Manipulating Citation Indices In A Social Context

The h-index [6] is a popular measure of a researcher's publication activity: a researcher's h-index is the largest number x such that she has at least x papers that have received at least x citations each. It has been observed that one can manipulate her h-index by strategically merging one or more articles, and the complexity of finding a successful/optimal manipulation has been investigated for a variety of models [3, 11]. In this paper, we extend this line of research to two other popular citation indices, namely, the g-index [4] and the i10-index, and show that these indices are somewhat easier to manipulate than the h-index. We then consider settings where the manipulator would like to take into account the impact of her actions on other researchers ( she may want to make sure that her manipulation does not harm her friends or that it hurts her competitors) or a group of researchers manipulate their indices simultaneously. We analyze the complexity of these problems, both in the worst-case and in the parameterized framework.