Automated Crowdturfing Attacks and Defenses in Online Review Systems
This addresses the problem of scalable misinformation in online review systems for platforms and users, presenting a novel attack and defense framework.
The paper tackles the problem of automated fake review generation using deep learning language models, showing that these attacks can produce reviews indistinguishable by state-of-the-art detectors and score high on user usefulness metrics. It also develops novel automated defenses based on the lossy transformation in RNN cycles, demonstrating that countermeasures have unattractive cost-benefit tradeoffs for attackers.
Malicious crowdsourcing forums are gaining traction as sources of spreading misinformation online, but are limited by the costs of hiring and managing human workers. In this paper, we identify a new class of attacks that leverage deep learning language models (Recurrent Neural Networks or RNNs) to automate the generation of fake online reviews for products and services. Not only are these attacks cheap and therefore more scalable, but they can control rate of content output to eliminate the signature burstiness that makes crowdsourced campaigns easy to detect. Using Yelp reviews as an example platform, we show how a two phased review generation and customization attack can produce reviews that are indistinguishable by state-of-the-art statistical detectors. We conduct a survey-based user study to show these reviews not only evade human detection, but also score high on "usefulness" metrics by users. Finally, we develop novel automated defenses against these attacks, by leveraging the lossy transformation introduced by the RNN training and generation cycle. We consider countermeasures against our mechanisms, show that they produce unattractive cost-benefit tradeoffs for attackers, and that they can be further curtailed by simple constraints imposed by online service providers.