Leveraging microblogs for resource ranking

Abstract

In order to compute page rankings, search algorithms primarily utilize information related to page content and link structure. Microblog as a phenomenon of today provides additional, potentially relevant, information - short messages often containing hypertext links to web resources. Such source is particularly valuable when considering a temporal aspect of information, which is being published every second. In this paper we present a method for resource ranking based on Twitter data structure processing. We apply various graph algorithms leveraging the notion of a node centrality in order to deduce microblog-based resource ranking. Our method ranks a microblog user based on his followers count with respect to a number of (re)posts and reflects it into resource ranking. The evaluation of the method showed that micro-based resource ranking a) can not be substituted by a common form of an explicit user rating, and b) has the great potential for search improvement. © 2012 Springer-Verlag.