PageRank without Hyperlinks: Reranking with Related Document Networks

Graph analysis algorithms such as PageRank and HITS have been successful in Web environments because they are able to extract important inter-document relationships from manually-created hyperlinks. We consider the application of these algorithms to related document networks comprised of automatically-generated content-similarity links. Specifically, this work tackles the problem of document retrieval in the biomedical domain, in the context of the PubMed search engine. A series of reranking experiments demonstrate that incorporating evidence extracted from link structure yields significant improvements in terms of standard ranked retrieval metrics. These results extend the applicability of link analysis algorithms to different environments