Energy policies are vital tools used by countries to regulate economic and social development as well as guarantee national security. To address the problems of fragmented policy objectives, conflicting tools, and overlapping initiatives, the internal logic and evolutionary trends of energy policies must be explored using the policy
content. This study uses 38,277 energy policies as a database and summarizes the four energy policy objectives:
clean, low-carbon, safe, and efficient. Using the TextCNN model to classify and deconstruct policies, the LDA +
Word2vec theme conceptualization and similarity calculations were compared with the EISMD evolution
framework to determine the energy policy theme evolution path. Results indicate that the density of energy
policies has increased. Policies have become more comprehensive, barriers between objectives have gradually
been broken, and low-carbon objectives have been strengthened. The evolution types are more diversified,
evolution paths are more complicated, and the evolution types are often related to technology, industry, and
market maturity. Traditional energy themes evolve through inheritance and merger; emerging technology and
industry themes evolve through innovation, inheritance, and splitting. Moreover, this study provides a replicable
analytical framework for the study of policy evolution in other sectors and evidence for optimizing energy policy
design
