Science is built upon scholarship consensus that changes over time. This raises the question of how revolutionary theories and assumptions are evaluated and accepted into the norm of science as the setting for the next scientific revolution. Yiling’s recent work analyzed millions of research articles published over four decades and revealed the scarcity (~2%) of highly novel work, the higher chance (25%) they disrupt existing research rather than developing it, and its long-shot nature (it takes ten years or longer to recognize a novel work, over which period only 20% scholars survived). Yiling’s work also provides one of the first computational models reformulating Uzzi Atypicality, a prominent novelty measure in the Science of Science, as the distance across the latent knowledge spaces learned by neural networks. The evolution of this knowledge space characterizes how yesterday’s novelty forms today’s scientific conventions, which condition the novelty–and surprise–of tomorrow’s breakthroughs. This paper is accepted to present at SciNLP, a community aiming to bring together researchers from diverse fields who are interested in building computational systems that extract, represent the knowledge in scientific texts, and/or provide humans better access to such knowledge.

Check out the paper!