H-index manipulation by merging articles: Models, theory, and experiments

Abstract

An author's profile on Google Scholar consists of indexed articles and associated data, such as the number of citations and the H-index. The author is allowed to merge articles; this may affect the H-index. We analyze the (parameterized) computational complexity of maximizing the H-index using article merges. Herein, to model realistic manipulation scenarios, we define a compatibility graph whose edges correspond to plausible merges. Moreover, we consider several different measures for computing the citation count of a merged article. For the measure used by Google Scholar, we give an algorithm that maximizes the H-index in linear time if the compatibility graph has constant-size connected components. In contrast, if we allow to merge arbitrary articles (that is, for compatibility graphs that are cliques), then already increasing the H-index by one is NP-hard. Experiments on Google Scholar profiles of AI researchers show that the H-index can be manipulated substantially only if one merges articles with highly dissimilar titles.

Over-optimization of academic publishing metrics: observing Goodhart’s Law in action

“…According to recent studies, and following Goodhart’s Law, these metrics have now become targets, ripe for manipulation [9, 10, 58]. All types of manipulative methods are used, such as increasing the number of self-citations [20], increasing the number of publications by slicing studies into the smallest quantum acceptable for publication [59], indexing false papers [60], and merging papers on Google Scholar [61]. Indeed, a recent study by Fong and Wilhite [58], which used data from >12,000 responses to a series of surveys sent to >110,000 scholars from 18 different disciplines, discovered “widespread misattribution in publications and in research proposals.” Fong and Wilhite’s findings revealed that the majority of researchers disapprove of this type of metric manipulation, yet many feel pressured to participate; other researchers blandly state “that it is just the way the game is played” [58].…”

h-Index manipulation by undoing merges

“…Indeed, we will closely follow the notation and concepts introduced by van Bevern et al (2016b) and we refer to this work for discussion of related work concerning strategic self-citations to manipulate the h-index (Bartneck & Kokkelmans, 2011;Delgado López-Cózar, Robinson-García, & Torres-Salinas, 2014;Vinkler, 2013), other citation indices (Egghe, 2006;Pavlou & Elkind, 2016;Woeginger, 2008), and manipulation in general (Faliszewski & Procaccia, 2010;Faliszewski, Hemaspaandra, & Hemaspaandra, 2010;Oravec, 2017). The main difference between this work and previous publications is that they focus on merging articles for increasing the h-index (Bodlaender & van Kreveld, 2015;de Keijzer & Apt, 2013;Pavlou & Elkind, 2016;van Bevern et al, 2016b) or other indices, such as g-index and the i10-index (Pavlou & Elkind, 2016), while we focus on splitting.…”

“…Our main points of reference are three publications dealing with the manipulation of the h-index, particularly motivated by Google Scholar author profile manipulation (de Keijzer & Apt, 2013;Pavlou & Elkind, 2016;van Bevern, Komusiewicz, et al, 2016b). Indeed, we will closely follow the notation and concepts introduced by van Bevern et al (2016b) and we refer to this work for discussion of related work concerning strategic self-citations to manipulate the h-index (Bartneck & Kokkelmans, 2011;Delgado López-Cózar, Robinson-García, & Torres-Salinas, 2014;Vinkler, 2013), other citation indices (Egghe, 2006;Pavlou & Elkind, 2016;Woeginger, 2008), and manipulation in general (Faliszewski & Procaccia, 2010;Faliszewski, Hemaspaandra, & Hemaspaandra, 2010;Oravec, 2017).…”

“…A further point that distinguishes manipulation by splitting from manipulation by merging is that for merging it is easier to detect whether someone cheats too much. This can be done by looking at the titles of merged articles (van Bevern et al, 2016b). In contrast, it is much harder to prove that someone is manipulating by splitting; the manipulator can always claim to be too busy or that he or she does not know how to operate the profile.…”

Should we hunt a metric?

“…Yet the moment a metric appears, so it is open to manipulation. One can increase self-citations [18, 19], index false papers [20], merge papers on Google Scholar [21, 22] and more. Merging is quite commonly undertaken, and should that be your thing, the h-index is best manipulated by merging articles with widely dissimilar titles.…”