Testing Ranking Algorithms on CiteSeer Data

This article describes how various ranking algorithms have been tested to evaluate researchers based on the data from a digital library called CiteSeer. We apply five well-known ranking methods such as citation counts, HITS, or PageRank and seven other methods derived from PageRank that take into account not only citation but also collaboration information to assess the importance of individual researchers. We compare the resulting rankings and show that some of them are highly correlated while others are not. Keywords— CiteSeer, researchers, rankings, PageRank, correlation, convergence