Detecting fake-review buyers using network structure: Direct evidence from Amazon.

Online reviews significantly impact consumers' decision-making process and firms' economic outcomes and are widely seen as crucial to the success of online markets. Firms, therefore, have a strong incentive to manipulate ratings using fake reviews. This presents a problem that academic researchers have tried to solve for over two decades and on which platforms expend a large amount of resources. Nevertheless, the prevalence of fake reviews is arguably higher than ever. To combat this, we collect a dataset of reviews for thousands of Amazon products and develop a general and highly accurate method for detecting fake reviews. A unique difference between previous datasets and ours is that we directly observe which sellers buy fake reviews. Thus, while prior research has trained models using laboratory-generated reviews or proxies for fake reviews, we are able to train a model using actual fake reviews. We show that products that buy fake reviews are highly clustered in the product reviewer network. Therefore, features constructed from this network are highly predictive of which products buy fake reviews. We show that our network-based approach is also successful at detecting fake review buyers even without ground truth data, as unsupervised clustering methods can accurately identify fake review buyers by identifying clusters of products that are closely connected in the network. While text or metadata can be manipulated to evade detection, network-based features are more costly to manipulate because these features result directly from the inherent limitations of buying reviews from online review marketplaces, making our detection approach more robust to manipulation.

Identification of Preference "Phenotypes" in Men With Prostate Cancer.

INTRODUCTION: Patient preference assessment is key to high-quality decision-making in men with prostate cancer. We aimed to determine if "phenotypes" could be identified among men with prostate cancer, with each phenotype representing a cohort with a distinct combination of preferences. We wished to learn if there was an association between phenotype and treatment selection. METHODS: A prospective cohort of men with prostate cancer received a pre-visit decision aid. This software used conjoint analysis to quantify relative patient preferences for treatment-associated survival, quality of life outcomes, and recovery time. We collected patient clinical data, physician recommendation for active treatment or surveillance, and treatments received. Preferences were analyzed using latent class analysis to identify distinct classes of preference phenotypes. We compared patient characteristics and treatment choice across phenotypes, both univariately and in a multivariable logistic regression. RESULTS: In 250 men who used the decision aid as part of routine care, latent class analysis revealed 3 phenotypic classes. Men in Class 1 had the highest concerns around recovery time and the lowest value on improving lifespan. Men in Class 2 had relatively evenly distributed concerns. Men in Class 3 had the lowest concerns around recovery time and risk of surgical complications. On multivariate analysis, treatment choice was not associated with preference-based phenotype. Only physician recommendation was associated with choice of active treatment. CONCLUSIONS: We identified the existence of 3 patient preference-based phenotypes in men with prostate cancer. Each phenotype had a unique combination of trade-offs when considering competing treatment outcomes. These phenotypes were not associated with treatment. Physician recommendation was the only factor determining treatment choice.