Abstract
In the past, the models accounting for the scale-invariant nature of the real-world network are unable to capture the individual connectivity dynamics for some common social networks, such as citation networks. To predict the citation dynamics of individual papers, a citation dynamic model was proposed, well capturing the citation history for scientific works. However, it did not offer explanations to the sources behind the papers towards significant impacts. To find out the potential behind impactful scientific works, we conducted citation-based measures on the Web-of-science dataset, finding that most exceptional papers cite other impactful but yet-to-be-recognized ideas in a very early stage, suggesting that a paper's impact depends on its references' impact and the timing it cites its references. Furthermore, by combining the references' impact and the citing timing, we define a unique metric representing the potential of scientific works. We discover that the papers' impacts solely depend on its potential, independent of other factors. We propose a model whose theoretical results quantitatively agree with the real-world network quantities based on the newly defined metric.