Journal of Informetrics 15 (2021) 101105 Contents lists available at ScienceDirect Journal of Informetrics jou rn al hom epage : www.elsevier.com/locate/joi Regular article Mitigating ageing bias in article level metrics using citation network analysis István Tóth a,1 , Zsolt I. Lázár a,1 , Levente Varga a,b , Ferenc Járai-Szabó a , István Papp a , R˘ azvan V. Florian c,** , Mária Ercsey-Ravasz a,* a Babes ¸ -Bolyai University, Faculty of Physics, Cluj-Napoca, 400084, Romania b Babes ¸ -Bolyai University, Faculty of Mathematics and Computer Science, Cluj-Napoca, 400084, Romania c Romanian Institute of Science and Technology, Cluj-Napoca, Romania a r t i c l e i n f o Article history: Received 24 April 2020 Received in revised form 23 September 2020 Accepted 13 October 2020 Keywords: Scientometric indicator Article level metric Citation network Ageing bias a b s t r a c t Article level scientometric indicators (ALMs) are usually of cumulative nature making arti- cles of different age hard to compare. Here, we introduce a new ALM, the Time Debiased Significance Score (TDSS), which measures the significance of a publication based on the structure of the whole citation network and eliminates the global ageing bias in the net- work: older publications should not be a priori privileged or disadvantaged compared to newer ones. The TDSS is based on a modified variant of the PageRank measure, incorporat- ing a mathematically consistent temporal detrending and ensuring a few key features: (i) the TDSS should not show any global trend as a function of the topological index (causal order in the citation network); (ii) the TDSS value of a publication should decrease as time passes (and the citation network grows) if no more citations are associated with it. The above definition is beneficial in multiple ways, including e.g. low computational complex- ity and weak domain dependence. Further, estimation of reliability of the TDSS and its extension to groups of items like overall score of a research group are also possible. © 2020 Elsevier Ltd. All rights reserved. 1. Introduction Due to the direct and increasingly decisive impact of science on fundamental components of society such as economy and politics the unbiased assessment of performance of scientific activity has become one of the greatest challenges of scientific communities, academies, universities, research grant committees (Fortunato et al., 2018). Science can be viewed as a complex and continuously evolving network of articles, researchers and institutions (Newman, 2018). A field where “scholarly citation remains the dominant measurable unit of credit” (Fortunato et al., 2018). Therefore, the most widespread methods for assessing scientific output rely on citations modeled as edges in the network of publications referencing each other (Radicchi, Fortunato, & Vespignani, 2012). Studies on the structure and dynamics of citation networks have revealed universal properties (Eom & Fortunato, 2011; Golosovsky & Solomon 2012a,2012b; Néda, Varga, & Biró, 2017; Parolo et al., 2015; Radicchi, Fortunato, & Castellano, 2008; Waltman, van Eck, & van Raan, 2011) determined most probably by general anthropological factors. Another direction of research fostered by the fast growing corpus of available data and a matur- ** Corresponding authors at: Romanian Institute of Science and Technology, Str. Virgil Fulicea, nr. 3, 400022, Cluj-Napoca, Romania. * Corresponding authors at: Babes ¸ -Bolyai University, Faculty of Physics, str. M. Kog˘ alniceanu 1, Cluj-Napoca, 400084, Romania. E-mail addresses: florian@rist.ro (R.V. Florian), maria.ercsey@ubbcluj.ro (M. Ercsey-Ravasz). 1 These authors contributed equally to this work. https://doi.org/10.1016/j.joi.2020.101105 1751-1577/© 2020 Elsevier Ltd. All rights reserved.