On characterizing and computing the diversity of hyperlinks for anti-spamming page ranking

摘要: With the advent of big data era, efficiently and effectively querying useful information on Web, largest heterogeneous source in world, is becoming increasingly challenging. Page ranking an essential component search engines because it determines presentation sequence tens millions returned pages associated with a single query. It therefore plays significant role regulating quality user experience for retrieval. When measuring authority web page, most methods focus quantity neighborhood that direct to using inbound hyperlinks. However, these ignore diversity such pages, which we believe important metric objectively evaluating page authority. In comparison true usually contain large number hyperlinks from wide variety sources, difficult fake authorities, boost their rank techniques as link farms, occupy high due prohibitively costs. We propose probabilistic counting-based method quantitatively compute then novel link-based algorithm, named Drank, by simultaneously analyzing quantity, The validations both synthetic real-world show Drank outperforms other state-of-the-art terms finding high-quality suppressing spams.