Time sensitive ranking with application to publication search

Abstract

Link-based ranking has contributed significantly to the success of Web search. PageRank and HITS are the most well-known link-based ranking algorithms. These algorithms are motivated by the observation that a hyperlink from a page to another is an implicit conveyance of authority to the target page. However, these algorithms do not consider an important dimension of search, the temporal dimension. These techniques favor older pages because these pages have many in-links accumulated over time. New pages, which may be of high quality, have few or no in-links and are left behind. Research publication search has the same problem. This project investigates the temporal aspect of search in the framework of PageRank with application to publication search. Existing remedies to PageRank are mostly heuristic approaches. This project proposes a principled method based on the stationary probability distribution of the Markov chain. The new algorithm, TS-Rank (for Time Sensitive Rank), generalizes PageRank. Methods are also presented to rank new papers that have few or no citations. The proposed methods are evaluated empirically; the results show the proposed methods are highly effective.