A visualization system for presenting industry maps based on scientific and technological innovation data

Abstract

The present invention relates to the field of data visualization technology, and aims to solve the problem that it is difficult to map the complex scientific and technological innovation ecological structure using conventional charts when presenting the scientific and technological innovation industry pattern in the prior art, which makes it difficult to provide strong support for regional industrial planning and precise resource allocation. To this end, the present invention provides a visualization system for presenting an industrial map based on scientific and technological innovation data, including: a data acquisition module collects scientific and technological innovation data of all scientific and technological innovation subjects in the target area; an associated graph construction module constructs an associated graph of scientific and technological innovation subjects; an associated graph processing module uses a graph convolutional network to process the associated graph of scientific and technological innovation subjects; a judgment module judges whether the target node has a scientific and technological innovation strength label; a label assignment module assigns matching scientific and technological innovation strength labels to the target node and the potential strength subject associated with it; and an industrial map visualization module constructs an industrial map and presents it visually. The present invention can present the association of scientific and technological innovation subjects in the form of an industrial map, providing a decision-making basis for decision makers.

Description

Technical Field

[0001] The present invention relates to the technical field of data visualization, and in particular to a visualization system for presenting an industrial map based on scientific and technological innovation data. Background Art

[0002] Despite the booming development of science and technology innovation and the massive accumulation of data, existing means are insufficient to deeply explore the value of data and accurately understand the pattern of science and technology innovation. Traditional data analysis focuses on single-dimensional static data such as financial indicators and number of patents of isolated science and technology innovation entities, ignoring the rich information contained in the complex network of associations between entities. For example, when evaluating the innovation strength of a technology company, only considering the number of its patent applications will not provide insight into the synergistic effect of industry-university-research cooperation behind the generation of patents, and the linkage relationship between technology iterations of upstream and downstream companies, which can easily misjudge the company's real innovation potential and industry influence.

[0003] When presenting the scientific and technological innovation industry structure, conventional charts (bar charts, line charts, etc.) can only show simple data trends and proportion relationships, and are unable to map the complex scientific and technological innovation ecological structure. It is impossible to intuitively present how the scientific research results of universities penetrate into the production end of enterprises through the industry-university-research chain, and the cross-field cooperation and innovation paths of scientific and technological innovation entities in different industries. It is difficult to provide strong visual support for regional industrial planning and precise resource allocation, making it difficult for decision makers to control the pulse of scientific and technological innovation from a global perspective and formulate highly compatible development strategies, which restricts the coordinated development of regional scientific and technological innovation and the improvement of intensive resource utilization efficiency.

[0004] In view of this, the art needs a visualization system that presents an industry map based on scientific and technological innovation data to solve the above problems. Summary of the invention

[0005] In order to solve the above technical problems, that is, to solve the problem that the conventional charts used in the existing technology to present the scientific and technological innovation industry structure are difficult to map the complex scientific and technological innovation ecological structure, which makes it difficult to provide strong support for regional industrial planning and precise allocation of resources.

[0006] The present invention provides a visualization system for presenting an industry map based on scientific and technological innovation data, the visualization system comprising:

[0007] A data collection module, which is used to collect scientific and technological innovation data of all scientific and technological innovation entities in the target area;

[0008] A correlation graph construction module, which is used to construct a correlation graph of scientific and technological innovation entities based on all the scientific and technological innovation entities and the scientific and technological innovation data, wherein all the scientific and technological innovation entities constitute nodes of the scientific and technological innovation entity correlation graph, the scientific and technological innovation data of each scientific and technological innovation entity constitutes the node attribute of the node, and the cooperation between the scientific and technological innovation entities constitutes the edges of the associated scientific and technological innovation entities;

[0009] A correlation graph processing module, which is used to process the correlation graph of the scientific and technological innovation subject using a graph convolutional network to obtain high-order node features of each node;

[0010] A judgment module, which is used to judge whether the target node of the processed association map has a scientific and technological innovation strength label;

[0011] A label assignment module, which is used to assign a matching scientific and technological innovation strength label to the target node when the target node of the processed association map does not have a scientific and technological innovation strength label, and to identify the potential strength subject associated with the target node by using label propagation and label update iteration when the target node of the processed association map has a scientific and technological innovation strength label and assign a matching scientific and technological innovation strength label to the potential strength subject;

[0012] An industry map visualization module is used to construct an industry map and visualize it based on the target area, the processed association map and all the scientific and technological innovation strength labels.

[0013] In some preferred embodiments, the graph convolutional network is used to process the scientific and technological innovation subject association graph to obtain the high-order node features of each node, including:

[0014] Build a 3-layer GCN network architecture;

[0015] Direct neighbor attribute aggregation is performed in the first GCN layer;

[0016] The output of the first GCN layer is used as the input of the second GCN layer to perform second-order neighbor information fusion;

[0017] The output of the second GCN layer is used as the input of the third GCN layer for global feature integration and output to obtain the high-order node features of each node.

[0018] In some preferred embodiments, the construction of a 3-layer GCN network architecture includes:

[0019] Construct a network architecture containing three layers of graph convolutional layers according to the following formula:

[0020] ;

[0021] in, It is The node feature representation matrix of the layer, It is The node feature representation matrix of the layer, is the activation function, is the degree matrix, , is the adjacency matrix, is the identity matrix, It is The learnable weight matrix of the layer is hour, is a node initial feature matrix containing all the scientific and technological innovation subjects and the scientific and technological innovation data;

[0022] In the first GCN layer, according to Perform calculations;

[0023] In the second GCN layer, according to Perform calculations;

[0024] In the third GCN layer, according to Perform calculations.

[0025] In certain preferred embodiments, when the target node of the processed association graph does not have a scientific and technological innovation strength label, assigning a matching scientific and technological innovation strength label to the target node includes:

[0026] The label matching score of the scientific and technological innovation subject corresponding to the target node is calculated using the following formula:

[0027] ;

[0028] in, is the calculated label matching score of the scientific and technological innovation entity, Score the innovation achievements of scientific and technological innovation entities. The score of R&D investment of scientific and technological innovation entities. Score the talent resources of the scientific and technological innovation entities. The market performance score of the scientific and technological innovation entity. Score the cooperation network of scientific and technological innovation entities. , , , and They are the weights of the scientific and technological innovation entities’ innovation achievements, R&D investment, talent resources, market performance and cooperation network;

[0029] Determine the score range of the calculated label matching score of the scientific and technological innovation entity;

[0030] Based on the score range of the label matching score of the scientific and technological innovation subject, the scientific and technological innovation subject is assigned a corresponding scientific and technological innovation strength label.

[0031] In certain preferred embodiments, the method of identifying potential strength subjects associated with a target node by adopting a label propagation and label update iteration method and assigning a matching scientific and technological innovation strength label to the potential strength subjects includes:

[0032] Perform unique hot encoding on the target node's scientific and technological innovation strength label and set the propagation attenuation coefficient ;

[0033] Node-based and nodes The edge weight between , according to the formula Compute Node From neighbor nodes The propagation probability of receiving label information ;

[0034] The label update iteration is performed according to the following formula:

[0035] ;

[0036] in, Is a node The updated label vector, Is a node The set of neighbor nodes of It is a node The sum of the propagation probabilities of all neighbor nodes, Neighbor node The known label vector of the neighbor nodes The known label vector is the one-hot encoding vector of the target node;

[0037] Based on the updated iteration Assign matching scientific and technological innovation strength labels to the potential strength entities.

[0038] In some preferred embodiments, the termination condition of the label update iteration is:

[0039] ;

[0040] in, is the label vector dimension, It is Iteration node label vector elements, It is Iteration node label vector elements, is the number of iterations, is the label vector convergence judgment threshold.

[0041] In some preferred embodiments, the label vector convergence determination threshold Equal to 0.01.

[0042] In certain preferred embodiments, the visualization system further includes a policy retrieval and storage module, which is used to retrieve and store the science and technology innovation support policy information in the target area.

[0043] In certain preferred embodiments, when a scientific and technological innovation subject is selected in the industrial map, the industrial map visualization module matches the support policy information corresponding to the scientific and technological innovation subject from the policy retrieval and storage module and presents it visually.

[0044] In some preferred embodiments, the industrial map supports regional zooming;

[0045] When the industry map is in a regional zoom state and a scientific and technological innovation subject is selected in the zoomed area:

[0046] If all other entities associated with the selected scientific and technological innovation entity are within the zoomed area, their associations will be presented by connecting them with highlighted lines;

[0047] If some of the other subjects associated with the selected scientific and technological innovation subject are within the zoomed area and the other subjects are not within the zoomed area, the other subjects within the zoomed area will be connected by highlight lines to show their association with the selected scientific and technological innovation subject, and the other subjects not within the zoomed area will be shown their association with the selected scientific and technological innovation subject in a text pop-up window;

[0048] If all other entities associated with the selected scientific and technological innovation entity are not within the zoomed area, their mutual relationships will be presented in the form of a text pop-up window.

[0049] It can be seen from the above that the visualization system for presenting an industrial map based on scientific and technological innovation data provided by the present invention has the following beneficial technical effects:

[0050] The present invention collects the scientific and technological data of all scientific and technological entities in the target area, and then weaves the fragmented scientific and technological information into a correlation map, and mines the complex cooperative relationship between the entities as the edge information of the map. It can break through the limitations of traditional single-dimensional static analysis, deeply understand the internal logic of the scientific and technological innovation ecosystem, and provide a data basis for subsequent accurate decision-making. The correlation map processing module uses a graph convolutional network (GCN) to deeply analyze the correlation map, extract the high-order node features of each node, and the judgment module and the label assignment module work together to innovate the logic of label assignment of scientific and technological innovation strength. For unlabeled nodes, it can scientifically evaluate and match labels; for labeled nodes, the label propagation and update iteration mechanism is based on edge weights and multiple iterations. The industry map visualization module integrates regional geographic information, processed related maps and scientific and technological innovation strength labels to present the industry map more intuitively and visually, allowing decision makers to get rid of the troubles of complicated data reports and directly understand the spatial distribution of the scientific and technological innovation industry, the strength pattern of the main body, and the density of cooperation. It guides relevant departments to accurately plan scientific and technological innovation parks and allocate innovative resources, helping enterprises to quickly find partners and benchmark industry benchmarks, and scientific research institutes to clarify the direction of results transformation, greatly improving the efficiency of regional scientific and technological innovation collaboration, reducing resource mismatch, and gathering innovation synergy to promote high-quality and sustainable development of regional scientific and technological innovation industries. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] The features and advantages of the present invention will be more clearly understood by referring to the accompanying drawings, which are schematic and should not be construed as limiting the present invention in any way. In the accompanying drawings:

[0052] Figure 1 It is a structural schematic diagram of the visualization system for presenting an industry map based on scientific and technological innovation data according to the present invention. DETAILED DESCRIPTION

[0053] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0054] Based on the background technology, it is difficult to map the complex scientific and technological innovation ecological structure by using conventional charts when presenting the scientific and technological innovation industry pattern in the prior art, which makes it difficult to provide strong support for regional industrial planning and precise resource allocation. The present invention provides a visualization system for presenting industrial maps based on scientific and technological innovation data, which can more intuitively and visually present industrial maps related to scientific and technological innovation entities. Decision makers can get rid of the troubles of complicated data reports, and can directly understand the spatial distribution of scientific and technological innovation industries, the strength pattern of entities, and the density of cooperation. It guides relevant departments to accurately plan scientific and technological innovation parks and allocate innovative resources, helping enterprises to quickly find partners and benchmark industry benchmarks, and scientific research institutes to clarify the direction of achievement transformation, greatly improving the efficiency of regional scientific and technological innovation collaboration, reducing resource mismatch, and gathering innovation synergy to promote high-quality and sustainable development of regional scientific and technological innovation industries.

[0055] like Figure 1 As shown, the visualization system for presenting an industrial map based on scientific and technological innovation data of the present invention includes: a data acquisition module 100, a correlation map construction module 200, a correlation map processing module 300, a judgment module 400, a label assignment module 500 and an industrial map visualization module 600. The function of each module is described below:

[0056] A data collection module 100, which is used to collect the scientific and technological innovation data of all scientific and technological innovation entities in the target area;

[0057] Among the above, the scientific and technological innovation data of scientific and technological innovation entities can be collected from a variety of data sources, including records of government science and technology departments, patent databases, corporate financial reports, university and research institute results release platforms, etc. The scientific and technological innovation entities include scientific and technological innovation enterprises, research institutes, universities, etc. The scientific and technological innovation data can include innovation results data (such as patent data, paper data), and can also include R&D investment data, major project participation data, etc.

[0058] The association graph construction module 200 is used to construct an association graph of scientific and technological innovation entities based on all scientific and technological innovation entities and scientific and technological innovation data, wherein all scientific and technological innovation entities constitute nodes of the scientific and technological innovation entity association graph, the scientific and technological innovation data of each scientific and technological innovation entity constitutes the node attribute of the node, and the cooperation between scientific and technological innovation entities constitutes the edges of the associated scientific and technological innovation entities;

[0059] Preferably, the node attributes of each node may include not only the scientific and technological innovation data of the scientific and technological innovation subject, but also the basic information of the scientific and technological innovation subject (such as name, establishment time, industry or subject field, etc.).

[0060] In a specific case, for the node of scientific and technological innovation subject, in the dimension of innovation results, the number of invention patents, utility model patents and design patents are counted separately; for example, the weight of invention patents is set to 0.5, and the number of PCT applications and domestic independent applications are distinguished when counting; the weight of utility model patents is set to 0.3, and they are subdivided by technical field; the weight of design patents is set to 0.2, and they are sorted according to design style and application product category; at the same time, the number of papers published is weighted, such as SCI papers are classified according to impact factor (high impact factor>5.0 has a weight of 0.4, high impact factor>5.0 has a weight of 0.5, high impact factor>5.0 has a weight of 0.6, high impact factor>5.0 has a weight of 0.7, high impact factor>5.0 has a weight of 0.8, high impact factor>5.0 has a weight of 0.9, high impact factor>5.0 has a weight of 0.10, high impact factor>5.0 has a weight of 0.2, high impact factor>5.0 has a weight of 0.11, high impact factor>5.0 has a weight of 0.9, high impact factor>5.0 has a weight of 0.12, high impact factor>5.0 has a weight of 0.13, high impact factor>5.0 has a weight of 0.14, high impact factor>5.0 has a weight of 0.15 The weight of the papers indexed by EI is set to 0.3, the weight of the papers in core journals is set to 0.2, and the weight of the papers published in ordinary journals is set to 0.1. The number of citations of the papers and the distribution of the institutions of the co-authors are tracked. In terms of R&D investment attributes, the annual R&D investment amount in the past five years is recorded, and a time series is constructed and normalized through Z-score. According to the proportion of investment to revenue, it is divided into high investment (>15%, weight 0.4), medium investment (10%-15%, weight The four levels are as follows: weight 0.3), low investment (5%-10%, weight 0.2), micro investment (<5%, weight 0.1); in terms of highlighting the attributes related to the level of scientific research projects, the participation of national major special projects is given a weight of 0.6 according to the project type (such as major projects of the National Natural Science Foundation of China, special projects of the National Key R&D Program, etc.), and the number of times the institutes are the leading units of the projects, the number of sub-topics undertaken and the proportion of funds are counted; the weight of provincial and ministerial projects is set at 0.4, and they are classified and counted according to the fields of local science and technology strategic priorities (such as marine science and technology projects in coastal areas, scientific research plans for ecological restoration in the western region, etc.); in addition, the value of scientific research equipment assets can be accurately counted, classified and evaluated by equipment type (large scientific research instruments such as nuclear magnetic resonance instruments, electron microscopes, etc., conventional experimental equipment such as centrifuges, chromatographs, etc.), and a comprehensive value score is given in combination with the purchase age of the equipment (considering the technological advancement of new equipment and the operational stability and experience accumulation of old equipment) and the frequency of shared use (internal service to scientific research teams, external industry-university-research cooperation testing services); through the above methods, the node attributes of scientific and technological innovation data can be assigned and quantified, which is convenient for constructing the initial feature matrix of the nodes.

[0061] In terms of edge construction and weights, for cooperation between enterprises, if two enterprises jointly participate in a major national science and technology project, the initial edge weight is assigned +5, and is multiplied by coefficients of 1.0, 1.2, and 1.5 respectively according to the project cycle (short-term 1-2 years, medium-term 3-4 years, long-term 5 years and above); each time there is a joint patent application, the weight is +1, and it is subdivided and added according to the patent type (invention patent joint application weight +1.5, utility model +1, design +0.5) and patent application region (international joint application weight is doubled, and domestic cross-regional joint application weight is moderately increased by 0.3); for universities and For the industry-university-research cooperation relationship between enterprises, 1 point is calculated for every 1 million yuan of the cooperation agreement amount, and then weighted according to the nature of the project (basic research cooperation coefficient is 0.8, applied technology research and development is 1.2, and achievement transformation and implementation is 1.5); the number of professional talents sent each year is calculated for each person, and points are accumulated based on the educational level of the talents (PhD 1.5 points / person, Master 1.0 point / person, undergraduate 0.5 point / person) and professional fit (close fit 1.2 times the score, related majors 1.0 times, cross-professional 0.8 times); through these rules, an association map with a weighted directed graph structure is constructed to represent the complex relationship network between scientific and technological innovation entities.

[0062] The association graph processing module 300 is used to process the association graph of the scientific and technological innovation subject using a graph convolutional network to obtain the high-order node features of each node;

[0063] In some preferred implementations, the above-mentioned use of graph convolutional networks to process the scientific and technological innovation subject association graph to obtain high-order node features of each node includes:

[0064] Construct a 3-layer GCN network architecture: Construct a network architecture containing 3 layers of graph convolutional layers according to the following formula:

[0065] ;

[0066] in, It is The node feature representation matrix of the layer, It is The node feature representation matrix of the layer, is the activation function, is the degree matrix, , is the adjacency matrix, is the identity matrix, It is The learnable weight matrix of the layer is hour, is a node initial feature matrix containing all the scientific and technological innovation subjects and the scientific and technological innovation data;

[0067] Direct neighbor attribute aggregation in the first GCN layer: In the first GCN layer, according to Perform calculations;

[0068] The output of the first GCN layer is used as the input of the second GCN layer to perform second-order neighbor information fusion: In the second GCN layer, according to Perform calculations;

[0069] The output of the second GCN layer is used as the input of the third GCN layer for global feature integration and output: In the third GCN layer, according to Calculations are performed to obtain high-order node features for each node.

[0070] Taking the regional biomedical entities as an example, in the biomedical map represented by the weighted directed graph, The value of reflects the relationship between subjects according to the set edge weight rule. For example, "Pharmaceutical Company A" and "Institute B" jointly participate in a major national biomedical project "New Tumor Immunotherapy Research and Development". The initial edge weight is +5, and the project cycle is 3 years (corresponding to a coefficient of 1.2). During this period, there are 2 joint patent applications (1 invention patent and 1 utility model patent, plus points according to the patent type). The comprehensive calculation is Corresponding weights, and so on, determine the weights of all edges in the graph to form an adjacency matrix , reflecting the closeness of cooperation among the subjects; by calculating For "Pharmaceutical Company A", after introducing self-loop information, it ensures that its own characteristics (such as unique R&D pipeline advantages, internal R&D team structure, etc.) participate in the calculation in the subsequent information dissemination, improves the basis for iterative update of node information, and makes Fusion of self-loop and inter-subject correlation information; calculation Get the degree matrix The sum of the corresponding row elements of "Pharmaceutical Company A" reflects the total amount of its association with surrounding entities and the accumulated weights, which is then normalized to obtain , balance the differences in the number of neighbors of different nodes in the graph, avoid information aggregation biased towards high-connectivity nodes, and ensure effective transmission and utilization of information from nodes that focus on niche areas, have relatively few connections but strong innovation capabilities.

[0071] In the first GCN layer, the adjacency matrix that integrates the self-loop information is constructed , degree matrix , node initial feature matrix and a randomly initialized weight matrix As input, according to the graph convolutional network formula For "Pharmaceutical Company A", the adjacency matrix can clearly show the connection with neighbors such as "Institute B", "Company C", and "University D". Combined with the node feature matrix, the first-level calculation can obtain the average patent conversion rate of neighbors and the average R&D investment intensity of surrounding cooperative enterprises. The essence of neighbor attributes is condensed and integrated into itself, and its own feature representation is updated to achieve the first step of information mining based on graph association. The second layer GCN layer uses the output of the first layer As input, continue to follow the graph convolutional network formula Calculations are performed. "Pharmaceutical Company A" examines the partners of its partners. For example, "Institute B" has in-depth cooperation with internationally renowned biomedical research centers, participates in cutting-edge gene editing anti-cancer therapy international projects, and accumulates advanced gene targeting technology and clinical transformation experience. Although "Pharmaceutical Company A" has no direct connection with the international research center, it is inspired by the second-layer GCN information transmission to adjust its R&D strategy, focusing on the development of traditional small molecule drugs, expanding the layout of gene therapy tracks, introducing new target screening technologies to optimize the R&D process, and using the innovative power of second-order neighbors to broaden horizons and resource allocation paths, and deepen information fusion and utilization in the graph ecosystem. The third layer, GCN, uses the output of the second layer As input, according to the formula Calculate and output high-dimensional node embedding representation after three layers of information condensation and feature learning From the perspective of the global biopharmaceutical industry, after this layer of calculation, the distance between the nodes of the graph reflects the similarity and closeness of scientific and technological innovation in the high-dimensional abstract space, so that similar pharmaceutical companies are clustered (such as pharmaceutical companies focusing on immunotherapy are clustered), and different industrial sectors are clearly distinguished (pharmaceutical companies, scientific research services, equipment research and development, etc. are clearly distinguished), providing an analytical basis for subsequent label propagation and other analyses.

[0072] A judgment module 400 is used to judge whether the target node of the processed association map has a scientific and technological innovation strength label;

[0073] The label assignment module 500 is used to assign a matching scientific and technological innovation strength label to the target node when the target node of the processed association map does not have a scientific and technological innovation strength label, and to identify the potential strength subject associated with the target node by using label propagation and label update iteration when the target node of the processed association map has a scientific and technological innovation strength label and assign a matching scientific and technological innovation strength label to the potential strength subject;

[0074] In the above, when the target node of the processed association graph does not have a scientific and technological innovation strength label, assigning a matching scientific and technological innovation strength label to the target node includes:

[0075] The following formula is used to calculate the label matching score of the scientific and technological innovation subject corresponding to the target node:

[0076] ;

[0077] in, is the calculated label matching score of the scientific and technological innovation subject, Score the innovative achievements of scientific and technological innovation entities. The score of R&D investment of scientific and technological innovation entities. Score the talent resources of the scientific and technological innovation entities. The market performance score of the scientific and technological innovation entity. Score the cooperation network of scientific and technological innovation entities. , , , and They are the weights of the scientific and technological innovation subject's innovation achievements, R&D investment, talent resources, market performance and cooperation network respectively; preferably, the innovation achievement score comprehensively considers the number of domestic invention applications, the number of PCT applications, the invention authorization rate, the number of utility model patent applications, the number of design patent applications, the number of papers, etc.; the R&D investment score comprehensively considers the intensity of capital investment and R&D equipment assets; the talent resource score comprehensively considers the scale and quality of scientific researchers and talent training and incentives; the market performance score comprehensively considers the product market share and the speed of product innovation iteration; the cooperation network score comprehensively considers the depth of industry-university-research cooperation and the breadth of cooperation between enterprises.

[0078] Determine the score range of the calculated label matching score of the scientific and technological innovation entity;

[0079] Based on the score range of the label matching score of the scientific and technological innovation subject, the corresponding scientific and technological innovation strength label is assigned to the scientific and technological innovation subject;

[0080] For example, the tag matching rule is:

[0081] like (full score 100), and awarded the label of "Excellent Innovation Leader", representing excellent performance in all dimensions and leading the innovation and development of the industry;

[0082] like , giving the label of "backbone innovation-driven enterprise", indicating that it has a solid innovation foundation and market competitiveness, and is the backbone of the industry;

[0083] like , giving it the label of "Growth-oriented Innovation Potential Enterprise", highlighting the innovation growth space and development potential, which need to be continuously nurtured and supported;

[0084] like , giving it the label of "start-up and incubation enterprise", which means that it is in the innovation and accumulation stage and needs to improve its strength in all aspects;

[0085] It should be noted that in order to ensure The scores can be calculated to be in the range of 1-100. For each sub-dimension in innovation results, R&D investment, talent resources, market performance and cooperation network, Z-score standardization and / or normalization can be performed to a specific range.

[0086] In the above, the label propagation and label update iteration methods are used to identify the potential strength subjects associated with the target node and assign matching scientific and technological innovation strength labels to the potential strength subjects, including:

[0087] Perform unique hot encoding on the target node's scientific and technological innovation strength label and set the propagation attenuation coefficient ;

[0088] Node-based and nodes The edge weight between , according to the formula Compute Node From neighbor nodes The propagation probability of receiving label information ;

[0089] The label update iteration is performed according to the following formula:

[0090] ;

[0091] in, Is a node The updated label vector, Is a node The set of neighbor nodes of It is a node The sum of the propagation probabilities of all neighbor nodes, Neighbor node The known label vector of neighbor nodes The known label vector of is the one-hot encoding vector of the target node;

[0092] Based on the updated iteration Assign matching scientific and technological innovation strength labels to potential strength entities;

[0093] The termination condition of the label update iteration is:

[0094] ;

[0095] in, is the label vector dimension, It is Iteration node label vector elements, It is Iteration node label vector elements, is the number of iterations, is the label vector convergence determination threshold, which is equal to 0.01.

[0096] Still taking the aforementioned regional biomedical entities as an example, assuming that in the regional biomedical map, some leading pharmaceutical companies (such as "Pharmaceutical Company E") are rated as "Excellent Innovation Leading Enterprises", and excellent scientific research institutes (such as "Institute F") have been awarded the "Excellent Innovation Leading Enterprise" label due to their undertaking of multiple major national projects, top scientific research equipment, and high-impact paper publications. The encoding is in the form of a unique hot vector and is set as . Set the propagation attenuation coefficient , according to the edge weight By formula Calculate the probability of label propagation between nodes. For example, the edge weight between "Pharmaceutical Company A" and "Pharmaceutical Company E" is 10 (due to joint research and development and frequent talent exchanges), then , ensuring that high-weight edges efficiently transfer labels. According to the formula For Node (Here, "Pharmaceutical Company A" is taken as an example) The label is updated. "Pharmaceutical Company A" works closely with other entities and other neighbors, and has multiple high-weight edge connections. Its invention patents have grown steadily in the past three years (annual average growth rate of 12%), and two innovative drugs have entered the clinical trial stage, with outstanding innovation output. In the first round of label propagation, the weighted sum is calculated based on the probability formula to update the label score after the propagation probability of each neighbor is calculated. After multiple rounds of iterations, based on the cumulative high-confidence label score, if the converged label score fits the "backbone innovation-driven enterprise" range, this label is assigned. Relevant departments can use this to accurately identify, allocate special R&D funds, and grant tax incentives to activate the vitality of regional biomedical innovation and promote the advanced and high-quality development of the industry.

[0097] The industry map visualization module 600 is used to construct and visualize the industry map based on the target area, the processed association map and all scientific and technological innovation strength labels.

[0098] Preferably, the industry map supports regional zooming;

[0099] When the industry map is in the regional zoom state and a scientific and technological innovation entity is selected in the zoomed area:

[0100] If all other entities associated with the selected scientific and technological innovation entity are within the zoomed area, their associations will be presented by connecting them with highlighted lines;

[0101] If some of the other subjects associated with the selected scientific and technological innovation subject are within the zoomed area and the other subjects are not within the zoomed area, the other subjects within the zoomed area will be connected by highlight lines to show their association with the selected scientific and technological innovation subject, and the other subjects not within the zoomed area will be shown their association with the selected scientific and technological innovation subject in a text pop-up window;

[0102] If all other entities associated with the selected scientific and technological innovation entity are not within the zoomed area, their mutual relationships will be presented in the form of a text pop-up window.

[0103] In some preferred cases, the visualization system further includes a policy retrieval and storage module, which is used to retrieve and store the science and technology innovation support policy information in the target area. Further, when the science and technology innovation subject in the industrial map is selected, the industrial map visualization module matches the support policy information corresponding to the science and technology innovation subject from the policy retrieval and storage module and presents it visually.

[0104] Specifically, the following policy information matching model is used to match the corresponding policy information on scientific and technological innovation support for scientific and technological innovation entities:

[0105] ;

[0106] in, The degree of match between scientific and technological innovation entities and support policies. is the number of dimensions considered in the matching process. For the The weight of each matching dimension in the overall matching evaluation. For scientific and technological innovation In the Standardized scores on the matching dimensions, To support policy In the The fit score on the matching dimension.

[0107] In the above, The value range is between 0 and 1. The closer to 1, the higher the matching degree, which means that the policy is more applicable to the sci-tech innovation subject. The closer to 0, the lower the matching degree and the weaker the applicability. The matching degree will be used as the core indicator for screening and recommending the policy to the sci-tech innovation subject. When it is close to 1, it indicates that the sci-tech innovation subject and the support policy are highly consistent in multiple key dimensions, which means that the various clauses and support measures of the support policy are likely to be urgently needed by the sci-tech innovation subject; when the matching degree is When the degree of match is between 0.4 and 0.6, the science and technology innovation subject and the support policy have a certain degree of fit in some dimensions, but there are also some mismatches, which suggests that it is necessary to further screen the specific clauses in the policy to determine the applicable part; when the degree of match is between 0.4 and 0.6, the science and technology innovation subject and the support policy have a certain degree of fit in some dimensions, but there are also some mismatches, which suggests that it is necessary to further screen the specific clauses in the policy to determine the applicable part. When it is close to 0, it indicates that the scientific and technological innovation entities and the support policies do not match in most key dimensions.

[0108] It should be noted that the above describes a specific embodiment of the present invention. Other embodiments are within the scope of the present invention. In some cases, the actions or steps described above can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the process depicted in the accompanying drawings does not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0109] For the convenience of description, the above device is described as being divided into various modules according to their functions. Of course, when implementing the present invention, the functions of each module can be implemented in the same or multiple software and / or hardware.

[0110] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0111] Each embodiment of the present invention is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0112] Those skilled in the art should understand that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of the present disclosure is limited to these examples. Based on the concept of the present disclosure, the technical features in the above embodiments or different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of different aspects of one or more embodiments of the present invention as described above, which are not provided in detail for the sake of simplicity.

[0113] While the disclosure has been described in conjunction with specific embodiments thereof, many alternatives, modifications and variations of these embodiments will be apparent to those skilled in the art in light of the foregoing description.

[0114] One or more embodiments of the present invention are intended to cover all such replacements, modifications and variations that fall within the broad scope of the present invention. Therefore, any omissions, modifications, equivalent replacements, improvements, etc. made within the spirit and principles of one or more embodiments of the present invention should be included in the protection scope of this disclosure.

Claims

1. A visualization system for presenting an industrial map based on scientific and technological innovation data, characterized in that: The visualization system comprises: A data collection module, which is used to collect scientific and technological innovation data of all scientific and technological innovation entities in the target area; A correlation graph construction module, which is used to construct a correlation graph of scientific and technological innovation entities based on all the scientific and technological innovation entities and the scientific and technological innovation data, wherein all the scientific and technological innovation entities constitute nodes of the scientific and technological innovation entity correlation graph, the scientific and technological innovation data of each scientific and technological innovation entity constitutes the node attribute of the node, and the cooperation between the scientific and technological innovation entities constitutes the edges of the associated scientific and technological innovation entities; A correlation graph processing module, which is used to process the correlation graph of the scientific and technological innovation subject using a graph convolutional network to obtain high-order node features of each node; A judgment module, which is used to judge whether the target node of the processed association map has a scientific and technological innovation strength label; A label assignment module, which is used to assign a matching scientific and technological innovation strength label to the target node when the target node of the processed association map does not have a scientific and technological innovation strength label, and to identify the potential strength subject associated with the target node by using label propagation and label update iteration when the target node of the processed association map has a scientific and technological innovation strength label and assign a matching scientific and technological innovation strength label to the potential strength subject; The method of using label propagation and label update iteration to identify potential strength subjects associated with the target node and assigning matching scientific and technological innovation strength labels to the potential strength subjects includes: Perform unique hot encoding on the target node's scientific and technological innovation strength label and set the propagation attenuation coefficient ; Node-based and nodes The edge weight between , according to the formula Compute Node From neighbor nodes The propagation probability of receiving label information ; The label update iteration is performed according to the following formula: ； in, Is a node The updated label vector, Is a node The set of neighbor nodes of It is a node The sum of the propagation probabilities of all neighbor nodes, Neighbor node The known label vector of the neighbor nodes The known label vector is the one-hot encoding vector of the target node; Based on the updated iteration Assign matching scientific and technological innovation strength labels to the potential strength entities; An industry map visualization module is used to construct an industry map and visualize it based on the target area, the processed association map and all the scientific and technological innovation strength labels.

2. The visualization system for presenting an industrial map based on scientific and technological innovation data according to claim 1 is characterized in that: The graph convolutional network is used to process the scientific and technological innovation subject association graph to obtain the high-order node features of each node, including: Build a 3-layer GCN network architecture; Direct neighbor attribute aggregation is performed in the first GCN layer; The output of the first GCN layer is used as the input of the second GCN layer to perform second-order neighbor information fusion; The output of the second GCN layer is used as the input of the third GCN layer for global feature integration and output to obtain the high-order node features of each node.

3. The visualization system for presenting an industrial map based on scientific and technological innovation data according to claim 2 is characterized in that: The construction of the 3-layer GCN network architecture includes: Construct a network architecture containing three layers of graph convolutional layers according to the following formula: ； in, It is The node feature representation matrix of the layer, It is The node feature representation matrix of the layer, is the activation function, is the degree matrix, , is the adjacency matrix, is the identity matrix, It is The learnable weight matrix of the layer is hour, is a node initial feature matrix containing all the scientific and technological innovation subjects and the scientific and technological innovation data; In the first GCN layer, according to Perform calculations; In the second GCN layer, according to Perform calculations; In the third GCN layer, according to Perform calculations.

4. The visualization system for presenting an industrial map based on scientific and technological innovation data according to claim 1 is characterized in that: When the target node of the processed association graph does not have a scientific and technological innovation strength label, the target node is assigned a matching scientific and technological innovation strength label including: The label matching score of the scientific and technological innovation subject corresponding to the target node is calculated using the following formula: ； in, is the calculated label matching score of the scientific and technological innovation entity, Score the innovation achievements of scientific and technological innovation entities. The score of R&D investment of scientific and technological innovation entities. Score the talent resources of the scientific and technological innovation entities. The market performance score of the scientific and technological innovation entity. Score the cooperation network of scientific and technological innovation entities. , , , and They are the weights of the scientific and technological innovation entities’ innovation achievements, R&D investment, talent resources, market performance and cooperation network; Determine the score range of the calculated label matching score of the scientific and technological innovation entity; Based on the score range of the label matching score of the scientific and technological innovation subject, the scientific and technological innovation subject is assigned a corresponding scientific and technological innovation strength label.

5. The visualization system for presenting an industrial map based on scientific and technological innovation data according to claim 1 is characterized in that: The termination condition of the label update iteration is: ； in, is the label vector dimension, It is Iteration node label vector elements, It is Iteration node label vector elements, is the number of iterations, is the label vector convergence judgment threshold.

6. The visualization system for presenting an industrial map based on scientific and technological innovation data according to claim 5 is characterized in that: The label vector convergence determination threshold Equal to 0.

01.

7. The visualization system for presenting an industrial map based on scientific and technological innovation data according to claim 1 is characterized in that: The visualization system also includes a policy retrieval and storage module, which is used to retrieve and store science and technology innovation support policy information in the target area.

8. The visualization system for presenting an industrial map based on scientific and technological innovation data according to claim 7 is characterized in that: When a scientific and technological innovation subject is selected in the industrial map, the industrial map visualization module matches the support policy information corresponding to the scientific and technological innovation subject from the policy retrieval and storage module and presents it visually.

9. The visualization system for presenting an industrial map based on scientific and technological innovation data according to any one of claims 1 to 8, characterized in that: The industrial map supports regional zooming; When the industry map is in a regional zoom state and a scientific and technological innovation subject is selected in the zoomed area: If all other entities associated with the selected scientific and technological innovation entity are within the zoomed area, their associations will be presented by connecting them with highlighted lines; If some of the other subjects associated with the selected scientific and technological innovation subject are within the zoomed area and the other subjects are not within the zoomed area, the other subjects within the zoomed area will be connected by highlight lines to show their association with the selected scientific and technological innovation subject, and the other subjects not within the zoomed area will be shown their association with the selected scientific and technological innovation subject in a text pop-up window; If all other entities associated with the selected scientific and technological innovation entity are not within the zoomed area, their mutual relationships will be presented in the form of a text pop-up window.

Images