Assessing impact and quality from local dynamics of citation networks

We show that essentially local dynamics of citation networks bring special information about the relevance/quality of a paper. Up to some rescaling, they exhibit universal behavior in citation dynamics: temporal patterns are remarkably consistent across disciplines, and uncover a prediction method for citations based on the structure of references only, at publication time. Above-average cited papers universally focus extensively on their own recent subfield – as such, citation counts essentially select what may plausibly be considered as the most disciplinary and normal science; whereas papers which have a peculiar dynamics, such as re-birthing scientific works – ‘rediscovered classics’ or ‘early birds’ – are comparatively poorly cited, despite their plausible relevance for the underlying communities. The “rebirth index” that we propose to quantify this phenomenon may be used as a complementary quality-defining criterion, in addition to final citation counts. ⺠We study relationships between impact and local, dynamical patterns of citation networks. ⺠Citation counts essentially captures normal science, from the recent past. ⺠Citation data immediately available at publication time has predictive properties. ⺠We define rebirth-index to quantitatively detect ‘sleeping beauties’ or ‘early birds’. ⺠‘Re-birthing’ articles have significantly poorer-than-average citation metrics.