Method and system for predicting evolution and emergence of science and technology hotspots based on network evolution

Abstract

The invention discloses a prediction method and system for science and technology hotspot evolution and emergencies based on network evolution, and the method comprises the steps: S1, constructing a time sequence knowledge network: obtaining time information containing scientific research projects, author information and structured data of theme keywords from a science and technology database, abstracting the author information into network nodes, and storing the network nodes in a database; abstracting the cooperation relationship into an edge of the network, and constructing an initial time sequence knowledge network carrying time and subject attributes; s2, network dynamic evolution simulation: based on the initial time sequence knowledge network, utilizing a dynamic evolution simulation engine to carry out forward iteration simulation so as to simulate dynamic evolution of a network structure; and S3, future hotspot identification and characterization: after simulation in the step S2 reaches a preset future time point, based on a future knowledge network generated by simulation, identifying a high-density node sub-graph in the future knowledge network through a community discovery algorithm, marking the high-density node sub-graph as a predicted future science and technology hotspot, and performing technical characterization on the future science and technology hotspot.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of computer technology and artificial intelligence application, in particular to a science and technology hotspot prediction method, and more particularly to a science and technology hotspot evolution and emergence prediction method based on network evolution and related system. BACKGROUND

[0002] In the field of cutting-edge science and technology, it is very important to identify and layout the key direction and emerging hotspots of future scientific and technological development in advance and build a forward-looking prediction capability for academic research and scientific innovation. At present, the mainstream technical solutions for analyzing science and technology trends and identifying hotspots in the field mainly include: index statistical analysis method based on literature metrology, discovery algorithm based on static network, and trend analysis method based on time series, etc. These methods can identify hotspots to some extent. However, these methods are essentially trend summary and static, and they cannot model and dynamically predict the emergence process of science and technology hotspots "from nothing to something", so their forward-looking prediction capability is insufficient. Specifically, the limitations of these methods mainly include: 1. Index statistical analysis method based on literature metrology: this method identifies the current hot field by analyzing indicators such as keyword appearance frequency, paper citation number, and patent growth rate. The core technical defect of this method is that it can only summarize the existing trend, and although it can well describe the fields that have become hot, it is difficult to predict how new hotspots "emerge from nothing". This method lacks the modeling capability of the internal dynamic mechanism behind the formation of hotspots.

[0003] 2. Discovery algorithm based on static network: this method constructs a static cooperation network or citation network from scholars or papers, and uses community discovery algorithms to identify the "research community" in the network that already exists and has a tight structure as a hotspot. However, its technical limitation is static features, it can only provide the network structure state at a certain point in time, and cannot simulate the dynamic process of network evolution over time, nor can it predict the changes in future network topology.

[0004] 3. Trend analysis based on traditional time series: this method establishes a time series model for indicators such as the appearance frequency of a certain technical keyword for prediction and analysis. However, its defect is that the model is relatively simple. It regards each technical field as an isolated variable, ignoring the fact that scientific and technological development is driven by complex cooperation among researchers. Therefore, it is difficult to predict new hotspots that are driven by cross-disciplinary cooperation, nonlinear, and sudden changes.

[0005] In summary, the prior art lacks a technical solution capable of simulating the internal generation mechanism of a scientific cooperation network and dynamically deducing the future structural evolution thereof. Therefore, there is an urgent need in the field for a new technology capable of shifting from static and retrospective statistical analysis to dynamic and global simulation evolution to achieve a powerful prediction of future scientific hotspots. SUMMARY

[0006] In view of this, the present application proposes a prediction method and system for the evolution and emergence of scientific hotspots based on network evolution, which can prospectively predict the emergence time and core theme of future scientific hotspots, while overcoming the defects of "locality", "staticity" and "simplicity" of the prior art, thereby solving the technical problem that the prior art cannot dynamically and generatively predict the future topological structure of a scientific knowledge network.

[0007] To achieve the above object, one aspect of the present application proposes the following technical solution: A prediction method for the evolution and emergence of scientific hotspots based on network evolution, comprising the following steps: step S1, construction of a time-series knowledge network: obtaining structured data containing time information, author information and theme keywords of scientific research projects from a scientific database, abstracting the author information as network nodes and the cooperation relationship as network edges to construct an initial time-series knowledge network carrying time and theme attributes; step S2, dynamic evolution simulation of the network: based on the initial time-series knowledge network, using a dynamic evolution simulation engine to perform forward iteration simulation to simulate the dynamic evolution of the network structure; step S3, identification and characterization of future hotspots: after the simulation of step S2 reaches a preset future time point, based on the future knowledge network generated by the simulation, identifying a high-density node subgraph in the future knowledge network by a community discovery algorithm, marking it as a predicted future scientific hotspot, and performing technical characterization on the future scientific hotspot.

[0008] Further, the author information in step S1 includes the author's name and the institution he / she belongs to, and the cooperation relationship includes co-published papers or serving the same scientific research project.

[0009] Further, step S2 specifically includes: given a probability parameter p that regulates the network evolution mechanism, the simulation engine performs forward iteration at discrete time steps by computer simulation to execute the following mechanisms to simulate the dynamic evolution of the network structure: Mechanism A with probability p: the mechanism A is a new node introduction and theme attachment mechanism, used to simulate the expansion of the research field; Mechanism B with probability 1-p: the mechanism B is an internal connection and community formation mechanism, used to simulate the emergence of scientific hotspots.

[0010] Furthermore, mechanism A simulates the expansion of research fields by simulating the dynamic process of new research forces entering existing research fields; mechanism B simulates the emergence of scientific and technological hotspots by simulating the dynamic process of existing researchers establishing new cooperative relationships through collaborators, leading to an increasingly close internal structure of the research community.

[0011] Furthermore, the mechanism A specifically includes: calculating a new node attraction parameter λ based on node i's seniority τ(i) in the network and a seniority influence coefficient α. i (t); based on this parameter λ i (t), through a probability distribution model, calculate the number Z of new nodes that node i will connect to at the current time step; then, generate Z new nodes in the network and establish their connection with node i. The new nodes inherit or associate with the topic attributes of node i.

[0012] Furthermore, mechanism B specifically includes: identifying all two-connected neighbor nodes j of node i, where a two-connected neighbor is a non-direct neighbor connected through a common neighbor node; for each two-connected neighbor j of node i, calculating the connection probability μ of node i establishing a new connection with it based on its current network influence and an influence coefficient β. ij .

[0013] Furthermore, step S3 specifically includes: The simulation engine, after executing step S2, reaches the preset future time point T. n Then, a predictive future knowledge network G(T) is generated. n According to this future knowledge network G(T) n The size and topology of network G(T) are determined, and a community detection algorithm is run to identify the network. n High-density node subgraphs with clustering coefficients and connection densities higher than preset values ​​are identified; these high-density node subgraphs are marked as predicted future technological hotspots. For each predicted future technology hotspot, semantic vector technology is used to extract the most concentrated thematic keywords within its internal nodes, which are then used as the core technology keywords for that future technology hotspot.

[0014] Furthermore, step S3 also includes: for each predicted future technological hotspot, calculating its network structure index to assess its potential impact, and transforming the abstract network subgraph into structured, understandable information; wherein the network structure index includes network size, connection density, and average path length.

[0015] In another aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, can implement the steps of the aforementioned prediction method.

[0016] In another aspect, this invention proposes a prediction system for the evolution and emergence of scientific and technological hotspots based on network evolution, comprising: a data acquisition module configured to acquire structured data containing time information, author information, and topic keywords of scientific research projects from a scientific and technological database; a network construction module configured to abstract the author information into network nodes and the cooperation relationships into network edges based on the structured data, thereby constructing an initial temporal knowledge network carrying time and topic attributes; a dynamic evolution simulation engine module configured to perform forward iterative simulation based on the initial temporal knowledge network using the dynamic evolution simulation engine to simulate the dynamic evolution of the network structure, and to form a predicted future knowledge network after the simulation reaches a preset future time point; and a hotspot identification and analysis module configured to identify high-density node subgraphs in the future knowledge network through a community discovery algorithm, mark them as predicted future scientific and technological hotspots, and perform technical characterization on the future scientific and technological hotspots.

[0017] The beneficial effects of this invention are as follows: Compared with existing technologies, the prediction method and system for the evolution and emergence of scientific and technological hotspots based on network evolution provided by this invention, by constructing an initial temporal knowledge network and using a dynamic evolution simulation engine that integrates mechanism A (new node introduction and topic attachment) and mechanism B (internal connection and community formation) for forward iterative simulation, can simulate the dynamic generation and evolution process of scientific and technological cooperation networks from a mechanistic perspective. Mechanism A simulates the introduction of new research forces and domain expansion, while mechanism B simulates the strengthening of internal connections based on common neighbors and community formation. The probabilistic synergy of these two mechanisms enables the simulation process to reproduce the nonlinear emergence phenomenon of scientific and technological hotspots in the real world, from nothing to something, and from small to large. Finally, by performing community discovery and semantic analysis on the simulated future network structure, high-density potential hotspot communities can be identified and their core themes extracted, thereby achieving forward-looking and generative prediction of scientific and technological hotspots. This overall technical solution overcomes the fundamental deficiency of existing static analysis methods in predicting future network structures and emerging hotspots, provides interpretable analysis of hotspot formation mechanisms, enhances the robustness and accuracy of predicting emerging hotspots driven by complex cooperation, and provides dynamic and global decision support for science and technology strategic planning.

[0018] Furthermore, by limiting author information to include name and institution, and collaboration to include co-authored papers or joint projects, the network construction is made closer to real-world scientific research collaboration scenarios, thus improving the accuracy and representativeness of the initial network.

[0019] Furthermore, by specifying the execution logic and parameters of mechanism A and mechanism B, the dynamic evolution process has a clear and implementable mathematical basis, which enhances the reliability of the simulation and the repeatability of the results.

[0020] Furthermore, by using clustering coefficients and connection density as screening criteria in future hotspot identification, and combining semantic vector technology to extract core keywords, we can more accurately locate and characterize the real technological hotspots from complex network structures, thereby improving the understandability and practicality of the prediction results.

[0021] Furthermore, by calculating the structured indicators (size, density, path length) of hotspot communities, it is possible to quantitatively assess the potential influence and development stage of different hotspots, providing data support for prioritizing resource allocation. Attached Figure Description

[0022] Figure 1 This is a flowchart of a prediction method for the evolution and emergence of technological hotspots based on network evolution, proposed in an embodiment of the present invention.

[0023] Figure 2 This is a schematic diagram of the core dual mechanism (mechanism A and mechanism B) used by the simulation engine in one embodiment of the present invention.

[0024] Figure 3 This is a network visualization diagram in one embodiment of the present invention, showing the emergence process of a certain technological hotspot from its initial state to its future state. Detailed Implementation

[0025] The inventive concept, technical solution, specific embodiments, and resulting technical effects of the present invention will be described in detail below with reference to the accompanying drawings, so as to fully understand the purpose, features, and effects of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0026] This invention aims to provide a novel technical solution capable of simulating the global dynamics of a scientific and technological knowledge network, revealing its intrinsic evolutionary mechanisms, and generatively extrapolating future network topologies. This solution primarily integrates two key technologies: complex network modeling and computer simulation. It abstracts the complex scientific research ecosystem into a dynamically evolving temporal knowledge network and utilizes a simulation engine based on its intrinsic generative mechanism to simulate and extrapolate the network's future structural evolution, particularly the emergence and evolution of new technological hotspots. It identifies subgraphs emerging due to structural compaction, predicts them as future technological hotspots, and provides technical characterization of these hotspots. Ultimately, this invention transforms the simulation results into clear predictions and quantitative representations of future technological hotspots, thus forming an intelligent analysis technology applicable to forward-looking technology forecasting and strategic R&D planning.

[0027] One embodiment of the present invention provides a method for predicting the evolution and emergence of technological hotspots based on network evolution. Figure 1 This is a flowchart of the prediction method. Please refer to it. Figure 1 The prediction method includes the following steps: Step S1: Initial Knowledge Network Construction. This step is mainly used to construct an initial network based on existing scientific and technological data. Specifically, it includes: First, obtaining structured data containing timestamps of achievements and projects, authors and their institutions, and subject keywords from one or more scientific and technological databases; then, abstracting the collaborating entities (such as researchers and research institutions) in this structured data into network nodes, and abstracting their collaborative relationships (such as jointly publishing papers or jointly serving the same project) into network edges, thus constructing an initial temporal knowledge network G(T0) carrying time and subject attributes.

[0028] Step S2: Dynamic Evolution Simulation Engine Based on Dual Generation Mechanism. This step is the core of the prediction method of this invention. The engine iterates forward in discrete time steps. Given a probability parameter p (ranging from 0 to 1) that regulates the network evolution mechanism, for each existing node i in the aforementioned constructed temporal knowledge network, it executes the following mechanism probabilistically using computer simulation, thereby simulating a complete evolutionary network. The principle of this dual generation mechanism is explained in [reference needed]. Figure 2 Specifically: (1) Mechanism A based on probability p: New node introduction and topic attachment mechanism. This mechanism simulates the dynamic process of new research forces entering the existing research field, realizing the simulation of field expansion. The computer calculates a new node attraction parameter λ based on the seniority τ(i) of node i in the network and a seniority influence coefficient α. i (t); based on this parameter λ i (t), using a probability distribution model (such as Poisson distribution), calculate the number Z of newly introduced nodes that node i will connect to at the current time step; subsequently, generate Z new nodes in the network and establish connections between them and node i, with the new nodes inheriting or associating with the topic attributes of node i. For example... Figure 2 As shown, the new node attraction parameter λ i The formula for calculating (t) is: , Wherein, the seniority of node i in the network τ(i) represents the existence time of the node in the network, t represents the total time that the network has evolved, and the seniority influence coefficient α ranges from 0 to 5.

[0029] (2) Mechanism B based on probability 1-p: Internal connection and community formation mechanism. This mechanism simulates the dynamic process by which existing researchers establish new cooperative relationships through collaborators, leading to an increasingly close internal structure of the research community (or scientific hotspot), thus simulating the emergence of hotspots. The computer identifies all 2-connected neighbor nodes j of node i, where a 2-connected neighbor is a non-direct neighbor connected through a common neighbor node; for each 2-connected neighbor j of node i, the connection probability μ of node i establishing a new connection with it is calculated based on its current network influence and an influence coefficient β. ij This mechanism is a key technological step driving the increase in local network density and ultimately forming high-density communities (scientific and technological hotspots). It achieves a dynamic simulation of the "triangular closure" phenomenon in networks. The network influence of node j is quantified through its degree value d(j) and other centrality indicators. Figure 2 As shown, the connection probability μ of node i establishing a new connection with its 2-connected neighbor is... ij The calculation formula is: , in, , Let i and j represent the degrees of node i and node j, respectively. This represents the maximum degree of a node in the network at time t-1, and the influence coefficient β ranges from 0 to 5.

[0030] Step S3: Future Hotspot Identification and Characterization. After the simulation engine in step S2 reaches the preset future time point T... n Then, a predictive future knowledge network G(T) is generated. n According to the network G(T) n The network's size and topology are analyzed, and community detection algorithms (such as Louvain's algorithm and clustering algorithms) are run to identify high-density subgraphs with tight internal connections. These identified high-density subgraphs are then marked as predicted future technological hotspots. Subsequently, for each predicted hotspot, further technological characterization is performed, using semantic vector technology to extract the most concentrated thematic keywords from its internal nodes, thereby defining the core technological keywords of the hotspot. Simultaneously, network structure indicators (size, density, average path length, etc.) are calculated for the hotspot to assess its potential influence, transforming the abstract network subgraph into structured and understandable information. Identifying high-density subgraphs with tight internal connections specifically includes: classifying the network G(T) into high-density subgraphs with tight internal connections. n Subgraphs with a clustering coefficient greater than 0.7 and a connection density greater than 0.5 are identified as high-density node subgraphs.

[0031] In this embodiment of the invention, the values ​​of the three parameters p, α, and β can be determined as follows: First, the knowledge network is derived as shown in equations (1), (2), and (3), where N(t) represents the number of nodes, M(t) represents the number of edges, and C(t) represents the number of triangular loops. Then, given a real network, the values ​​of the three parameters p, α, and β can be calculated by inversely based on the number of nodes, the number of edges, and the number of triangular loops in the real network.

[0032] (1) (2) (3) in, e Represents the natural logarithm. This represents the average degree of nodes in the network at time t-1; It is a combination number, which can also be written as .

[0033] The ranges of the three parameters p, α, and β can all be determined in the manner described above. In this embodiment of the invention, the effect is better when the three parameters are within the following ranges obtained by the above method: 0 < α ≤ 5, 0 < β ≤ 5, 0 < α ≤ 5, 0 < β ≤ 5. <p≤1。

[0034] Compared with the prior art, the prediction method of the present invention has the following advantages: (1) A technological breakthrough has been achieved, moving from summarizing and generalizing to forward-looking prediction. This invention is no longer limited to finding existing hotspots from historical data, but creates quantitative predictions of future network structures through a generative model with an inherent mechanism, which greatly improves the forward-looking nature and effectiveness of technology trend analysis.

[0035] (2) It provides an evolutionary analysis of the hotspot emergence mechanism. This invention simulates the underlying dynamics of domain expansion and community formation, making prediction no longer a black box and enabling it to have mechanistic interpretability, which helps to understand the internal logic of the formation of scientific and technological hotspots.

[0036] (3) Improved robustness for hotspot prediction. Compared with simple time series models and trend extrapolation models, this invention models the complex and nonlinear dynamic evolution of network cooperation, which can more accurately capture the emergence of hotspots driven by complex factors such as cross-domain cooperation, thus greatly improving the reliability of prediction results.

[0037] (4) It realizes the full-cycle simulation of hotspot emergence. The dynamic simulation method of the present invention can simulate the entire evolution path of a hotspot from its emergence, development to maturity, providing a brand-new technical means for judging the development stage of a hotspot.

[0038] The present invention will be described in more detail below through a specific embodiment.

[0039] Example 1 Step 1 (corresponding to step S1): Construction and preprocessing of the temporal knowledge network.

[0040] First, in selecting and collecting data sources, this specific embodiment of the invention chose the following three publicly available scientific and technological databases as data sources: Information (for obtaining academic paper data in the field of transportation and logistics), arXiv (for obtaining paper data in the field of computer science, especially artificial intelligence), Google Patents, and USPTO (for obtaining patent data related to related technologies). We set the collection time range to January 2018 (denoted as time T0) to January 2024 (denoted as time T1), and the collected metadata fields mainly include: title, abstract, author list, author affiliation, publication / publication date, keywords, etc. of the paper / research project / patent.

[0041] Subsequently, entity recognition and relationship construction were performed on the collected data. Specifically, recognition techniques from natural language processing were used to accurately extract the names of researchers from the author list and treat them as nodes in the network. Co-author relationships were then considered as collaborations, and an edge was established between the corresponding nodes. The weight of each edge was initially set to the number of collaborations. Each edge also included a corresponding timestamp indicating the year the paper was published.

[0042] Then, the topics / domains are labeled, and the abstracts of all papers / patents are analyzed using existing topic extraction models to automatically extract several topic clusters. Then, a label of the main research field is added to each node in the network. There can be multiple labels to reflect its cross-domain research characteristics.

[0043] Ultimately, we constructed an initial temporal knowledge network G(T1) containing approximately 10,000 nodes and 50,000 edges, with time and topic labels.

[0044] Step 2 (corresponding to step S2): Configuring and running the dynamic evolution simulation engine.

[0045] First, we set a set of parameters that align with the general situation in academia. We set the probability parameter p=0.8, based on the general observation that in academic environments, the vast majority of new collaborations arise from senior researchers mentoring new students or attracting new postdoctoral fellows, corresponding to the introduction of new nodes in the network; while the proportion of entirely new collaborations established through "collaborators' collaborators" is relatively low. The seniority influence coefficient is set at α=1.2, indicating a slight Matthew effect in a researcher's ability to attract collaborators. For example, a scholar active in the field for 10 years is more than twice as attractive as one active for 5 years, which is consistent with reality. The influence coefficient is set at β=1.5, indicating that when establishing connections within a field, people tend to collaborate with "2-connected neighbors" with high influence and many collaborators. This is a strong tendency towards clustering and a key driver of hotspot formation.

[0046] Evolution and simulation were then performed. The simulation period was set from January 2024 to January 2027 (denoted as time T2). The time step was set to one month, resulting in a total of 36 time steps. The simulation program was launched on a computer with high-performance memory. Starting from T1=1 (January 2024), the program sequentially traversed all nodes in G(T1), executing mechanism A with probability p and mechanism B with probability 1-p at each node, causing the network to continuously evolve. At each time step, the total number of nodes and edges in the network dynamically increased, and the network topology and local density underwent significant changes.

[0047] Step 3 (corresponding to step S3): Identification and technical characterization of future hotspot topologies.

[0048] After 36 time steps of simulation, we obtained a predicted future knowledge network G(T3) that has significantly evolved in size and structure. We then executed the efficient Louvain community detection algorithm, which quickly identifies sets of nodes in network G(T3) where internal connections are much denser than external connections—the so-called hotspot communities. We filtered out mature communities that already existed in the initial network G(T1), focusing instead on newly emerging communities or those formed by the merging of several older communities, exhibiting extremely high network modularity scores. Network modularity score is a classic statistical indicator in networking: it is calculated by comparing the proportion of edges within a community in a real network with the expected proportion of edges in a random network that maintains constant node degree, and calculating the difference between the two. The final prediction results are as follows: The system identified a new community that was almost non-existent in G(T2) but was large and extremely densely structured in G(T3). After removing nodes from the evolutionary network that were not part of this new community, we obtained the temporal evolution of this new community, as follows:Figure 3 As shown, the network evolution process from time step 0 (containing only one node) to time step 11 is illustrated. By analyzing the topic tags of the nodes within this community, we found that its core keywords are highly concentrated on "intelligent agent" and "logistics delivery". The system automatically named it the predicted hotspot: "Agent-based logistics order delivery". Then, the system further analyzed its formation path and found that the community was formed by key nodes that initially belonged to two remote communities, "deep learning" and "logistics scheduling", which were gradually connected and merged during the simulation through the "2-connected neighbor" mechanism. Finally, the system used AIGC (Artificial Intelligence Content Generation Technology) to output a prediction report, pointing out that "within the next three years, 'agent-based logistics order delivery' is very likely to develop into a highly influential interdisciplinary scientific and technological hotspot". The report can also list the core technological components of this predicted hotspot and a list of noteworthy researchers who played a key bridging role in the formation of this hotspot.

[0049] Through this specific example, we have put into practice the technical solution of "a prediction method for the evolution and emergence of technological hotspots based on network evolution." Practical results show that the prediction method of this invention can effectively predict short- to medium-term technological hotspots using dynamic, global simulation evolution. This demonstrates that this invention not only provides a new perspective for tracking and predicting technological hotspots but also provides decision support for the precise allocation of technological resources.

[0050] Another embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, can implement the steps of the aforementioned prediction method for the evolution and emergence of technological hotspots based on network evolution.

[0051] Another embodiment of the present invention provides a prediction system for the evolution and emergence of technological hotspots based on network evolution, comprising: a data acquisition module, a network construction module, a dynamic evolution simulation engine module, and a hotspot identification and analysis module. Wherein: The data acquisition module includes a memory that stores computer programs for acquiring and preprocessing scientific and technological data. This acquisition and preprocessing includes: acquiring structured data from a scientific and technological database that contains time information, author information, and subject keywords related to research projects. Specific acquisition and preprocessing methods are described in step S1 above and will not be repeated here.

[0052] The network construction module includes a memory that stores a computer program for constructing a temporal knowledge network. The construction of the temporal knowledge network includes: based on the structured data, abstracting the author information into network nodes and the cooperation relationship into network edges to construct an initial temporal knowledge network carrying time and topic attributes; the specific construction method is described in step S1 above and will not be repeated here.

[0053] The dynamic evolution simulation engine module includes a memory that stores a dynamic evolution simulation program for executing a dual-generation mechanism. This program can simulate the dynamic evolution of the network structure by performing forward iterative simulation based on the initial temporal knowledge network and using a dynamic evolution simulation engine based on mechanisms A and B. Mechanism A is a new node introduction and topic attachment mechanism used to simulate the expansion of research fields; mechanism B is an internal connection and community formation mechanism used to simulate the emergence of scientific and technological hotspots. The program can also form a predicted future knowledge network after the simulation reaches a preset future time point. The specific simulation process is described in step S2 above and will not be repeated here.

[0054] The hotspot identification and analysis module includes a memory storing computer programs for implementing hotspot identification and analysis. These programs are used to: identify high-density node subgraphs in the future knowledge network using a community discovery algorithm, mark them as predicted future technological hotspots, and perform technical characterization on these hotspots. Specific identification, analysis, and characterization methods are detailed in step S3 above and will not be repeated here.

Claims

1. A predictive method for the evolution and emergence of technological hotspots based on network evolution, characterized in that, Includes the following steps: Step S1: Construction of Temporal Knowledge Network: Obtain structured data containing time information, author information and topic keywords of scientific research projects from the science and technology database, abstract the author information as network nodes, and abstract the cooperation relationship as the edge of the network to construct an initial temporal knowledge network carrying time and topic attributes; Step S2, Network Dynamic Evolution Simulation: Based on the initial temporal knowledge network, a dynamic evolution simulation engine is used to perform forward iterative simulation to simulate the dynamic evolution of the network structure. Step S3, Future Hotspot Identification and Characterization: After the simulation in step S2 reaches the preset future time point, based on the future knowledge network generated by the simulation, the high-density node subgraph in the future knowledge network is identified by the community discovery algorithm, marked as the predicted future technology hotspot, and the future technology hotspot is technically characterized.

2. The prediction method as described in claim 1, characterized in that, The author information mentioned in step S1 includes the author's name and institution, and the collaboration includes jointly publishing papers or jointly serving the same research project.

3. The prediction method as described in claim 1, characterized in that, Step S2 specifically includes: given a probability parameter p that regulates the network evolution mechanism, the simulation engine simulates the dynamic evolution of the network structure by iterating forward in discrete time steps through computer simulation, executing the following mechanism: Mechanism A is executed based on probability p: Mechanism A is a new node introduction and topic attachment mechanism used to simulate the expansion of the research field; Mechanism B is executed according to probability 1-p: Mechanism B is an internal connection and community formation mechanism used to simulate the emergence of technological hotspots.

4. The prediction method as described in claim 3, characterized in that, Mechanism A simulates the expansion of research fields by simulating the dynamic process of new research forces entering existing research fields; Mechanism B simulates the emergence of scientific and technological hotspots by simulating the dynamic process of existing researchers establishing new cooperative relationships through collaborators, leading to an increasingly close internal structure of the research community.

5. The prediction method as described in claim 4, characterized in that, The mechanism A specifically includes: calculating a new node attraction parameter λ based on node i's seniority τ(i) in the network and a seniority influence coefficient α. i (t); based on this parameter λ i (t), through a probability distribution model, calculate the number Z of new nodes that node i will connect to at the current time step; then, generate Z new nodes in the network and establish their connection with node i. The new nodes inherit or associate with the topic attributes of node i.

6. The prediction method as described in claim 4, characterized in that, The mechanism B specifically includes: identifying all two-connected neighbor nodes j of node i, where a two-connected neighbor is a non-direct neighbor connected through a common neighbor node; for each two-connected neighbor j of node i, calculating the connection probability μ of node i establishing a new connection with it based on its current network influence and an influence coefficient β. ij .

7. The prediction method as described in claim 1, characterized in that, Step S3 specifically includes: The simulation engine, after executing step S2, reaches the preset future time point T. n Then, a predictive future knowledge network G(T) is generated. n According to this future knowledge network G(T) n The size and topology of network G(T) are determined, and a community detection algorithm is run to identify the network. n High-density node subgraphs with clustering coefficients and connection densities higher than preset values ​​are identified; these high-density node subgraphs are marked as predicted future technological hotspots. For each predicted future technology hotspot, semantic vector technology is used to extract the most concentrated thematic keywords within its internal nodes, which are then used as the core technology keywords for that future technology hotspot.

8. The prediction method as described in claim 7, characterized in that, Step S3 further includes: for each predicted future technology hotspot, calculating its network structure index to assess its potential impact, and transforming the abstract network subgraph into structured, understandable information; wherein the network structure index includes network size, connection density, and average path length.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program can implement the steps of the prediction method according to any one of claims 1-8.

10. A predictive system for the evolution and emergence of technological hotspots based on network evolution, characterized in that, include: The data acquisition module is configured to retrieve structured data from scientific and technological databases, including time information, author information, and subject keywords for scientific research projects. The network construction module is configured to abstract the author information into network nodes and the cooperation relationship into network edges based on the structured data, thereby constructing an initial temporal knowledge network carrying time and topic attributes. The dynamic evolution simulation engine module is configured to simulate the dynamic evolution of the network structure by performing forward iterative simulation based on the initial temporal knowledge network using the dynamic evolution simulation engine. And after the simulation reaches the preset future time point, a predicted future knowledge network is formed; as well as The hotspot identification and analysis module is configured to identify high-density node subgraphs in the future knowledge network through a community discovery algorithm, mark them as predicted future technology hotspots, and perform technical characterization on the future technology hotspots.

Images