FACTOR PARTICIPATING AND IMPACTING E-MARKETS PIONEERS

BEHAVIORAL TRACKING

RIYADH MEJBELYASEEN AL-DULAIMI 1, AMEL MEDDEB MAKHLOUF 2 and

AHMED FAKHFAKH 3
1, 2
NTS'COM National School of Electronic and Telecommunication of Sfax, University of Sfax, Tunisia.
3
NTS'COM National School of Electronic and Telecommunication of Sfax, Laboratory of Signals, System,
Artificial Intelligence and Networks, Digital Research Center of Sfax, University of Sfax, Tunisia.
Email: 1 [email protected], 2 [email protected], 3 [email protected]

Abstract
Due to the emerging of internet technology and development of mobile communications, word economic has
tended to digital and soft markets. Accordingly, all the necessary steps were taken to establish a virtual market
ensuring stunning shopping experience for each customer. The customer’s behaviors are the most important factor
in E-Commerce platforms. Marketing team (company) is desperately willing to know how dose each category of
customers react to their product. Data mining technologies including the clustering techniques, classification and
prediction are deployed to mining the knowledge of marketing data. The data mining technology influence on the
E-Commerce has been analyzed in this study; a questionnaire base review has made to collect the data required
in this study. SPSS 23 software has been used to analysis the questionnaire data. According to the results observed
in this study, data mining can enhance the performance of E-Commerce by providing a smart and efficient
customer behaviors prediction. Furthermore, this technology is realized with similar advantages for the end-
customer also.
Keywords: E-marketing, Shopping, Clustering, Neural Network, Mean, Standard Deviation, HTML, Event
Prediction.

1. INTRODUCTION
Today, the world has witnessed a wave of rapid changes and developments, which have added
its effects on all levels of the economy, whether partial or total, and at various levels [1].
Specialists have agreed to call it globalization, or inclusivity, and if we reflect on this
phenomenon that is in continuous development, especially starting from the half second of the
twentieth century, we find that the most important component of making this privacy and
excellence in it: information and communications technology [2][3].
Although this term is made up of two broad concepts: information and communication, and
despite their vastness and depth, they are closely related and interconnected, and without the
information and its technologies, there would be no communication, and there would be no
development of the means of communication, and without communication and its technologies,
the information could not reach the importance it is now [4].
Also, their cohesion has the greatest impact on the phenomenon that the universe lives today,
and one of its most important features is the elimination of spatial and temporal barriers thanks
to the Internet, the fruit of information technology, which enabled the whole world to
communicate and exchange information everywhere and at any time with the whole world, as

657 | V 1 8 . I 1 2
it had great credit in changing the style and management of institutions. It enabled information
technology, especially the Internet, to impose new rules in the field of business [5] [6].
The most important one is betting on the customer as the only way to excel [16][17][18]. The
institutions have worked for creating value for him, in a manner that ensures that his needs and
renewable needs are constantly met and his tendencies are known even before they are
revealed, something that will not be achieved of course unless these institutions are able to
make optimal use of technology Information, by establishing E-Commerce in its organization
as the most cliental job in the organization, and the most closely aligned and most appropriate
to adapt to information technology, in order to achieve this institution efficiently and
effectively, as information technology is a link between the institution and its environment, and
because the via the Internet in particular, and information technology in general requires the
availability of certain physical and knowledge infrastructures [7][8].

2. PROBLEM STATEMENT
Since the todays economy tends towards E-commerce and online transactions, the term E-
Commerce gained extended importance. Generally, marketing strategies can be established by
understand the customer ways of thinking. Each category of customers is having different
requirements and hence, marketing planers need to address out each factor associated with
customer’s needs. Customer needs and ways of thinking are understandable by analyzing data
related to customer behaviors. Many ways are available for gathering the data such as web
logging information, surveys on the web networks, personal interviews in public places, etc.
Eventually, large amounts of data are witnessed for analysis prior making any marketing plans.
The amount of this data is seemed not possible to be analyzed in the conventional methods and
normal computer programs. Extraction of knowledge from the big data is one of the important
challenges facing the marketing professionals. Data mining technology can help for addressing
this issue; but other considerations alike, cost, privacy, worthiness, etc. are came to the image.

3. THEORETICAL CONCEPTS AND HYPOTHESIS

The E-Commerce had gained wide interest in the resent years due to economic changes and
life development [9]. The revolution of communication has impacted the individuals as well as
service provider’s routines [10]. New applications alike smart mobile platforms and artificial
intelligence have made great impact on business and economy [11]. Internet development as
well as mobile communications popularity has made noteworthy impact on the e-commerce
and marketing applications [19]. That imposed large amount of data that being exchanged
hourly wised over the e-commerce and E-Commerce platforms [12]. This type of data is being
gathered from various sources more likely, short time surveys, cookies, customer behaviors,
etc. Using the E-Commerce platform is required to understand the factors impacting on its
performance [13] [14]. Data mining technology has made outstanding impact on the
performance of e-marketing [20] [21]. The factors that directly and indirectly impacting the
performance of E-Commerce are enlisted in the following subsections with [15] [16].

658 | V 1 8 . I 1 2
3.1 Samples Collection
Sample a grope of people who are representing the population; so-to-say, sample of this
population is to be taken to represent the population itself in this research study. A grope of
thirty individuals (n=30) are invited to participate in this study by filling up a list of
questionnaires. Those individuals are selected carefully to participate this study, the selection
of each is made according to the following criteria:
a) Considering the age of candidate to be from 23 years to 70 years.
b) Every candidate might have adequate knowledge of E-Commerce as well as the data
mining technology.
c) Expecting each candidate is having diploma as minimum academic education
requirements to participate the survey.
d) Candidates participating this research study must be a combination of males and
females.
The candidates participating in this research and fulfilling the above criteria are reliable enough
to provide the information for drawing a conclusion and testing the hypothesizes. Because of
that, the above points are kept in mind while inviting the candidates to fill-up the survey
questions. Each candidate is invited to participate the survey by sharing a link that direct
him/her into the questionaries’ (survey) web page. a Google form are used to enlist all the
questions in the survey, candidate may require to open that link and provide the answers using
the radio buttons (multi-choices answer). Table below is demonstrating the demographic
quintenary, the Table below provides the responses of the candidates about the biometrical
questions in the survey such as gender, education, age and experience.
3.2 Survey Implementation
As soon as the validation or pilot study is over, the draft of each question involved in the
questionnaire is made. The survey questionnaire was constructed of four groups of questions
as illustrated in Table 1. However, the responses or the perception of taken from each candidate
(participant) is made according to Likert scale such as strongly disagree to strongly agree.
However, some questions are to be responded using yes or no answers. However, the
questionnaire used in this survey is illustrated in Tables (2) through (4).
Table 1: Demographic and Biological Questions in the Questionnaire List
Item Variable name Questions
DB1 Provide your age.
Demographic
DB2 Provide your gender.
and
DB3 Provide your experience in E-Commerce and data mining.
biometrics
DB4 Provide your educational qualification.
Short survey in other websites (while exploring the internet) can help
DS1 to build a shopping DB.
Data sources
DS2 Cookies usage in E-commerce can provide complete idea about customer

659 | V 1 8 . I 1 2
 shopping tendency

Table 2: Customer Profiling Related Questions in the Questionnaire List

Item Variable name Questions
Web logs information can be used for determining the behaviors of the
CP1
costumer
DM can improve the delivery of the values associated with shopping
CP2
experience to the costumer end
DM can determine whether costumer is really willing to purchase products
CP3
or he/she are only exploring the products
DM can find out whether the customers are familiar with the products they
CP4
Customer are exploring
profiling With the use of DM, an idea is driven about the location that customers are
CP5
shopping from likely, offices, home, etc.
DM find out whither customer is accessing the E-portal from mobile tablet,
CP6
computer, smart phone
DM is used to propose other products (add-on items) upon customer
CP7
interest in particular product.
DM is lead to generate various interactive HTML (web pages) for various
CP8
types of customers

Table 3: Events Prediction Related Questions in the Questionnaire List

Item Variable name Questions
EP1 Clustering algorithms has vital role in event prediction system
Event prediction system is benefiting for both customer as well as e-comm.
EP2
Merchant
Event prediction system can provide the E-Commerce company with the
way that customers are interacting with the new offers so that, E-Commerce
EP3
company can understand how to trigger customers to make more
purchasing.
Event prediction system helps customer to get maximum profit alike cost
EP4
cutting by comparing the prices offered by various merchants.
If-then-else roles can provide straight forward logic with less latency for
EP5
implementation enhanced event prediction system
Fuzzy logic can be used to implement an enhanced event management
EP6
Events system inspire of the computational cost (e.g. latency)
prediction Artificial intelligence (AI) can be used to implement a smart event
EP7
prediction system irrespective of revenue volume
Artificial intelligence (AI) can be used to implement a smart event
EP8
prediction system for small businesses revenue volume
Artificial intelligence (AI) can be used to implement a smart event
EP9
prediction system for big businesses revenue volume
Customer profiling is strongly linked to the process of event prediction
EP10
system implementation
Customer profiling can be achieved using only basic demographic
EP11 information provided by the customer while he/she signed up into E-
Commerce portal
Customer profiling can be implemented using several sources of data
EP12
including demographic (basic information), behaviors data and event

660 | V 1 8 . I 1 2
 prediction system data

Table 4: Technical Aspects Related Questions in the Questionnaire List

Item Variable name Questions
DM is associated with E-Commerce portal hosting in terms of the size
TA1
required to accommodate the data considering the future extension.
The essential demand of DM (e.g. classification or clustering) is to
TA2 segregate the large logging data into separated classes depending on the
Technical
nature of each data source
aspects
View transection matrix can be used for determining the time spent by
TA3
customer on each page of the E-Commerce portal
Information taken from view transection matrix plays essential role in
TA4
mining the knowledge of E-Commerce portal graphical design
Algorithms such as KNN and Associated role discovery algorithm can be
TA5 used for processing of view transection matrix information to gain
knowledge of shopping interest similarity among the customers

3.3 Hypothesis Test

After analyzing the survey questions in the aforementioned sections, the hypothesis that were
applied in this study are checked. The hypothesis correlation level with the study results are
listed in Table 5.
Table 5: Hypothesis Test and Result Decision
Hypothesis Decision
Basic biometrical and personal data can have big influence in the E-Commerce strategy. Accepted
Dataset of E-Commerce portal can be established using web logging information as well
Accepted
as physical surveys conducting on face to face bases.
Customer profiling is vital step in improvement of marketing strategy by providing a
unique content for each customer by using it as a backbone for dynamic HTML (web Accepted
pages) generation system.
Guessing or event prediction system is most required for successful pre launching and
Accepted
post launching marketing.
Smart marketing might enhance the product sell irrespective of the quality of that
Not accepted
product.

4. CONCLUSION
Data mining played vital role in today’s business as a result to communication and technology
development. Mining of large amounts of data that being collected daily wise from the
customer’s activities over the shopping portals is challengeable task. The conventional models
and techniques for extraction a knowledge from this large amount of data is no longer useful.
Using the data mining technologies i.e. (clustering, classification and prediction) as well as
tools alike neural network will required completely understand of both technical and
economical consideration. In this study, methodology and impact of datamining on the E-
Commerce is been discussed.
A sample of thirty people from the industry of data mining and E-Commerce specialists are

661 | V 1 8 . I 1 2
been invited to participate this study by providing their opinions about list of questionnaires
that been shared with them using online form from Google. The questionnaire was consisted
from four sections namely: General biometrical data, customer profiling data, event prediction
data, technical aspects data and eventually some miscellaneous data. Responses of each section
were collected and analyzed using the SPSS version 23 software.
Analysis of the responses were done in section wised where question from each section have
been treated individually. Statistical alike, mean, median, variance, standard deviation,
frequency, percentage, minimum value and maximum value are determined for each question
in each section of the questionnaire. Results reveled that candidates were agreeing to the
questions that used in hypothesis formulation.
In other word, data mining technologies adaptation in E-Commerce context is likely preferred
by the candidates, some restrictions might be considered while so alike the cost of computing
as well as economical cost should be considered carefully. Some other concern alike customer’s
behaviors data collection is suggested to be collected from the cookies used by the site pages.
Cookies is cost efficient and highly trusted for behaviors prediction.

References
1) Albrecht, M.R., Player, R., Scott, S., 2019. On the concrete hardness of learning with errors. J. Mathe.
Cryptol. 9 (3), 169–203.
2) Brakerski, Z., Gentry, C., Vaikuntanathan, V., 2021. (leveled) fully homomorphic encryption without
bootstrapping. ACM Trans. Comput. Theory (TOCT) 6 (3), 1– 36.
3) Chen, H., Chillotti, I., Song, Y., 2022. Multi-key homomorphic encryption from tfhe. In: International
Conference on the Theory and Application of Cryptology and Information Security. Springer, pp. 446–472.
4) Cheng, K., Wang, L., Shen, Y., Wang, H., Wang, Y., Jiang, X., Zhong, H., 2018. Secure k k-nn query on
encrypted cloud data with multiple keys. IEEE Trans. Big Data 7 (4), 689–702.
5) Chillotti, I., Gama, N., Georgieva, M., Izabachene, M., 2020. Faster fully homomorphic encryption:
Bootstrapping in less than 0.1 seconds. In: International Conference on the Theory and Application of
Cryptology and Information Security. Springer, pp. 3–33.
6) Chillotti, I., Gama, N., Georgieva, M., Izabachène, M., 2020. Tfhe: fast fully homomorphic encryption over
the torus. J. Cryptol. 33 (1), 34–91.
7) Creighton, C., Hanash, S., 2023. Mining gene expression databases for association rules. Bioinformatics 19
(1), 79–86.
8) Dam, K.H.T., Given-Wilson, T., Legay, A., Veroneze, R., 2022. Packer classification based on association
rule mining. Appl. Soft Comput. 127, 109373.
9) da Silveira Barcellos, D., de Souza, F.T., 2022. Optimization of water quality monitoring programs by data
mining. Water Res. 221, 118805.
10) Gentry, C., Sahai, A., Waters, B., 2023. Homomorphic encryption from learning with errors: Conceptually-
simpler, asymptotically-faster, attribute-based. Annual Cryptology Conference. Springer, pp. 75–92.
11) Goldreich, O., 2019. Foundations of Cryptography: Volume 2, Basic Applications. Cambridge University
Press.
12) Hardy, S., Van Hertem, D., Ergun, H., 2022. Application of association rule mining in offshore hvac

662 | V 1 8 . I 1 2
 transmission topology optimization. Electric Power Syst. Res. 211, 108358.

13) Kantarcioglu, M., Clifton, C., 2023. Privacy-preserving distributed mining of association rules on
horizontally partitioned data. IEEE Trans. Knowledge Data Eng. 16 (9), 1026–1037.
14) Kaosar, M.G., Paulet, R., Yi, X., 2021. Secure two-party association rule mining. In: Proceedings of the
Ninth Australasian Information Security Conference-Volume 116, pp. 15–22.
15) Kaur, M., Kang, S., 2022. Market basket analysis: Identify the changing trends of market data using
association rule mining. Procedia Comput. Sci. 85, 78–85.
16) Khedr, A.M., Al Aghbari, Z., Al Ali, A., Eljamil, M., 2021. An efficient association rule mining from
distributed medical databases for predicting heart diseases. IEEE Access 9, 15320–15333.
17) Li, L., Lu, R., Choo, K.-K.R., Datta, A., Shao, J., 2022. Privacy-preserving-outsourced association rule
mining on vertically partitioned databases. IEEE Trans. Inf. Forensics Secur. 11 (8), 1847–1861.
18) Liu, X., Deng, R.H., Choo, K.-K.R., Weng, J., 2023. An efficient privacy-preserving outsourced calculation
toolkit with multiple keys. IEEE Trans. Inf. Forensics Secur. 11 (11), 2401–2414.
19) Liu, L., Su, J., Chen, R., Liu, X., Wang, X., Chen, S., Leung, H., 2022. Privacy-preserving mining of
association rule on outsourced cloud data from multiple parties. In: Australasian Conference on Information
Security and Privacy. Springer, pp. 431–451.
20) Pang, H., Wang, B., 2020. Privacy-preserving association rule mining using homomorphic encryption in a
multikey environment. IEEE Syst. J. 15 (2), 3131–3141.
21) Pradel, G., Mitchell, C., 2021. Privacy-preserving biometric matching using homomorphic encryption. In:
2021 IEEE 20th International Conference on Trust, Security and Privacy in Computing and Communications
(TrustCom). IEEE, pp. 494–505.

663 | V 1 8 . I 1 2