Discovering and Leveraging the Most Valuable Links for Ranking
This work addresses ranking for search engines and site owners by providing a faster alternative to PageRank, though it appears incremental as it builds on existing link analysis methods.
The paper tackles the problem of ranking web pages by discovering and leveraging the most valuable backlinks, proposing MaxRank as an algorithm that updates scores recursively with a parameter λ to balance between using the best backlink and random ones. Empirical results on Wikipedia show that MaxRank converges dramatically faster than PageRank while maintaining comparable performance, with large λ values (but less than 1) performing best.
On the Web, visits of a page are often introduced by one or more valuable linking sources. Indeed, good back links are valuable resources for Web pages and sites. We propose to discovering and leveraging the best backlinks of pages for ranking. Similar to PageRank, MaxRank scores are updated {recursively}. In particular, with probability $λ$, the MaxRank of a document is updated from the backlink source with the maximum score; with probability $1-λ$, the MaxRank of a document is updated from a random backlink source. MaxRank has an interesting relation to PageRank. When $λ=0$, MaxRank reduces to PageRank; when $λ=1$, MaxRank only looks at the best backlink it thinks. Empirical results on Wikipedia shows that the global authorities are very influential; Overall large $λ$s (but smaller than 1) perform best: the convergence is dramatically faster than PageRank, but the performance is still comparable. We study the influence of these sources and propose a few measures such as the times of being the best backlink for others, and related properties of the proposed algorithm. The introduction of best backlink sources provides new insights for link analysis. Besides ranking, our method can be used to discover the most valuable linking sources for a page or Website, which is useful for both search engines and site owners.