Collective Opinion Spam Detection: Bridging Review Networks and Metadata
October 19, 2016
We propose a new holistic approach called SpEagle that utilizes metadata (text, timestamp, rating) and relational data (network) under a unified framework to spot suspicious users and reviews, as well as targeted products. Moreover, our method can integrate labels if available, without training. We demonstrate the effectiveness and scalability of SpEagle on three review datasets from Yelp.com.