Article Preview
TopIntroduction
Due to advancements in technologies, online banking services via mobile applications and desktop applications hold continued growth among customers (Teutsch, Jain, & Saxena, 2016). This growth may increase the complexity of the banking system and raise security concerns among users. Also, there are many currencies exhibit in the real world, and each country has its currency, which again increases the complexity of the financial system (Sklavos, & Koufopavlou, 2005). To bypass the banking systems' complexity, a new type of currency came into the picture called cryptocurrency (Ahmad, Kumar, Shrivastava, & Bouhlel, 2018), which has no walls or boundaries and used globally anywhere in the world. The concept of cryptocurrency is new; it may raise security concerns when using mobile applications or desktop applications to prevent fraud and protect personal information.
A cryptocurrency is a decentralized form of money compared to any central banking system. In this age, where cryptocurrencies are slowly creeping into the banking services and making a name for them, it is becoming crucially essential to figure out the security concerns when users make transactions. With the high instability of cryptocurrencies and highly competitive central bank currencies, most of the population is hesitant when investing in cryptocurrencies and skeptical of how to differentiate the trusted user, untrusted users. Cryptocurrencies are mainly working with blockchain mechanism, initially proposed by Satoshi Nakamoto in 2009. Satoshi Nakamoto named the proposed bitcoin (Schwartz, Youngs, & Britto, 2014) cryptocurrency. Figure 1 shows the basic workflow of the cryptocurrency using the blockchain mechanism. Here, client A sends money to B without the involvement of any other thirst party.
This paper investigates the untrusted users of cryptocurrency transaction services, which are connected using smartphones and computers. Though the cryptocurrency users are anonymous, it is necessary to maintain the users' reputation information to identify risky users.
As technology is increasing, transaction frauds are growing, and there is a need to detect vulnerabilities in such systems. Several types of risk exist in financial systems, such as hacking, fraud, password leak, and so on (Niu, Ji, & Tan, 2005). Various kinds of attacks can be conducted by attackers to breach smart devices' security and obtain illegal access (Upadhyaya, & Jain, 2016). Such as existing web pages are replaced with fraudulent web pages, broken firewalls for mobile apps using financial services, old desktop apps that are not updated with technologies, and encourage fraud security parameters. Many third parties are involved in financial services that may promote risks, Hackers, or un-trusted users. Hence, authors pick up one of such risks involved in the latest cryptocurrency financial system, i.e., identifying untrusted users from the networks. Such problems motivate us to find chances by collaborating with different domains to find the optimal and reliable solution. Here, the authors combine social networks, centrality, and machine learning with seeing untrusted users.
The term Social network includes the set of users connected via some mode of relationships. This relationship could be anything such as friendship, business, and so on (Kuncheva & Montana, 2015). In cryptocurrency (Swan, 2015), users are anonymous to each other, and there is direct trading takes place between each other. Based on this relationship, the authors measure the trust factor and find suspicious users in the network.
In-network theories, measuring centralities are among the pre-eminent ways of computing the existence of nodes in the network. Here, the authors utilize centrality measures with machine learning techniques to find suspicious cryptocurrency traders. Centralities are used to find the importance score of individuals in a given network. Many algorithms are defined for measuring the centralities of nodes in the system like degree, eigenvector, stress, page-rank, and so on. Each of these centrality measures has its technique to measure the score of nodes. A degree centrality measure is the amount of all connections of a node. Eigenvector centrality is computed by calculating the relative result of neighboring nodes. Page-rank centrality is given by Google for its search engine to rank the web page.