From informal to formal: scientific knowledge role transition prediction

Comprehending the patterns of knowledge evolution benefits funding agencies, policymakers, and researchers in developing creative ideas. We introduce the notation of scientific knowledge role transition as an evolution from informal to formal. We investigate how different factors affect the role transition of scientific knowledge, considering the two primary levels—transition pace and transition possibility. The interpretive machine learning models are conducted to discover that the Gradient Boosting classifier performs better for predicting transition possibility, and Random Forests regression is the most effective for predicting transition pace. Specifically, knowledge attribute features have a more obvious effect on the transition probability, while knowledge network structure has a greater effect on the transition pace. We further find that knowledge relatedness and citation number have negative effects on knowledge role transition, while adoption frequency, indegree centrality in the knowledge citation network, node number of the egocentric co-occurrence network, and journal impact of scientific knowledge have positive effects. The aforementioned discoveries enhance our comprehension of scientific knowledge evolution patterns and provide insight into the trajectory of scientific and technological advancement.