Purchased Fame: Exploring the Ecosystem of Private Blog Networks

Abstract

For many, a browsing session starts by entering relevant keywords in a popular search engine. The websites that users thereafter land on are often determined by their position in the search results. Although little is known about the proprietary ranking algorithms employed by popular search engines, it is strongly suspected that the incoming links have a significant influence on the outcome. This has lead to the inception of various black-hat SEO techniques that aim to deceive search engines to promote a specific website. In this paper, we present the first extensive study on the ecosystem of a novel type of black-hat SEO, namely the trade of artificially created backlinks through private blog networks (PBNs). Our study is three-pronged: first, we perform an exploratory analysis, through which we capture intrinsic information of the ecosystem and measure the effectiveness of backlinks. Next, we develop and present an ML-driven methodology that detects PBN sites with an accuracy of 98.7% by leveraging various content-based and linking-based features intrinsic to the operation of the ecosystem. Finally, in a large-scale experiment involving more than 50,000 websites, we expose large networks of backlink operations, finding thousands of websites engaged in PBNs.