Trust Evaluation and Management in E-commerce Systems using Artificial Intelligence and Emerging Technologies
Abstract
The internet market has a lack of physical presence and physical indication; therefore,
it is more complex and difficult to bring trust between sellers and buyers. This
is what distinguishes it as a big obstacle for businesses engaged in the several applications
of e-commerce platforms. Moreover, it is essential to recognize influences,
which enables to create customer interest in e-commerce, and consumers can feel
reliable and get interested to buy on an e-commerce platform frequently. Therefore,
there is an alarming demand to deal with the aforementioned issues and how
to create a trust evaluation-based model that provides feasible support for both
current and past sales. The proposed research comprises several trust hypotheses,
a business case review, and primary data selections to establish a more reliable and
diverse database. The purpose was to examine customer trust from a personal perspective
and to decide if the variables of trust are positively influenced by personal
characteristics, culture, and experience.
The overall objectives of this proposed research are to examine that how to build
trust in online markets such as in e-commerce applications, also to investigate the
impact of trust between sellers and buyers during their active interaction and online
trust formation process. This work is particularly concerned with how different
types of users (online stores and internet users) use trust in their trust-building
practices. The relationship between trust variables and risks associated with trust
adoption was not provided by researchers in existing work. Researchers have failed
to explore the dilemma in a multifaceted and multidimensional manner. To deal with these issues, the first contribution provides an effective method that considers
three dimensions that play important roles in any online transaction to help the
buyers to detect fraud. This method measures the similarity between the new
transaction and the past transactions in the products types dimension, the number
of the products sold dimension, and the transactions amount dimension.
Our second contribution utilises the trust evaluation between the transactions by
measuring the similarities between the new transactions of a purchaser and the
previous sales record of a buyer. The proposed model is capable to normalise several
user evaluations after following their evaluation criteria, which continuously changes
with time. This model also investigates trust and reputation for all users based on
the general merchant rating and the aspects of the transaction time, both have
a significant role in any online transaction to help customers gain trust in online
shopping and identify the frauds. Overall, our model based on eBay and Epinions
transaction datasets derives stable trust values from immutable transactions.
Our third contribution is based on mining customer’s comments and feedback expressed
in free text. We introduce an algorithm for extracting input on measurements
weights, integrating natural language processing methods, information extraction,
and subject/topic modelling. Evaluation based on Amazon data sets enables
us to demonstrate that order of importance for our predefined dimensions is
Rating, Product Type, Price, Time interval, and Quantity Sold.
The final contribution is designed in collaboration with the Software-Defined Networking
(SDN) and blockchain. SDN allows versatile implementation and advancement
of modern networking applications due to the programmability approach, and
decoupling of the control plane from the data plane, the concept and way of developing
and operating interconnected networks has evolved dramatically and reduced
the obstacles in the e-commerce markets. Moreover, SDN also experiences a variety
of challenges, such as trust limitation and single point failure in a distributed net