Leveraging Text Mining and Network Analysis for the Diffusion of Agricultural Science and Technology Policies in China

Abstract

Agricultural science and technology policies (ASTPs) have played a pivotal role in shaping agricultural innovation, sustainability, and cleaner production practices. Understanding how ASTPs diffuse is essential for optimizing policy design and advancing the green transition in agriculture. This study aims to investigate the diffusion of ASTPs in China, using a quantitative citation-based approach. The goal is to explore diffusion patterns, topic characteristics, and historical trajectories of ASTPs, thereby providing insights into policy transmission mechanisms that can inform future policy improvements. We analyze 3207 ASTP documents, focusing on policy citation links to examine the distribution, diffusion characteristics, and dynamics of policies. The analysis includes tracking topic evolution and identifying key policies while estimating the main diffusion paths. The results show that the top-down diffusion model is the dominant pattern of policy transmission, exhibiting the highest diffusion speed and both short- and long-term impacts. ASTPs have progressively expanded toward industrialization, informatization, and green development, with increased policy transmission efficiency. The diffusion process has formed three primary pathways: (i) enhancing agricultural innovation capacity, (ii) accelerating the transformation of technological achievements, and (iii) improving the agricultural science and technology innovation system. These pathways are critical to advancing sustainable and cleaner agricultural production. This study provides valuable insights into the diffusion of ASTPs and highlights key pathways for policy optimization. The findings suggest that enhancing policy frameworks and improving policy implementation efficiency will be crucial for facilitating the transition toward sustainable, low-carbon, and environmentally friendly agricultural practices. Future research should refine data sources and incorporate semantic analysis to capture more detailed policy transmission mechanisms.

1. Introduction

Scientific and technological innovation is a vital foundation for economic development and a decisive force in maintaining a nation’s sustained competitive advantage [1,2]. In the process of promoting agricultural science and technology innovation, policy formulation and diffusion not only influence the transformation of agricultural production methods but also play a crucial role in building a cleaner production and sustainable agricultural system [3,4,5,6]. Cleaner production emphasizes reducing pollution emissions and improving resource use efficiency, and ASTPs play a key guiding role in this process [7,8]. By supporting the research and promotion of green agricultural technologies, such as precision agriculture, bio-based inputs, and low-carbon agricultural practices, ASTPs not only facilitate the modernization of agricultural production models but also effectively reduce the negative environmental impacts of agricultural production [9,10]. As China’s agricultural science and technology policies continue to evolve, their contribution to cleaner production goals has gradually become evident, including promoting green agricultural development, reducing the use of chemical pesticides and fertilizers, optimizing agricultural waste management, and advancing agricultural carbon reduction [11,12,13]. Therefore, systematically analyzing the diffusion process of ASTPs helps to understand how they promote agricultural green transformation and provides scientific evidence for optimizing policy design, thereby better supporting sustainable agriculture and cleaner production practices.

Over the last forty years, since the implementation of reform and opening-up policies, China has witnessed significant progress in agricultural science and technology. These accomplishments have clearly demonstrated that the key to success lies in the innovation-driven development of agricultural science and technology [14]. Agricultural science and technology innovation play a crucial role in promoting the transformation of traditional agriculture into modern agriculture, as well as accelerating the shift from extensive to intensive agricultural development. Particularly as China enters a new phase of expanded openness, the development of agricultural science and technology has transitioned from a “crossing the river by feeling the stones” approach to one driven by innovation based on top-level institutional design. This shift places higher demands on the country’s capacity for top-level policy design, with ASTPs’ formulation serving as a critical component of this process [15]. As a key component of science and technology policy design, policy diffusion reflects the process and direction of policy learning. It plays a significant role in enhancing the efficiency and effectiveness of ASTPs, thus fostering the prosperous development of agricultural science and technology. Policy diffusion refers to the process by which a policy is adopted and implemented by new public policy actors [16]. As a theoretical framework within policy process studies, its core value lies in explaining how policies are transmitted and adopted. This is particularly relevant given the ongoing deepening of China’s agricultural science and technology reforms and the frequent diffusion of related policies [17]. Tracking the historical development and propagation of ASTPs reveals transmission trends and informs more effective policy design.

To date, research on ASTPs diffusion has been primarily conducted from three aspects. First, the measurement of policy diffusion. Most studies typically use national-level ASTPs as research objects and base the analysis of policy diffusion on the sequence of ASTPs’ promulgation [18]. This approach implies that the diffusion of ASTPs is solely dependent on time. However, policy diffusion is not a random process. Particularly within the context of China’s unitary system, disparities in the power hierarchy play a crucial role in driving policy adoption [19]. To quantify the extent of policy diffusion, some studies employ various indicators to calculate the differences in the number of policies adopted at different administrative levels [20]. These indicators tend to be relatively independent and cannot be effectively linked, thus limiting our comprehensive understanding of the complex mechanisms underlying the diffusion of ASTPs [21]. In recent years, some scholars have attempted to explore the differential characteristics in the process of policy diffusion by examining the coordination and interaction among various policies [22].

Second, the topic characteristics of ASTPs diffusion. Current research predominantly emphasizes topic evolution by focusing on the semantic comprehension of abstract texts and capturing the associated temporal trends [23,24]. However, this line of inquiry often neglects the interrelationships between policy topics, which can directly reflect policy implementation. Therefore, recently, some scholars have attempted to utilize text—similarity algorithms to investigate degrees of topic coordination throughout the policy diffusion process [25]. They have calculated metrics such as topic inheritance ratio, diffusion ratio, and innovation ratio in order to elucidate how well the central spirit is implemented locally [26]. Nevertheless, due to the inherent limitations of text—analysis methodologies, where only single-topic relationships can be measured—it remains essential for future research endeavors to further uncover complex structures and mechanisms capable of coordinating multiple interrelated topics [27].

Third, mapping the path of ASTPs diffusion. Scholars frequently employ qualitative methods such as event history analysis and case studies to examine the developmental context, ideological framework, and construction processes of ASTP systems [28]. These studies converge on a common conclusion: the characteristics of national agriculture and rural areas during different periods significantly determine the quantity and structure of ASTPs at those periods. The objectives and measures associated with these policies in each stage are highly consistent with the strategic goals for national agriculture and rural development during the corresponding periods. Such findings offer valuable insights into understanding the evolution and underlying logic of ASTPs. In recent years, as demand for agricultural science and technology has continued to grow, there has been a substantial increase in available textual information regarding policies for researchers [29,30,31]. Consequently, the framework surrounding these policies has become increasingly complex, making it challenging for qualitative methods, which focus on subjective interpretation and qualitative analysis, to analyze a large amount of policy literature to generate objective and representative insights. Quantitative methods have thus become a new trend in current policy diffusion research [32,33].

Although the above literature has conducted extensive research on ASTPs diffusion in terms of diffusion measures, thematic differences, and diffusion paths, there are still some deficiencies and room for extension: (i) Existing studies mostly assess the probability of policy diffusion events based on experience or the order of time release. The vague definition of policy adoption fails to thoroughly investigate the overall distribution and hierarchical characteristics of policy diffusion [34,35]. To address this gap, this paper innovatively adopts a quantitative policy citation-based approach to construct a policy diffusion citation network, comprehensively observe the citation distribution, measure the speed and impact of policy diffusion from temporal and hierarchical perspectives, thereby providing a more precise and data-driven understanding of the ASTPs diffusion model. (ii) Regarding the thematic evolution of ASTPs diffusion, it is essential not only to comprehend the evolution and life cycle of themes but also to discern their intercorrelation characteristics. Most studies have employed the topic model approach to identify singular topic relationships, which cannot comprehensively examine the thematic dynamics of policy diffusion. By applying policy citation network analysis in conjunction with a network modularization algorithm, we can explore the topic structure and location distribution of policy diffusion globally and dynamically. (iii) In analyzing the diffusion pathways for ASTPs, the previous literature predominantly relies on qualitative methods. Given that such policies encompass multiple focal points, there is a need for improved representativeness and objectivity in qualitative research conclusions that primarily depend on subjective judgment or expert experience [36,37]. Thus, we introduce a main-path approach to quantitatively and visually track the diffusion trajectories of ASTPs.

In this context, this study aims to explore the diffusion patterns, topic dynamics, and structural trajectories of agricultural science and technology policies (ASTPs) in China by constructing a citation-based policy diffusion network.

The remainder of this paper proceeds as follows. Section 2 details data collection and research methods. Section 3 presents empirical results, covering the overall pattern characteristics, topic dynamics, and historical trajectory of ASTPs diffusion. Section 4 conducts a comprehensive discussion. Section 5 combines the discussion of policy implications, summarizes conclusions, and outlines directions for future research.

2. Materials and Methods

This study adopts a quantitative research design grounded in policy citation data analysis. Given the scale and complexity of China’s agricultural science and technology policies (ASTPs), a quantitative approach enables us to systematically capture structural characteristics, diffusion patterns, and topic evolution across large volumes of policy documents. It facilitates the objective measurement of inter-policy relationships, diffusion trajectories, and thematic dynamics.

Drawing upon theories of social networks and knowledge diffusion, we propose a research framework based on citation relationships to analyze the diffusion patterns and evolutionary processes of ASTPs in China, as illustrated in Figure 1. Firstly, we organize and archive the policy texts, extracting metadata such as issuing agencies, implementation dates, and validity periods. Secondly, we identify citation relationships by recognizing quoted titles within the main text of the documents and use this information to construct a policy citation network. Then, we uncover diffusion patterns, topic dynamics, and developmental paths of ASTPs by applying network topology indicators, policy diffusion metrics, modularity-based clustering algorithms, and the Search Path Link Count (SPLC) method.

2.1. Data Collection

Policy data for this study were primarily collected from the Peking University Law Database (www.pkulaw.com, accessed on 3 January 2024), which covers all laws and regulations issued in China from 1949 to the present and is widely recognized as the most comprehensive and authoritative legal and regulatory database in the country. We began by searching the legal and regulatory section of the database using the keyword “agricultural science and technology”, with a time span from 1978 to 2023. The initial dataset included metadata fields such as serial number, policy title, date of issuance, and issuing authority.

To ensure the relevance and consistency of the dataset, we performed a multi-step filtering process. First, we manually removed irrelevant or non-normative documents, including project approvals, announcements, planning documents, judicial interpretations, and internal work reports. We retained only those documents with explicit policy intent, such as regulations, guidelines, opinions, implementation measures, notices, decisions, and resolutions. Next, we applied a content relevance filter to exclude documents unrelated to agricultural science and technology, such as personnel appointments, reward and punishment notices, and digital formatting standards. After completing these screening procedures, we obtained a final dataset of 3207 valid ASTPs and 3776 citation links, which served as the basis for the analysis in this study.

2.2. Policy Citation Network Construction

Policy citation analysis, rooted in bibliometric research, serves as an effective means of tracking developments in scientific thinking and technological innovation [38,39]. Similar to the references in scientific papers, citations also exist among policy texts. A policy citation occurs when a policy references other policies as its guidance, reference standards, and implementation rules [40]. Policy citations embedded in policy documents represent a tangible manifestation of political value transmission and the diffusion of ideas. They not only reflect the foundations and origins of policy formulation but also illustrate the continuity, developmental context, and mutual influences between central and local policy institutions [41]. Unlike a traditional quantitative approach, which often relies on temporal evolution to reveal policy diffusion, policy citation analysis quantitatively records diffusion footprint by citing/cited links among policies, discerns detailed policy diffusion relationships by grasping the preference of government behavior, and ultimately achieves a comprehensive understanding of the logic of policy diffusion.

The policy citation network is defined as N = (U, R), where node U represents policy documents, the size of the node indicates the frequency of citations, and edges between nodes represent the citation relationships among policy documents. The direction of the edges indicates the direction of policy diffusion (i.e., from the cited policy to the citing policy). Figure 2a illustrates the seven citation relationships from Policy A to Policy G, while Figure 2b depicts the corresponding policy citation network based on Figure 2a. The nodes represent policy documents, and the lines between nodes indicate citation relationships among policies, with the direction of the arrows indicating the direction of policy diffusion from the cited policy to the citing policy.

2.3. Diffusion Pattern Analysis

2.3.1. Network Topological Structure Metrics

Rogers, et al. [42] suggests that the structure of a network composed of innovation adopters significantly influences the pattern of innovation diffusion. We propose “interlocking” and “radiating, two typical interpersonal network structures, to characterize policy diffusion patterns. In the “radiating” diffusion model, citing (local) policies are only closely connected to the central policy, and there are few connections among citing policies themselves. It represents a vertical diffusion model characterized by coercion, competition, and learning, which can be specifically understood as a high push by the central government and a positive response by the local government, but limited innovation. In contrast, there are strong correlations between local policies and central policies as well as local policies themselves, as shown in Figure 2d. We employ a few citation network topology structure indicators, such as network density, average clustering coefficient, connectivity, and power-law distribution exponent, to observe citation distributions.

(1)

Network Density (D): Network density, which characterizes the density of the citation network, is defined as the ratio of the actual value of citation count to the theoretical maximum.

For a policy citation network, a lower network density indicates fewer cross-references among policies, suggesting a diffusion pattern closer to “radiating”, where the network structure is more open. Conversely, a higher network density suggests more citation connections among policies, indicating a diffusion pattern closer to “interlocking”, with more diverse policy diffusion pathways.

$$D={\displaystyle \frac{2\left|E\right|}{\left|V\right|(\left|V\right|-1)}}$$

(1)

In the formula, $V$ represents the set of nodes, and $E$ represents the set of edges.

(2)

Average Clustering Coefficient (C): The average clustering coefficient, which reflects the agglomeration degree of the network, is calculated as the mean clustering coefficient of all nodes.

Unlike network density, which focuses on describing the overall characteristics of the network, the average clustering coefficient emphasizes the average of local clustering behaviors across the network. In general, a higher average clustering coefficient indicates a larger spread of policy diffusion within a unit of time, suggesting a diffusion pattern closer to “interlocked”. Conversely, it is more likely to be a “radiation” diffusion pattern.

$$C={\displaystyle \frac{1}{\left|V\right|}}={\sum }_{v_i\in V}C\left(v_i\right)$$

(2)

$$C\left(v_i\right)={\displaystyle \frac{\mathrm{Number} \mathrm{of} \mathrm{closed} \mathrm{triplets} \mathrm{connected} \mathrm{to} v_i}{\mathrm{Number} \mathrm{of} \mathrm{triplets} \mathrm{centered} \mathrm{on} v_i}}$$

(3)

In this formula, $C\left(v_i\right)$ represents the clustering coefficient of $v_i$.

(3)

Connectivity is used to measure the connectivity of each node in the network, which is an important indicator of the performance of the network structure. It is defined as the proportion of nodes contained in the maximum connected subgraph relative to the total number of nodes in the network. For the policy citation network, we obtain the maximum connected subgraph by the assumption that two nodes are connected by a citation link.

(4)

Power-law Distribution Exponent, which refers to the index when the degree distribution of network nodes conforms to the power-law distribution, is defined as the probability that a randomly selected node in the network has a degree of $x$:

$$P\left(x\right)=a{x}^{-k}$$

(4)

In this formula, $a$ and $k$ are any constants greater than zero.

2.3.2. Policy Diffusion Indicators

By analyzing citation relationships, this study captures intergovernmental responsiveness. It further measures policy diffusion characteristics through diffusion speed, long-term impact, and short-term impact indicators.

(1)

Diffusion Speed $S_{ij}$ characterizes the dissemination dynamics of policy documents from the perspective of the policy implementation timeline. The shorter the time for a policy to be adopted, the faster its diffusion speed, and this indicates a quicker policy response. The following formula, adapted from the Walker [43] innovation diffusion approach, is used to calculate the diffusion speed of policy i in year j:

$$S_{ij}={\displaystyle \frac{{\sum }_1^j{N}_{ij}}{T_e-T_s}}$$

(5)

In the formula, $N_{ij}$ represents the number of citations received by policy $i$ during its j-th year of issuance. $T_e$ and $T_s$ denote the latest and earliest execution years of policy $i$, respectively.

(2)

Short-term impact describes policy diffusion from an immediate response. Based on the time interval characteristics of China’s science and technology planning, the short-term impact of a policy is calculated using its cited count for the first five years after it was enacted. It is calculated as the cited count for policy $i$ within five years of its issuance to the total number of policies issued in its promulgation year, calculated as follows:

$${ST}_{it}={\sum }_{t=1}^5{\displaystyle \frac{t{C}_{it}}{N_t}}$$

(6)

$$t=\left\{\begin{array}{c}1, Y_j=Y_i\\ Y_j-Y_i,Y_j\ne Y_i\end{array}\right.,and,t\le 5$$

(7)

In the formula,$t$ represents years of execution, where $Y_i$ is the execution year of policy $i$, and $Y_j$ is the execution year of the policy, which cites policy $i$. $C_{it}$ refers to the cited count of policy $i$ in the $t$-th year after its issuance, and $N_t$ represents the total cited count in the $t$-th year.

(3)

Long-term impact is used to describe policy diffusion from the perspective of its lasting influence. The longer the time interval between a policy and its cited policies, the greater its long-term impact. In contrast to short-term impact, long-term impact is calculated by the cited count after five years of its issuance. The calculation formula is as follows:

$$L{T}_{it}={\sum }_t{\displaystyle \frac{t\times C_{it}}{N_t}},t>5$$

(8)

In the formula, $t$ represents the year of promulgation year of policy $i$.

2.4. Policy Topic Clustering

The policy citation network exhibits characteristics of a social network, which could obtain policy topic clustering by community segmentation based on the closeness between nodes. Modularity is an optimal method for analyzing social networks. It can accurately describe the group structure distribution and underlying knowledge subgroups of the network through clustering algorithms based on the network’s characteristics, making it more aligned with real-world networks. In this study, we applied the Modularity optimization algorithm for complex social networks proposed by Blondel et al. [44] to perform policy topic clustering and identify community knowledge clusters. The Modularity gain value $\Delta Q$ for assigning network node $i$ to module $C$ is calculated as follows:

$$\Delta Q=\left[{\displaystyle \frac{{\displaystyle {\sum }_{in}+2k_{i,in}}}{2m}}-{\left({\displaystyle \frac{{\displaystyle {\sum }_{tot}+k_i}}{2m}}\right)}^2\right]-\left[{\displaystyle \frac{{\sum }_{in}}{2m}}-{\left({\displaystyle \frac{{\sum }_{tot}}{2m}}\right)}^2-\left({\displaystyle \frac{k_i}{2m}}\right)\right]$$

(9)

In this formula, ${\sum }_{in}$ denotes the total weight of all edges within module $C$; ${\sum }_{tot}$ refers to the total weight of edges that link to nodes inside module $C$; K_i,in is the sum of the weights of the edges connecting node $i$ to the nodes within module $C$; and $K_i$ is the sum of the weights of all edges connected to node $i$. It is important to note that only when ΔQ > 0, the network node i will be classified into module C.

In this study, we first excavate the overall layout of China’s ASTPs by conducting a modular analysis of the entire policy citation network. Then, we further detect topic dynamics of ASTPs during each stage by clustering temporal policy citation networks and measuring citation relationships among them.

2.5. Tracking Diffusion Paths

Tracking the diffusion paths is beneficial for exploring the evolutionary trajectory and clarifying the formation process of policies. This study employs the Main Path Method (MPM) to characterize important pathways of policy diffusion. MPM is a quantitative and visualization approach based on “connectivity” that reduces the complexity of networks and extracts key paths, which has significant practical significance in the path detection of science and technology development. There are two key steps and algorithms for MPM recognition: traversal count and path search. To traversal count, this study primarily employs the Search Path Link Count (SPLC) method [45], which is widely applicable in knowledge diffusion scenarios within science and technology development. The path search mainly utilizes the Key-Route Local Main Path. Unlike traditional approaches that start forward or backward searching from a fixed node, the key path search begins from the key path (usually the link with the highest traversal weight) and simultaneously searches toward both ends of the path until reaching the source and sink points. Finally, the extracted paths are fitted to form the main path of the citation network [46].

3. Results

3.1. Patterns of ASTPs Diffusion

Figure 3a illustrates the temporal development trends of the number of ASTPs issued and cited in China. In the early stages of the reform and opening-up, the number of such policies was relatively low. However, following the 16th CPC National Congress in 2002, which emphasized the acceleration of agricultural science and technology advancements, the number of China’s ASTPs began to surge. The 18th CPC National Congress in 2012 proposed the establishment of a modern agricultural science and technology innovation system, further stimulating Chinese government departments to formulate ASTPs, leading to a peak in policy issuance that has since stabilized. The increase in China’s ASTPs formulation has also been accompanied by ongoing policy diffusion, causing the trends in issuance and citations to align. The policy citation network exhibits significant sparsity, reflected in its low average degree (1.179) and minimal density (0.0001), with 2570 policies showing no citation ties.

As shown in Figure 3b, the network density significantly decreases monotonically with the addition of new nodes, indicating a shift towards a radial rather than interlocked network structure in policy diffusion. Figure 3c displays the evolution of the average clustering coefficient of the citation network, which was essentially zero before the 18th CPC National Congress but has shown a fluctuating upward trend since, increasing from 0.021 in 2012 to 0.0417 in 2023. This suggests that policy diffusion within localized time frames is relatively dense, resembling an interlocked structure.

Figure 3d depicts the trend in connectivity within the citation network, which has fluctuated downwards before showing an increasing trend. Post-18th CPC National Congress, network connectivity has begun to rise, with values increasing from 0.057 in 2012 to 0.385 in 2023, indicating improved diffusion efficiency of China’s ASTPs after this point.

Figure 3e shows that policy citations conform to an exponential power-law distribution [47]. Using the maximum likelihood estimation method, the out-degree and in-degree power-law indices (k) of the policy citation network are determined to be 1.411 and 4.092, respectively. The goodness-of-fit is verified through the Kolmogorov–Smirnov test, yielding p-values of 0.6487 and 0.1243, both greater than 0.1, thus confirming the unevenness in policy diffusion.

The imbalance in policy diffusion is also reflected in the citation patterns across different administrative levels.

Table 1. Policy issuance at different administrative levels.

Policy Source	Major Issuing Authorities	No. of ASTPs	Citing Count	Proportion of Citing (%)	No. of Cited	Proportion of Cited (%)
National Level	Central Committee of the Communist Party of China, State Council, Standing Committee of the National People’s Congress, National People’s Congress	186	126	3.38	1707	45.7
Ministerial Level	National Development and Reform Commission, Ministry of Science and Technology, Ministry of Agriculture and Rural Affairs, Ministry of Education, Ministry of Human Resources and Social Security, Ministry of Industry and Information Technology, Ministry of Commerce, State Education Commission, State Science and Technology Commission	840	1096	29.36	1835	49.2
Provincial Level	Beijing, Chongqing, Anhui, Zhejiang, Yunnan, Sichuan, Shanghai, Shaanxi, Shandong, Qinghai, Liaoning, Jiangxi, Jiangsu, Hunan, Hubei, Heilongjiang, Henan, Hainan, Guangxi, Guangdong, Gansu, Fujian, Xinjiang, Xizang, Tianjin, Shanxi, Hebei, Ningxia, Inner Mongolia, Jilin, Guizhou	2171	2511	67.26	191	5.1

shows the characteristics of policy issuance and citation distribution at different administrative levels. As the administrative level increases from provincial and ministerial to national, ASTPs have lower issued and citing counts and higher cited counts. Specifically, national and ministerial ASTPs tend to receive more cited counts, while provincial ASTPs have more citing counts. Figure 4a illustrates the citation characteristics of China’s ASTPs at different administrative levels. Citation patterns show that 75.54% of ASTPs are cited across administrative levels, with lower-level policies predominantly referencing those from higher authorities. Sibling citations are secondary, while only 1.23% of citations prefer to quote lower-administrative levels, suggesting that there is a significant hierarchical pressure effect in policy diffusion [48]. The detailed citation distribution of different administrative levels is shown in Figure 4b.

Figure 4c shows the diffusion speed and policy impact of different administrative levels. The results suggest that higher-administrative-level ASTPs enjoy advantages in diffusion speed, short-term impact, and long-term impact. To examine the relationship between administrative level (1 for provincial, 2 for ministerial, and 3 for national) and both short-term impact and long-term impact, a correlation analysis was conducted. The Pearson coefficients were 0.3000 and 0.7210, respectively, with p < 0.05, indicating that the administrative level is significantly correlated with both short-term and long-term policy impacts.

As shown in Figure 4d, the diffusion of ASTPs is relatively timely, with more citations generated when the time interval between policy issuance and citation is shorter. Policies are cited most frequently in the year of issuance and the following year, with proportions of 31.72% and 30.75%, respectively.

3.2. Topic Dynamics of ASTPs Diffusion

Figure 5 presents nine major ASTPs topics identified through modularity-based analysis of the policy citation network. ASTPs on “Basic research”, “Technology transfer of scientific achievements”, “Industrialization of agricultural science and technology”, and “Rural technological innovation” are the core topics in the policy citation network.

Table 2. Diffusion characteristics of the main ASTP topics.

Policy Topics	Internal Structure						External Links		Administrative Levels
	Node	Edge	Network Density	Average Degree	Average Clustering Coefficient	Average Path Length	In-Degree	Out-Degree	National	Ministerial	Provincial
1. Basic research (BR)	546	688	0.002	1.26	0.033	1.709	757	190	34	159	351
2. Transformation of scientific and technological achievements (TSTAs)	417	487	0.003	1.168	0.042	1.415	542	55	19	83	315
3. Agricultural Science and Technology Industrialization (ASTI)	405	457	0.003	1.128	0.024	1.662	478	21	21	96	288
4. Rural Technological Innovation (RTI)	391	444	0.003	1.136	0.027	1.482	458	14	11	90	290
5. Intellectual Property Protection (IPP)	294	337	0.004	1.146	0.035	1.405	385	48	28	92	174
6. Talent Incentives (TIs)	231	253	0.005	1.095	0.026	1.648	272	19	12	46	173
7. Agricultural Technology Promotion (ATP)	226	258	0.005	1.142	0.032	1.723	286	28	5	58	163
8. Science and Technology System Reform (STSR)	179	213	0.007	1.19	0.053	1.354	230	17	25	70	84
9. Agricultural Machinery Equipment Innovation (AMEI)	113	133	0.011	1.177	0.045	1.498	154	21	4	32	77

provides nine main citation links and network structural characteristics. To enhance interpretability, representative policy examples for each topic have been provided in Supplementary Table S1. The diffusion among these ASTPs topics remains minimal, which can be attributed to their relatively low in-degree and out-degree values. Each topic exhibits high average path lengths, along with low network density and average clustering coefficients, indicating a low diffusion efficiency. Additionally, according to administrative levels, most ASTPs come from provincial institutions, and the number of ASTPs decreases as the effectiveness level increases from provincial to ministerial to national.

The topic dynamics of the ASTPs diffusion are examined by dividing the developmental stages of China’s agricultural science and technology in relation to shifts in policy quantity: the Initial Establishment Stage (1978–2001), the Steady Advancement Stage (2002–2011), and the Innovation-Driven Development Stage (2012 to 2023). A temporal citation network of policy diffusion is constructed based on policy citation (edge) relationships. At each stage, modular algorithms are employed to identify key policy topics from dynamic citation networks, and variations in their structural attributes are analyzed to uncover the evolution of China’s agricultural science and technology priorities. Three temporal policy citation networks are constructed using Gephi software (version 0.10.0) and are presented in Figure 6.

In the Preliminary Establishment Stage (1979–2001), during the initial stage of reform and opening-up, China’s agricultural science and technology experienced comprehensive and rapid recovery and development, and China’s ASTPs entered the Preliminary Establishment Stage. A total of 415 ASTPs were issued by 148 administrative agencies, with 55.90% originating from the State Council and its ministries, whereas local governments contributed a relatively smaller portion. Additionally, only 64 joint policies were published. These features reflect insufficient policy coordination at the national level in the development of ASTPs. During this period, policy themes were mainly concentrated on strategic frameworks, the dissemination of technology, and the promotion of agricultural innovations. Key policies such as the “Decision of the Central Committee of the Communist Party of China on the Reform of the Science and Technology System” (ID: 149), “Law of the People’s Republic of China on Scientific and Technological Progress” (ID: 88), and “Law of the People’s Republic of China on Agricultural Technology Promotion” (ID: 83), were promulgated to enhance the technological planning system and promote agricultural technology dissemination. As shown in

Table 3. Dynamic diffusion of ASTPs in different periods.

Periods		1978–2001	2002–2011	2012–2023
Number of ASTPs		415	914	1860
Number of ASTPs with citations (Nodes)		402	807	1403
Number of internal citations		464	916	1621
Number of external citations	Cited count	358	1289	0
	Citation count	0	145	587
Agency distribution	Number of national agencies	6	6	6
	Number of ministerial agencies	54	37	70
	Number of provincial agencies	88	176	368
Average time interval (Years)		1.7	2.23	3.14
Average diffusion speed		0.071	0.058	0.045
Average short-term impact		0.207	0.035	0.029
Average long-term impact		0.264	0.043	0.017

, although the number of ASTPs released during this period was relatively low, the diffusion activities were more active and frequent. A total of 405 ASTPs were issued during this stage, generating 464 citations within the same period and 358 cross-stage diffusion links. To more effectively illustrate topic associations across different stages, we constructed a dynamic diffusion flow chart for ASTPs. As depicted in Figure 7, a majority of these policies—59.49%—were cited by policies issued between 2012 and 2023. Notably, this stage demonstrates the shortest average time interval of policy diffusion, while exhibiting the highest average diffusion speed, as well as the most significant short-term and long-term impacts.

During the Steady Advancement Stage (2002–2011), 219 administrative bodies issued 914 ASTPs to support the continued development of agricultural science and technology in China, with provincial-level policies accounting for 67.18% of the total. This reflects a strong commitment from local governments to pursue policy innovation and strengthen localized ASTP support systems tailored to regional needs, thereby amplifying local policy diffusion effects during this phase. Policy themes in this period were largely oriented toward enhancing and refining the existing ASTP framework. Key areas of focus included basic research, the development of grassroots agricultural extension systems, talent retention in agricultural science and technology, and the promotion of agri-tech and related equipment. Notably, policy diffusion was markedly more extensive than in the previous period (1979–2001), with 807 ASTPs generating 2205 citations. Among these, 391 were referenced by policies enacted between 2012 and 2023.

In the Innovation-Driven Development Stage (2012–2023), following the successful convening of the 18th CPC National Congress in 2012, agricultural science and technology in China entered a new opportunity for innovation-driven development. During this period, 1860 ASTPs were issued by 444 administrative entities, comprising 1368 provincial-level policies and 312 from the State Council and its ministries, among which 145 were jointly promulgated. ASTPs priorities included three important changes to promote agricultural science and technology innovation-driven development during this period: the scope of agricultural science and technology gradually extended to encompass rural agricultural technology, shifting the development focus from basic research to industrialization, and changing the development goal from increasing grain production to promoting green and sustainable agricultural development [49]. At this stage, the number of ASTPs reached its peak, with the most active ASTP diffusion, including 1403 internal diffusions and 587 external diffusions. This stage exhibited the longest average time interval and relatively moderate levels of policy diffusion speed, as well as short- and long-term impacts, primarily due to insufficient cumulative citations compared to other stages.

3.3. Historical Trajectory of ASTPs Diffusion

Mapping the historical trajectory of policy diffusion provides a richer, more intuitive data analysis for studying policy processes and identifying key policy impacts. To identify key policies at each developmental stage, we extracted the largest connected components from the temporal policy citation networks and employed both PageRank and Bridging Centrality algorithms to evaluate the significance of individual ASTPs. The top 20 influential policies are presented in Supplementary Table S2. These policies typically exhibit high citation frequency and strong policy impact, despite receiving relatively few outbound citations. Serving as foundational directives, they were issued across various stages of China’s agricultural development to support the advancement of agricultural science and technology. Notably, the “Law of the People’s Republic of China on Scientific and Technological Progress” (promulgated in 1993 and amended in 2007 and 2021) receives a substantial number of citations, PageRank value, diffusion speed, and short-term and long-term impacts. This indicates that the law plays a critical role in China’s ASTPs framework and occupies a critical position for the policy knowledge flow viewpoint. According to the issuance order of these key ASTPs, we further map the evolution path of China’s agricultural science and technology sector. As shown in Figure 8, the focuses of China’s ASTPs are different at each stage, which experience the following three major transformations: advancing agricultural technology dissemination, deepening agricultural science and technology system reforms, and supporting rural quality improvement and efficiency gains.

To visually trace the historical diffusion trajectory of China’s ASTPs, the SPLC main path of the policy citation network was extracted using Pajek software (version 5.19) (Figure 9). This main path highlights 60 ASTPs identified as part of the key-route backbone. Each of these policies corresponds to a node in the citation network and is annotated with its policy ID and year of issuance. The direction of the arrows points from the cited policy to the citing policy, representing the direction of policy diffusion. China’s ASTPs diffusion main path is divided into the following three sub-paths. Path A focuses on enhancing innovation capacity in agriculture and rural areas, formed from nodes 189-2004 and 176-2005. Path B involves the transformation of agricultural technology achievements, while Path C relates to the agricultural science and technology innovation system. They share similar path characteristics. The special process is that all three sub-paths follow a top-down diffusion model, which has relatively low node count, and diffusion mainly occurs between national policies and ministerial-level policies at the early stages of the paths. As time progresses, local diffusion gradually becomes more frequent.

Path A, containing 15 ASTPs, primarily concerns the promotion of new socialist countryside construction, enhancement of comprehensive agricultural production capacity, and acceleration of rural talent resource development. Rural issues have been a longstanding challenge for China [50]. In China, the State Council has implemented a new round of institutional reform to optimize the agricultural science and technology responsibilities of the Ministry of Agriculture and Rural Affairs since 2023. The details are transferring the responsibility for organizing and drafting policies and plans for promoting agricultural and rural development from the Ministry of Science and Technology to the Ministry of Agriculture and Rural Affairs. This reflects a coordinated approach to strengthening agricultural science and technology innovation. Path A originates from multiple nodes (189-2004, 176-2005), experiences a process from fusion to separation, and presents numerous branches at the end, indicating that the current policy topic of using agricultural science and technology to promote rural innovation and development is in a highly innovative split stage, offering more opportunities for policy innovation.

Path B includes 17 ASTPs, the policy target of which focuses on agricultural science and technology parks, science and technology commissioners, the construction of agricultural technology service systems, and the cultivation of agricultural technology promotion talents. The transformation of agricultural science and technology achievements is the “last mile” of the entire agricultural science and technology innovation process [51]. Agricultural technology service systems, agricultural science and technology parks, and science and technology commissioners are important carriers for the transformation of agricultural technology achievements, and building a high-quality team for the transformation of talents is a key measure to accelerate the transformation of scientific and technological achievements [52]. Path B starts from three nodes in 1993 (85-1993, 83-1993, and 200-1993), and it presents multiple branches at the end, indicating that policies related to agricultural achievement transformation still exhibit strong innovation potential after several decades of evolution.

Path C is composed of 28 ASTPs and starts from 88-1993, which was issued by the National People’s Congress to stimulate scientific and technological progress. The law is China’s first Science and Technology Progress Law, which aims to comprehensively promote scientific and technological progress and drive technological innovation to support and lead economic and social development, with the ultimate goal of building a modern socialist country. Specifically, as the path progresses, it separates into two sub-paths: Sub-path C-1 contains 15 ASTPs, most of which are guiding policies in such outlooks as “outlines” and “implementation plans”. The focus of these ASTPs can be summarized as the cultivation of technological innovation talents, specifically addressing strategies to improve the scientific literacy of various groups, including youth, farmers, leaders, and civil servants. Improving the scientific literacy of farmers is an important measure to strengthen the technological support for rural revitalization [53]. The continuous emphasis on cultivating innovative science and technology talents from the national to the local levels ensures that the talent system meets the development needs of agricultural science and technology innovation. Sub-path C-2 focuses on the construction of the agricultural industry technology system. Specifically, at the beginning of the path, relevant policies were mainly regarding building and improving the agricultural industry technology systems. Subsequently, and as the path progresses, the policy focus shifts to the management of funds, project evaluation, and talent selection within the agricultural industry technology system. The improvement and development of the agricultural science and technology innovation system are receiving increasing attention from both the national and regional levels.

4. Discussion

This research advances the existing body of knowledge by introducing a quantitative, citation-based framework for analyzing policy diffusion. As an early effort to investigate diffusion mechanisms within the domain of agricultural science and technology through the lens of policy citation, it expands methodological approaches to policy process studies and enhances understanding of the field’s knowledge structure. By uncovering extensive citation relationships embedded in ASTP documents, we construct policy citation networks and apply both topological and diffusion-based indicators to systematically quantify diffusion patterns. Then, the topics’ features and dynamic correlation of ASTPs diffusion are analyzed based on the Modularity algorithm and temporal deconstruction methods. Finally, utilizing PageRank, Bridging Centrality algorithms, and the Search Path Link Count (SPLC) method, we recognize key policies and key relations in the process of ASTPs diffusion, and portray the historical trajectory.

Of note is that the top-down pattern is the most critical pattern of ASTPs diffusion in China, with the highest diffusion speed, long-term impact, and short-term impact. An imbalance in the distribution of policy citations reveals the effect of preferential attachment, or the Matthew Effect, in ASTPs diffusion. Specifically, it displays an upward priority link based on the ASTPs reference. With the increase of the administrative level of the government, the number of policies promulgated decreases sharply, but the number of policies cited is higher; that is, only a few important policies achieve a high citation, while most policies have a limited citation. Policies at lower administrative levels, or coerced by hierarchical pressure, or integrated with their own development needs [54], usually choose to cite higher administrative level policies, which also accelerate the diffusion of high administrative level policies and lead to more obvious advantages of diffusion speed, long-term and short-term impact [55]. This is consistent with other public domain research, such as climate [56], land [57], and new energy [58], which further proves that top-down hierarchical pressure is an important determinant of policy adoption in China’s special institutional context.

The field of ASTPs’ topic diffusion is gradually expanding and possesses obvious time characteristics. With the gradual realization of China’s economic goals, the topic areas of ASTPs continues to broaden, from experimental technological innovation and agricultural scientific development in the early stage of reform and opening up, to technological innovation and agricultural industrialization development in the socialist market economy stage, to agricultural informatization and agricultural modernization development under the background of “four modernizations synchronization” [59,60]. ASTPs have always been designed to meet the latest needs of social and economic development, and have laid a solid foundation for China’s economic development, such as nurturing an innovative ecosystem and significantly increasing the size of the educated workforce [61,62]. In addition, studies have confirmed that digital agriculture policies have obviously promoted agricultural economic resilience [63].

Although ASTP diffusion between different development stages and different themes is severely sparse, the diffusion between different ASTP topics within the same development stage is extremely active, as presented in Figure 10. The citation degree (including in-degree and out-degree) of policy themes within the stage is significantly higher than that of the overall themes, and with the progress of the economic development stage, the degree of theme within the stage is also gradually increasing, which reflects that the transmission efficiency of social economy to ASTPs is constantly improving [64].

There are three main paths of ASTP diffusion, which are independent of each other but have the same trajectory trend. Main path analysis results reported in the study demonstrate that a series of key ASTPs, such as the policy “National Medium- and Long-Term Science and Technology Development Plan Outline (2006–2020)”, occupy a dominant position in the citation network, which play a vital role in improving the agricultural science and technology system [65,66]. Meanwhile, these key policies have promoted the flow and innovation of agricultural science and technology elements, and gradually developed into three main independent diffusion paths. Their formulation often depends on the cooperation among multiple departments, while policy adoption is more dependent on the government’s decision-making view. Some studies have found that socioeconomic and contextual factors would lead to more cautious policy adoption [67,68].

The three main path trends further reveal that ASTPs follow a strict top-down diffusion pattern, and future local policies will continue to revolve around key aspects such as enhancing innovation capacity in agriculture and rural areas [69], promoting the transformation of agricultural technological achievements [70], and improving the agricultural science and technology innovation system [59].

Investigating the diffusion of China’s ASTPs holds substantial significance for both academic inquiry and policy practice in the domain of agricultural science and technology governance. Such research enables scholars to gain a clearer picture of the field’s development, monitor policy evolution over time, and enhance the empirical foundation for evidence-based policymaking. For decision-makers, these insights offer timely guidance, support informed judgment, and facilitate strategic planning in response to ongoing changes. Theoretically, this study contributes to three key areas. First, it enriches the understanding of indigenous policy diffusion and transformation processes by examining patterns of citation inequality and preference within the policy system. It further confirms the imbalance of China’s ASTPs diffusion, which is mainly due to the differences in the long-standing pyramid power structure in China [71,72] and the domain development stage [73].

Second, by employing a citation-based analytical approach, this study contributes to the development of a new theoretical lens for examining ASTPs adoption and implementation. This framework offers alternative perspectives and adds dynamic insights to the discourse on agricultural science and technology governance. For instance, it could enhance information trust in government public content and media coverage, thereby strengthening confidence in the advancement of local agricultural science and technology [74,75]. Third, this study provides more extensive evidence in the fields of agricultural science and technology governance and economy about the ability of ASTPs in shaping policy practice to meet the needs of economic and social development [76]. Such ASTPs’ practice, including broadening the scope of services, deepening institutional reform, and strengthening independent innovation ability, can have a significant impact on market economy development [77].

Practically, our results have some significant implications for policy makers optimizing ASTPs and promoting agricultural science and technology innovation. There are three practical implications of studying ASTPs’ diffusion. First, the study’s identification of attention points and dominant trends in China’s ASTPs, which is conducive to policy makers to gain insight into policy topic allocation and optimize policy system [78], such as focusing on supporting key areas and key industries of agricultural science and technology innovation by resource concentration and environmental optimization [79], paying more attention to innovation practices in sub-key areas, including agricultural science and technology input, innovation output, achievement application, and environmental support covered by agricultural science and technology innovation activities [80]. Second, the citation-based analysis used in this study can serve as a useful tool for the government authorities and practitioners to systematically analyze issues on China’s ASTPs diffusion and obtain policy inspiration. For instance, when the central government implements ASTPs, while highlighting the timely transmission of enteral policies from top to bottom, it should also focus on enhancing the “ reproduction “ ability of ministries and local government policies, so as to promote the policy impact of policy implementation agencies [81,82]. Third, quantitative tracking of ASTPs’ trajectory contributes to promoting the policy common understanding of core issues among policy participants with diverse interests [83], encourages them to actively invest various resources needed in the process of policy implementation, and accelerates the achievement of policy intentions such as high-quality development and rural innovation application.

While this study provides meaningful insights, certain limitations remain. Specifically, the current analysis focuses exclusively on policies issued at the central, ministerial, and provincial levels. Enriching and refining the policy dataset, such as complementing local policies, may uncover more diffusion patterns [84]. Future studies can expand the scale of hierarchical analysis to the county and municipal levels to reveal the complex process and differentiated impact of ASTPs diffusion in a more fine-grained way. In addition, we simulated policy diffusion by citation relationships between policy texts. Influenced by the model of intergovernmental relations, we can only track policy diffusion activities between central agencies and from central to local. Multiple studies have shown that bottom-up absorption of radiation is also an important path of public policy diffusion [54,85,86]. A more semantic approach can uncover implicit diffusion information and also provide deeper insights into the identified thematic areas.

Furthermore, our dataset was sourced exclusively from the Peking University legal database. While it is one of the most comprehensive and authoritative legal policy databases in China, it may not capture the full universe of issued policies, particularly at the grassroots level. We acknowledge this as a limitation and suggest that future work incorporate additional data sources to improve policy coverage and representativeness.

While citation-based analysis provides valuable insights into the structural pathways and influence patterns of policy diffusion, it is important to acknowledge its limitations. In the Chinese political and administrative context, policy citations do not necessarily equate to substantive implementation. Some citations may be symbolic or procedural in nature, serving to align with superior directives rather than reflecting actual adoption or enforcement. Moreover, this study does not distinguish between symbolic and functional citations, as most policy texts lack explicit semantic cues or metadata that would enable such classification. We recognize this as a methodological limitation. Future research could incorporate administrative outcome indicators, implementation records, or advanced semantic analysis techniques to better assess the depth of diffusion and differentiate between citation intents.

5. Conclusions

Our study distinguishes itself in China’s ASTPs diffusion research by employing a rigorous citation-based analysis. Unlike conventional quantitative approaches, which often rely on predefined theme models or single-indicator frameworks extracted from policy documents, our approach is more comprehensive, accurate, and visual, identifying key topics, key policies, and the main paths of policy diffusion. This methodological approach allows for a deeper identification of policy diffusion characteristics while also shedding light on potential trajectories of future change.

As evidence of our methodological rigor, our citation-based analysis enriches and deepens the understanding of the policy diffusion process by highlighting the synergy of policy interaction and time evolution. Our citation-based analysis with interactive relationship observes policy diffusion bias at different administrative levels. For instance, higher administrative level ASTPs have higher diffusion speed, short-term and long-term impacts, and lower-level governments respond promptly and effectively to higher-level policies. Similarly, our citation-based analysis with time evolution emphasizes the thematic dynamics and historical trajectory of policy diffusion. For instance, with the advance of time, cross-topic policy diffusion has become more frequent, and the connotation and extension of ASTPs have been continuously enriched. As a whole, three main paths of ASTPs diffusion have been formed, including the improvement of agricultural and rural innovation ability, the transformation of agricultural technology achievements, and the improvement of the agricultural science and technology innovation system.

This citation-based investigation significantly enhances the ASTPs literature, offering researchers and policy makers valuable insights and potential trends. As information technology continues to shape the policy understanding, future research should consider supplementing the ASTPs dataset and merging advanced text-mining techniques for more nuanced analysis. Investigating ASTPs’ diffusion mechanisms, effects, and influences could provide valuable directions for further research. Moreover, understanding the economic development, ecological environment, and innovation level implications of ASTPs diffusion should be a priority for future investigations.