This paper addresses the issue of opinion spam in product reviews, a critical concern given the widespread use of such reviews by consumers and manufacturers. The authors analyze 5.8 million reviews and 2.14 million reviewers from Amazon.com to study the prevalence and characteristics of opinion spam. They identify three types of spam: untruthful opinions, reviews on brands only, and non-reviews (advertisements and irrelevant reviews). The main contributions of the paper include a novel approach to detect untruthful opinion spam using duplicate and near-duplicate reviews, and the development of a classification model based on logistic regression. The model effectively distinguishes spam from non-spam reviews, with an average AUC of 98.7%. The study also highlights the importance of reviewer behavior and product characteristics in identifying spam, such as reviewer rank, feedback levels, and product ratings. The findings suggest that top-ranked reviewers and reviews with high feedback levels are more likely to be spam, contrary to common beliefs. The paper provides valuable insights and techniques for detecting and mitigating opinion spam in online reviews.This paper addresses the issue of opinion spam in product reviews, a critical concern given the widespread use of such reviews by consumers and manufacturers. The authors analyze 5.8 million reviews and 2.14 million reviewers from Amazon.com to study the prevalence and characteristics of opinion spam. They identify three types of spam: untruthful opinions, reviews on brands only, and non-reviews (advertisements and irrelevant reviews). The main contributions of the paper include a novel approach to detect untruthful opinion spam using duplicate and near-duplicate reviews, and the development of a classification model based on logistic regression. The model effectively distinguishes spam from non-spam reviews, with an average AUC of 98.7%. The study also highlights the importance of reviewer behavior and product characteristics in identifying spam, such as reviewer rank, feedback levels, and product ratings. The findings suggest that top-ranked reviewers and reviews with high feedback levels are more likely to be spam, contrary to common beliefs. The paper provides valuable insights and techniques for detecting and mitigating opinion spam in online reviews.