DTM-Based Analysis of Hot Topics and Evolution of China’s Energy Policy

Abstract

Quantitative research on the evolution and transformation of topics in China’s energy policy can enhance the theoretical and methodological framework of policy document analysis. Utilizing dynamic topic modeling (DTM) and social network analysis, this study examined 1872 energy policy documents issued in China between 1980 and 2023, focusing on detecting hot topics and analyzing trend evolution. DTM identified five core topics: State Grid and new energy, comprehensive energy conservation and emission reduction, intelligent building energy management, promotion of energy-saving products and new energy vehicles, and standardization of energy industry management. Temporal analysis of these core topics reveals a shift in policy focus over time, moving from infrastructure development and standardization management to new energy development and modernization of the energy system. The co-occurrence network of thematic terms transitions from an “independent and loose” structure to a “concentrated and balanced” one, with increasing network scale and frequency. The conclusions of this study offer valuable insights for establishing a dynamic monitoring and real-time updating mechanism for energy policies, enhancing the integration and coordination of energy policy topics, and effectively supporting national energy strategies in response to global energy market challenges.

1. Introduction

Global climate change and the energy crisis are increasingly prominent issues, and the decoupling of economic growth from environmental pollution and the pursuit of low-carbon, green development have become universally recognized goals [1]. Climate change, fundamentally caused by greenhouse gas emissions from energy consumption [2], necessitates global efforts to adjust energy structures and enhance energy efficiency as key strategies for energy transition [3]. The shift from traditional fossil fuels to a clean, low-carbon, safe, and efficient energy system has risen to the top of the agenda for energy policymakers and stakeholders worldwide [4]. To address market failures and stimulate innovation in clean energy technologies, a variety of energy policy tools have been developed and implemented [5]. However, the dynamic energy environment, diverse regulatory frameworks, and variations in economic systems, resource endowments, and environmental awareness across countries [6] have led to challenges in harmonizing policy objectives, behaviors, and tools. Thus, it is crucial to quantitatively analyze the objectives and thematic evolution of existing energy policies, explore the interconnections between different policy themes [7], and uncover the underlying logic and values that drive these policies.

China, the world’s largest energy consumer and carbon dioxide emitter, is vigorously accelerating the pace of low-carbon energy transition to achieving its “carbon peaking and carbon neutrality” goals [8]. However, due to the tightening constraints on traditional energy resources and the difficulties in developing renewable energy, China’s energy transition policy framework must balance energy security, regional energy poverty alleviation, and global climate change mitigation [9]. Comprehensive policies such as the “Energy system revolution action plan”, “Action plan for further supporting energy development in poverty-stricken areas to promote poverty alleviation (2018–2020)”, and “Opinions on improving institutional mechanisms and policy measures for green and low-carbon energy transition” have been implemented to drive the low-carbon transformation of the energy system through coordinated efforts at multiple levels [10]. Analyzing the key themes within these energy policies helps to clarify the processes of policy formulation, evolution, and dissemination, summarize the focus areas and characteristics of each stage, and outline scenarios for interdepartmental collaboration and coordination, thereby providing insights into future policy trends [11].

This study utilizes 1872 energy policy texts issued in China from 1980 and 2023 using DTM and other quantitative text tools to segment the texts and extract comprehensive energy policy themes. We conduct comparative analysis and assess the popularity of these themes. Additionally, social network analysis (SNA) is then used to map the co-occurrence of thematic terms across different development stages of China, identifying the evolution patterns and forecasting trends in energy policy themes.

This study offers several innovations compared to previous research on energy policies. First, it utilizes policy text coding strategies and machine learning techniques to construct an extensive and comprehensive panel dataset of Chinese energy policies, addressing limitations in qualitative analysis [12] and case studies [13] that often fail to fully capture the underlying value logic and evolutionary dynamics of energy policies. Second, the use of DTM for topic modeling and trend analysis enables precise identification of key energy policy themes, calculation of theme popularity scores, and generation of corresponding heat maps. This approach overcomes the limitations of latent Dirichlet allocation (LDA), which is typically restricted to static analyses and unable to dynamically assess the theme-word matrix, word weights, and changes in thematic prominence [14], thereby providing deeper insights into the evolving roles of core themes over time. Third, the SNA is employed to extract key themes from China’s energy policies and build a co-occurrence matrix, drawing network maps of policy themes at different stages of development. These maps visually depict the collaborative evolution of themes, identifying key nodes and weak links within the network [15]. Furthermore, the findings offer valuable insights for optimizing the objectives of China’s current energy policies and enhancing coordination among various policy tools, providing guidance for the scientific formulation of energy industry policies [16] and contributing to the collaborative governance of energy management strategies.

This article is structured as follows. Section 2 provides an overview of the current research status on energy policies. The materials and methods are described in Section 3. Section 4 analyzes the results. Section 5 comprises a discussion and conclusion based on the research findings and puts forward policy implications.

2. Literature Review

Currently, scholars focusing on the logic and evolutionary trajectory behind energy policies mainly concentrate on three aspects. First of all, in policy theory analysis, scholars primarily use qualitative analysis and case studies to conduct political reviews of the origins of energy policies, such as their legal bases, attributes, and effectiveness [17]. They also examine international energy policy models by referencing practices from the European Union, the United States, France, and other regions [18,19]. These studies compare and draw lessons regarding the specific roles and impacts of energy policies on economic development, environmental protection, and climate change [11,20]. Scholars identify issues in policy elements, behavioral patterns, and implementation paths, making real-time adjustments to policy directions and objectives while proposing improvements to the policy framework and enhancing implementation efficiency [21,22]. For instance, progressive feedback mechanisms in energy policy and the use of carbon taxes are suggested as solutions to overcoming carbon lock-in and promoting renewable energy technologies [23,24]. However, such studies often neglect the continuity and synergy in the evolution of energy policies [25], leading to partial conclusions.

In addition, in terms of research methodology, to balance the differing interests of multiple stakeholders within the energy policy framework, evolutionary game theory can focus on the conflicts of interest and implementation barriers faced by policy implementers, while fuzzy consistency analysis can enhance the effectiveness of policy tools [26,27]. For quantitative analysis of energy policies, techniques such as co-word analysis and cluster analysis can be employed to identify policy objectives and core concepts [28]. These methods allow for a quantitative assessment of energy security policies, highlighting regional differences and preferences in policy tool usage, thereby providing a basis for optimizing the energy security policy framework [29]. Using keyword co-occurrence and panel VAR models, the interactions between policy themes are clearly delineated, with significant positive mutual influences observed in the “energy–water” policy domain [30]. Cluster analysis facilitates multidimensional quantitative analysis of energy policy texts, uncovering trends in the intensity and breadth of policy attention, as well as clarity of objectives and the coverage of policy tools [31,32]. To analyze the network structure of energy policies and the evolving roles of implementing entities, SNA is useful in examining the relationships among policymakers, implementers, beneficiaries, and other stakeholders, providing insights into the social structure and dynamic changes in energy policy. For example, using CiteSpace knowledge mapping and cluster analysis to explore the synergy effects of renewable energy policies reveals that synergy networks and innovation are critical drivers of energy industry development [33]. This approach can also be applied to analyze the temporal evolution of cooperation networks within the new energy vehicle industry [34] or to explore the spatial correlation network of China’s interprovincial energy consumption, describing the overall network density and the evolving roles of different provinces [35].

Finally, in evaluating the effectiveness of policy implementation, the success of energy policies relies on support from both the energy sector and the general public. Assessing the effectiveness and fairness of these policies is crucial for reducing fossil fuel consumption and fostering nationwide participation in green and low-carbon initiatives [36]. To build an efficient and secure energy system based on renewable energy, scholars have focused on evaluating the impact of green energy pilot policies on renewable energy use and environmental pollution [37], with the goal of achieving net-zero emissions by 2050 and limiting global warming to below 1.5 °C [38]. Additionally, some scholars have examined decarbonization effects in the industrial, transportation, and construction sectors from the perspective of innovation ecosystems [39]. These studies emphasize the reorganization and decarbonization of the energy system and advance innovative technologies for the development and utilization of low-carbon energy. When it comes to the dynamic evaluation of both single policies and policy combinations, the academic consensus is that combined policy effects are more effective in addressing challenges within the energy industry [40]. Such policy combinations tend to be more strategic and systematic, offering greater synergistic effects [41].

In summary, while existing research has explored the effects and impacts of energy policies from various perspectives and methods, there are still some shortcomings. Firstly, regarding the content of energy policies, current studies primarily focus on theoretical analyses related to policy implementation, issues, and optimization, often neglecting the quantitative evaluation of policy content. Secondly, in terms of research methodology, most studies employ comparative analysis, text analysis, and social network analysis to describe the network characteristics of policy implementation entities, lacking analysis of the evolution of policy themes and network synergy over time. Moreover, the application of text mining methods, such as DTM, in energy policy research is still insufficient. In response to these gaps, this study focuses on energy policy texts, utilizing DTM models, topic heat calculations, and social network analysis to explore the thematic evolution trends in and dynamics of China’s energy policies. The study aims to identify the key areas and critical issues in the formulation and implementation of energy policies across different periods, analyze the evolutionary paths and strategic logic behind these policies, and offer optimization suggestions for future energy policy development in China, contributing to the realization of the “dual carbon” goals.

3. Research Methods and Materials

3.1. Research Methods

Topic mining employs various algorithms to identify and extract research themes from sources like academic papers, policy documents, and e-books, enabling a comprehensive understanding of current research status and trends in the field. While models such as ATM, LDA, and BiLSTM are commonly utilized, DTM extends the capabilities of the LDA model by incorporating temporal analysis. Unlike other models, DTM can analyze how the distribution of topics and words evolves over time. The model divides the text into multiple time intervals, assuming that while the overall dimensional distribution of topics remains consistent, the distribution of topics and associated words within each interval varies over time. By integrating steps such as text preprocessing, stop-word removal, text vectorization, time segmentation, and model training, DTM introduces a temporal dimension into topic modeling, allowing for the dynamic tracking of topic and word distribution changes over different time periods [42]. In energy policy analysis, DTM is particularly effective in capturing the dynamic shifts in policy themes, uncovering the logical progression and structural changes within policies [43]. Furthermore, DTM aids in discerning policymakers’ intentions and the selection of policy instruments. By tracking the evolution of topics over time, it provides a deeper insight into adjustments in policy goals, thereby offering a scientific foundation for crafting more effective policies. This makes DTM an invaluable tool for analyzing and understanding long-term policy developments [15].

3.1.1. Topic Popularity Calculation

The support metric is primarily used to explain the characteristics of topic intensity and reflects the level of attention a topic receives during a specific period. Using DTM, a document-to-topic matrix can be obtained, which enables the calculation of the probability that each document belongs to a specific topic. Following the approach of Abdullah et al. [44], this study used a 10% threshold to calculate topic intensity based on the support metric. Once the topics have been identified and their intensity calculated, an intensity threshold must be set. Comparing the topic intensity to this threshold allows for the identification of hot topics that capture the attention of scholars.

3.1.2. Social Network Analysis

SNA is a tool used to study social structures by analyzing the relationships (referred to as edges) between entities (referred to as nodes) to construct a network. This method quantifies and describes the relationships between nodes, revealing the characteristics of social structures [45]. SNA is mainly used to predict the decisions and behaviors of nodes within a network and to examine how interactions between nodes form and evolve. It also uncovers interaction patterns among individuals or groups within a system, analyzes the underlying reasons for these patterns, and attempts to identify the impact and outcomes of these structural patterns [46]. In this study, Python was primarily used for DTM modeling to dynamically analyze the evolution of energy policy topics and construct a co-occurrence matrix of these topics, and Gephi was utilized to visualize the collaborative network among policy topics.

3.2. Research Materials

3.2.1. Data Acquisition

The policy document data for this study were sourced from the Peking University Law (PKULaw) database (https://www.pkulaw.com/ (accessed on 1 March 2024.)), an authoritative Chinese legal retrieval system designed to provide convenient, accurate, and comprehensive legal information services [47]. Since its establishment in 1949, the system has accumulated around 760,000 entries, including laws, regulations, departmental rules, and local regulations. The data collection processed as follows: First, the “Energy” category was selected under the “Regulation Category” section of the retrieved documents were focused on energy-related laws and policies. The search was confined to documents issued from the “Report on strengthening energy conservation work” in 1980 to the end of 2023, yielding 1970 policy documents. To enhance the accuracy and reliability of the analysis, these documents were cross-verified using energy policies from the Wanfang Data platform [48]. Furthermore, keywords related to the intensity, objectives, and measures of energy policies were expanded based on the policy coordination and evolution evaluation model of Peng et al. [49]. After eliminating duplicates and documents with weak relevance or impact, 1872 documents were finalized for analysis. The distribution of documents by year of publication is shown in Figure 1. This rigorous process ensured that the analyzed documents were highly relevant to energy policy themes and possessed significant policy influence, providing a robust foundation for subsequent dynamic topic modeling analysis.

The analysis of effectiveness levels across different types of China’s energy policy documents highlights the diversity of the nation’s energy policy landscape, as demonstrated in

Table 1. Number of policies at various effectiveness levels.

Effectiveness Level	Number of Policies	Summary of Content
Departmental Working Documents	1138	Issued by various national ministries, such as the National Development and Reform Commission, Ministry of Industry and Information Technology, and Ministry of Finance. Focus on energy conservation, emission reduction, clean energy promotion, industrial safety, and environmental protection, aiming to advance green, low-carbon development and efficient resource utilization through targeted management and technical guidance.
Departmental Normative Documents	537	Issued by institutions like the National Energy Commission and Administrative Bureaus of the CPC Central Committee and the State Council. Cover regulations on energy-saving loans, surcharges for exceeding fuel quotas, tax treatments, standards for energy projects, and infrastructure funding, aimed at strengthening energy management and ensuring efficient and equitable energy use.
Ministerial Regulations	70	Include regulations on energy management and monitoring in sectors like energy and transportation. Key issuers include the Ministry of Energy and Ministry of Transportation, focusing on energy efficiency in power grids, transportation, and building sectors, promoting sustainable development through regulation and technical standards.
State Council Normative Documents	56	Broad energy and environmental policy guidelines covering oil consumption, coal utilization, biogas projects, small refinery management, energy saving, emission reduction, and new energy vehicle promotion. Jointly issued by the State Council and related departments, these documents guide energy use and environmental protection efforts.
Administrative Approvals	25	Issued by the National Energy Administration and Ministry of Industry and Information Technology, covering approvals for standardization committees, pilot projects, and safety registrations, ensuring standardized and safe industry practices.
Laws	15	Enacted by the Standing Committee of the National People’s Congress, including key laws like the Coal Law, Energy Conservation Law, Cleaner Production Promotion Law, Renewable Energy Law, and Circular Economy Promotion Law, ensuring comprehensive legal frameworks for energy management and environmental protection.
Others	36	Include administrative and group regulations, addressing standards and safety in power, coal, and gas industries, along with operational and oversight roles in national policies and resource conservation.

. Departmental working documents, primarily issued by various national ministries and commissions, emphasize energy conservation, emission reduction, and the promotion of clean energy, showcasing the government’s targeted efforts in advancing green, low-carbon development and resource efficiency. Departmental normative documents and regulations take a more formal approach to energy management, focusing on areas such as funding for energy infrastructure and energy-saving monitoring. Normative documents from the State Council provide overarching guidelines for energy use and environmental protection, underscoring the importance of optimizing the energy mix and safeguarding the environment. Administrative approvals play a critical role in ensuring the standardization and safety management of energy and industrial projects through rigorous approval processes. Legal documents, enacted by the Standing Committee of the National People’s Congress, reinforce the authority and comprehensiveness of energy-related laws, covering a broad spectrum from the Coal Law to the Renewable Energy Law. Additionally, industry regulations and group standards highlight the involvement of professional organizations and authoritative institutions in fine-tuning the implementation and oversight of energy policies. Collectively, these multi-tiered policy documents form a comprehensive framework of China’s energy policy, offering crucial guidance for the sustainable development of the energy sector and the rational utilization of resources.

3.2.2. Sample Stage Division

The Five-Year Plans are foundational documents formulated every five years by the Chinese government to guide economic and social development. Each plan is shaped by its distinct historical context and development objectives [50], reflecting the strategic trajectory of China’s growth during different periods. Each Five-Year Plan also outlines specific energy policies, and by analyzing these—especially those concerning energy structure adjustments, technological innovations, and efficiency improvements—one can comprehensively explore the evolution of China’s energy policy themes [51]. Based on adjustments to energy policies and critical milestones within the Five-Year Plans, China’s energy policy development can be categorized into four stages, as follows.

Energy Strategy Initiation Stage (1980–1990): This phase includes the Sixth and Seventh Five-Year Plans. The 1980 issuance of the “Report on strengthening energy conservation work” marked the beginning of the government’s emphasis on addressing inefficient energy management and significant energy wastage. This led to the adoption of effective measures to enhance energy management and conservation, thereby improving energy utilization efficiency. The introduction of various policies, such as the “State Council directive on the development of coal washing and processing and the rational use of energy”, provided actionable guidelines for the efficient exploitation and conservation of traditional energy sources like coal and oil. During this stage, the focus of traditional energy policies was relatively narrow.

Energy Development Acceleration Stage (1991–2005): Encompassing the Eighth Five-Year Plan to the Tenth Five-Year Plan, this stage saw a significant expansion in the scope and variety of energy policies. The technological advancement in the energy industry was noteworthy, with improvements in energy security metrics and equipment technology. However, the focus on energy management largely remained on coal production and processing enterprises, although there was a noticeable increase in the proportion of hydropower, nuclear power, and wind power enterprises within the new energy sector. This stage marked a diversification in the topics and goals of China’s energy policies.

Critical Energy Transition Stage (2006–2015): Covering the Eleventh Five-Year Plan to the Twelfth Five-Year Plan, this period coincided with China’s critical phase of industrialization and urbanization, which required substantial energy support for urban development and improving living standards. Faced with emerging energy supply challenges, the government prioritized the development of renewable energy, improvement in energy efficiency, and optimization of the energy structure as major strategic goals. The implementation of the Renewable Energy Law of the People’s Republic of China in January 2006 aimed to remove legal and policy barriers hindering the development and utilization of renewable energy. During this stage, policy goals and priorities gradually shifted towards the development, utilization, and management of renewable and new energy sources.

Modern Energy System Optimization Stage (2016–2023): This stage includes the Thirteenth Five-Year Plan and the Fourteenth Five-Year Plan. Guided by the energy strategy of “Four Revolutions, One Cooperation”, the primary objectives of energy policies during this period were to promote clean energy use, improve energy efficiency, and optimize the energy structure, aiming to establish a modern, green, low-carbon, efficient, and secure energy system. The implementation of the Energy Production and Consumption Revolution Strategy (2016–2030) accelerated the diversification of energy supply and governance, laying a solid foundation for a fundamental transformation in energy production and consumption practices. This stage also saw an increase in policies related to environmental protection, carbon emission reduction, carbon neutrality, and energy conservation. The collaborative and networked nature of energy policy implementation became more pronounced, with an emphasis on achieving win–win outcomes and a growing focus on energy security, energy-saving practices, and ecological civilization.

3.2.3. Text Data Preprocessing

Topic modeling is widely recognized as a powerful statistical tool for uncovering latent variables within large unstructured text datasets [52]. Before constructing the topic model, it was essential to systematically preprocess the 1872 text documents, as detailed in Figure 2.

Obtaining and Cleaning Energy Policy Texts. Using Python, we trawled energy policy texts and information from PKULaw, extracting policy documents containing the keyword “energy”. We matched these with policies from the National Energy Administration and the Ministry of Ecology and Environment, manually removing policies with low relevance to “energy” to obtain cleaned policy information and text data.

Further Processing and Vectorizing Policy Texts. The cleaned text data were converted into a unified format based on the fields of title, publication date, issuing department, and main text. We then performed tokenization, lemmatization, word frequency filtering, and temporal slicing on the data, generating time-stamped keywords. Using the corpora module in the Gensim library and the bag-of-words model, we represented the tokenized texts as word frequency vectors.

Identifying Topics Based on DTM. Python programs were used to traverse the corpus and terms under each topic, generating a topic-term matrix. We conducted DTM temporal analysis and modeling, combining word2vec embedding technology with the softmax output of neural networks to perform topic semantic calculations, constructing a topic model and identifying all the topics contained in the text.

Identifying Topic Evolution Patterns. We used knowledge mapping software such as Gephi 0.10.1 to draw co-occurrence network maps of topics at different stages, conducting visual analysis of policy topic evolution.

3.2.4. Determining the Optimal Number of Topics

In the quantitative analysis of 1872 policy texts, determining the optimal number of topics is crucial, as it directly influences the interpretability and practical utility of the topic model. This study employed a coherence score to assess the performance of the model under different topic numbers. The coherence score reflects the statistical correlation among the words within each topic. A higher coherence score indicates that the model produces more statistically cohesive and closely related topics, enhancing the clarity and comprehensibility of the topics. However, “Remember that, while popular, topic coherence scores are not an absolute indicator of topic quality” [53]. Due to challenges in controlling for assumptions about key parameters and potential biases in the data sources, “… some techniques are also not representative of real-world data relationships. This is due to assumptions regarding key parameters in the calculation process and inefficiency of many optimization methods, which often attempt to overcome uncertainty by performing many time-consuming iterations to determine the best value for the parameter” [54]. As shown in Figure 3, when the number of topics is set to 5, the average coherence score across four different time periods reaches its peak (0.504), indicating that this configuration yields the best performance in terms of the language model. Consequently, we selected 5 topics for in-depth analysis to ensure the accuracy and reliability of the findings.

Using Python’s library Gensim 4.3.1, we trained the LDA model on the preprocessed policy texts and visualized the LDA topic model using the library pyLDAvis 3.4.0. With num_topics = 5, Figure 4 demonstrates clearly defined topics and effective classification.

4. Results

4.1. Horizontal Topic Analysis

This study utilized DTM to derive the “topic-word matrix and word weights” for energy policies across different stages. We organized the 10 highest-probability characteristic words for each topic to identify the most representative topic labels, as shown in

Table 2. Keywords for energy policy topics.

ID	Topics	Keywords
1	State Grid and New Energy	renewable energy, dispatched agency, group companies, new energy, State Grid, distributed, southern grid, solar energy, wind farms, energy industry
2	Comprehensive Energy Conservation and Emission Reduction	energy saving and emission reduction, dissemination week, water-saving, energy saving, implementation plan, seriously implement, transportation, water conservation, overachievement, storage battery
3	Intelligent Building Energy Management	building energy conservation, energy conservation, energy saving and emission reduction, informatization, energy management, comprehensive utilization, special device, service companies, energy-saving products, mechanical and electrical equipment
4	Promotion of Energy-Saving Products and New Energy Vehicles	energy-saving products, new energy, passenger cars, transportation, consumption, promotion and application, implementation rules, vehicle and vessel tax, financial subsidies, computers
5	Standardization of Energy Industry Management	standardization, energy industry, industry standards, management system, natural gas, washing machines, after review, implementation rules, energy-saving products, coal industry

. The topic distribution revealed that at each stage, China’s energy policies encompass a variety of fields, including State Grid and new energy, comprehensive energy conservation and emission reduction, intelligent building energy management, promotion of energy-saving products and new energy vehicles, and standardization of energy industry management. This comprehensive and systematic approach underscores China’s strategic planning in energy application and development. Examining the relationships between topics, it is evident that some topics are closely interconnected or mutually influential. For instance, the regulation and implementation of energy policies and the standardization and implementation of energy-saving products are core drivers of energy management. The promotion of new energy and transportation, along with the marketization and standardization of renewable energy, form the foundation of the energy transition. The development and utilization of clean energy, coupled with energy conservation and emission reduction and building energy efficiency, serve as institutional guarantees for sustainable energy development. Meanwhile, energy-saving products and resource conservation, along with the implementation of policies and management of the energy industry, constitute the external environment for enhancing energy efficiency. This analysis highlights the logical and hierarchical structure of China’s energy strategy.

After analyzing different topics in the energy sector, we observed significant differences in the number of policy texts for each topic, as shown in Figure 5. Firstly, the “State Grid and new energy” topic had the most documents, with 572, highlighting the central role of the State Grid in energy transition and new energy promotion. Secondly, “Intelligent building energy management” had 481 documents, reflecting the crucial role of intelligent building technologies in improving energy efficiency and promoting sustainable development. Additionally, “Comprehensive energy conservation and emission reduction” and “Promotion of energy-saving products and new energy vehicles” had 333 and 279 documents, respectively, indicating that these areas are relatively new in policy exploration and practice or they hold a supplementary position within the overall energy strategy. Lastly, the “Standardization of energy industry management” topic had 207 documents, which, although fewer in number, are essential for ensuring the healthy and orderly development of the energy industry.

4.2. Temporal Analysis of Core Topic Popularity in Energy Policies

To reflect the focus of energy policies during different development stages, we calculated the core topic popularity for four stages and plotted them as a heatmap, as shown in

Table 3. Heatmap of core topic popularity across different stages.

Core Topic	Energy Strategy Initiation Stage	Energy Development Acceleration Stage	Critical Energy Transition Stage	Modern Energy System Optimization Stage	Topic Popularity
State Grid and New Energy	0.096	0.217	0.213	0.444
Comprehensive Energy Conservation and Emission Reduction	0.101	0.149	0.233	0.117
Intelligent Building Energy Management	0.485	0.41	0.291	0.148
Promotion of Energy-Saving Products and New Energy Vehicles	0.114	0.081	0.168	0.142
Standardization of Energy Industry Management	0.194	0.132	0.082	0.136

. The values in the figure represent the proportion of supporting texts for a specific topic at the current stage compared to the total number of texts. A higher value indicates greater popularity of the topic. The popularity values of each topic vary across the four stages, indicating shifts in the strategic focus of energy policies over time. Specifically, the details are as follows.

Energy Strategy Initiation Stage (1980–1990). During this stage, China was in the early stage of its economic reform and opening up, facing dual pressures of energy structure adjustment and economic development. At this time, the development and utilization of energy were first explicitly recognized as part of the national strategy. The most emphasized topic during this stage was “Intelligent building energy management”, with a popularity score of 0.485. This indicates that in the early stage of economic transition, policy focus was on improving building energy efficiency, likely related to large-scale urban construction and infrastructure development. The second-most significant topic was “Standardization of energy industry management”, with a popularity score of 0.194. This suggests that during the initial recognition of energy as a strategic resource, establishing industry standards and management systems was a key policy focus. In contrast, “State Grid and new energy” and “Promotion of energy-saving products and new energy vehicles” received relatively less attention. This was likely due to the focus at that time being more on the development of traditional energy and foundational energy infrastructure.

Energy Development Acceleration Stage (1991–2005). During this period, the focus of China’s socialist modernization shifted to enhancing the quality of economic development and optimizing the economic structure. The attention to “State Grid and new energy” rose significantly to 0.217, reflecting the increased demand for grid expansion and new energy sources such as hydropower and nuclear power due to rapid economic growth. Meanwhile, although the popularity of “Intelligent building energy management” decreased to 0.410, it remained high, indicating that energy-efficient buildings continued to be a policy priority. During this period, the extensive extraction and use of traditional energy sources like coal led to rising environmental issues, making energy conservation and emission reduction more prominent, with its popularity increasing to 0.149.

Critical Energy Transition Stage (2006–2015). Urbanization during this period led to a significant increase in buildings, causing energy demand to surge. The popularity of “Comprehensive energy conservation and emission reduction” rose sharply to 0.233, making it the second-most important topic after “State Grid and new energy”. This increase was a direct response to both domestic and international environmental pressures and the need for sustainable development. Concurrently, “State Grid and new energy” saw its popularity increase to 0.213, reflecting the government’s focus on diversifying energy supply and developing renewable energy. In contrast, the attention paid to the standardization management of the energy industry and intelligent building energy management declined, indicating a shift in policy priorities.

Modern Energy System Optimization Stage (2016–2023). In this latest stage, “State Grid and new energy” reached its peak popularity at 0.444, showing a concentrated investment in building a modern energy system, particularly in new energy technology research and application. However, the focus on “Comprehensive energy conservation and emission reduction” and “Intelligent building energy management” significantly declined, likely because foundational policies had already been established earlier or because the policy focus had shifted to broader energy structure adjustments and technological innovations.

Overall, these data show a gradual shift in China’s energy policy focus from early infrastructure construction and standardization management to later new energy development and comprehensive energy system modernization. This evolution reflects strategic adjustments made by the state in response to economic, environmental, and social needs at different development stages.

4.3. Analysis of Topic Evolution Paths

Based on the topic heatmap analysis results, this study outlines the heat values of core topics at each stage. To further examine the annual evolution of each topic (whether hot or cold), a regression analysis was conducted using the topic weight for each corresponding year as the dependent variable and time as the independent variable, employing the least squares method. The results are shown in

Table 4. Topic types and regression results.

Topic	Slope	p-Value	Type
State Grid and New Energy	0.0098	0.0276	Hot
Comprehensive Energy Conservation and Emission Reduction	−0.0068	0.0063	Cold
Intelligent Building Energy Management	−0.0111	0.0009	Cold
Promotion of Energy-Saving Products and New Energy Vehicles	0.0048	0.0096	Hot
Standardization of Energy Industry Management	0.0033	0.0556	-

, categorizing the topics as “hot”, “cold”, or having no significant change based on the p-value (significance level). Specifically, the topics “State grid and new energy” and “Promotion of energy-saving products and new energy vehicles” demonstrate positive slopes with statistical significance, indicating a significant rise in policy focus on these areas, and are thus classified as “hot” topics. In contrast, the topics “Comprehensive energy conservation and emission reduction” and “intelligent building energy management” show negative slopes with statistical significance, suggesting a decline in policy focus in these areas, leading to their classification as “cold” topics. Meanwhile, the topic “Standardization of energy industry management” does not exhibit a significant trend in either direction and therefore is not classified as either “hot” or “cold”. These findings offer policymakers and researchers valuable insights into how policy priorities shift over time.

To understand the evolution patterns and future trends of energy policy topics, we used Gephi software to create co-occurrence maps for energy policy keywords across four periods. Each node represents a keyword, and the lines between nodes indicate their co-occurrence within the same policy text. Thicker lines denote a higher frequency of co-occurrence. A visual analysis of policy topic evolution is presented in Figure 6.

Energy Strategy Initiation Stage (1980–1990). During this stage, energy policies primarily focused on “Intelligent building energy management” and “Standardization of energy industry management”, with frequent and strong connections between these topics. The government aimed to enhance energy efficiency and sustainability by optimizing building energy management and standardizing industry practices. “Intelligent building energy management” appeared frequently across policy texts, highlighting the government’s emphasis on improving energy management and reducing consumption. The “Comprehensive energy conservation and emission reduction” topic had weaker links to “Intelligent building energy management” and “Standardization of energy industry management”, suggesting that while broad in scope, its implementation relied heavily on specific measures outlined in specialized policies. The “Promotion of energy-saving products and new energy vehicles” topic was connected to “Intelligent building energy management”, reflecting a comprehensive approach to energy savings in buildings and transportation. The “State Grid and new energy” topic had limited presence, indicating that the focus remained on traditional energy development and infrastructure, with less emphasis on new energy and grid modernization.

Energy Development Acceleration Stage (1991–2005). In this stage, energy policies mainly focused on “Intelligent building energy management”, which had strong and frequent connections with the other topics, highlighting its central role in policy formulation and wide application. The “State Grid and new energy” topic emerged during this period, closely linked with “Intelligent building energy management” and “Promotion of energy-saving products and new energy vehicles” indicating the State Grid’s expanded role in promoting renewable energy in buildings and transportation. The weak connection between “Comprehensive energy conservation and emission reduction” and “Standardization of energy industry management” suggests that while standardization focused on macro-level industry practices, specific conservation measures relied on local implementation, leading to less effective outcomes. This reflects a disconnect between policy design and execution. The lack of connection between “Comprehensive energy conservation and emission reduction” and “State Grid and new energy” indicates that renewable energy development and grid modernization were not fully integrated into the conservation framework, revealing a limited perspective in policy priorities.

Critical Energy Transition Stage (2006–2015). During this stage, the connections between energy policy topics became stronger and more complex, marked by an increase in linkages and their intensifying strength. The size of the nodes in the graph indicates that “Intelligent building energy management” held a slightly more prominent position than the other topics, reflecting its continued policy support as a core area for energy conservation and emission reduction. The “State Grid and new energy” topic frequently connected with other topics, emerging as a crucial driver of the energy transition. The modernization and intelligent upgrade of the State Grid, especially through integrating renewable and distributed energy systems, enhanced the grid’s flexibility and sustainability. This shift improved energy supply stability and facilitated the widespread adoption of new energy sources like wind and solar power. The “Comprehensive energy conservation and emission reduction policy” also gained prominence, showing connections with all other topics, signifying its adoption as a common approach in policy formulation and implementation. The policy’s comprehensiveness and flexibility allowed it to better meet the specific needs of different industries and regions.

Modern Energy System Optimization Stage (2016–2023). Energy policies in this stage primarily focused on “State Grid and new energy”, “Intelligent building energy management”, and “Promotion of energy-saving products and new energy vehicles”, reflecting a balanced development strategy. These topics highlight the country’s commitment to energy efficiency and sustainable development, as well as the strengthening of cross-sectoral synergies. The role of “State Grid and new energy” was significantly enhanced through the integration and improvement of distributed energy systems, which increased stability and efficiency. This advancement supported “Intelligent building energy management” and “Promotion of energy-saving products and new energy vehicles” making energy supply more reliable and efficient. The State Grid played a key role in coordinating and distributing new energy sources such as wind and solar power. “Intelligent building energy management” remained central, optimizing energy consumption and enhancing building energy efficiency through digital and intelligent methods. The “Promotion of energy-saving products and new energy vehicles” also gained prominence, significantly reducing transportation sector energy consumption via advanced information technology and practical strategies. This promotion reduced reliance on fossil fuels, lowered environmental pollution, and fostered green transportation. Although the importance of “Comprehensive energy conservation and emission reduction” and “Standardization of energy industry management” declined, they continued to provide essential guidance and standards for energy-saving practices, ensuring sustainable development and improved resource efficiency.

Table 5. Characteristics of the network structure of China’s energy policy themes.

Indicator	1980–1990	1991–2005	2006–2015	2016–2023
Sample Size	72	142	924	733
Network Size	12	27	57	58
Number of Network Relations	27	91	616	547
Network Connection Frequency	165	560	8683	5720
Network Cohesion Index	0.13	0.106	0.108	0.126
Overall Network Density	0.409	0.259	0.386	0.331
Average Node Distance	1.591	0.254	1.638	1.717

outlines the characteristics of the network structure of China’s energy policy topics, comparing network metrics across different periods to trace the evolution and dynamics of these topics. The data reveal distinct trends over time in indicators such as sample size, network scale, number of connections, connection frequency, network cohesion, overall density, and average node distance. From 1980 to 1990, the network was relatively small, but exhibited high density, suggesting strong interconnections among early energy policy topics. From 1991 to 2005, the network expanded in scale, but its density decreased, indicating a dispersion of policy focus. The period from 2006 to 2015 saw a significant increase in both network scale and connection frequency, reflecting growing complexity and diversity in policy topics. From 2016 to 2023, the network structure continued to stabilize, with a slight decline in density and an increase in average node distance, suggesting that while connections became more widespread, they also became less direct. Overall, these metrics illustrate the evolving nature of China’s energy policy network, highlighting a trend towards greater diversity and systematization of policy topics over time, while also reflecting the deepening adjustments and optimizations made by policymakers in the energy sector.

5. Discussion and Conclusions

5.1. Discussion

Based on China’s energy policy texts from 1980 to 2023, this study used DTM and social network analysis to dynamically explore hot energy policy topics from three perspectives: hot topic-word matrix, topic popularity analysis, and topic evolution path. In the context of China’s current goals to achieve carbon peak and carbon neutrality (“dual carbon” goals) [2,37], studying the behavior of energy policy actors, the evolution of content topics, and the transformation of policy tools has significant practical value for overcoming resource and environmental constraints, solving energy-saving and low-carbon issues, and improving the implementation efficiency of China’s energy policies [8,16].

From the perspective of the coherence analysis of the model and the extraction of hotspot themes, the highest average coherence value for the number of different themes across the four stages is 0.504, indicating that the topic language model’s performance meets the research needs [54]. The ideal model depends on one’s dataset and intended overall purpose. We noted that the low coherence value is attributed to the large amount of noise from tens of thousands of energy-related terms when stop words are not completely selected, making effective clustering difficult. It may also be due to the insufficient capture of the number of hotspot energy policy terms. Therefore, future research should focus on improving the model’s coherence.

From the content dimension of the results, China’s energy policy topics are diverse, including six major categories such as comprehensive energy conservation and emission reduction. Unlike theoretical analyses that summarize the past forty years of China’s energy policy and propose the necessity of energy transition [50], the analysis of topic popularity in this text reveals a clear trend in the evolution of China’s energy policy topics, mainly in the directions of “low carbonization” and “intelligentization”. This aligns with the “Ten actions for carbon peaking” in the Action Plan for Carbon Peaking Before 2030, which emphasizes “promoting green and low-carbon energy transition actions, green and low-carbon technology innovation actions”, reflecting China’s willingness to comprehensively strengthen energy cooperation with countries worldwide, maintain energy security, address climate change, protect the ecological environment, and better benefit people around the world [39,47].

From the time dimension of the results, based on adjustments to energy policies and critical milestones within the Five-Year Plans, China’s energy policy development from 1980 to 2023 can be categorized into four stages: energy strategy initiation stage, energy development acceleration stage, critical energy transition stage, and modern energy system optimization stage. This classification aligns with earlier studies that analyzed changes in energy-related content within the Five-Year Plans, focusing on targets for energy efficiency or carbon intensity, the energy structure, and investments in renewable energy [20,50], reflecting the continuous deepening and upgrading of China’s energy industry [3,14]. The heatmap analysis results indicate that thematic heat values fluctuate at each stage. For instance, the value for “State Grid and new energy” was merely 0.096 in the first stage, but rose to 0.444 in the fourth stage, reflecting the dynamic adjustment and adaptability of China’s energy policies. These findings align with the conclusions of Li and Cheng [2] and Wang et al. [9]. Notably, this study employs the methodology of Jung and Kim [52], treating the theme weights for each corresponding year as the dependent variable and time as the independent variable in the regression analysis. This empirical approach highlights the transition of energy policy hotspot themes between “hot topics” and “cold topics” across various stages [52]. Furthermore, this method not only examines the evolution of policy hotspot themes over four stages but also captures the overall evolution of hotspot themes throughout the entire study period, effectively combining local and overall time series analysis. This enhances the comprehensiveness and robustness of the research findings.

The spatial dimension of the results aligns with previous studies by Wu et al. [32] and Peng et al. [49], indicating that China’s energy policy themes span the entire industrial chain. These themes provide comprehensive support and guidance in areas such as energy development, technological innovation, standard setting, market regulation, and fiscal and tax support, promoting the coordinated development of energy in China. Unlike Xu et al. [34], who focused solely on a network analysis of new energy vehicle policy themes, this study expands the scope to include a network spatial analysis of China’s energy policy themes. The network evolution diagrams of hotspot themes across four stages reveal evolutionary relationships such as emergence, splitting, merging, and disappearance, highlighting trends and directions between hotspot themes and addressing research gaps identified by Zou et al. [4] and Wang et al. [15]. Furthermore, the numerical values of network structure characteristics show that the overall network scale is continuously expanding, with a significant increase in network connection frequency in the latter two stages. Although overall network density and network cohesion index values have generally increased, they remain relatively low, indicating weak synergy between policy themes [47]. This finding is consistent with Chen et al. [7], who explored the connections between different policies and potential evolutionary directions. The loose connections and weak overall effects between policy themes may be attributed to a lack of coordination and communication between internal execution entities of energy policies and external environmental factors [10].

5.2. Conclusions

This study employed DTM for a cross-sectional analysis of energy policy topics. By calculating coherence scores to determine the optimal number of topics at each stage, five core topics were identified: State Grid and new energy, comprehensive energy conservation and emission reduction, intelligent building energy management, promotion of energy-saving products and new energy vehicles, and standardization of energy industry management. This approach not only elucidates the underlying logic of China’s energy policy evolution but also offers a quantitative tool for future policy formulation and adjustment, thereby deepening the understanding of the factors influencing energy policy changes.

Secondly, in the time-series-based analysis of the topic popularity of China’s energy policies, this study systematically plotted heat maps for four stages and conducted regression analysis on hot topics, thereby deeply exploring the functional evolution of core policy topics over time. It reveals how the focus of energy policy shifted from infrastructure construction and standardization management to new energy development and energy system modernization from 1980 to 2023. Particularly during the critical energy transition stage and the modern energy system optimization stage, the integration of new energy and the State Grid emerged as significant policy hotspots, reflecting the evolution of each topic (hot/cold) annually and indicating an increased emphasis on sustainable development and technological innovation. These findings provide a quantitative basis for understanding policy dynamics and offer references for the formulation and adjustment of future energy policies.

Finally, by applying Gephi knowledge graph software to analyze the evolutionary trends of the network structure of China’s energy policy topics, the study observes, from the co-occurrence analysis of thematic terms in policy texts, that the energy policy has gradually evolved from the initial “Intelligent building energy management” and “Standardization of energy industry management” to a more synergistic network encompassing “State Grid and new energy”, ”Intelligent building energy management” and “Promotion of energy-saving products and new energy vehicles”. This evolution forms an efficient and sustainable energy management network, where “Comprehensive energy conservation and emission reduction” and “Standardization of energy industry management” provide systematic guidance and regulation for the overall framework, ensuring the comprehensiveness and consistency of policy implementation. The calculation of the characteristics of the energy policy topic network structure further confirms the changes in the synergy and stability of the topic network structure. Over time, these policy topics have not only become richer in content but also more closely interconnected, reflecting the comprehensiveness and foresight of policy formulation. This cross-period topic evolution and interaction highlights the profound impact of policies in promoting technological innovation and environmental protection, providing valuable insights and experiences for the formulation of future energy policies.

5.3. Limitations and Avenues for Future Research

Although this study has achieved certain research outcomes, it also has some limitations. To address these, future research will focus on the following areas. First of all, the collection of energy policy texts will be expanded. Through the analysis of high-frequency words and topic identification, it is evident that the research scope of energy policies is relatively broad, and the keywords that reflect hotspots and themes are somewhat scattered, lacking a certain degree of systematicity. Therefore, based on existing research text materials, future studies will incorporate energy policy data from provincial levels and even other countries through web scraping to conduct comparative analyses and uncover more diverse and comprehensive hot topics.

Moreover, although this paper trained the LDA model on preprocessed policy texts and conducted a visual analysis of the LDA topic model using the pyLDAvis library, as well as performed robustness checks on the average topic coherence value, future work can employ latent semantic analysis (LSA) to improve the average topic coherence value of the model. LSA uses singular value decomposition (SVD) to reduce the dimensions of the document-term matrix generated in bag-of-words TF-IDF. It then calculates the cosine similarity between documents in this reduced matrix. By analyzing the frequency and co-occurrence of words, LSA determines the similarity between documents and generates keyword topic lists for document classification and sorting.

In addition, this paper primarily focuses on the hot topics of energy policies, but lacks discussion on the diversified policy implementation entities. The formulation and implementation of energy policies involve complex entities, including policy communities and intergovernmental networks, and their roles in governance networks also vary. Therefore, it is important to actively study the role evolution of diverse entities such as government, market, and society within the collaborative implementation network.

Finally, there is considerable room for exploration in the empirical analysis of energy policies. In the process of calculating policy popularity and analyzing social network characteristics, this study did not include their impacts on the economy, society, or technology within the analytical framework. Therefore, it is necessary in future research to conduct empirical analyses on the relationships between hot policy topics, implementing entities, and fields such as the economy and environment, further promoting studies on the synergy of entities in energy policies, environmental improvement, and high-quality economic development.

5.4. Countermeasures and Suggestions

Enhance dynamic monitoring and real-time updating mechanisms for energy policies. Given the distinct stages and topics observed in China’s energy policies from 1980 to 2023, advanced tools like artificial intelligence and big data should be used to continuously evaluate policy effectiveness and make timely adjustments based on changes in the energy environment. For instance, a dedicated policy evaluation team could be established to regularly assess current policies and adjust them as needed. This approach ensures that energy policies remain responsive to market and technological shifts, effectively supporting national energy strategies and addressing global challenges.

Foster integration and synergy among energy policy topics. Analysis using Gephi knowledge graphs has shown synergistic evolution among various energy policy topics. To build an efficient and sustainable energy management network, it is crucial to enhance policy integration, ensuring that different measures support and complement each other. Comprehensive policies should be developed, especially in areas like new energy integration with the State Grid, intelligent building energy management, and the promotion of energy-saving products and new energy vehicles. Policymakers should align and support various policies to form a cohesive system, optimizing overall policy effects and maximizing synergies.

Improve the systematic and forward-looking nature of energy policy. Given the long-term impact of energy policies and the rapid pace of technological innovation, policy formulation should be systematic and forward-thinking. Policymakers should regularly review global energy policy trends and anticipate future changes to better prepare for new challenges. Policies should provide unified guidance and standards across all aspects of energy policy through frameworks like “Comprehensive energy conservation and emission reduction” and “Standardization of energy industry management”. Additionally, transparency and public participation should be increased through open discussions and consultations, strengthening the social foundation and enforcement of policies to ensure coherence and continuity.