A method and system for dynamically evaluating innovation capability of a technology enterprise

Abstract

This invention belongs to the field of Internet and cloud computing, big data service technology; specifically, it relates to a method and system for dynamically evaluating the innovation capabilities of technology enterprises. The method includes the following steps: acquiring multi-source data; converting first unstructured data into text information data, and extracting second structured data from the text information data; fusing the first structured data and the second structured data to obtain standardized data; the standardized data includes first quantitative data and first unquantified data; quantifying the first unquantified data to obtain second quantitative data; normalizing the first quantitative data and the second quantitative data according to historical dataset data to obtain evaluation data; and calculating the evaluation result based on the evaluation data. This invention can realize the dynamic evaluation of the innovation capabilities of technology enterprises and can achieve a balance of data from different dimensions.

Description

Technical Field

[0001] This invention belongs to the field of Internet and cloud computing, big data service technology; specifically, it relates to a method and system for dynamic evaluation of the innovation capabilities of technology enterprises. Background Technology

[0002] The enterprise technological innovation capability evaluation system takes the improvement of core technology research and development capabilities and the ability to transform scientific and technological achievements as its basic goal. Based on a full consideration of the internal and external environment of enterprises, it integrates innovation elements to achieve the added value and spillover of innovation.

[0003] There are various existing systems related to the evaluation of science and technology enterprises. For example, Chinese patent application CN202210299828.6 discloses a method, apparatus, equipment, and medium for evaluating science and technology enterprises. The method includes: obtaining attribute information of the science and technology enterprise to be evaluated; determining the weight and adjustment coefficient of the science and technology innovation evaluation index corresponding to the science and technology enterprise to be evaluated based on the technical field information, wherein the adjustment coefficient is related to the technical field and credit policy; and evaluating the scientific and technological innovation capability of the science and technology enterprise to be evaluated based on at least one science and technology innovation evaluation index value, the weight of the science and technology innovation evaluation index, and the adjustment coefficient. Chinese patent application CN202211430691.X discloses a dynamic evaluation method for the innovation capability of a science and technology innovation platform based on a radar chart model. This method includes establishing evaluation directions with no fewer than three evaluation dimensions for each evaluation point and determining evaluation reference data for each direction; setting evaluation limits for each direction and evaluating each direction to obtain a score; converting the evaluation direction scores according to the weights of different evaluation directions to form the evaluation point score; and conducting a multi-dimensional evaluation of innovation capability based on the evaluation point score, using a radar chart comprehensive evaluation model and a first-level evaluation index radar chart evaluation model. Chinese patent application CN202110895689.9 discloses a method for evaluating the creditworthiness of science and technology enterprises based on big data mining. The method includes collecting credit-related data of science and technology enterprises, transmitting the data to the cloud, storing and preprocessing the data, using a Generative Model (GMM) to cluster the data, using labeled data to train a Backpropagation (BP) neural network model that can accurately predict the credit rating of science and technology enterprises, and finally accepting enterprise data input from users, using the BP neural network to make predictions, and returning the results to the user from the cloud. Chinese patent application CN202311623569.9 discloses a novel intelligent evaluation method for scientific and technological talents based on a hypergraph attention mechanism, including the following steps: S1, constructing a knowledge hypergraph for scientific and technological talents; S2, introducing an attention mechanism to design a network model for classifying and evaluating scientific and technological talents; S3: constructing an intelligent recommendation model for scientific and technological talents.

[0004] Currently, there are several methods for evaluating enterprise technological innovation capabilities, including before-and-after comparison, benchmarking, cost-benefit analysis, data analysis, expert review, stakeholder survey and analysis, and multi-indicator comprehensive evaluation. However, in reality, due to differences in the availability of evaluation data, the feasibility of methods, and the different evaluation purposes and needs, there are many obstacles to achieving a unified and dynamic system tool for evaluation. Summary of the Invention

[0005] In order to overcome the shortcomings of existing technology innovation capability evaluation systems, such as poor dynamic evaluation performance, low data utilization, and difficulty in balancing data from different dimensions, this invention provides a method and system for dynamic evaluation of the innovation capability of technology enterprises.

[0006] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: a method for dynamic evaluation of the innovation capability of technology enterprises, comprising the following steps: Step 1: Obtain multi-source data, which includes first structured data and first unstructured data; Step 2: Convert the first unstructured data into text information data, and extract the second structured data from the text information data; Step 3: Perform data fusion on the first structured data and the second structured data to obtain standardized data; the standardized data includes first quantized data and first non-quantized data; Step 4: Quantize the first non-quantized data to obtain the second quantized data; normalize the first quantized data and the second quantized data according to the historical dataset to obtain the evaluation data; Step 5: Calculate the evaluation results based on the evaluation data.

[0007] Step 2 further includes a step of matching the second structured data, specifically including: Step 21: Extract enterprise name data from the second structured data and match the extracted enterprise name data with the verification database; Step 22: Perform secondary similarity matching on the company name data that failed to match; Secondary similarity matching includes: splitting unmatched company name data into three fields: region, trade name / industry, and organization form; if a unique company name with identical region and trade name / industry fields exists in the verification database, the match is successful; if a company name with identical region and organization form fields exists in the verification database, it is further determined whether a unique company name exists with identical pinyin for the trade name / industry field and only one different character in the trade name / industry field. If such a company name exists, the match is successful. Step 23: Mark the data corresponding to the company names that failed the secondary similarity match as erroneous data.

[0008] The first set of quantitative data includes R&D expenses, operating profit, R&D intensity, and operating profit margin; The second set of quantitative data includes intellectual property quantitative data and science and technology management quantitative data; the intellectual property quantitative data is obtained by calculating data from patents, copyrights, standards, and integrated circuit layouts; the science and technology management quantitative data is obtained by calculating data from innovation platforms, science and technology awards, and government science and technology projects.

[0009] In step 4, during the normalization process, the historical dataset is dynamically adjusted and updated based on the acquisition progress of the first quantized data and the second quantized data.

[0010] In step 4, the method for normalizing the first quantized data and the second quantized data to obtain evaluation data is as follows: For the data to be processed in the first quantized data and the second quantized data, obtain the data within the corresponding historical time period as the historical dataset; determine the maximum and minimum values ​​in the historical dataset; and calculate the normalized data of the data to be processed. As evaluation data; the calculation formula is: ; in, This indicates the data to be processed. This represents the minimum value in the historical dataset. This represents the maximum value in the historical dataset. This represents a historical dataset.

[0011] In step 4, the method for normalizing the first quantized data and the second quantized data to obtain the evaluation data includes: Step 41: For any of the first quantized data and the second quantized data to be processed, obtain the data of the historical time period corresponding to the data to be processed as the historical dataset; obtain the maximum and minimum values ​​in the historical dataset and calculate the normalized data of the data to be processed. The calculation formula is: ; in, This indicates the data to be processed. This represents the minimum value in the historical dataset. This represents the maximum value in the historical dataset. Represents historical datasets; Step 42: Perform min-max normalization on the data in the historical dataset to obtain the historical normalized data corresponding to the historical dataset; Step 43: After arranging the historical normalized data from largest to smallest, divide the historical normalized data into m segments, determine the data of each segment node and assign a weight to each segment node, and combine the segment nodes and the normalized data of the data to be processed to calculate the evaluation data corresponding to the data to be processed.

[0012] In step 43, the calculation formula for the evaluation data corresponding to the data to be processed is: ; in, This represents the normalized data of the data to be processed. These represent the historical normalized data corresponding to the first and second segment nodes, respectively. and These represent the historical normalized data corresponding to the j-th and (j+1)-th segment nodes, respectively; , m represents the number of segments, represents the weight assignment of the j-th and j+1-th segment nodes, respectively, and X represents the evaluation data corresponding to the data to be processed.

[0013] In step 5, the formula for calculating the evaluation result is: U; This represents the evaluation data for R&D expenses. This represents the quantitative evaluation data of intellectual property rights. This represents quantitative evaluation data for science and technology management. This represents the data used to evaluate operating profit. This represents the data for evaluating R&D investment intensity. This represents the data used to evaluate operating profit margin. Represents the weighting constant. This indicates an adjustment of the weights.

[0014] In step 5, the formula for calculating the adjusted weights is: ; Where y represents R&D expenses, z represents intellectual property quantitative data, M represents the dataset formed by the ratio of R&D expenses to intellectual property quantitative data within a historical period, max represents the maximum value, and β represents the preset adjustment constant.

[0015] Furthermore, this invention also provides a dynamic evaluation system for the innovation capability of technology enterprises, used to implement the aforementioned dynamic evaluation method for the innovation capability of technology enterprises, comprising: Data acquisition module: used to acquire the multi-source data; Data processing module: used to convert the first unstructured data into text information data, and extract the second structured data from the text information data; Data fusion module: used to fuse the first structured data and the second structured data to obtain standardized data; The dynamic evaluation module is used to quantize the first non-quantized data to obtain the second quantized data; it is also used to normalize the first quantized data and the second quantized data using historical datasets to obtain evaluation data. Evaluation result output module: used to calculate and output the evaluation results based on the evaluation data.

[0016] Compared with the prior art, the present invention has the following advantages: 1. This invention provides a method for dynamically evaluating the innovation capabilities of technology companies by acquiring multi-source data and transforming it into multi-dimensional quantitative data through steps such as data transformation, data extraction, data fusion, and data quantification. Evaluation data is then obtained through dynamic adjustment of the normalization model, and the evaluation results are calculated through the evaluation model.

[0017] 2. In the data extraction stage, this invention further ensures the reliability of extracting key data such as company names by introducing a verification database for comparison. This invention further identifies and classifies cases of mismatched or incorrect data names using algorithms, significantly reducing the loss of valid data and improving the efficiency of key data extraction.

[0018] 3. This invention employs an optimized normalization model for data processing and utilizes data segmentation and scaling preprocessing techniques to effectively eliminate extreme values, ensuring a better evaluation effect for the data distribution. This ensures both data comparability and balance across different dimensions of data.

[0019] 4. The comprehensive index method evaluation model of the present invention optimizes the evaluation data of R&D expenses by introducing adjustment weights, ensuring that the evaluation results of the comprehensive index method are more in line with the actual situation, and improving the adaptability and conformity of the model for different evaluation subjects, different evaluation objectives, and different evaluation objective scenarios.

[0020] 5. By calculating and adjusting multidimensional data, this invention generates six dimensions of data: R&D expenses, operating profit, R&D investment intensity, operating profit margin, intellectual property quantitative data, and science and technology management quantitative data. This greatly reduces redundant data, improves data usability, and can meet different evaluation needs by adjusting the weights of different dimensions. Attached Figure Description

[0021] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments; Figure 1 This is a flowchart illustrating a method for dynamically evaluating the innovation capabilities of technology enterprises, provided in Embodiment 1 of the present invention. Figure 2 This is a schematic diagram comparing the evaluation data obtained in Embodiment 3 and Embodiment 4 of the present invention; where (a) is the evaluation data corresponding to Embodiment 3 and (b) is the evaluation data corresponding to Embodiment 4.

[0022] Figure 3 This is a schematic diagram of the structure of a dynamic evaluation system for the innovation capability of technology enterprises provided in Embodiment Six of the present invention. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] Example 1 like Figure 1 As shown, Embodiment 1 of the present invention provides a method for dynamically evaluating the innovation capability of technology enterprises, including the following steps: Step 1: Obtain multi-source data, which includes first structured data and first unstructured data; the first structured data comes from standard databases or data files, and the first unstructured data mainly comes from web robot collection.

[0025] Step 2: Convert the first unstructured data into text information data, and extract the second structured data from the text information data.

[0026] Specifically, in this embodiment, a recognition algorithm is used to uniformly identify and convert PDF files, image files, WORD files, etc. in the first unstructured data collected by the network robot into text information data, and then extracts the second structured data through existing large model algorithms.

[0027] Step 3: Perform data fusion on the first structured data and the second structured data to obtain standardized data; the standardized data includes first quantized data and first non-quantized data.

[0028] Specifically, in this embodiment, the first structured data and the second structured data of each data entity are merged using the enterprise name, unified organization code, etc.

[0029] Step 4: Quantize the first non-quantized data to obtain the second quantized data; normalize the first quantized data and the second quantized data according to the historical dataset to obtain the evaluation data.

[0030] Specifically, in this embodiment, the first non-quantized data is assigned a value using the assignment method, and then quantized by addition and merging to obtain the second quantized data.

[0031] Step 5: Calculate the evaluation results based on the evaluation data.

[0032] In this embodiment, evaluation results are obtained by calculating the evaluation data using methods such as the comprehensive index method, entropy method, or principal component analysis.

[0033] Example 2 Embodiment 2 of the present invention provides a method for dynamic evaluation of the innovation capability of technology enterprises, including the following steps: Step 1: Obtain multi-source data, which includes first structured data and first unstructured data; the first structured data comes from standard databases or data files, and the first unstructured data mainly comes from web robot collection.

[0034] Step 2: Convert the first unstructured data into text information data, and extract the second structured data from the text information data.

[0035] Specifically, in this embodiment, a recognition algorithm is used to uniformly identify and convert PDF files, image files, WORD files, etc. in the first unstructured data collected by the network robot into text information data, and then extracts the second structured data through existing large model algorithms.

[0036] Specifically, unlike Embodiment 1, step 3 in this embodiment further includes a step of matching the second structured data, specifically including: Step 21: Extract enterprise name data from the second structured data and match the extracted enterprise name data with the verification database; Step 22: Perform secondary similarity matching on the company name data that failed to match; Secondary similarity matching includes: splitting unmatched company name data into three fields: region, trade name / industry, and organization form; if a unique company name with identical region and trade name / industry fields exists in the verification database, the match is successful; if a company name with identical region and organization form fields exists in the verification database, it is further determined whether a unique company name exists with identical pinyin for the trade name / industry field and only one different character in the trade name / industry field. If such a company name exists, the match is successful. Step 23: Mark the data corresponding to the company names that failed the secondary similarity match as erroneous data.

[0037] The above matching process can eliminate erroneous data in the second structured data, providing a foundation for the accurate fusion of the second structured data and the first structured data.

[0038] Step 3: Perform data fusion on the first structured data and the second structured data to obtain standardized data; the standardized data includes first quantized data and first non-quantized data.

[0039] Specifically, in this embodiment, the first and second structured data of each data entity are merged using enterprise names, unified organization codes, etc. Since each data entity is matched by its enterprise name in step 2, invalid data can be removed, improving data fusion efficiency.

[0040] Step 4: Quantize the first non-quantized data to obtain the second quantized data; normalize the first quantized data and the second quantized data according to the historical dataset to obtain the evaluation data.

[0041] Specifically, in this embodiment, the first non-quantized data is assigned a value using the assignment method, and then quantized by addition and merging to obtain the second quantized data.

[0042] Step 5: Calculate the evaluation results based on the evaluation data.

[0043] Example 3 Embodiment 3 of the present invention provides a method for dynamic evaluation of the innovation capability of technology enterprises, including the following steps: Step 1: Obtain multi-source data, which includes first structured data and first unstructured data; the first structured data comes from standard databases or data files, and the first unstructured data mainly comes from web robot collection.

[0044] Step 2: Convert the first unstructured data into text information data, and extract the second structured data from the text information data.

[0045] Specifically, in this embodiment, an identification algorithm is used to uniformly identify and convert PDF files, image files, WORD files, etc. in the first unstructured data collected by the network robot into text information data, and then extracts the second structured data through algorithms such as large model.

[0046] Step 3: Perform data fusion on the first structured data and the second structured data to obtain standardized data; the standardized data includes first quantized data and first non-quantized data.

[0047] Specifically, in this embodiment, the first structured data and the second structured data of each data entity are merged using the enterprise name, unified organization code, etc.

[0048] Step 4: Quantize the first non-quantized data to obtain the second quantized data; normalize the first quantized data and the second quantized data according to the historical dataset to obtain the evaluation data.

[0049] Specifically, in this embodiment, the first non-quantized data is assigned a value using the assignment method, and then quantized by addition and merging to obtain the second quantized data.

[0050] The first set of quantitative data includes R&D expenses, operating profit, R&D intensity, and operating profit margin; the second set of quantitative data includes intellectual property quantitative data and science and technology management quantitative data.

[0051] The intellectual property quantification data is obtained through calculations using data from patents, copyrights, established standards, and integrated circuit layouts. Preferably, this data is calculated by assigning values ​​to each individual patent (utility model and invention patents), copyrights, established standards, and integrated circuit layouts. The sum of these summed values ​​is then accumulated to obtain the intellectual property quantification data.

[0052] The quantitative data for science and technology management is obtained through data calculations from innovation platforms, science and technology awards, and government science and technology projects. Preferably, values ​​are assigned to each innovation platform, science and technology award, and government science and technology project separately and then added together. The sum of these values ​​is then accumulated to form the quantitative data for science and technology management.

[0053] Unlike Embodiment 1, in this embodiment, the method for normalizing the first quantized data and the second quantized data is as follows: For the data to be processed in the first quantized data and the second quantized data, data within the corresponding historical time period is obtained as a historical dataset. For example, for the R&D expense data to be processed, the R&D expense amounts of all data entities in the previous year or the most recent three years are obtained to obtain the historical dataset; the maximum and minimum values ​​in the historical dataset are determined; and the normalized data of the data to be processed is calculated. As evaluation data; the calculation formula is: (1) in, This indicates the data to be processed. This represents the minimum value in the historical dataset. This represents the maximum value in the historical dataset. Represents historical datasets; X This represents the normalized data to be processed.

[0054] Step 5: Calculate the evaluation results based on the evaluation data.

[0055] Example 4 Embodiment 4 of the present invention provides a method for dynamic evaluation of the innovation capability of technology enterprises, including the following steps: Step 1: Obtain multi-source data, which includes first structured data and first unstructured data; the first structured data comes from standard databases or data files, and the first unstructured data mainly comes from web robot collection.

[0056] Step 2: Convert the first unstructured data into text information data, and extract the second structured data from the text information data.

[0057] Specifically, in this embodiment, an identification algorithm is used to uniformly identify and convert PDF files, image files, WORD files, etc. in the first unstructured data collected by the network robot into text information data, and then extracts the second structured data through algorithms such as large model.

[0058] Step 3: Perform data fusion on the first structured data and the second structured data to obtain standardized data; the standardized data includes first quantized data and first non-quantized data.

[0059] Specifically, in this embodiment, the first structured data and the second structured data of each data entity are merged using the enterprise name, unified organization code, etc.

[0060] Step 4: Quantize the first non-quantized data to obtain the second quantized data; normalize the first quantized data and the second quantized data according to the historical dataset to obtain the evaluation data.

[0061] Specifically, in this embodiment, the first non-quantified data is assigned a value using an assignment method, and then quantified through addition and merging to obtain the second quantified data. The first quantified data includes R&D expenses, operating profit, R&D intensity, and operating profit margin; the second quantified data includes intellectual property quantified data and technology management quantified data.

[0062] Unlike Example 3, in this example, the normalized data of the data to be processed is calculated. Subsequently, instead of directly using it as evaluation data X, the evaluation data X is obtained by assigning values ​​to segmented nodes and then performing function transformation. Specifically, in this embodiment, the first quantized data and the second quantized data are normalized respectively. The method for obtaining evaluation data X is as follows: Step 41: For the data to be processed in the first quantified data and the second quantified data, obtain the data within the corresponding historical time period as the historical dataset. For example, for the R&D expense data to be processed, obtain the R&D expense of all data entities in the previous year or the most recent three years to obtain the historical dataset; determine the maximum and minimum values ​​in the historical dataset; calculate the normalized data of the data to be processed. The calculation formula is the above formula (1).

[0063] Step 42: Perform min-max normalization on the data in the historical dataset to obtain the historical normalized data corresponding to the historical dataset.

[0064] Step 43: After arranging the historical normalized data from largest to smallest, divide the historical normalized data into m segments, determine the data of each segment node and assign a weight to each segment node, and combine the segment nodes and the normalized data of the data to be processed to calculate the evaluation data X corresponding to the data to be processed.

[0065] Specifically, in step 43, the calculation formula is as follows: ; (2) in, This represents the normalized data of the data to be processed. These represent the historical normalized data corresponding to the first and second segment nodes, respectively. and These represent the historical normalized data corresponding to the j-th and (j+1)-th segment nodes, respectively; , m represents the number of segments, represents the weight assignment of the j-th and j+1-th segment nodes, respectively, and X represents the evaluation data corresponding to the data to be processed.

[0066] Furthermore, in step 42 of this embodiment, the calculation formula for performing min-max normalization on each data point in the historical dataset is as follows: (3) in, This represents the i-th data point in historical dataset A. The corresponding historical normalized data, This represents the minimum value in the historical dataset. This represents the maximum value in the historical dataset.

[0067] Furthermore, in this embodiment, historical normalized data After sorting from largest to smallest, determine 5 segment nodes, i.e., m=5. These 5 segment nodes are respectively set as: the first data entity in the sorted order, the data entities in the top 1% of the sorted order, the data entities in the top 16% of the sorted order, the data entities in the top 64% of the sorted order, and the last data entity in the sorted order. Extract the historical normalized data corresponding to these segment nodes, and denote them as follows: , , , , ;in, Weights are assigned to the five segment nodes, as follows: , , , , .

[0068] Substituting the segmented node data and weights into formula (2), the data to be processed is... Corresponding evaluation data for: (4) Net profit data of 2100 data entities, historical normalized data. After sorting from largest to smallest, five segment nodes are determined: 1st, 21st, 336th, 1344th, and 2100th. The historical normalized data corresponding to the data entities of these segment nodes are then extracted. , , , , The evaluation data were calculated using the methods in Examples 3 and 4, and the results are as follows: Figure 2 As shown, by Figure 2 As can be seen, the data comparability is increased after processing in this embodiment, and the balance of various indicators in the comprehensive evaluation is stronger.

[0069] Step 5: Calculate the evaluation results based on the evaluation data.

[0070] Example 5 Embodiment 5 of the present invention provides a method for dynamic evaluation of the innovation capability of technology enterprises, including: Step 1: Obtain multi-source data, which includes first-level structured data and first-level unstructured data; Step 2: Convert the first unstructured data into text information data, and extract the second structured data from the text information data; Step 3: Perform data fusion on the first structured data and the second structured data to obtain standardized data; the standardized data includes the first quantized data and the first non-quantized data; Step 4: Quantize the first non-quantized data to obtain the second quantized data; normalize the first and second quantized data based on historical datasets to obtain evaluation data.

[0071] Specifically, in this embodiment, during normalization, the historical dataset is dynamically adjusted and updated based on the latest acquisition progress of the first quantized data and the second quantized data. That is, the historical dataset is adjusted according to the latest data acquisition progress, thereby changing the model parameters for normalization. For example, the initially selected historical dataset can be set to data from 2022-2023. When 90% of the data collection progress in 2024 is completed, the historical dataset is adjusted to data from 2023-2024, and this adjustment is repeated year by year.

[0072] Step 5: Calculate the evaluation results using the evaluation model based on the evaluation data; Specifically, in this embodiment, the evaluation results are calculated using a comprehensive index method evaluation model. Using the comprehensive index method can make the evaluation results closer to the actual evaluation objectives of experts; the calculation formula is: U; (5) in, This represents the evaluation data for R&D expenses. This represents the quantitative evaluation data of intellectual property rights. This represents quantitative evaluation data for science and technology management. This represents the data used to evaluate operating profit. This represents the data for evaluating R&D investment intensity. This represents the data used to evaluate operating profit margin. The weighting constants can be determined through expert brainstorming or the Delphi method. This indicates an adjustment of the weights.

[0073] Furthermore, in this embodiment, the weights are adjusted. The calculation formula is: (6) Where y represents the amount of R&D expenses, and z represents the quantitative data of intellectual property rights. M This dataset represents the ratio of R&D expenditure to intellectual property quantification data over a historical period. max represents the maximum value. For example, the dataset represents the ratio of R&D expenditure to intellectual property quantification data for each technology company in 2023. β represents a preset adjustment constant, which can be determined through expert brainstorming or the Delphi method.

[0074] The normalization model is dynamically adjusted based on the latest acquisition progress of the first and second quantified data. In other words, the historical dataset is adjusted according to the latest data acquisition progress, thereby changing the normalization model parameters. For example, the initial historical dataset can be set to include data from 2022-2023. Once 90% of the data collection progress in 2024 is completed, the historical dataset is adjusted to include data from 2023-2024, and this adjustment continues year by year.

[0075] Example 6 like Figure 3 As shown, Embodiment Six of the present invention provides a dynamic evaluation system for the innovation capability of technology enterprises, used to implement the evaluation method in any one of Embodiments One to Five, including: Data acquisition module: used to acquire the multi-source data; Data processing module: used to convert the first unstructured data into text information data, and extract the second structured data from the text information data; Data fusion module: used to fuse the first structured data and the second structured data to obtain standardized data; The dynamic evaluation module is used to quantize the first non-quantized data to obtain the second quantized data; it is also used to normalize the first quantized data and the second quantized data using historical datasets to obtain evaluation data. Evaluation result output module: used to calculate and output the evaluation results based on the evaluation data.

[0076] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for dynamically evaluating the innovation capability of technology enterprises, characterized in that, Includes the following steps: Step 1: Obtain multi-source data, which includes first structured data and first unstructured data; Step 2: Convert the first unstructured data into text information data, and extract the second structured data from the text information data; Step 3: Perform data fusion on the first structured data and the second structured data to obtain standardized data; the standardized data includes first quantized data and first non-quantized data; Step 4: Quantize the first unquantized data to obtain the second quantized data; The first quantified data and the second quantified data are normalized based on historical dataset data to obtain evaluation data. Step 5: Calculate the evaluation results based on the evaluation data.

2. The method for dynamic evaluation of the innovation capability of technology enterprises according to claim 1, characterized in that, Step 2 further includes a step of matching the second structured data, specifically including: Step 21: Extract enterprise name data from the second structured data and match the extracted enterprise name data with the verification database; Step 22: Perform secondary similarity matching on the company name data that failed to match; Secondary similarity matching includes: splitting unmatched company name data into three fields: region, trade name / industry, and organization form; if a unique company name with identical region and trade name / industry fields exists in the verification database, the match is successful; if a company name with identical region and organization form fields exists in the verification database, it is further determined whether a unique company name exists with identical pinyin for the trade name / industry field and only one different character in the trade name / industry field. If such a company name exists, the match is successful. Step 23: Mark the data corresponding to the company names that failed the secondary similarity match as erroneous data.

3. The method for dynamic evaluation of the innovation capability of technology enterprises according to claim 1, characterized in that, The first set of quantitative data includes R&D expenses, operating profit, R&D intensity, and operating profit margin; The second set of quantitative data includes intellectual property quantitative data and science and technology management quantitative data; the intellectual property quantitative data is obtained by calculating data from patents, copyrights, standards, and integrated circuit layouts; the science and technology management quantitative data is obtained by calculating data from innovation platforms, science and technology awards, and government science and technology projects.

4. The method for dynamic evaluation of the innovation capability of technology enterprises according to claim 1, characterized in that, In step 4, during the normalization process, the historical dataset is dynamically adjusted and updated based on the acquisition progress of the first quantized data and the second quantized data.

5. The method for dynamic evaluation of the innovation capability of technology enterprises according to claim 1, characterized in that, In step 4, the method for normalizing the first quantized data and the second quantized data to obtain the evaluation data is as follows: for the data to be processed in the first quantized data and the second quantized data, the data within the corresponding historical time period of the data to be processed is obtained as the historical dataset; the maximum and minimum values ​​in the historical dataset are determined. Calculate the normalized data of the data to be processed As evaluation data; The calculation formula is: ； in, This indicates the data to be processed. This represents the minimum value in the historical dataset. This represents the maximum value in the historical dataset. This represents a historical dataset.

6. The method for dynamic evaluation of the innovation capability of technology enterprises according to claim 1, characterized in that, In step 4, the method for normalizing the first quantized data and the second quantized data to obtain the evaluation data includes: Step 41: For any of the first quantized data and the second quantized data to be processed, obtain the data of the historical time period corresponding to the data to be processed as the historical dataset; obtain the maximum and minimum values ​​in the historical dataset and calculate the normalized data of the data to be processed. The calculation formula is: ； in, This indicates the data to be processed. This represents the minimum value in the historical dataset. This represents the maximum value in the historical dataset. Represents historical datasets; Step 42: Perform min-max normalization on the data in the historical dataset to obtain the historical normalized data corresponding to the historical dataset; Step 43: After arranging the historical normalized data from largest to smallest, divide the historical normalized data into m segments, determine the data of each segment node and assign a weight to each segment node, and combine the segment nodes and the normalized data of the data to be processed to calculate the evaluation data corresponding to the data to be processed.

7. The method for dynamic evaluation of the innovation capability of technology enterprises according to claim 6, characterized in that, In step 43, the calculation formula for the evaluation data corresponding to the data to be processed is: ； in, This represents the normalized data of the data to be processed. These represent the historical normalized data corresponding to the first and second segment nodes, respectively. and These represent the historical normalized data corresponding to the j-th and (j+1)-th segment nodes, respectively; , m represents the number of segments, represents the weight assignment of the j-th and j+1-th segment nodes, respectively, and X represents the evaluation data corresponding to the data to be processed.

8. The method for dynamic evaluation of the innovation capability of technology enterprises according to claim 1, characterized in that: In step 5, the formula for calculating the evaluation result is: U； This represents the evaluation data for R&D expenses. This represents the quantitative evaluation data of intellectual property rights. This represents quantitative evaluation data for science and technology management. This represents the data used to evaluate operating profit. This represents the data for evaluating R&D investment intensity. This represents the data used to evaluate operating profit margin. Represents the weighting constant. This indicates an adjustment of the weights.

9. The method for dynamic evaluation of the innovation capability of technology enterprises according to claim 8, characterized in that: In step 5, the formula for calculating the adjusted weights is: ； Where y represents R&D expenses, z represents intellectual property quantitative data, M represents the dataset formed by the ratio of R&D expenses to intellectual property quantitative data within a historical period, max represents the maximum value, and β represents the preset adjustment constant.

10. A dynamic evaluation system for the innovation capability of technology enterprises, used to implement the dynamic evaluation method for the innovation capability of technology enterprises as described in any one of claims 1-9, characterized in that, include: Data acquisition module: used to acquire the multi-source data; Data processing module: used to convert the first unstructured data into text information data, and extract the second structured data from the text information data; Data fusion module: used to fuse the first structured data and the second structured data to obtain standardized data; Dynamic evaluation module: used to quantize the first non-quantized data to obtain the second quantized data; It is also used to normalize the first quantized data and the second quantized data using historical datasets to obtain evaluation data; Evaluation result output module: used to calculate and output the evaluation results based on the evaluation data.

Images