Detecting problematic transactions in a consumer-to-consumer e-commerce network

摘要: Providers of online marketplaces are constantly combatting against problematic transactions, such as selling illegal items and posting fictive items, exercised by some their users. A typical approach to detect fraud activity has been analyze registered user profiles, user's behavior, texts attached individual transactions the user. However, this traditional may be limited because malicious users can easily conceal information. Given background, network indices have exploited for detecting frauds in various transaction platforms. In present study, we analyzed networks an consumer-to-consumer marketplace which a seller corresponding buyer connected directed edge. We constructed egocentric each several hundreds fraudulent those similar number normal calculated eight local based on up connectivity between neighbors focal node. Based descriptive analysis these indices, fed twelve features that from random forest classifiers with aim distinguishing engaged one four types transactions. found classifier accurately distinguished classification performance did not depend type transaction.