TRUST EVALUATION VERSUS RISK MANAGEMENT

There is a common misconception that trust evaluation and risk management refer to the same thing, which is assessing the degree of confidence in completing a goal. Indeed, trust evaluation and risk management have a strong correlation. However, both terms encompass different factors within them. When we talk about managing risk in the digital space, we refer to the complex process of investigating vulnerabilities, threats, and risks in physical networking systems. This process includes framing the risk, assessing it, responding to it once determined and monitoring it.

Although all these steps are also common to the trust evaluation process, they are seen and performed from a different perspective than that of risk management. Trust Evaluation is more human-oriented; thus, takes into account a lot of social factors like opinions, experiences, etc. of the participants. Risk Management, on the other hand, is more system-oriented and assesses only the computer system as an entity. Trust considers the entity from under the lens of the trustor and is used to express a very subjective opinion about the trustee in a specific social context, while in the case of Risk Management; the context is only the digital system (see Figure 5.1).

FIGURE 5.1 Comparison of risk management and trust evaluation. (Reprinted from Truong et al., 2017. http://creativecommons.Org/licenses/by/4.0/).

ONLINE TRUST EVALUATION IN E COMMERCE

DEFINING E COMMERCE AND IMPLICATIONS OF TRUST?

E-commerce is a medium through which consumers and merchants may purchase or sell products and/or services online, i.e., over the Internet, using digital communication as an option for payment (see Figure 5.2.) Since the development of the internet, several models have been identified to define the different types of e-commerce such as B2B (business to business), B2C (business to consumer), C2B (consumer to business), C2C (consumer to consumer), B2G (business to government), and Mobile Commerce. Over the years, B2C e-commerce has seen phenomenal growth as many organizations have shown growing interest to use Electronic Commerce to enhance their competitiveness and cater to a broader audience.

FIGURE 5.2 Depiction of a typical e-commerce process. (Reprinted from https://www. pinterest.com/pin/322359285822573233)

E-commerce has gained wide popularity because of the benefits it offers in contrast to traditional commerce. These benefits may be attributed to the availability of more quantity and quality of infonnation to the customers. For Example, they may gather as much information required via user- friendly websites, track their orders, and/or provide feedback.

However, the Internet’s open system environment along with the innovation that comes with remote shopping is also prone to some hindrances in terms of generating consumer loyalty, the most significant being lack of trust.

The concept of trust is thus, is given much importance by researchers as if there is no general notion of online trust, doing business online would be a challenge. The various threats that exist in the e-commerce space can be seen from two perspectives: business-related, and technology-related (see Figure 5.3). Technology-related risks include security, integrity, and privacy concerns, while on the other hand; the business-related risks include those that involve the misuse of personal information and fraudulent transactions. E-commerce systems tend to be unstable because of such risks. Consumers, therefore, become apprehensive of the low level of personal data security, hidden charges, and difficulties in refunds/cancellations and of the possibilities of non-delivery of goods. With these risks and challenges prevalent in the online sphere, all the online merchants are majority dependent on their web presence via websites and applications only to represent themselves to prospective clients. This makes them focus on web interface extremely integral to online selling. It also calls for the use of different frameworks like reputation-based algorithms and feedback analysis by the sellers to enhance their perceived trustworthiness to potential customers. All major e-commerce websites today like Amazon, E-bay are actively integrating these models within their systems to evaluate customer trust and enhance their processes, accordingly, thereby making this a popular field of study within their research teams.

REPUTATION BASED ONLINE TRUST EVALUATION

In applications involving e-commerce, the reputation of the seller is a major concern for customers while making a purchase or payment. This has made researchers extremely interested in reputation-based search algorithms. Trust evaluation using reputation-based schemes usually focuses on parameters like sendee type, transaction, and interaction history. Socially oriented environments demand that the focus while evaluating trust should lie on the nature of the relationship between the interacting parties while sendee- oriented environments require that there is reasoning and maintenance about quality of services (QoS) in addition to relationships when evaluating trust. The latter also requires that recommendation-based trust is evaluated and incoiporated in the context.

FIGURE 5.3 Various risks associated with e-commerce.

Reputation-based trust evaluation can be applied not only to socially- oriented trust environments but also to service-oriented trust environments. This can be understood by observing how a service can gain a good reputation. This is usually done via customer ratings, once services of good quality are accumulated over some time; ratings improve thereby signaling a good reputation. However, this does not solely constitute the final reputation value. To come to a final reputation value, the relationship between the trustee (who gets rated) and the trustor (who rates) is studied, and more objective trust results are obtained.

We further discuss the reputation-based analysis in different trustcomputing categories.

5.6.2.1 REPUTATION EVALUATION IN C2C E-COMMERCE

To study the reputation evaluation in C2C e-conimerce, we consider one of the earliest yet best existing algorithms, i.e., that of e-commerce giant eBay. In eBay's trust management mechanism, customer feedback is given the utmost importance. After every purchase, the customer can rate the service quality as one of the three available options of positive, neutral, or negative. This rating is then stored along with all other customers' ratings and used to calculate a feedback score for the firm. The feedback score calculation is simply the difference between the number of positive ratings and Negative Ratings left by the customers. This score is then published and updated regularly on the web page of the seller, acting as plain data to the buyers for coming to a judgment on their own. Similarly, it uses encouragement metrics like a positive feedback rate to reward sellers with a positive feedback rate greater than a specified threshold, also referring to them as power sellers. In this manner, we can see that this reputation- based scheme is fairly simple yet very effective and gives the users a fair chance to know about sellers and their credibility.

5.6.2.2 TRUST EVALUATION IN PEER TO PEER NETWORKS

In P2P information-sharing networks, before downloading data, the client peer should have the relevant information to know which service peer can fulfill its needs for files. For this, the method of trust evaluation may use polling algorithms, a system for calculating final trust value by binary rating or a voting reputation mechanism that requests the experience of other peers with the given peer and combines the values given by the responders to come to a final value of trust. In the case of systems involving file sharing, binaiy-value ratings are used to signify if the file is a complete version or not. For more complex applications, like the ones that are service-oriented a rating in the range of [0,1] is used.

5.6.2.3 TRUST EVALUATION IN MULTIAGENT ENVIRONMENTS

When we talk about evaluating trust in multi-agent environments, i.e., in cases where autonomous and self-interested software agents finish the tasks given by their owner or even other agents, other issues crop up in online trust evaluation. These include evaluating the agent's motivation and also the dependency relationships between them. Various models like the multidimensional trust model, generic trust vector models, and others that may be based on the rank and reputation of QoS-based services are used in multiagent collaboration situations.

With the above-given models and mechanisms for evaluating trust online in e-commerce related services and transactions, the user may use a suitable architecture based on factors of choice like workload, scalability, reliability, and nature of tasks.

OLX.NG: A CASE STUDY ON TRUST IN THE E MARKETPLACE

5.6.3.1 OLX.com: THE WORLD'S ONLINE MARKETPLACE

With the boom in the e-commerce industry, several companies started to come up since the early 2000s to act as e-marketplaces, with OLX being a major disruptor. Founded in 2006, OLX.com is an online platform for users to buy, sell, and exchange a variety of products. From advertising, it has come up with novel ways through which users can easily design catchy advertisements and use the concepts of control buying-selling and community engagement through the addition of pictures, videos, and display options on social networking sites like Facebook, Linkedln, etc. Spread across over 100 countries and available in more than 30 languages, OLX.com has grown to be of the world’s largest online marketplaces today.

• 5.6.3.2 A SURVEY ON TRUST BETWEEN BUYERS AND SELLERS
• 5.6.3.2.1 The Survey

A group of final year students of an E-Business Course at the University of Ilorin, Nigeria conducted a survey to investigate the degree of online trust in e-commerce. Each student of the class posted items to sell on OLX. ng such as handbags, wristwatches, shoes, laptops, mobile phones, and toasters. The relationship between the students, i.e., the sellers and the buyers were then closely monitored.

5.6.3.2.2 Findings

Within a week, the sellers were able to identify at least two or more prospective buyers. However, by the end of the second week, only about 11% of students were able to sell their products. This also depended upon the kind of goods being sold, with electronics being more comfortable to sell as compared to consumer goods like handbags and shoes, showing that it is difficult to generate trust online in these cases, where there is a significant dependency on the quality, look, and feel. Similarly, about 68% of sellers received two or more calls for their products, but more than half of the buyers still did not trust the seller. One reason for this was the mode of payment provided. If the seller would not provide Cash on Delivery option, the e-vendor could be judged as one having the wrong intentions or being untrustworthy. On the other hand, this also poses a risk to the seller because there is no guarantee of payment on delivery. Therefore, there needs to be a mechanism of generating this trust online within the buyer-seller community. For example, if a seller wants to prove his trustworthiness, he may provide full information regarding his cost of goods, delivery-payment arrangements, after-sales agreements, and also his complete proof of identification on the website. Similarly, if there is a duly signed contract regarding the completion of payment, the seller's trust in the buyer would also increase. This can be achieved using digital signatures and authentication mechanisms. Hence, we see that a critical factor in the success of an e-marketplace is generating trust online. If the sellers can guarantee timely delivery of goods in good condition and the buyers can guarantee their payment at the time of the online transaction, the online trust environment can prosper.