Construction engineering in China under green Transition: Policy research Topic clusters and development forecasts

• Summarize topics in policy research directions for construction projects in China. • Clustering Chinese policy research topics using Latent Dirichlet Allocation. • Predicting policy research hotspots based on time series analysis. • Construction industry is more inclined to shift to sustainable construction. China’s construction engineering market is gradually transitioning towards high-quality development, driven by policies such as green and low-carbon. Identifying research topics for China’s construction engineering policy and predicting future research trends are crucial for policymakers to make informed decisions and promote the transformation of the construction industry. This study combines Latent Dirichlet Allocation (LDA) and Autoregressive Integrated Moving Average Model (ARIMA) to analyze more than 6,500 relevant literature abstracts. The LDA model identified 11 key topics, with the five hottest topics conceptualized with trend prediction. ARIMA-based trend forecasts show that these topics will continue to grow over the next two years. Energy Consumption in Construction (Topic 2) is caused by the urgency of transforming China’s construction industry. The environmental impacts of green construction (Topic 6) are due to the sustainability potential of the green construction model. The relationship between urbanization, the economy and the environment (Topic 7) was prompted by the discussion that the relationship between them is still unclear. The use of recyclable materials in construction (Topic 8) is a result of resource depletion and environmental degradation. The relationship between urbanization and ecosystems (Topic 9) was prompted by the environmental problems associated with China’s high rate of urbanization. These findings reveal the focus and development trend of construction engineering transformation in China and provide feedback to policy makers and companies from the perspective of experts and scholars. The accuracy of the predictions can be further improved in the future through data updates and multi-modeling combinations.