A Randomized Reputation System in the Presence of Unfair Ratings

摘要: With the increasing popularity of online shopping markets, a significant number consumers rely on these venues to meet their demands while choosing different products based ratings provided by others. Simultaneously, feel confident in expressing opinions through ratings. As result, millions are generated web for products, services, and dealers. Nonetheless, noticeable users post unfair feedback. Recent studies have shown that reputation escalation is emerging as new service, which dealers pay receive good feedback escalate markets. Therefore, finding robust reliable ways distinguish between fake trustworthy from crucial task every market. Moreover, with dramatic increase consumers, scalability has arisen another issue existing methods systems. To tackle issues, we propose randomized algorithm calculates random sample Since randomly selected logarithmic size, it guarantees feasible large-scale review In addition, randomness nature makes against We provide thorough theoretical analysis proposed validate its effectiveness extensive empirical evaluation using real-world synthetically datasets. Our experimental results show method provides high accuracy running much faster than iterative filtering approaches.