A Novel Framework for Identifying Hot Spots in Coal Research

Abstract

The global imperative for a low-carbon energy transition is prompting significant shifts in the coal industry, driving the need to identify and analyze emerging research hot spots in coal-related research. Traditional methods that rely on domain knowledge to identify hot spots may have limitations, such as time costs and incomplete coverage. Moreover, a comprehensive analysis of coal-related research has yet to be conducted. Therefore, in this paper, a novel framework consisting of the semantic part and the word frequency part is proposed to analyze hot spots of coal-related research. Initially, a dataset consisting of 40,120 coal-related paper information from the Scopus database was constructed. Then, the novel framework was employed to analyze coal-related research. In the semantic part, bidirectional encoder representations from transformers and K-means algorithms were combined to conduct the hot spot analysis, and six hot spots are obtained. In the word frequency part, the bag-of-words and the latent Dirichlet allocation algorithms were combined to conduct hot spot analysis, and six hot spots were obtained. Finally, through the framework analysis, this study found that the 12 coal-related hot spots mainly revealed four main research directions: efficient coal utilization and resource recovery, carbon dioxide capture and emission reduction, environmental impact assessment and pollution control, and coal mine safety and geological modeling.