Identifying citation patterns of scientific breakthroughs: A perspective of dynamic citation process

Abstract This paper introduces the perspective of dynamic citation process to identify citation patterns of scientific breakthroughs. We construct a series of citation metrics and apply them to over 100 pairs of Nobel and non-Nobel papers with millions of citations. As expected, we find that most metrics cannot distinguish the two groups under similar conditions of discipline, publication year, venue, and citation impact. Some metrics, however, not only show significant discriminative power, but also reflect scientific breakthroughs’ temporal and structural characteristics—namely, prematurity and fruitfulness. Breakthrough works, that is, have long-lasting impact, but recognition lags behind; they do not just solve a problem, but more importantly open up new questions. Three metrics—average clustering coefficient, connectivity, and density of citing literature networks—show particular promise for early identification of breakthrough works. Our findings bear significant implications for science and technology management practices: from a science policy standpoint, our work demonstrates the utility of this citation process-based approach and provides a new dimension for both innovation researchers and decision makers in search of emerging scientific breakthroughs.