It is defined as “the affect the emotions of a potential customer. Internal
study of adaptive mechanisms to enable or facilitate disturbances may include processor speed, user’s lack of

14
technical expeerience, systeem malfunctio on due to a Speeech Recognitioon – People eexpress their ffeelings by
ccomputer viruss, etc. Externall disturbances would refer to
o the aacoustic featurees and by the ccontent of whatt they want
thhe elements arround the custoomer’s environnment that willl to saay (words, phhrases and syyntactic structuures). The
aaffect their emotional state, e.g.
e loud music playing. Thee user’’s tone and inntonation wouuld depict the emotional
ggoal of the em motion-oriented eCommercee system is to o state..
ffacilitate the switch from m negative and a unwanted d
Faciaal Expression – Expressionss like a smile oor a nod are
eemotions to a positive em motion which will be moree
used to carry out a semantic fuunction, to com mmunicate
rreceptive to purrchasing a prod
duct or service.
emottions. The basiic emotions desscribed by Ekmman 13 can
MMeasuring the emotions of the consumer is a valuablee be rrelated to movvement in the mouth, cheeeks, eyes,
ccomponent in interface
i desig
gn and usabilitty testing. Thee eyebrrows and forehhead.
kknowledge off consumers’ emotions will enable thee
Motoor Behaviouraal Patterns - B Behavioural meeasurement
ssystem to respo
ond appropriateely, thus satisffying the needss
methhods are basedd on the abilityy of the body to respond
oof the consumeer.
motion (e.g. chaanges in muscle tension,
physiically to an em
SScheirer et. all. 12 describeed an approacch to building g coorddination, strenngth, frequenccy) and that the motor
ccomputers thatt recognises aspects of usser frustration. systeem acts as a caarrier for comm
municating affeective state
SSocial interfacees should be abble to recognisse and respond d 14.
to emotions sho own by users in n order to effecctively executee
Bodyy Gestures an nd Movementts – There arre different
rreal-world in nterpersonal interaction strategies
s 11.
posittions of the boddy and it changges over a period of time.
EEmotional sign nals, whether visual
v or audittory responses,,
It coonveys differennt meanings, ffor example. a clenched
aare important during
d the sociaal communication process.
fist m
may indicate ann emotion of feear, anger, or exxcitement.
TThere are a number of waysw humans express theirr
Biosiignals - A serries of biologiical changes ooccur when
eemotions. Leoon and Nikov 8 summarrizes the key y
peopple have emotioons. Biosignals such as hearrt rate, skin
technologies in
n emotion recognition startiing with inputt
condductivity, respiiration, tempeerature, pulse,, electrical
ssignals, data collection and processing, emotion n
activvity in the musscles, brain acttivity, etc. can be used to
rrecognition based
b on computational
c intelligencee
identtify the emotional states in a uuser.
mmodelling and emotion expression.

Fiigure 1. Intellig
gent Emotion Re
Recognition 8

15
SIMULATION MODEL can assist experts in understanding precisely how
Computational Models of emotion would assist in consumers really think, act and behave. This theory can
building models of intelligent behaviour. In an emotion- be extended to any online shopping environment. EEG
oriented eCommerce environment, if the customer is allows for the gathering of detailed information on a
confused or angry whilst using the system, this is an user’s emotional relationship to a particular product or
indication that there is some form limitation in using the service by measuring electrical activity in different parts
system. Designers are immediately alerted that design of the brain in response to certain stimuli like eCommerce
modification needs to be done as the system is creating website elements.
negative emotions in the customer. The combination of design elements (typography, layout,
The use of Neural Networks (NN) provides the navigation, graphics arts and images, video, theme and
opportunity to gather and identify which combination of audio) would be trained by the neural network to form the
design elements will elicit a positive reaction from the interface layer. The output variables are the EEG signals
customer. Neural Networks are able to adapt weights to a given by the eCommerce customer. These signals would
particular environment and retrain easily. Once the give an indication of the type of decision the customer
desired goal is achieved, the system will know that the would possibly make, whether positive or negative.
intended target was met. Positive feedback, may lead to the eventual purchasing of
a product, whilst negative feedback will lead to product
A model for simulation of emotion-oriented eCommerce evaluation and system design modification. The decision
systems following the basic control structure shown in made by the customer, will establish if further emphasis
Figure 2 is proposed. The target rules identifies what the need to be placed on the input variables.
system should do and the input variables refers to the
design elements required for the design of an emotion- SIMULINK was used to create the model for simulation
oriented eCommerce System. of the emotion-oriented eCommerce System. The
SIMULINK control diagram shown in Figure 3 is made
Electroencephalography (EEG) was selected to measure up of six (6) subsystem blocks. These blocks individually
the emotional responses of the customer when faced with carry the elements and processes needed to carry out the
an eCommerce environment. EEG measure the frequency simulation as outline in Figure 2.
and amplitude of electrical activity generated from the
human brain. The benefits of utilizing EEG measurements The six (6) main areas that establish the proper technical
to determine emotions are non-invasive, simple to design for implementation are as follows:
understand and can quickly be processed 15. 1. Design Elements – incorporates the design elements
In this study EEG would be used to assess the user’s and the interface of the eCommerce system. The
experience in an eCommerce environment. It will assess design elements chosen for analysis are typography,
the user emotional experiences by recording customers’ navigation, layout, graphics and images, video, theme
reaction during the interaction with an eCommerce and audio.
website. By measuring the consumers’ state of mind,
emotions and other subconscious responses, EEG signals

Figure 2 - Simulation Model for Emotion-Oriented eCommerce Systems

16
1. DE-NN (Design
( Elemments Neural Network) – Emottion-oriented eeCommerce is relatively a neew field of
receives th
he values generrated by the design
d elementss studyy which presennts many oppoortunities to reesearch the
subsystem block and trrains the netwo ork to get onee abilitty of the comp
mputer to recoggnise and resppond to the
value. userss’ reaction.
2. EEG Userr Features– in ncoporates the EEG Signals,,
The business com mmunity wouldd reap the bennefits from
Alpha, Betta, Delta, Theeta and Gamm ma that can bee
undeerstanding and applying the riight combinatiion of tools
used to dettermine a posittive or negativee emotion.
in ann effort to satisffy the needs off the customer.
3. UF–NN (U User Features Neural Netwo ork) - receivess
the valuees generated by EEG User U Featuress Conssumers are em motional beings and the emotiions which
subsystem block and trrains the netwo ork to get onee they exhibit will afffect their purcchase decision.. Designers
value shoulld focus on designing an eC Commerce intterface that
4. Emotion Recognition
R – iff a positive vallue is received,, woulld elicit positivve emotions. Inntelligence cann be viewed
the value is analysed as a a positive emotion. If a and understood ffrom many perspectives. T This paper
negative vaalue is receiveed, the value iss analysed as a highllights the stuudy of emotioon-oriented eCommerce
negative emmotion. usingg computationnal intelligencce. EEG has been the
5. Expert - receives
r the poositive or nega ative value andd identtified as the preferred m method to iddentify the
determiness whether the loop should co ontinue until a emottional states off the consumerr. A model for simulation
positive va
alue is achievvedelements su ubsystem blockk of inntelligent emootion-oriented eCommerce systems is
and trains the network too get one value.. propoosed. This model has been iddentified as ann important
tool for supportingg the experimeental study andd design of
T n will be used to predict thee
The results of the simulation emottion-oriented eCommerce sysstems.
bbehaviour of consumers
c in a dynamic env vironment likee
eeCommerce. Futurre work in this study w would involvve website
devellopment, perfforming experriments in a controlled
envirronment and coomparative anaalysis.
C
CONCULSION AND FUTURE
E WORK

Figure 3 - MATLAB-bassed Control Structure for Emootion-Oriented eeCommerce Syystems

REFERENCES intelligencce of business ggraduates, Jourrnal of

1. Anumba,
A C.J., Ruikar,
R K., Elecctronic Educationn for Business, Vol. 75, No. 66,
co
ommerce in construction—treends and (2000), 3331-341.
prrospects, Auto omation in Connstruction, Vol. 4. Engelbreccht, A. P., Comp
mputational
11
1, No. 3, (2002 2), 265-275. Intelligencce: An Introducction, Wiley, 22007.
2. What
W is Ecomm merce and the Im
mportance of 5. Leon, Sim
mone, Nikov, AAlexander, “Em motion-
Ecommerce. Oriented eeCommerce Syystems”, WSEA AS
htttp://webdesign
[Link]//what-is- Transactioons on Systemss, Vol. 9, 2010,
eccommerce-imp [Link]. Accessed
A 7
Jaanuary 2011. 6. [Link]
[Link]/journals/systeems/s
[Link].
3. Tu ucker, M.L., ett al., Training tomorrow's
t
leeaders: Enhancing the emotio onal

17
 18

7. Michaud, F. EMIB – computational interaction, New York: Lawrence Erlbaum

architecture based on emotion and Associates, (2002), 251-271.
motivation for intentional selection and
11. Picard, R., Affective Computing, MIT Press,
configuration of behavior-producing
1997.
modules. Cognitive Science Quarterly, Vol.
3, No. 4, (2002) 340–361. 12. Scheirer, J., Fernandez, R., Klein, J., Picard,
R. W., Frustrating the User on Purpose: A
8. Leon S., Nikov, A. Intelligent Emotion-
Step toward Building an Affective
Oriented eCommerce Systems. Recent
Computer, Interacting with Computers, Vol.
Advances in Artificial Intelligence
14, No. 2, 2001.
Knowledge Engineering & Data Bases.
Proc. 9th WSEAS International Conference 13. Ekman, P., Friesen, W.V., Unmasking the
on Artificial Intelligence, Knowledge Face, Prentice Hall, 1975.
Engineering and Data Bases (AIKED’10),
WSEAS Press (2010), 202-207. 14. Zimmermann, P., Guttormsen S., Danuser
B. Gomez P., Affective Computing - A
9. Zhang, P., Benbasat I., Carey J., Davis F., Rationale for Measuring Mood with Mouse
Galletta D., Strong D., Human-Computer and Keyboard, sournal of Occupational
Interaction Research in the MIS Discipline," Safety and Ergonomics, (2003), 539-551.
Communications of the AIS, Vol. 9, No. 20,
15. Li, M., et. al. EEG Emotion Recognition
(2002), 334-355.
System. Vehicle Corpus and Signal
10. Brave, S., Nass, C., Emotion in human- Processing for Driver Behavior, (2008), 125.
computer interaction in J. Jacko & A. Sears
(Eds.), Handbook of human-computer

18