Monitoring online reviews for reputation fraud campaigns

摘要: Abstract Online reviews are critical for both purchasers and sellers in the era of E-commerce. Praiseful and/or 5-star ratings can yield remarkable profit gains, on other hand, a bad-mouth review or low rating score often incurs sales decrease. Therefore, fake detection has attracted lots research interests recent years. While most existing approaches detect an offline fashion, i.e., finding suspicious from large volume historical data, few efforts have been made to online detecting data streams. many more benefits than that damages be significantly reduced by removing them as early possible. In this paper, we propose novel monitoring technique reputation fraud campaigns product reviews. The includes two phases. First, it monitors generate abnormal subsequences (MARSs), which considered candidate campaigns. Second, conditional random fields exploited label each MARS genuine. Experiments show our proposed methods highly effective efficient, with advantages compared approaches.