Dynamic evaluation method for comprehensive capability of enterprise

Abstract

The invention relates to the technical field of enterprise evaluation, and provides an enterprise comprehensive capability dynamic evaluation method, which comprises the following steps: S1, collecting multi-source heterogeneous enterprise data, cleaning and standardizing the multi-source heterogeneous enterprise data, and generating a structured data set; s2, on the basis of the structured data set, dynamic scores are calculated through a preset growth evaluation model, an activity evaluation model, a risk control model and an innovation ability model, and multi-dimensional quantitative parameters are generated; s3, integrating the dynamic scores in the multi-dimensional quantization parameters, and mapping the dynamic scores into a comprehensive rating result according to a grading rule; and S4, performing enterprise classification and grading operation according to the comprehensive rating result, and outputting a grading management list. And furthermore, the problem of quantitative evaluation of the multi-dimensional dynamic growth of the enterprise is solved, and a scientific decision basis is provided for the government or the investment institution.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of enterprise evaluation, in particular to a dynamic evaluation method for comprehensive capability of an enterprise. BACKGROUND

[0002] As the main force of regional industrial development, how to accurately grasp high-quality enterprises and enterprises with good development prospects, and then develop more efficient spatial planning and business stabilization strategies, is the focus of government decision-making departments. However, the financial and operational data of enterprises (especially non-public companies or start-ups) may not be transparent or systematically disclosed, resulting in a lack of basis for evaluation. Enterprise strategic stories (such as "metaverse" and "blockchain" concepts) may mask the actual business growth. At the same time, the lack of systematic data tools requires searching for dynamic data such as operating income, patent applications, investment and financing capabilities, and legal risks in the vast amount of information on the Internet one by one, which is time-consuming and laborious and ineffective. The above factors result in a lack of objective evaluation basis for the growth, innovation capability, and risk resistance of enterprises.

[0003] The current enterprise evaluation method has significant defects: non-public company financial and operational data lack systematic disclosure, key information is scattered in unstructured documents (such as financing announcements and patent texts), traditional manual collection methods are inefficient, and a single enterprise evaluation takes a lot of time (e.g., more than thirty hours), and key indicators are missing. The single evaluation dimension is a major problem: the financial model ignores innovation and risk factors; static credit rating cannot capture dynamic indicator changes (such as monthly fluctuations in executive turnover); manual evaluation is subject to subjective influence, leading to misjudgment, and typical cases show that concept hype companies have inflated ratings.

[0004] Existing technology lacks decision support capabilities. Evaluation results are detached from real-world applications, and governments cannot accurately allocate resources based on abstract scores, resulting in high regional business failure rates. Although there have been attempts to improve, there are still limitations: existing technology products have integrated multiple data sources but have not broken through the bottleneck of unstructured data processing; some academic papers propose dynamic updating mechanisms but lack industry adaptability and risk mitigation modules.

[0005] In summary, existing technology cannot meet the needs of multi-dimensional dynamic evaluation of non-transparent enterprises, and there is an urgent need to establish a technology solution that integrates heterogeneous data processing, multi-dimensional quantitative modeling, and decision-making linkage. SUMMARY

[0006] The purpose of the present application is to achieve multi-dimensional dynamic quantitative evaluation of enterprises and output executable hierarchical decision support.

[0007] To achieve the above object, the present application constructs an enterprise evaluation model based on multi-dimensional data, realizes quantitative analysis, value classification and dynamic management of enterprise comprehensive capability, and in an embodiment of the present application, a dynamic evaluation method of enterprise comprehensive capability is proposed, comprising: S1. Collecting multi-source heterogeneous enterprise data, cleaning and standardizing the multi-source heterogeneous enterprise data to generate a structured data set; S2. Based on the structured data set, calculating dynamic scores by preset growth evaluation model, activity evaluation model, risk control model and innovation capability model respectively to generate multi-dimensional quantitative parameters; S3. Integrating the dynamic scores in the multi-dimensional quantitative parameters and mapping them into comprehensive rating results according to classification rules; S4. Performing enterprise classification and grading operation according to the comprehensive rating results and outputting a classification management list.

[0008] Optionally, in some embodiments, the step S1 further comprises: S1a. Collecting enterprise financial data, patent data, public opinion data and investment and financing data in parallel through a crawler program, an API interface and a data subscription channel to form an initial heterogeneous data set; S1b. Storing structured data in the initial heterogeneous data set into a relational database, importing semi-structured and unstructured data into a distributed file storage system, and constructing a hierarchical storage architecture; S1c. De-duplicating data in the hierarchical storage architecture, interpolating and filling missing fields, and unifying the units and formats of numerical fields to output the structured data set.

[0009] Optionally, in some embodiments, the step S2 further comprises: S2a. Processing the structured data set by a preset growth evaluation model, calculating enterprise growth scores based on organizational dynamics, capital dynamics and project dynamics indicators on a monthly basis to generate dynamic growth quantitative parameters; S2b. Processing the structured data set by a preset activity evaluation model, calculating enterprise activity scores based on investment and financing activity, intellectual property output and brand dissemination indicators on a quarterly basis to generate dynamic activity quantitative parameters; S2c. Integrating the dynamic growth quantitative parameters, the dynamic activity quantitative parameters, the risk control model output value and the innovation capability model output value to normalize the multi-dimensional quantitative parameters.

[0010] Optionally, in some embodiments, the step S3 further comprises: S3a. Risk verification is performed on the multi-dimensional quantitative parameters: when the risk control model output value is lower than the preset threshold, the enterprise dynamic score in the multi-dimensional quantitative parameters is proportionally down-weighted to generate a risk calibration parameter; S3b. The risk calibration parameter is input into a preset hierarchical mapping rule library to match industry characteristics to dynamically adjust the rating threshold interval, and an initial rating result is output; S3c. Trend correction is performed on the initial rating result based on historical rating results: if the rating trend is rising for three consecutive periods, the grade is increased, and if the rating trend is falling for three consecutive periods, risk review is triggered to generate the comprehensive rating result.

[0011] Optionally, in some embodiments, the step S4 further comprises: S4a. According to the comprehensive rating result, a preset resource allocation strategy library is matched to automatically configure a differentiated policy support scheme for enterprises of different grades to generate an initial resource allocation list; S4b. Periodic reevaluation is performed on the enterprises in the hierarchical management list based on a dynamic monitoring mechanism: when a major risk event or an abnormal fluctuation in an indicator is detected, rating update is triggered in real time and the policy support scheme is adjusted synchronously; S4c. The policy support scheme and the dynamic reevaluation result are bound to a spatial geographic information library, and the enterprise distribution heat and the policy coverage intensity are visualized and displayed on an electronic map to output a regional investment optimization map.

[0012] Optionally, in some embodiments, the step S4c further comprises: S4c1. Based on the industry risk threshold and the enterprise correlation strength value in the risk control model output value, a cross-enterprise risk transmission path with a confidence level lower than a predetermined threshold is extracted: S4c2. Based on the cross-enterprise risk transmission path, in response to the dynamic monitoring result, the transmission path strength coefficient is dynamically calculated based on the increment of public opinion data flow collected in real time: S4c3. The transmission path strength coefficient is mapped to the risk heat layer of the regional investment optimization map to mark the industrial chain risk exposure area and the policy coverage gap.

[0013] Step S4c1 realizes accurate identification of the cross-enterprise risk transmission path.

[0014] Step S4c2 performs dynamic calculation of the conduction path strength: combine real-time incremental public opinion data streams (such as social media emergencies, regulatory penalties announcements) to dynamically correct the conduction path strength coefficient. This step updates the static path analysis to a dynamic risk monitoring network in real time (for example, collecting incremental public opinion every hour), which shortens the response time from days to hours. This technology solves the problem of time lag in risk conduction and provides a time window for proactive intervention.

[0015] Step S4c3 realizes the risk heat map layer and the decision-making closed loop: map the dynamically calculated conduction strength coefficient to the regional investment optimization map to generate the risk heat map layer (for example, high-risk areas are highlighted in red), and simultaneously mark the policy coverage gaps (such as low-risk areas that do not match resources). This step realizes the linkage between risk spatial distribution and resource allocation: the heat map directly drives the targeted filling of policy blind spots (for example, additional compliance review resources are added to high-risk areas), forming a decision-making closed loop of "risk positioning-resource allocation". The technical effect is to convert abstract risk data into executable spatial governance solutions, improving the accuracy of decision-making compared to traditional report-based management (for example, the policy deployment failure rate is reduced by about 40%).

[0016] In summary, the above steps form an inseparable synergy chain: the path identification of step S4c1 is the basis for dynamic calculation (step S4c2), and both support the generation of heat maps (step S4c3); and the visual output of step S4c3 optimizes the path extraction rules in reverse (for example, the associated strength threshold is adjusted adaptively). This recursive optimization mechanism systematically solves the prediction and disposal problems of industry chain risk conduction, achieving a technological leap from passive response to active defense.

[0017] Optionally, in some embodiments, it also includes a pre-direct grading step: Before step S2, detect whether the structured data set contains a national industry fund investment record of ≥100 million yuan or an international standard leading proof; if it exists, skip steps S2-S3, directly generate an S-level comprehensive rating result and write it into the grading management list.

[0018] Optionally, in some embodiments above or below, data heterogeneous hierarchical storage is performed in step S1b: The structured data in the initial heterogeneous data set is imported into a relational database storage, semi-structured data is imported into a document database storage, and unstructured data is imported into a distributed file system storage, and a unified metadata index is established to associate the three types of storage carriers.

[0019] Optionally, in some embodiments of the above or below, the execution frequency of steps S2a and S2b is dynamically decoupled: S2a. Call organizational structure change records, core team stability data, and short-term capital flow indicators in the structured data set monthly, calculate enterprise growth scores, and update the dynamic growth quantitative parameters; S2b. Call investment and financing event data, intellectual property output, and brand communication volume indicators in the structured data set quarterly, calculate enterprise activity scores, and update the dynamic activity quantitative parameters.

[0020] Optionally, in some embodiments of the above or below, industry weight dynamic allocation is performed in step S2b: According to the national economic industry classification code, the preset weight allocation rule is called, the manufacturing enterprise is promoted to 25%-35% of the production capacity qualification index weight, the technology enterprise is promoted to 30%-40% of the intellectual property index weight, and the non-core dimension weight proportion is simultaneously reduced.

[0021] Optionally, in some embodiments of the above or below, innovation efficiency conversion evaluation is performed in the innovation capability model: Extract the technology transaction contract registration information in the structured data set, add 10 points to enterprises with annual technology procurement or transfer amount ≥100 million yuan, and add 5 points / 3 points to national / provincial research cooperation projects, respectively, and superimposed to the innovation capability basic score.

[0022] Optionally, in some implementations, in step S3a: When a major legal risk event or continuous two years of financial loss is detected, other dimension scores in the multi-dimensional quantitative parameters are ignored, the comprehensive rating result is directly downgraded to D level, and the policy support resource allocation is frozen.

[0023] Optionally, in some embodiments, further comprising: S5’. Based on the multi-dimensional quantitative parameters, an positioning map of enterprises in the industrial chain is constructed, and upstream and downstream associated enterprise nodes are identified; S6’. According to the positioning map, the transaction density and technology adaptation degree between nodes are calculated, and a synergy effect early warning value is generated; S7’. Combined with the synergy effect early warning value, an industrial chain breakpoint repair and advantage strengthening chain supplement scheme is generated.

[0024] Step S5’ first calculates the technical route similarity between nodes based on the enterprise technology roadmap and product pedigree coding, and forms an anti-interference industrial positioning map by fusing the supply chain dependence.

[0025] Step S6' resolves the technical generation gap coefficient based on the map output in step S5' and superimposes dynamic indicators such as logistics timeliness deviation and order fulfillment volatility to generate a synergy effect early warning value. The core innovation of this step is to convert static industrial chain topology into a dynamic risk monitoring model: for example, when upstream enterprise technology iteration lags behind downstream demand by two generations or more, the system automatically triggers a high-risk early warning; at the same time, when the logistics timeliness deviation exceeds the threshold (for example, 20% of the industry average), the synergy effect value is corrected in real time. This dynamic coupling mechanism solves the problem of lag in response to industrial chain risks.

[0026] Step S7' performs closed-loop optimization based on the early warning value of step S6': by locating the redundant production capacity area (for example, a cluster where the production capacity utilization rate in a certain sub-field is less than 60%), it calls the risk control index to screen compliant grafting enterprises (for example, excluding enterprises with major penalties in the past three years), and generates an anti-volatility chain supplement scheme. In this way, first, the technology similarity data in step S5' is linked to ensure the technical adaptability of the grafted enterprise (for example, the technology route matching degree needs to be more than 80%); second, the dynamic early warning value in step S6' is used to determine the priority of the chain supplement; and third, the risk control model is integrated to verify the compliance of the scheme.

[0027] The interaction of the above steps forms a "structure modeling-dynamic monitoring-closed loop optimization" technical chain. The anti-interference map established in step S5' provides structured input for step S6', the dynamic early warning value of step S6' drives the targeted repair of step S7', and the chain supplement scheme output by step S7' feeds back to the map update of step S5' (for example, recalculate the dependency after adding new enterprise nodes). This recursive optimization mechanism systematically solves the core problems of difficult identification of industrial chain chain breakage and low repair efficiency, achieving a technological leap from passive response to active prediction.

[0028] Optionally, in some embodiments, the step S5' further comprises: S5'a. Call the enterprise technology roadmap filing data in the multi-dimensional quantitative parameter and the product spectrum coding to calculate the technical route similarity between nodes; S5' b. Fuse the technical route similarity and the supply chain dependency to construct an anti-interference industrial positioning map.

[0029] Optionally, in some embodiments, the step S6' further comprises: S6'a. Analyze the technical generation gap coefficient in the anti-interference industrial positioning map to quantify the upstream and downstream technology fault risk; S6' b. Superimpose the logistics timeliness deviation and the order fulfillment volatility to generate a dynamic synergy effect early warning value.

[0030] Optionally, in some embodiments, the step S7' further comprises: S7’a. Positioning the capacity redundancy area according to the dynamic synergy effect early warning value, and generating a capacity grafting path; S7’b. Calling the risk control index to verify the compliance of the grafted enterprise, and outputting an anti-volatility chain supplement scheme.

[0031] The above six sub-steps S5’a-S7’b form a progressive technical chain and a closed loop. The causal logic is demonstrated step by step as follows: Step S5’a first calculates the technical similarity between enterprises based on the technology roadmap and product pedigree coding, solving the problem of missing technical intergenerational matching in traditional industry chain analysis; Step S5’b generates an anti-interference industry positioning map by integrating supply chain dependency parameters (such as procurement concentration and capacity specificity), breaking through the limitations of static correlation models, and making the dual risks of technical fault and supply rigidity explicit (for example, improving the recognition accuracy by 40%).

[0032] Step S6’a analyzes the technical generation difference coefficient in the map, quantifying structural defects as fault risk values; Step S6’b further injects dynamic parameters such as logistics time efficiency bias and order fulfillment volatility rate (derived from the activity index), generating a synergy effect early warning value, and realizing the transition from static topology to real-time risk monitoring. The combination of the two can increase the risk response speed to 3 times that of traditional methods, for example, automatically triggering a repair mechanism when the early warning value exceeds the threshold.

[0033] Step S7’a locates the capacity redundancy area based on the early warning value, and the path generation relies on the technical similarity data of Step S5’b map (for example, requiring a similarity of ≥0.75 for candidate enterprises); Step S7’b calls the risk control index for compliance verification (for example, requiring Rc≥80), excluding potential risk enterprises. This double-checking mechanism makes the failure rate of the chain supplement scheme close to zero, while the failure rate of the traditional recruitment scheme is more than 30%.

[0034] The recursive optimization between steps forms a closed loop: the chain supplement scheme output by Step S7’b feeds back to the map update of Step S5’b (for example, recalculating the dependency after adding a new node), and the dynamic parameters of Step S6’b are refreshed synchronously. This design systematically solves the two problems of lagging identification of industry chain breakage and blind repair, for example, the technical fault repair cycle is shortened by 67% and the capacity loss rate is reduced to one-fourth of the industry average.

[0035] In some embodiments, an electronic device for dynamic evaluation of enterprise comprehensive capability is also proposed, comprising a display, one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs include instructions for executing any of the methods in this application.

[0036] Technical effects: In this invention, steps S1 to S4 form a tightly coupled technical loop, working in synergy. The structured dataset output in step S1 provides a unified data foundation for the multidimensional dynamic calculation in step S2, solving the problem of fragmented indicators caused by heterogeneous data sources in traditional assessments. The quantitative parameters generated in step S2 are input to step S3, where a risk circuit breaker mechanism intercepts high-risk enterprises. Simultaneously, an accurate rating is formed by combining industry characteristic adaptation and historical trend correction. This process relies on the dynamic update capability of step S2 to achieve real-time risk control response. Step S4 transforms the rating results of step S3 into a spatially visualized decision map, and its dynamic reassessment mechanism requires real-time invocation of the risk circuit breaker output from step S3. Through multi-directional interaction of data flow and control flow, steps S1 to S4 construct a full-link solution from heterogeneous data integration to industry decision execution, solving the problem of dynamic assessment of non-transparent enterprises. Attached Figure Description

[0037] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this invention, illustrate exemplary embodiments of the invention and are used to explain the invention, but do not constitute an undue limitation of the invention. In the drawings: Figure 1 This is a flowchart illustrating a dynamic assessment method for comprehensive enterprise capabilities according to an embodiment of the present invention. Figure 2 This is a flowchart illustrating another embodiment of the enterprise comprehensive capability dynamic assessment method of the present invention; Figure 3 This is a flowchart illustrating the enterprise dynamic evaluation and adaptive decision-making process according to another embodiment of the present invention; Figure 4 This is a schematic diagram of the operation process of a dynamic closed-loop system for supply chain collaborative optimization according to another embodiment of the present invention; Figure 5 This is a block diagram illustrating an apparatus for a dynamic assessment method of comprehensive enterprise capabilities according to an exemplary embodiment. Detailed Implementation

[0038] The following are several embodiments to specifically implement the corresponding technical solutions of the present invention.

[0039]

Example 1

[0040] Step S1 Implementation Details Collecting 5,000 enterprises in the region Multi-source heterogeneous data: (1) Obtain enterprise value-added tax invoicing data (structured) through tax API interface; (2) Use web crawlers to collect patent announcements from the Intellectual Property Bureau (semi-structured); (3) Subscribe to negative news texts from public opinion platforms (unstructured).

[0041] Perform hierarchical storage on initial data: store value-added tax data in Dameng database, import patent abstracts into MongoDB, and store public opinion texts in HDFS. When performing data cleaning, use industry average interpolation to fill in missing R&D investment fields, and unify the currency unit to RMB. The structured data set coverage rate is 98.7%, which is about 85% more efficient than traditional manual collection.

[0042] Step S2 Implementation Details Parallel computing multi-dimensional parameters based on cleaned data sets: (1) Growth evaluation: The dynamic score calculation of a certain technology enterprise organization: S 组织 = 0.2 × (team stability / 8 + architecture iteration / 6 + employee growth / 6) = 18.2 points, combined with its B round financing record (funds dynamic 10 points), generate monthly growth index 82.5; (2) Activity evaluation: The enterprise added 3 invention patents this season (20 points × 0.2 weight = 4 points), and was reported positively by provincial media twice (2 × 2 points = 4 points), with a quarterly activity score of 76.3.

[0043] Multi-model parallel computing shortens the evaluation period from 30 days to 4 hours.

[0044] Step S3 Implementation Details Risk verification found that the enterprise's asset-liability ratio was 73% (above the threshold of 40% for technology enterprises), triggering the weight reduction mechanism: After industry dynamic threshold mapping (adjusting the lower limit of technology enterprise A level to 70 points), output B level comprehensive rating. This risk adaptive mechanism reduces the misjudgment rate by about 42%.

[0045] Step S4 Implementation Details According to the B-level rating, match the resource strategy: (1) Automatically configure the R&D expense add-back policy; (2) Dynamically monitor the system to detect its new environmental penalties, and trigger real-time reevaluation and downgrade to C level and freeze subsidies; (3) Bind the enterprise coordinates with the policy list to generate a regional heat map to show that the B-level enterprises in the northwest area are concentrated, but the policy coverage rate is less than 12%, which drives the targeted optimization of resource allocation by the investment promotion department.

[0046] Through the above-mentioned embodiments, the data heterogeneous hierarchical storage improves the efficiency of unstructured data processing by about 3 times, and the missing value interpolation algorithm improves the field integrity from 81% to 98.7%, which is much higher than the level of about 90% of similar systems. Moreover, the risk dynamic weight reduction mechanism reduces the misjudgment rate of high-risk enterprises by 42% through industry adaptive thresholds (such as setting the debt ratio threshold of technology enterprises to 40%), while the traditional fixed threshold (such as uniform 60%) leads to about 28% of missed judgments. Furthermore, the spatial resource binding technology associates the geographical location of enterprises with dynamic ratings to generate heat maps that accurately identify policy blind spots (such as coverage gaps in the northwest area), saving about 75% of decision-making time compared to manual analysis. Finally, the pre-position direct rating rule skips the calculation steps for 2 enterprises that have obtained investment from the national industrial fund, directly outputting S-level ratings, reducing the evaluation time from an average of 4 hours to 5 minutes, and reducing resource consumption by about 87%.

[0047] Those skilled in the art should understand that the meanings of several technical terms such as "multi-dimensional quantitative parameters" in some of the above-mentioned embodiments can be understood in combination with the following aspects: Multi-dimensional quantitative parameters: a set of standardized calculation results output by four index models of growth, activity, risk control, and innovation capability, used for enterprise comprehensive rating. This parameter is integrated by weighting to obtain a comparable value of 0-100 points, solving the problem of dimension fragmentation in traditional evaluation. Specifically, it includes: (1) Growth parameter: monthly score based on organizational dynamics (8 points for a senior management turnover rate of ≤10%), capital dynamics (10 points for B-round financing), and project dynamics (8 points for R&D investment ≥8%); (2) Activity parameter: quarterly calculation of investment and financing activity (5 points per 1 billion yuan of financing), intellectual property output (5 points per invention patent), and brand dissemination (5 points per central media report); (3) Risk parameter: financial risk (1 point deduction for every 1% above the industry threshold), compliance risk (10 points deduction for major lawsuits), etc.; (4) Innovation parameter: R&D investment ratio bonus (20 points for ≥20%), and industry-university-research cooperation bonus (5 points for national level).

[0048] Dynamic score: refers to the enterprise capability indicators updated over time, and its "dynamic" is reflected in three aspects: 1) update mechanism: growth score is updated monthly (such as capital flow monitoring), and activity score is updated quarterly (such as patent output statistics); 2) data driven: relying on real-time input of structured data sets (such as negative news detected by public opinion system triggering score reduction immediately); 3) historical correlation: automatically upgraded when the trend of the last three periods rises. Unlike static snapshot evaluation, dynamic score makes enterprise rating response speed increase by 20 times.

[0049] Example 2 The following takes a manufacturing enterprise as an example to illustrate the implementation process of the technical solution of the application.

[0050] Step S1: Multi-source heterogeneous enterprise data integration Step S1a. Data collection: Obtain the financial statements of Cloud X Technology from 2021 to 2023 through the API interface of the stock exchange, synchronously crawl the patent data (including invention patent CN202110XXX and utility model CN202230XXX) disclosed by the State Intellectual Property Office, and subscribe to the public opinion platform to capture media reports related to its "intelligent production line". The data subscription channel pushes the record of its B-round financing of 150 million yuan in real time, forming an initial heterogeneous data set.

[0051] Step S1b. Hierarchical storage architecture: Structured financial data is stored in the MySQL database table fin_data, patent JSON data is imported into the MongoDB collection patent_coll, and unstructured public opinion text is stored in the HDFS file path / news / 2023 / yunchuang. Establish a unified metadata index table (Table 1) to associate the three types of storage carriers: Table 1: Metadata index table Step S1c. Data standardization: For the missing "R&D investment ratio" field in 2022, use the industry average of 12.5% to interpolate and fill in; convert the dollar financing amount to RMB according to the real-time exchange rate of 7.2; merge the employee growth rate data from the industrial and commercial system and the social security platform, and output the structured data set (Table 2): Table 2: Example of standardized data set Step S2: Multi-dimensional dynamic score calculation Step S2a. Growth evaluation (executed monthly) Call Table 2 data to execute the growth evaluation model: 1) Organization dynamic score calculation (1) Team stability: 8% turnover rate (≤10%), 8 points (2) Architecture iteration: Add "Intelligent Equipment Division", which meets the definition of "Strategic Department", and get 2 points Organizational dynamic score = 8 + 2 = 10 2) Calculate the dynamic score of funds B round financing record triggers scoring rules (≥100 million yuan), 10 points Output: Dynamic growth quantification parameter = 20 points (full score 25 points) Step S2b. Activity assessment (per quarter) 1) Industry weight dynamic allocation (claim 9) (1) Manufacturing enterprises: production capacity qualification weight increased to 30% (original 25%) (2) Technology enterprises: intellectual property weight increased to 40% (original 30%) 2) Investment and financing activity Financing event amount 150 million → trigger grading rules (≥100 million), 5 points 3) Intellectual property output New utility model patents 2 (2 points per piece x 2), 4 points Activity score = (5 x 0.3) + (4 x 0.3) = 2.7 Step S3: Comprehensive rating generation Step S3a. Risk verification Detect asset-liability ratio > 80% for two consecutive years, trigger D-level grading rules (claim 11), directly downgrade the rating and freeze policy resources: Comprehensive rating = D level (original B level) Step S3b. Grading mapping 1) Match the manufacturing threshold interval (Table 3): Table 3: Manufacturing rating threshold 2) Risk calibration score 65 points → mapped to B level Step S3c. Trend correction Historical rating sequence: 2023Q1 (B level) → 2023Q2 (B level) → 2023Q3 (B level). Trigger risk review for three consecutive periods, maintain B level but mark "insufficient growth momentum" warning.

[0052] Step S4: Dynamic management implementation Step S4a. Policy matching B-level enterprises match "technology transformation interest subsidy" policy, generate resource allocation list: (1) Subsidy quota: 2 million yuan (2) Technical consultation: 10 times / year Step S4b. Periodic reevaluation 2023Q4 detected environmental administrative punishment: (1) Real-time trigger rating update (B level → C level) (2) Dynamic adjustment policy: cancel interest subsidy and start "green production support plan" Step S4c. Spatial visualization Bind to Baiyun District geographic information library, and output heat map on electronic map Figure 1 ): [Cloud X technology coordinate point] Policy coverage intensity: Core plant area: 85% Logistics center: 60% Note: The heat value is calculated by kernel density estimation method, and the bandwidth parameter is set to h=0.5.

[0053] Through the above technical solution, through multi-dimensional index dynamic coupling, industry adaptive weight distribution and real-time topology analysis, the problems of data fragmentation and static lag in traditional evaluation are solved, and technical level support is provided for financial institution credit decision-making. The following technical problems are solved: 1) Dynamic weight mechanism improves industry adaptability After the weight of manufacturing industry capacity is increased to 30%, the evaluation error rate of cloud X technology is reduced from 15% to 8% (compared with traditional fixed weight model).

[0054] 2) Real-time risk interception Through public opinion monitoring, risk signals are captured within 48 hours before administrative punishment publicity, which improves the efficiency of manual screening by 40%.

[0055] 3) Topology correlation analysis accuracy The correlation coefficient R² of supply chain stability index in predicting customer churn rate is 0.89 (the traditional method is only 0.62).

[0056] It should be understood that the semi-structured data in this embodiment refers to data with partial organizational structure but not strictly following the form of table (such as JSON format patent abstract).

[0057]

Embodiment 3

[0058] 1) In step S1, collect multi-source heterogeneous enterprise data, clean and standardize the multi-source heterogeneous enterprise data, and generate a structured data set.

[0059] Step S1a collects enterprise financial data, patent data, public opinion data, and investment and financing data in parallel through a crawler program, an API interface, and a data subscription channel to form an initial heterogeneous data set, including: (1) Financial data: Obtain the 2022-2024 value-added tax return form (structured) through the tax API interface, including the revenue compound growth rate of 41%, the R&D expense ratio of 18%, etc. (2) Patent data: Crawl the State Intellectual Property Office announcement (semi-structured) to extract 23 invention patents and 5 PCT international patents. (3) Public opinion data: Subscribe to the text of the public opinion platform (unstructured) to identify the proportion of positive public opinion on "solid-state battery technology breakthrough" as 82%. (4) Investment and financing data: Retrieve B-round financing of 250 million yuan through the Qichacha API.

[0060] As a result, the initial data set contains 152 fields, and the data volume is about 90% higher than manual collection.

[0061] Step S1b stores the structured data in the initial heterogeneous data set into a relational database, imports semi-structured and unstructured data into a distributed file storage system, and constructs a hierarchical storage architecture: financial data is stored in a MySQL database (relational), patent abstracts are imported into a MongoDB document database (semi-structured), and public opinion text is stored in a HDFS distributed system (unstructured). Establish metadata index association with three types of storage carriers, and shorten the query response time to 0.8 seconds (traditional architecture ≥ 3 seconds).

[0062] Step S1c performs a deduplication operation on the data in the hierarchical storage architecture, fills in missing fields by interpolation, and unifies the units and formats of numerical fields, outputting the structured data set, wherein: (1) Deduplication: Merge 3 duplicate financing records. (2) Interpolation: Fill in the missing 2023 social security number as 286 people according to the industry average. (3) Standardization: Currency unit is unified to RMB ten thousand yuan, and R&D investment ratio is unified to percentage format. The output structured data set has a completeness of 99.1% (industry average ≤ 92%).

[0063] 2) In step S2, based on the structured data set, calculate dynamic scores through preset growth evaluation model, activity evaluation model, risk control model and innovation ability model respectively to generate multi-dimensional quantitative parameters.

[0064] The structured data set is processed by a preset growth evaluation model, enterprise growth scores are calculated monthly based on organization dynamics, capital dynamics and project dynamics indicators, dynamic growth quantitative parameters are generated, and organization dynamics scores are calculated monthly: S 组织 = 0.2 x (team stability / 8 + architecture iteration / 6 + employee growth / 6), wherein the core team turnover rate is 5% (8 points), a new solid-state battery research and development center is added (architecture iteration + 2 points), and the employee annual increase is 35% (6 points), thereby outputting a monthly growth score of 84.3.

[0065] Step S2b processes the structured data set by a preset activity evaluation model, calculates enterprise activity scores quarterly based on investment and financing activity, intellectual property output and brand dissemination indicators, generates dynamic activity quantitative parameters, and specifically: Perform industry weight dynamic allocation (step S9) S 活跃 = 0.35 x (intellectual property score) + 0.25 x (investment and financing score), wherein the intellectual property: 2 new PCT patents (10 points x 2 x 0.35 = 7 points); investment and financing: B round financing 250 million yuan (amount classification score 8 points); thereby outputting quarterly activity 89.6.

[0066] Step S10 performs innovation efficiency conversion evaluation in the innovation capability model, extracts technology transaction contracts, annual technology transfer amount 180 million yuan (≥ 100 million yuan additional 10 points), 2 national university-industry research projects (additional 5 points x 2 = 10 points), and the innovation capability basic score is increased from 70 points to 90 points.

[0067] In step S2c, the dynamic growth quantitative parameters, the dynamic activity quantitative parameters, the risk control model output value and the innovation capability model output value are integrated and normalized into the multi-dimensional quantitative parameters.

[0068] Weighted integration formula: P 综合 = 0.3 x S 成长 + 0.3 x S 活跃 + 0.2 x S 风险 + 0.2 x S 创新 , thereby outputting multi-dimensional parameter values: growth 84.3, activity 89.6, risk 75, innovation 90.

[0069] 3) In step S3, the dynamic scores in the multi-dimensional quantitative parameters are integrated and mapped into comprehensive rating results according to the grading rules.

[0070] Step S3a performs risk verification on the multi-dimensional quantitative parameters: when the risk control model output value is lower than the preset threshold, the enterprise dynamic score in the multi-dimensional quantitative parameters is proportionally down-weighted to generate a risk calibration parameter: Wherein, the risk control model detects the asset-liability ratio of 68% (manufacturing industry threshold < 60%): Thus, the growth score is down-weighted to 77.56.

[0071] Step S3b inputs the risk calibration parameter into the preset grading mapping rule library, dynamically adjusts the rating threshold interval according to the industry characteristics, and outputs the initial rating result. In the grading rule library matching the new energy automobile industry: A-level threshold: 80-89 points (traditional manufacturing is 75-84 points), after calibration, growth score 77.56 + active score 89.6 = 167.16 → mapped to B-level.

[0072] Step S3c trend corrects the initial rating result based on historical rating results: if the rating trend is rising for three consecutive periods, the grade is raised, if it is continuously falling, the risk review is triggered, and the comprehensive rating result is generated: historical rating: 2023Q4 (B+), 2024Q1 (A-), 2024Q2 (B), two consecutive periods of decline triggered risk review, confirmed pending patent litigation, finally downgraded to B-level.

[0073] 4) In step S4, according to the comprehensive rating result, the enterprise classification and grading operation is performed, and the grading management list is output.

[0074] Step S4a matches the preset resource allocation strategy library according to the comprehensive rating result, automatically configures a differentiated policy support scheme for enterprises of different grades, generates an initial resource allocation list, and specifically: Match B-level strategy, automatically configure R&D expense add-back ratio 150%, provide green factory declaration channel; thus generate resource list containing 5 kinds of support policies.

[0075] Step S4b periodically re-evaluates the enterprises in the grading management list based on the dynamic monitoring mechanism: when major risk events or abnormal fluctuations in indicators are detected, real-time rating update is triggered and the policy support scheme is adjusted synchronously. Monitor that in July 2024, it obtains the qualification of national technology innovation demonstration enterprise (step S10 adds 30 points), real-time re-evaluation is triggered: innovation capability score rises to 120 points, comprehensive rating rises to A-level, and intellectual property pledge financing permission is unlocked.

[0076] Step S4c binds the policy support scheme and the dynamic reevaluation result to the spatial geographic information library, visualizes the enterprise distribution heat and policy coverage intensity in the electronic map, and outputs the regional investment optimization atlas. Binding the enterprise coordinates (116.41°E, 39.92°N), the heat map shows that the density of A-level enterprises in the new energy cluster of Beijing-Tianjin-Hebei is 38 per hundred square kilometers, and the policy coverage rate is only 45%, identifying the Langfang area as a resource investment blind area.

[0077] Technical effects achieved by the embodiment 1) Heterogeneous data fusion efficiency: (1) Hierarchical storage of semi-structured patent data and unstructured public opinion (S1b) reduces data cleaning time to 28% of traditional methods; 2) The accuracy of the missing field interpolation algorithm is 99.1% (industry average ≤ 92%).

[0078] 2) Dynamic decoupling calculation advantage: (1) Monthly growth points (S2a) and quarterly active points (S2b) asynchronous update mechanism, evaluation period from 30 days to 5.2 hours; 2) Industry weight dynamic allocation (S9) improves the accuracy of new energy vehicle intellectual property evaluation by about 43%.

[0079] 3) Risk adaptive mechanism: Asset-liability ratio trigger ratio weight reduction formula (S3a): (Δ R Risk deviation amount, K Industry adjustment coefficient), the false positive rate is reduced by 39.7% compared with the fixed threshold model.

[0080] 4) Spatial resource linkage efficiency: (1) Policy coverage heat map (S4c) identifies the gap in the Langfang area, and the investment department adjusts resource allocation within 48 hours, with a response speed improved by about 8 times; 2) The statistical error of regional A-level enterprise density is reduced from ±15% of manual analysis to ±2.3%.

[0081] The present application realizes comprehensive capability evaluation with a false positive rate of less than 4.7% in the field of new energy vehicles through multi-source data fusion, dynamic weight allocation, risk adaptive calibration, and spatial resource linkage, which is about 60% higher than the existing technology (average false positive rate ≥ 12%), providing irreplaceable technical support for regional industrial decision-making.

[0082] Among them, those skilled in the art should understand that some concepts in some embodiments can be explained as follows: Dynamic allocation of industry weight: automatically adjust the evaluation dimension weight coefficient according to the national economic industry classification code, for example, the intellectual property weight of technology enterprises is increased to 30%-40%, and the manufacturing capacity qualification weight is increased to 25%-35%.

[0083] Spatial geographic information library: refers to a GIS database that stores enterprise geographic coordinates and policy coverage intensity, wherein the data input is the GPS coordinates of the enterprise registered address, and the policy coverage radius (such as 5 kilometers within an industrial park); the output form is an electronic map heat map (red: D-level enterprise gathering area, green: S-level enterprise dense area); the application scenario is to identify weak industrial areas and automatically push upgrade solutions.

[0084] Example 4 As Figure 3 shown below, the following examples take the biopharmaceutical enterprise "XX Biopharmaceutical Co., Ltd." (referred to as XX Biopharmaceutical) as the evaluation object, and the implementation process of the technical solution of the present application is described in detail. The enterprise was established in 2019 and focuses on the research and development of new anticancer drugs. It has completed C round financing and holds 3 clinical batch items of Class I new drugs. The compound growth rate of revenue in 2024 reached 58%.

[0085] 1) Step S1a: Multi-source data collection Obtain the enterprise's financial statements for 2022-2024 through API interface (structured data), extract key fields: R&D investment proportion 32% (industry average 15%), revenue compound growth rate 58%; collect patent announcements from the State Intellectual Property Office through a crawler program (semi-structured data): 18 invention patents (including 6 PCT international patents), 2 Class II medical device registration certificates; obtain public opinion monitoring reports through data subscription channels (unstructured data): "PD-1 inhibitor clinical breakthrough" positive public opinion proportion 91%, FDA fast track qualification news communication volume: 120,000 times.

[0086] Through multi-source data collection, an initial data set covering 152 indicators of enterprise operation is formed, and the data collection efficiency is improved by 87% compared with traditional manual methods.

[0087] 2) Step S1c: Cleaning and standardization Perform data deduplication: merge 2 duplicate financing records (2023 C round financing 1.8 billion yuan and 1.5 billion yuan announcement deduplication) and fill in missing fields: The missing number of social security in 2023 is interpolated as 217 according to the average of the biopharmaceutical industry, and the calculation formula is: Where ω i is the weight of the same scale enterprise, x i is the actual value In addition, unify the data units and formats: unify the currency units to RMB ten thousand yuan, and unify the R&D investment proportion to percentage format (such as 0.32→32%) After cleaning and standardization, the output structured dataset completeness reached 99.3%, and the field abnormal value ratio decreased to 0.7% (industry average ≥8%).

[0088] 3) Step S6: Rating condition verification and direct rating Key qualification data detected: 1) National Industry Fund "Big Health Industry Investment Fund" strategic investment of 250 million yuan (≥ 100 million yuan threshold); 2) Leading the development of "CAR-T Cell Therapy Quality Control International Standard" (ISO / TC 276); System automatically triggers S-level direct rating, comprehensive rating result is S-level (95 points), generates enterprise rating report containing strategic value label: "breakthrough technology leading enterprise", and skips subsequent assessment steps. This process takes only 3.2 seconds (the average time required for complete assessment process is 5.5 hours).

[0089] 4) Step S2a: Growth assessment Note: Due to the satisfaction of S6 conditions, this step is not actually executed, but the system maintains the calculation ability for standby, implementation details (simulation scenario): a) Organizational dynamic score calculated by month: S 组织 =0.2x(8 / 8+6 / 4+6 / 6)=0.2x(1+0.67+1)=0.534, where team stability-core team turnover rate 3% (8 points), architecture iteration-add new gene therapy R&D center (+4 points), employee growth-yearly increase 45% (6 points); b) Calculate the funding dynamic score: C round financing 250 million yuan (amount classification 10 points), cash flow growth rate 41% (8 points).

[0090] Finally, the output monthly growth score is 86.7 points.

[0091] 5) Step S9: Dynamic allocation of industry weight According to the national economic industry classification code (C2760 biological drug manufacturing): increase the weight of technological innovation to 38% (baseline value 25%), reduce the weight of market promotion to 12% (baseline value 20%), the weight adjustment formula is: Where α is the industry adjustment coefficient (0.35 for biological medicine), I industry is the industry innovation index.

[0092] 6) Step S2b: Activity assessment Quarterly indicators: Intellectual property output - 3 new PCT patents (10 points x 3 x 0.38 = 11.4 points), investment and financing activity - completed Pre-IPO financing of 500 million yuan (amount classification 10 points), brand communication - 2 international medical journal cover stories (8 points each). Therefore, the activity score calculation: S Activity = 0.38 x 11.4 + 0.25 x 10 + 0.2 x 16 = 89.32.

[0093] 7) Step S3a: Risk Weighting Process The risk control model monitors a major event: a patent infringement lawsuit in Q3 2024 (minus 10 points), and the asset-liability ratio rises to 65% (biomedical threshold <40%), triggering the proportion of weight reduction mechanism: .

[0094] The system automatically generates a risk warning: "Intellectual property risk leads to temporary rating reduction".

[0095] 8) Step S4b: Dynamic Monitoring and Reevaluation Key positive events are monitored: (1) Obtained FDA orphan drug qualification in December 2024 (additional 15 points); (2) Completed patent settlement and received compensation of 120 million yuan; Real-time trigger reevaluation process: (1) Clear original lawsuit deduction record; (2) Add qualification bonus items; (3) Recalculate the comprehensive score: S new =71.25+15+5(settlement bonus)=91.25.

[0096] Thus, the rating is restored to A+ level, and the entire process is completed within 1.8 hours after the event.

[0097] Technical effects achieved by this embodiment Data collection and processing efficiency: Multi-source heterogeneous data fusion mechanism (S1a+S1c) compresses the data preparation period from traditional 7 days to 4.5 hours, and the data completeness is 99.3% (industry average ≤90%); Direct grading rules (S6) save 98.7% of evaluation resources for strategic innovative enterprises, with a response speed of 3.2 seconds / enterprise.

[0098] Dynamic assessment accuracy improvement: Industry weight distribution (S9) improves the precision of biopharmaceutical technology innovation evaluation by 47.6%, with a misjudgment rate of 3.8%; risk self-adaptive weight reduction (S3a) through a linear compensation model: (Wherein Lambda The high-risk enterprise underreporting rate is reduced by 52% by using the industry risk coefficient.

[0099] Decision support breakthrough improvement: dynamic reevaluation mechanism (S4b) realizes major event response within 2 hours, which is about 108 times faster than traditional quarterly assessment; the system automatically completes rating adjustment 23 times during the 2024 monitoring period, keeping the real-time deviation of assessment results <1.5 points.

[0100] The technical solution realizes enterprise comprehensive capability assessment precision of 98.2% in the biomedicine field through multi-source data fusion, intelligent grading and shunting, industry adaptive assessment, and real-time risk response technology synergy, which is 13.2 percentage points higher than the existing technology (average precision ≤85%), providing an irreplaceable technical foundation for precise support of technology innovation enterprises.

[0101] [Example 5] Here, as Figure 4 shown, the technical solution formed by the above steps S5', S6', and S7' can be applied to an application scenario to describe the specific implementation of the three steps in detail.

[0102] Specific implementation scenario description In a certain regional new energy automobile industry chain, power battery enterprise A causes idle capacity of downstream vehicle enterprise B due to lagging behind in technology iteration. Combined with the enterprise growth index (used for supply chain dependence calculation), risk control index, and innovation capability index (affecting technology route similarity), the technical solution is implemented for chain break repair.

[0103] Step S5' implementation details: industry chain positioning map construction The technology route similarity is calculated by calling the technology route maps of enterprises A and B (such as battery energy density route and BMS control system version): technology route similarity σ = number of common technology features / total number of maximum technology features, wherein the i-th generation technology feature vector of enterprise A contains elements such as solid-state battery patent proportion ≥30%, and the calculation result is σ = 0.62 (industry threshold σ min = 0.75); Fusion of "supply chain dependence" index: 1) enterprise B's procurement proportion of A (R 采购 = 85%); 2) enterprise A's capacity specificity coefficient (α = 0.9) to generate an anti-interference map, indicating that A→B is a high-risk single-point dependence link.

[0104] Through the above steps, the traditional method only counts transaction frequency, while the coupling of σ and α quantitatively reveals the dual risks of technology generation difference (2 generations) and supply rigidity, and the fault identification accuracy is improved by more than 40%.

[0105] Step S6' implementation details: dynamic quantification of synergistic effect The analysis technique generation gap coefficient δ = generation gap years / industry maximum tolerance period, calculated δ = 1.8 (critical value 1.5); The dynamic parameters injected into the "activity index" are: 1) logistics timeliness deviation β = + 22% (industry average ± 10%); 2) order fulfillment volatility γ = 35%, thus generating a synergistic effect warning value Ew = δ × (0.6β + 0.4γ) = 86 (> 80 triggers a red warning). Therefore, the linear weighting of δ and the operation dynamic parameters converts the static risk into real-time warning, which is 3 times faster than the pure financial analysis model.

[0106] Step S7' implementation details: breakpoint repair scheme generation Locate the redundant area of production capacity: search for enterprise clusters with power battery production capacity utilization rate < 65%, and lock enterprise C (production capacity utilization rate 58%, technical route similarity σ C-B = 0.82); Call risk control index verification: (1) Enterprise C risk index R c = 92 (excellent) (2) Historical litigation times = 0 Output the chain supplement scheme: migrate 40% of the orders of enterprise B to enterprise C to form A-C-B backup link.

[0107] R c ≥ 90 to avoid qualification fraud risk; technical adaptation and risk control double verification make the failure probability of the chain supplement scheme drop to less than 5%.

[0108] Through the above technical scheme, the following technical effects are achieved: Dynamic closed-loop optimization: after the implementation of the scheme, the production capacity utilization rate of enterprise B increases from 63% to 82%, the positioning map is updated and the A→B dependence is reduced to 45%; the technical fault repair period is shortened to 45 days (industry average 120 days); the production capacity loss rate in the simulation of supply chain interruption is reduced from 32% to 9%; the annual revenue of enterprise C increases by 120 million yuan.

[0109] This embodiment shows that the structured output of step S5b' drives the dynamic risk quantification of step S6', and the risk control model closed-loop verification of step S7b' forms an inseparable technical chain, and its synergistic effect far exceeds the independent implementation effect of each step.

[0110] It should be noted that some of the above terms "technical route similarity type" "production capacity specificity coefficient" "anti-interference industry positioning map" and the like can be understood as follows: 1. Technical route similarity (σ) The quantitative indicator characterizing the degree of technology generation matching between upstream and downstream enterprises in the industrial chain is calculated by comparing the commonness and difference of technology feature vectors (such as patent layout, R&D direction, product parameters).

[0111] Calculation formula: Where, T X,i is the i-th generation technology feature vector of enterprise X (for example: solid-state battery patent proportion ≥ 30%, BMS system version number), threshold σ min = 0.75: below this value is considered as technology fault risk.

[0112] 2. Capacity specificity coefficient (α) Measures the degree of irreplaceability of enterprise production equipment / technology for specific customers, with a value range of [0, 1].

[0113] Calculation method: For example: α of enterprise A = 0.9, indicating that 90% of its capacity cannot be quickly diverted to other customers, increasing the risk of supply chain rigidity.

[0114] 3. Technology generation difference coefficient (δ) Quantifies the imbalance of technology iteration speed between upstream and downstream enterprises, reflecting the severity of technology fault.

[0115] Calculation formula: δ = ΔT / τ max , where ΔT is the technology generation difference in years (for example, enterprise A's battery energy density lags behind the industry mainstream by 2 generations), τ max is the industry's maximum tolerance period (1.2 years in the field of power batteries).

[0116] Risk classification: δ > 1.5 triggers a red alert.

[0117] 4. Collaboration effect early warning value (Ew) Fusion of technology fault and operational volatility dynamic risk index, used to predict the probability of industrial chain breakage.

[0118] Synthesis formula: Ew = δ × (k1β + k2γ), where k1 = 0.6, k2 = 0.4, β is the logistics time efficiency deviation rate (actual time efficiency / industry benchmark time efficiency - 1), γ is the order fulfillment fluctuation rate (standard deviation / mean).

[0119] Action threshold: E w > 80 needs to be immediately started to supplement the chain.

[0120] 5. Risk control index (R c ) Enterprise compliance score based on risk control model output, range 0-100 points.

[0121] Core check items: (1) Financial risk: consecutive losses, asset-liability ratio > 80%; (2) Legal risk: major lawsuit defeat, administrative penalty; (3) Safety risk: environmental accidents, production safety incidents.

[0122] Classification standard: R c ≥90 is "excellent level", which is allowed to be included in the chain supplement scheme.

[0123] 6. Anti-interference industry positioning map The industry chain topology network constructed by fusing the similarity of technology routes (σ) and the supply chain dependence (α) has the following characteristics: Dynamic weight: mark single-point dependent links (such as α>0.8 and σ<0.75) as red high-risk nodes; Anti-interference design: identify backup path candidate enterprises (need to meet σ>0.8 and Rc≥80); Recursive update: when the chain supplement scheme is implemented, the dependence relationship of the map is automatically reconstructed.

[0124] [Example 6] Here, the technical solution formed by the above steps S4c1-S4c3 can be applied to an application scenario to describe the specific implementation of the three steps in detail.

[0125] Specific implementation scenario description Taking the new energy automobile industry chain risk monitoring of a province as an example, the industry chain includes lithium battery manufacturing enterprise A (upstream), electric control system enterprise B (midstream), and whole vehicle assembly enterprise C (downstream). In March 2025, enterprise A suddenly has a major environmental penalty event, triggering the dynamic risk assessment process of the present scheme.

[0126] Step S4c1 implementation details: cross-enterprise risk transmission path extraction 1) Call the industry risk threshold (lithium battery manufacturing environmental risk threshold = 75 points) and enterprise correlation strength value output by the risk control model.

[0127] 2) Calculate the dependence coefficient based on the supply chain database: (1) Enterprise B's dependence on enterprise A's raw material procurement = 0.85 (> threshold 0.8) (2) Enterprise C's dependence on enterprise B's module supply = 0.78 (> threshold 0.7) 3) Extract the transmission path A→B→C with a confidence level lower than the predetermined threshold (0.6).

[0128] Technical effects achieved: Identify core transmission links within 10 minutes, traditional manual analysis requires more than 3 days; Break through the limitations of static single risk assessment, and make the chain risk of "environmental punishment -> supply chain interruption -> vehicle production stop" explicit.

[0129] Step S4c2 implementation details: Dynamic calculation of transmission path strength 1) Real-time collection of incremental public opinion data stream: (1) Environmental protection department official website publicizes enterprise A punishment decision (weight coefficient 0.3) (2) Social media spreads "lithium battery factory shutdown" topic (peak volume 120,000 messages / hour, weight coefficient 0.2) 2) Dynamically update transmission strength coefficient: (1) Where α = 0.7 (historical decay factor), β = 1.2 (event impact coefficient) (2) Input public opinion events I1 (punishment announcement), I2 (shutdown rumor), weight ω1 = 0.3, ω2 = 0.2 3) Output path strength: A→B strength rises from 0.35 to 0.92, B→C strength rises from 0.28 to 0.81.

[0130] Technical effects achieved: Complete coefficient jump determination within 1 hour, traditional solution requires waiting for quarterly audit report; Fusion of unstructured public opinion data (such as social media topic volume) improves risk prediction accuracy by about 40%.

[0131] Step S4c3 implementation details: Risk heat map layer generation and decision optimization 1) Map transmission strength coefficient to electronic map: (1) Label the area where enterprise A is located as a red high-risk area (strength > 0.9) Label the areas where enterprises B and C are located as orange medium-high risk areas (strength > 0.8) 2) Superimpose policy coverage database: (1) Identify that enterprise C is located in an economic development zone without "supply chain emergency loan" policy coverage (gap rate 100%) The policy coverage rate of enterprise B in the high-tech zone is only 30% (lower than the threshold of 60%) 3) Automatically generate decision instructions: (1) Urgently inject supply chain financial support to the area where enterprise C is located (2) Start compliance review green channel for enterprise B Technical effects achieved: Risk event response speed is shortened to 2 hours (traditional process requires 7 days); Policy resource allocation accuracy is improved to 95% (average manual decision error rate is about 35%) Through the technical synergy between the above three steps, the following technical effects are achieved. First, in terms of dynamic recursive optimization mechanism, the policy adjustment instruction output by step S4c3 (such as adding compliance review for enterprise B) is fed back to the correlation strength calculation of step S4c1 in real time, triggering the risk threshold recalibration of enterprise B (from 0.7 to 0.85); forming a "risk identification → decision intervention → model self-update" closed loop, making the system false positive rate continue to decline (after 3 months of implementation, the false positive rate decreased from 12% to 3.5%). Second, the industry chain level risk blocking is realized: by injecting emergency funds to enterprise C in advance, avoiding the loss of about 230 million yuan due to the shutdown of the whole vehicle (the actual loss caused by the response lag of the traditional scheme); the heat map driven directional policy delivery makes the regional industry chain recovery period shorten by 67% (from 45 days to 15 days). Steps S4c1-S4c3 realize the three-level synergy of risk transmission quantification (step S4c1), dynamic strength calculation (step S4c2), and spatial decision mapping (step S4c3), which converts abstract industry chain risks into executable space-time governance schemes. The technical breakthrough lies in: solving the identification blind area of cross-enterprise implicit risk transmission for traditional methods; eliminating the problem of loss expansion caused by risk response lag; realizing the paradigm shift of regional industry governance from passive disaster relief to active defense.

[0132] It should be understood that in the embodiment, the enterprise correlation strength value is a parameter for quantifying the dependence relationship between enterprises, including supply chain dependence (purchase proportion), equity correlation degree (cross-holding proportion), etc., which is calculated in real time through an industry chain database. The confidence threshold is set according to the historical risk event transmission probability.

[0133]

Embodiment 7

[0134] Taking "Digital Technology Co., Ltd." as an example, the implementation process of the application in the construction and application of enterprise evaluation index system is described in detail. The company was founded in 2020 and focuses on industrial internet platform development. It has completed A round of financing, has 15 softwares, and the revenue growth rate reached 65% in 2024.

[0135] Step S1: Multi-source data collection and cleaning Obtain the financial statements of the enterprise for 2022-2024 through the tax interface, extract the finance and tax dimension indicators: revenue compound growth rate 65%, R&D expense additional deduction 2.8 million yuan, asset-liability ratio 55%; collect the "special, sophisticated and new" enterprise public data (semi-structured) from the Ministry of Industry and Information Technology through crawling, extract the size dimension indicators: employee number increased from 80 to 210, registered capital 50 million yuan; obtain 15 soft writings and 5 invention patents through the intellectual property bureau API; obtain A-round financing 60 million yuan record through investment and financing platform. Interpolate the missing 2023 profit rate field according to the industry average of 12%, unify the currency unit to ten thousand yuan, and the output structured data set completeness reaches 98.9%.

[0136] Step S2: Multi-dimensional index system construction and dynamic scoring An evaluation index system covering four core dimensions of finance and tax, innovation, size, and growth is constructed: Finance and tax dimension (weight 30%): revenue growth rate (65%→ score 20 / 25), profit margin (12%→ score 15 / 20), asset-liability ratio (55%→ score 18 / 20), weighted score 78.5; Innovation dimension (weight 30%): R&D input proportion (22%→ score 20 / 25), number of intellectual property (20 items→ score 22 / 25), industry-university-research cooperation (2 provincial projects→ bonus 6 points), weighted score 84.3; Size dimension (weight 20%): employee growth rate 162.5% (score 18 / 20), registered capital 50 million yuan (score 15 / 20), weighted score 66; Growth dimension (weight 20%): financing ability (A-round 60 million yuan→ score 20 / 25), market expansion speed (new customers increased by 80% per year→ score 22 / 25), weighted score 84.

[0137] Integrate the above dimension scores to generate multi-dimensional quantitative parameters: P comprehensive = 0.3 x 78.5 + 0.3 x 84.3 + 0.2 x 66 + 0.2 x 84 = 79.2 Step S3: Comprehensive rating and industry adaptation According to the rating threshold of the technology service industry (A level: 80-100 points), the initial score 79.2 is mapped to B+ level. Combined with the historical rating trend (B level in 2023Q4, B+ level in 2024Q1), the trend rising rule is triggered, and finally it is upgraded to A- level.

[0138] Step S4: Graded management and resource matching According to the A-rating matching resource policy: automatically configure "high-tech enterprise tax reduction" "talent introduction subsidy" and other 5 policies, generate resource list; bind enterprise geographic coordinates to regional industry map, and display the enterprise in the core area of the scientific and technological corridor on the heat map. The policy coverage strength reaches 90%, and no additional resource injection is needed.

[0139] Technical effects The embodiment realizes accurate description of the comprehensive ability of the enterprise by constructing a scientific index system covering four dimensions of finance and tax, innovation, scale and growth: The index system weight dynamically adapts to the industry characteristics (such as the innovation weight of a technology enterprise increasing to 30%), and the evaluation error rate is reduced to 5.2%; Under the support of multi-source data, the index field completeness reaches 98.9%, and the missing value interpolation algorithm is better than the industry average level; The dynamic rating mechanism responds to the growth trend of the enterprise, and the trend upward triggers the grade promotion, avoiding underestimating high-growth enterprises; The resource matching strategy is linked with spatial visualization to realize "one enterprise one strategy" precise support, and the policy investment accuracy is improved to 96%.

[0140]

Embodiment 8

[0141] I. Multi-source data collection and preprocessing

[0142] II. Periodic dynamic trend modeling Organizational dynamics (20% weight): core team stability (turnover rate ≤5% 10 points), new strategic departments (each +2 points), monthly statistics; Fund dynamics (25%): financing progress (complete B round + 8 points), cash flow growth rate (≥20% year-on-year + 10 points); Project dynamics (20%): R&D investment ratio (≥15% + 10 points), national project approval (each + 5 points); Social dynamics (15%): industry summit speech frequency (each + 1 point), number of new partners (each 5 + 2 points); Media dynamics (10%): central media coverage (each + 3 points), industry in-depth interview (each + 2 points); Securities trading dynamics (10%): stock price monthly volatility (<10% + 5 points), institutional research frequency (≥2 times per month + 3 points).

[0143] The growth index is calculated monthly by weighting: Where S i,t is the score of the ith dimension in month t, and the history value of the last 36 consecutive months is retained to form a time series.

[0144] III. Dynamic visualization and trend analysis A visualization board is built through the Dash framework to display the following content: A line chart presents the monthly change curve of the growth index, with key events (such as financing and project approval) marked; A radar chart shows the current score and industry average for each dimension; A heat map displays the score change trend of the six dimensions by quarter; A historical backtracking function is provided to support comparative analysis of any time interval.

[0145] This model realizes the following effects through multi-dimensional dynamic data fusion: The monthly update mechanism ensures the timeliness of the index, with a response speed 12 times faster than traditional annual assessment; Time series analysis reveals the periodicity of enterprise growth (such as the innovation peak of technology-based enterprises every 2 years); The visualization board helps investment institutions identify high-growth nodes, and the decision-making efficiency of pilot users is improved.

[0146]

Example 9

[0147] Step S1: Market dimension multi-source data collection and processing Through the API interface of the e-commerce platform, the sales data of the enterprise in 2022-2024 is obtained (structured), and key indicators such as market share, customer repurchase rate, and channel coverage rate are extracted; through web crawler, user reviews and star ratings on platforms such as Jingdong and Tmall are captured (semi-structured); through the public opinion system, industry exhibition records and new product launch conference communication volume are captured (unstructured). The missing quarterly channel growth rate field is interpolated and filled according to the industry moving average method, and the market data unit is unified to percentage format, and the market dimension structured data set is output, and the completeness is improved.

[0148] Step S2: Market dimension index system construction and dynamic scoring An evaluation index system covering market performance, channel capacity, brand influence, and customer behavior is constructed: Market performance (weight 35%): market share annual growth rate (18% → score 22 / 25), sales revenue compound growth rate (42% → score 20 / 25); Channel capacity (weight 25%): city coverage rate (180 / 250 → score 18 / 20), number of offline experience stores annual growth rate (+30% → score 16 / 20); Brand influence (weight 20%): industry media exposure times (annual increase 50% → score 17 / 20), user positive review rate (92% → score 18 / 20); Customer behavior (weight 20%): repurchase rate (35% → score 15 / 20), new product customer conversion rate (28% → score 14 / 20).

[0149] The analytic hierarchy process (AHP) is used to determine the weight of each index, and the market dimension score is calculated by weighting: P 市场 = 0.35×84 + 0.25×72 + 0.20×70 + 0.20×58 = 74.1 Step S3: Market growth comprehensive evaluation and dynamic grading Integrate the market dimension score with other dimensions (innovation, finance, risk, etc.) and normalize it into a comprehensive growth index. According to the smart home industry rating rules (A-level threshold ≥ 75 points), the market dimension score of 74.1 in this period is temporarily rated as B+. Combined with the historical three-period trend (two consecutive periods of increase), the trend correction rule is triggered, and the final comprehensive rating is upgraded to A-.

[0150] Step S4: Market strategy matching and resource optimization According to the A- rating result, automatically match policy resources such as "market development subsidies" and "channel construction low-interest loans"; bind enterprise market heat data with geographic information system to generate channel coverage blind area map (such as southwest coverage rate of only 35%), and drive regional investment departments to targetedly put in channel support resources.

[0151] This embodiment realizes the precise quantification and dynamic response of enterprise market growth by constructing market dimension index system and comprehensive evaluation model, and achieves the following technical effects: The coverage and depth of market dimension indicators are significantly improved, with data completeness reaching 97.6%; The AHP weighting model is adapted to industry thresholds, and the evaluation error rate is reduced to 5.8%; The dynamic trend correction mechanism avoids underestimating high-growth enterprises, and the response speed is improved to the hour level; The market heat map is linked with policy resources to realize "one area one strategy" precise placement, and the resource utilization rate is improved to 89%.

[0152]

Embodiment 10

[0153] I. Data preparation and structured processing Through tax API, intellectual property bureau database, public opinion monitoring platform and investment information interface, collect multi-source data of the enterprise from 2023 to 2024, including: Financial data: revenue growth rate, R&D investment proportion, cash flow status; Organization data: core team stability, number of new R&D departments; Intellectual property data: invention patents, utility model patents, software copyrights; Public opinion data: industry media reports, social media volume; Investment data: financing rounds, amount, investment background.

[0154] After data cleaning, deduplication, interpolation (e.g. missing 2023 R&D input filled with industry average), and unit conversion, structured datasets are generated, improving completeness.

[0155] II. Growth Index Calculation Model Growth Index (G index ) is calculated monthly, covering three dimensions: organizational dynamics, financial dynamics, and project dynamics, with weights of 25%, 35%, and 40% respectively. The specific calculation formula is as follows: Organizational Dynamics Score (S 组织 ): Core team turnover rate ≤5% gets 10 points, new "smart control laboratory" adds 2 points, total 12 points; Financial Dynamics Score (S 资金 ): Complete C round financing 200 million yuan (score 10 points), cash flow growth rate ≥30% (score 8 points), total 18 points; Project Dynamics Score (S 项目 ): R&D investment proportion 18% (score 9 points), leading one national intelligent manufacturing project (add 5 points), total 14 points.

[0156] Substitute into the formula: (Full score 25 points, converted to percentage system as 62.8 points) III. Active Index Calculation Model Active Index (A index ) is calculated quarterly, covering three dimensions: investment and financing activity, intellectual property output, and brand communication, with weights of 30%, 40%, and 30% respectively. The calculation formula is as follows: Investment and financing activity: complete strategic financing 100 million yuan this quarter (score 8 points); Intellectual property output: add 5 new invention patents (2 points each, total 10 points); Brand communication: get 1 CCTV financial report (score 5 points), release new products 2 times at industry exhibitions (3 points each, total 6 points), brand total score 11 points.

[0157] Substitute into the formula: (Full score 15 points, converted to percentage system as 60.7 points) IV. Comprehensive Score Calculation and Rating Mapping After normalizing the growth index and the activity index, the risk control model output value (72 points for this period) and the innovation capability score (85 points) are combined into a comprehensive score according to the weights: P 综合 =0.30× G index +0.30× A index +0.20× S 风险 +0.20× S 创新 =0.30×62.8+0.30×60.7+0.20×72+0.20×85=69.95 According to the high-end equipment manufacturing industry rating threshold (A level: 75-100 points, B level: 60-74 points), the comprehensive score of 69.95 is mapped to B+ level. Combined with the historical trend (rising for two consecutive periods), the rating upgrade mechanism is automatically triggered, and finally it is rated as A- level.

[0158] [Example 11] Identification and grading management of high-quality enterprises driven by enterprise growth algorithm model Taking the application of this program by the investment promotion department of Baiyun District, Guangzhou City as an example, this paper elaborates in detail the implementation process of enterprise comprehensive strength evaluation and grading management based on enterprise growth algorithm model. This example focuses on calculating key indicators such as enterprise growth index and risk resistance ability index through multi-dimensional dynamic algorithm model, and finally outputs the list of high-quality enterprises and classifies them by level.

[0159] I. Data integration and preprocessing Through tax API, intellectual property bureau crawler, public opinion platform and other multi-source channels, 5000 enterprises in Baiyun District from 2022 to 2024 are collected. Financial data, patent information, investment and financing records and negative public opinion data. The initial heterogeneous data is cleaned and standardized: missing fields are interpolated with industry average, currency units are unified to RMB, semi-structured patent data is stored in MongoDB, and unstructured public opinion text is imported into HDFS distributed system. Finally, a structured data set is generated, with a field completeness of 98.7%, providing a high-quality data foundation for model calculation.

[0160] II. Multi-dimensional algorithm model calculation The growth evaluation model is called to calculate the enterprise growth index (G_index) monthly, covering three types of indicators: organizational dynamics (team stability, architecture iteration), financial dynamics (financing progress, cash flow growth rate) and project dynamics (R&D investment, number of national projects), with the weighted formula as follows: G index= 0.3 x Organizational Dynamic Score + 0.4 x Financial Dynamic Score + 0.3 x Project Dynamic Score The anti-risk ability index (R index ) is calculated through the risk control model. The asset-liability ratio, legal proceedings, environmental penalties, and other risk events are monitored, and the scores of enterprises exceeding the standard are reduced: R index = Basic Risk Score x (1 - Risk Deviation Coefficient / Industry Tolerance) The final integration of the growth index, anti-risk index, and innovation ability score is normalized to the enterprise comprehensive score (P composite ), ranging from 0 to 100 points.

[0161] III. Enterprise Comprehensive Score and Grading Management According to the comprehensive score, the Baiyun District industry self-adaptive rating rules are matched: S level: ≥90 points (national strategic enterprises) A level: 80-89 points (high-growth quality enterprises) B level: 70-79 points (stable development enterprises) C level: 60-69 points (enterprises that need attention) D level: <60 points (high-risk enterprises) The 2024 assessment results show that Baiyun District identified 312 high-quality enterprises at level A and above, including 5 S-level enterprises (including 2 directly rated enterprises). The system automatically generates a grading management list and configures differentiated policies for enterprises at different levels: A-level enterprises enjoy R&D subsidies and preferential land supply, B-level enterprises are provided with low-interest loans and market expansion support, C-level enterprises are subject to financial counseling and risk monitoring, and D-level enterprises have their resource allocation frozen and trigger compliance review.

[0162] IV. Technical Effects This embodiment realizes the dynamic quantification and precise grading of enterprise comprehensive strength through algorithm models: The monthly update mechanism of the growth model improves the assessment response speed to 20 times that of traditional methods; The risk self-adaptive weight reduction algorithm reduces the misjudgment rate of high-risk enterprises by 42%; The grading management list is linked with the geographic information system to generate a policy coverage heat map, driving the investment promotion department to supplement the resource gap within 48 hours, and improving the decision-making efficiency by 75%.

[0163] In summary, this scheme provides a quantifiable, dynamically optimized, and spatially linked technical support system for enterprise evaluation and industrial governance.

[0164]

Example 12

[0165] After completing the multi-dimensional dynamic evaluation of enterprises and generating comprehensive rating results, the park performs the following classification and management operations: I. Enterprise classification mechanism The system divides enterprises into five categories according to the comprehensive rating: S level (90-100 points): national strategic leading enterprises, such as leading international standard setting or receiving significant industrial fund investment; A level (80-89 points): high-growth quality enterprises with outstanding innovation capabilities and controllable risks; B level (70-79 points): stable development enterprises with certain growth potential but individual shortcomings; C level (60-69 points): enterprises that need to be closely monitored, with certain operational or compliance risks; D level (<60 points): high-risk enterprises that may face serious financial or legal problems.

[0166] II. Differentiated allocation of policy support resources According to the enterprise level, automatically match the preset policy resource library: S-level enterprises: enjoy "one case one discussion" special support, including the highest amount of R&D subsidies, preferential land supply, talent introduction green channel, etc. A-level enterprises: automatically configure high-tech enterprise tax incentives, intellectual property pledge financing support, and innovation project priority recommendation; B-level enterprises: provide regular support such as low-interest loans, market expansion guidance, and technical transformation interest subsidies; C-level enterprises: start risk monitoring and compliance guidance, temporarily suspend resource allocation, and need regular review; D-level enterprises: freeze all policy resources, trigger compliance review, and be included in the key monitoring list.

[0167] III. Investment attraction scenarios The system outputs a "hierarchical management list" and a "regional investment attraction heat map" to support the park's precise investment attraction: The investment attraction department prioritizes A-level and above enterprises, targets key link enterprises in the industrial chain, and enhances cluster effects; Identify weak policy coverage areas (such as C-level enterprise clusters) based on the heat map, and intensify targeted investment promotion and service resource allocation; Conduct industrial chain risk investigation in D-level enterprise concentration areas to avoid attracting low-quality or high-risk projects.

[0168] IV. Implementation effects Through this scheme, the park realizes about 70% efficiency improvement in enterprise resource matching, and the policy allocation accuracy rate exceeds 95%; 42 A-level and above enterprises are successfully introduced in 2024, 18 D-level enterprises are eliminated, the overall industrial quality is significantly improved, the investment attraction failure rate is reduced to below 5%, and a closed-loop management mechanism of "evaluation-classification-measures-optimization" is formed.

[0169] This embodiment shows that relying on the dynamic evaluation results of enterprise comprehensive capability for classification and grading management can not only realize the precise and differentiated allocation of policy resources, but also provide scientific and dynamic decision support for regional investment attraction, and promote the transformation of industrial governance from "extensive" to "fine".

[0170] [Example 13] Application example of enterprise comprehensive capability dynamic evaluation algorithm - taking "Xinghai Intelligent Technology Co., Ltd." as an example This embodiment takes a technology enterprise specializing in artificial intelligence technology research and development, "Xinghai Intelligent Technology Co., Ltd." (hereinafter referred to as "Xinghai Intelligent"), as the evaluation object, and elaborates the specific application process of algorithm processing, index design and weight setting in this invention. The company was established in 2019, has completed B-round financing, the compound growth rate of revenue reached 55% in 2024, has 28 invention patents and 15 software copyrights.

[0171] I. Data collection and preprocessing Through multi-source data interface, the data of Xinghai Intelligent from 2022 to 2024 is collected: Financial data: compound growth rate of revenue 55%, R&D expense ratio 22%, asset-liability ratio 35%; Organizational data: core team turnover rate 8%, new "AI Research Institute", employee annual growth rate 30%; Intellectual property data: 28 invention patents, 4 PCT patents, 15 software copyrights; Public opinion data: reported by central media 3 times, reported by provincial media 8 times, positive public opinion rate 85%; Investment and financing data: B-round financing 180 million yuan, investment parties including national industrial fund.

[0172] After cleaning, deduplication, interpolation (such as filling the profit rate in 2023 according to the industry average of 12%), and unit unification, the structured data set is generated, with a completeness of 98.6%.

[0173] II. Multi-dimensional algorithm score calculation 1. Organizational dynamic score (full score 20) Core team stability: turnover rate 8% (≤10%), 8 points; Organizational architecture iteration: add "AI Research Institute", get 2 points; Employee growth rate: 30% per year (≥20%), 6 points; Organizational dynamics total score: 8 + 2 + 6 = 16 points.

[0174] 2. Capital dynamics score (full score 25 points) Financing capacity: B round financing 180 million yuan, 10 points; Cash flow growth rate (for listed companies): not applicable; Asset-liability ratio trend: continuous decline for 3 years, 7 points; Capital dynamics total score: 10 + 7 = 17 points.

[0175] 3. Project dynamics score (full score 20 points) R&D investment ratio: 22% (≥8%), 8 points; New project success rate: 70% commercialization success rate of projects in the past 3 years, 7 points; Industry-university-research cooperation: 2 national cooperation projects, 4 points; Project dynamics total score: 8 + 7 + 4 = 19 points.

[0176] 4. Media dynamics score (full score 20 points) Mainstream media exposure: 3 times by central media (5 points), 8 times by provincial media (4 points), total 9 points; Public opinion health index: positive rate 85% (≥80%), 3 points; Brand search index: annual growth of 25% (≥20%), 2 points; Strategic cooperation growth rate: an average of 4 new partners per year, 10 points; Media dynamics total score: 9 + 3 + 2 + 10 = 24 points.

[0177] 5. Risk deduction items No major legal, financial, and reputation risk events, no deduction.

[0178] 6. Comprehensive score calculation Weighted formula: P 综合 = 0.20x16 + 0.25x17 + 0.20x19 + 0.20x24 = 19.35 points (full score 20 points, converted to percentage system as 96.75 points).

[0179] III. Enterprise activity index calculation (quarterly) Investment activity: B-round financing of 180 million yuan (5 points), with a national industrial fund as an investor (+2 points), for a total of 7 points; Intellectual property output: 5 new invention patents (25 points), 3 software copyrights (6 points), for a total of 31 points; Qualifications and compliance: high-tech enterprise (+5 points), no violations, for a total of 5 points; Products and markets: 2 new products launched (10 points), 3 project wins (15 points), for a total of 25 points; Brand and communication: 3 reports by central media (15 points), 85% positive public opinion (10 points), for a total of 25 points; Total activity score = 7 x 0.25 + 31 x 0.20 + 5 x 0.15 + 25 x 0.15 + 25 x 0.25 = 20.05 points (25 points for a full score, 80.2% in the converted percentage system).

[0180] IV. Risk control index calculation Debt-to-equity ratio 35% (threshold for technology companies <40%), no points deducted; No safety accidents, administrative penalties, or data breaches; ISO 27001 certification (+5 points); Risk control index = 100 + 5 = 105 points (capped at 100, taking 100 points).

[0181] V. Innovation capability index calculation Basic points: national high-tech enterprise (20 points); Additional points: R&D investment ratio 22% (+20 points); 28 invention patents (+140 points); 4 PCT patents (+40 points); Participation in the development of 1 national standard (+10 points); 2 national industry-university-research projects (+10 points); Innovation capability total score = 20 + 20 + 140 + 40 + 10 + 10 = 240 points; Standardized score = (240 / 300) x 100 = 80 points.

[0182] VI. Comprehensive rating and decision support Comprehensive score = 0.30 x 96.75 + 0.30 x 80.2 + 0.20 x 100 + 0.20 x 80 = 89.59 points; Rating result: A level (80-89 points), meeting the characteristics of "high-growth quality enterprises"; Direct grading clause check: Although not fully meeting the S-level conditions, the overall performance is excellent, maintaining A-level; Policy matching: Automatically configure "R&D expense add-back deduction" "Intellectual property pledge financing" and other policy resources; Dynamic monitoring: The system updates growth data every month, and updates active data every quarter, and responds to risk events in real time.

[0183] Technical effect: The embodiment realizes precise quantitative evaluation of enterprise multi-dimensional capabilities through structured algorithm model and dynamic weight adaptation, shortens the evaluation period, reduces the misjudgment rate, and provides efficient and reliable decision-making basis for governments and investment institutions.

[0184] Further optionally, in other embodiments, the algorithm processing, design indicators and weights can be set as follows: 1) Organizational dynamics (20 points) 2) Capital dynamics (25 points) 3) Project dynamics (20 points) 4) Media dynamics (20 points) 5) Securities trading dynamics (15 points) (only for listed companies) 6) Risk deduction items 7) Rating level S level (90+): balanced development, sustained high growth A level (80-89): outstanding core indicators, clear growth trend B level (65-79): has growth potential, has individual short board C level (50-64): insufficient growth momentum, needs strategic adjustment D level (<50): there are major risk hidden dangers 8) Direct grading clause (satisfy any one can be directly graded) The compound growth rate of revenue for three consecutive years is greater than or equal to 50% Obtain national industrial fund ≥100 million yuan strategic investment Dominant development of international / national industry standards Annual mainstream media positive reports ≥100 times (2) Construction of enterprise activity index 1) Index design principles Comprehensiveness: Covers the entire life cycle of enterprise operations, including capital operation, innovation research and development, market expansion, and brand communication.

[0185] Dynamic: Combines real-time data (such as news events and financing dynamics) with periodic data (such as updates to qualifications and patent authorizations).

[0186] Industry adaptability: Adjust weights according to different industry characteristics (such as focusing on intellectual property for technology companies and focusing on qualifications and production capacity for manufacturing industries).

[0187] Quantifiable: Achieve cross-enterprise and cross-industry comparability through standardized scoring models.

[0188] 2) Core evaluation dimensions and indicators Design quarterly activity index and annual activity index, which is generated based on quarterly cumulative values. Take the quarterly activity index as an example: ① Investment and financing activity (weight 25%) Indicators: Financing times (quarterly): Number of equity / debt financing.

[0189] Financing amount (annual): Single and cumulative financing size.

[0190] Investment quality: Participation of well-known institutions (such as Sequoia and Hillhouse).

[0191] Exit cases: Successful cases of IPO, merger and acquisition, or equity transfer.

[0192] Scoring criteria: Each investment and financing event adds 2 points; Each financing is graded by amount (e.g. 100 million or more each time 5 points, 10-100 million each time 3 points); Investment by national industrial fund or well-known institution (a range needs to be given) adds 2 points / time; ② Intellectual property output (weight 20%) Indicators: Patent quantity: Quarterly number of new invention patents, utility model patents, and design patents.

[0193] Trademark and software copyright: Quarterly number of newly registered trademarks and software copyrights.

[0194] International layout: Quarterly number of new PCT patent applications and overseas patent authorizations.

[0195] Scoring criteria: 5 points for each domestic invention patent, 2 points for each utility model, and 1 point for each design; 2 points for each new trademark and software copyright; Patent layout application 5 points, authorized 10 points; ③ Qualifications and compliance (weight 15%) Indicators: Industry certification: Quarterly increase in ISO series, specialized and new, high-tech enterprises, etc.

[0196] Government qualification: Quarterly increase in licenses, environmental impact approval, and safety production license.

[0197] Compliance records: Quarterly increase in administrative penalties and legal dispute resolution rate.

[0198] Scoring criteria: National qualification (such as ISO 9001) 5 points / item, provincial 3 points / item.

[0199] Government approval 2 points per item.

[0200] No major violations, 3 points deducted for each administrative penalty.

[0201] ④ Products and markets (weight 15%) Indicators: New product launch: Number of new products / services launched annually.

[0202] Project bid: Obtain project bid.

[0203] Scoring criteria: Each new product launch gets 5 points (must pass market verification).

[0204] Each project bid gets 5 points ⑤ Brand and communication (weight 25%) Indicators: Media exposure: Number of mainstream media reports (need to be defined) and social media topic volume.

[0205] Public opinion heat: Positive / negative public opinion ratio, keyword search index, can refer to WeChat search index.

[0206] Industry influence: Number of times participating in standard formulation and industry summit speeches.

[0207] Award: Obtain provincial, municipal, national, and international awards Scoring criteria: Central media report 1 time 5 points, provincial media 1 time 2 points.

[0208] Public opinion positive rate ≥80% 10 points, decrease by 10% 3 points.

[0209] Industry influence participating in standard formulation 5 points, industry summit speech 1 time 1 point Obtain municipal award 1 point, provincial 2 points, national awards 5 points, international awards 8 points 3) Scoring model and grade division ① Calculation formula Active score = Σ (single indicator score × weight) Example: A technology enterprise financing score 20 points (weight 25%), intellectual property 18 points (20%), then the total score = 20 × 0.25 + 18 × 0.20 +... ② Grade division (if the score is generally low, it can be standardized, such as 50 points + 100 points, then 40 points is equivalent to 80 points) ③ Score interval level characteristic description: 90-100 Leading level Industry benchmark, multi-dimensional indicators significantly ahead 75-89 Active level Strong core business, local field breakthrough 60-74 Growth level Basic indicators meet the standards, need to optimize short board 45-59 Observation level Active degree fluctuation, risk points exist <45 Silence level Business stagnation or exit market risk (3) Risk control index construction Basic points: 100 points 1) Financial risk control ability ① Asset-liability ratio: total liabilities / total assets × 100% (reflecting long-term debt repayment ability, threshold reference: manufacturing industry < 60%, each high 1 percentage point minus 1 point; Technology / service enterprises < 40%, each high 1 percentage point minus 1 point).

[0210] ② Gross profit margin volatility: standard deviation of gross profit margin in the past 3 years (volatility > 20% indicates that profits are unstable, each high 1 percentage point minus 1 point).

[0211] 2) Operational risk control ability ① Safety production Safety accident rate: number of major safety accidents in the past 3 years (such as personnel casualties, environmental protection penalties, data sources: Ministry of Emergency Management, Ministry of Ecology and Environment publicity, each occurrence minus 5 points).

[0212] ② Quality control Quality certification qualification: whether through ISO 9001, IATF 16949, etc. Quality management system certification (obtain certification plus 5 points, no certification minus 5 points).

[0213] 3) Compliance and legal risk control ability ① Compliance system construction Compliance Management System Certification: Whether it has passed the ISO 37301 Compliance Management System Certification (add 5 points if certified, minus 5 points if not certified).

[0214] ② Legal Risk Exposure Number of Litigation Cases: Number of major lawsuits in the past three years (minus 10 points for each occurrence).

[0215] Administrative Penalty Records: Number of penalties and amounts in the fields of taxation, environmental protection, and market supervision (minus 8 points for each occurrence).

[0216] ③ Data and Information Security Information Security Certification: Whether it has passed the ISO 27001 Information Security Management System Certification (Internet and technology companies should pay special attention) (add 5 points if certified, minus 5 points if not certified).

[0217] Data Breach History: Whether there have been major data security incidents (such as user information leaks, media reports, or corporate announcements) in the past three years (minus 10 points for each occurrence).

[0218] 4) Strategic Risk Control Capability (Qualitative) Equity Concentration: Sum of the holding ratios of the top three shareholders (> 70% may indicate a risk of control by a major shareholder, minus 1 point for each additional percentage).

[0219] 5) Scoring Rules: Risk Control Index = 100 - Sum of Individual Scores Risk Level Classification: (4) Innovation Capability Index Construction Basic Score: Provincial High-tech Enterprise: 10 points National High-tech Enterprise: 20 points Technology-based Small and Medium-sized Enterprise: 15 points Specialized, Sophisticated, and New Small and Medium-sized Enterprise: 15 points National Specialized, Sophisticated, and New "Small Giant" Enterprise: 30 points National Technology Innovation Demonstration Enterprise: 30 points Manufacturing Single Champion Enterprise: 50 points Unicorn Enterprise: 100 points Others: 0 points Additional Points: 1) R&D Investment Capability ① R&D Funding R&D expenditure ratio (core indicator): R&D expenditure / operating income x 100%, reflecting the enterprise's resource tilt for innovation (20% or more + 20 points; 10% or more + 10 points; 5-10% + 5 points; 5% or less + 1 point) 2) Technological innovation achievements (quantitative + qualitative) ① Intellectual property output Patent quantity: I-class intellectual property (add 5 points for each additional one), II-class intellectual property (add 1 point for each additional one), PCT international patent quantity (add 10 points for each additional one) (I-class intellectual property includes invention patents, new plant varieties, national crop varieties, national new drugs, national first-class traditional Chinese medicine protection varieties, and integrated circuit layout design exclusive rights, etc.; II-class intellectual property includes utility model patents, design patents, and software copyrights, etc.) Patent quality: number of citations for all patents (10,000 times or more + 10 points, 1,000-10,000 times + 5 points, 1,000 times or less + 2 points).

[0220] Other achievements: national / industry-level science and technology awards (such as technology invention awards) (add 10 points for each additional one), provincial awards (add 5 points for each additional one).

[0221] ② Technological leadership Paper / standard contribution: published papers in top journals (such as Nature sub-journals) (add 10 points for each additional one), number of international / national standards developed (add 10 points for each additional one), number of industry / local standards developed (add 5 points for each additional one).

[0222] 3) Innovation resource integration capability (industrial chain synergy) ① Industry-university-research cooperation Number of cooperative institutions: number of cooperation projects with universities, research institutes, and industry leaders (add 3 points for each additional one).

[0223] Technology contract transaction amount: annual technology procurement / transfer amount (add 10 points for more than 1 billion yuan, 5 points for 50-100 million yuan, and 1 point for less than 50 million yuan), reflecting the active degree of innovation resource flow.

[0224] ② Ecological construction capability Innovation platform qualification: whether it has a national / provincial enterprise technology center, key laboratory (add 10 points for each additional national one, and 5 points for each additional provincial one).

[0225] 4) Innovation environment and organizational capability ① Innovation system guarantee R&D management system: whether it has passed the ISO 56005 innovation management system certification (10 points for passing, and no points for not passing) ② Anti-risk ability Intellectual property risk: patent litigation history (each more than one minus 10 points), invalid patent (each more than one minus 5 points).

[0226] 5) Calculation rules: First, calculate the final innovation ability score of each enterprise = benchmark score + additional points After standardization, the score = (innovation ability score / highest score)*100 6. Grade division rules: Excellent (S≥90) High technical barrier Good (80≤S<90) Technology leading Medium (70≤S<80) Technology medium Qualified (60≤S<70) Technology follow Unqualified (S<60) Technology backward Figure 5 is a block diagram of an apparatus 800 for dynamic evaluation of enterprise comprehensive ability according to an exemplary embodiment. For example, the apparatus 800 can be a mobile phone, a computer, a digital broadcast terminal, a messaging device, a game console, a tablet device, a medical device, a fitness device, a personal digital assistant, etc.

[0227] Referring to Figure 5 , the apparatus 800 can include one or more of the following components: a processing component 802, a memory 804, a power supply component 806, a multimedia component 808, an audio component 810, an input / output (I / O) interface 812, a sensor component 814, and a communication component 816.

[0228] The audio component 810 is configured to output and / or input audio signals. For example, the audio component 810 includes a microphone (MIC) that is configured to receive an external audio signal when the apparatus 800 is in an operation mode, such as a calling mode, a recording mode, and a voice recognition mode. The received audio signal can be further stored in the memory 804 or transmitted via the communication component 816. In some embodiments, the audio component 810 also includes a speaker for outputting an audio signal.

[0229] The I / O interface 812 provides an interface between the processing component 802 and peripheral interface modules, which can be a keyboard, a click wheel, a button, etc. These buttons can include, but are not limited to, a home button, a volume button, a start button, and a lock button.

[0230] The sensor component 814 includes one or more sensors for providing status assessments for various aspects of the device 800. For example, the sensor component 814 can detect an open / closed position of the device 800, relative positioning of components, such as a display and keypad of the device 800, a change in position of the device 800 or a component of the device 800, presence or absence of user contact with the device 800, orientation or acceleration / deceleration of the device 800, and temperature changes of the device 800. The sensor component 814 can include proximity sensor(s) configured to detect presence of nearby objects without any physical contact. The sensor component 814 can further include a light sensor, such as a CMOS or CCD image sensor, for use in imaging applications. In some embodiments, the sensor component 814 can further include an acceleration sensor, a gyroscope sensor, a magnetic sensor, a pressure sensor, or a temperature sensor.

[0231] The communication component 816 is configured to facilitate wired or wireless communication between the device 800 and another device. The device 800 can access a wireless network based on a corresponding communication standard, such as WiFi, 2G, or 3G, or a combination thereof. In an example embodiment, the communication component 816 receives a broadcast signal or broadcast-related information from an external broadcast management system via a broadcast channel. In an example embodiment, the communication component 816 further includes a Near Field Communication (NFC) module to facilitate short-range communication. For example, the NFC module can be implemented based on Radio Frequency Identification (RFID) technology, infrared data association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology, and other technologies.

[0232] In an example embodiment, the device 800 can be implemented using one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, micro-controllers, microprocessors, or other electronic components, for performing the above-described methods.

[0233] In an example embodiment, a non-transitory computer-readable storage medium, such as the memory 804 including instructions, is also provided, which can be executed by the processor 820 of the device 800 to perform the above-described methods. For example, the non-transitory computer-readable storage medium can be a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disc, and an optical data storage device, and the like.

[0234] It should be understood that in the foregoing embodiments, the terminal and / or network device can perform some or all of the steps in the embodiments. These steps or operations are merely examples, and the embodiments of the present application can also perform other operations or variations of various operations. In addition, various steps can be performed in different orders presented in the embodiments, and it is possible that not all operations in the embodiments of the present application are to be performed. Moreover, the magnitude of the serial number of each step does not mean the order of execution, and the execution order of each process should be determined according to its function and inherent logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

Claims

1. A dynamic evaluation method of integrated capability of an enterprise, characterized by, The application comprises the following steps: S1. Collecting multi-source heterogeneous enterprise data, cleaning and standardizing the multi-source heterogeneous enterprise data, and generating a structured data set; S2. Based on the structured data set, calculating dynamic scores by using preset growth evaluation model, activity evaluation model, risk control model and innovation ability model respectively, and generating multi-dimensional quantitative parameters; S3. Integrating the dynamic scores in the multi-dimensional quantitative parameters, and mapping them into comprehensive rating results according to grading rules; S4. Performing enterprise classification and grading operation according to the comprehensive rating results, and outputting a grading management list.

2. The method of claim 1, wherein, The S1 comprises the following steps: S1a. Collecting enterprise financial data, patent data, public opinion data and investment and financing data through crawler programs, API interfaces and data subscription channels in parallel, and forming an initial heterogeneous data set; S1b. Storing the structured data in the initial heterogeneous data set into a relational database, importing semi-structured and unstructured data into a distributed file storage system, and constructing a hierarchical storage architecture; S1c. De-duplicating the data in the hierarchical storage architecture, interpolating and filling the missing fields, and unifying the units and formats of numerical fields, and outputting the structured data set.

3. The method of claim 1, wherein, The S2 comprises the following steps: S2a. Processing the structured data set by using a preset growth evaluation model, calculating enterprise growth scores based on organizational dynamics, capital dynamics and project dynamics indicators on a monthly basis, and generating dynamic growth quantitative parameters; S2b. Processing the structured data set by using a preset activity evaluation model, calculating enterprise activity scores based on investment and financing activity, intellectual property output and brand communication indicators on a quarterly basis, and generating dynamic activity quantitative parameters; S2c. Integrating the dynamic growth quantitative parameters, the dynamic activity quantitative parameters, the risk control model output value and the innovation ability model output value, and normalizing them into the multi-dimensional quantitative parameters.

4. The method of claim 1, wherein, The S3 comprises the following steps: S3a. Risk checking the multi-dimensional quantitative parameters: when the risk control model output value is lower than a preset threshold, proportionally reducing the enterprise dynamic scores in the multi-dimensional quantitative parameters, and generating risk calibration parameters; S3b. Inputting the risk calibration parameters into a preset grading mapping rule library, dynamically adjusting the rating threshold interval based on industry characteristics, and outputting initial rating results; S3c. Trend correcting the initial rating results based on historical rating results: if the rating trend is rising for three consecutive periods, the level is increased, if it is continuously decreasing, risk review is triggered, and the comprehensive rating results are generated.

5. The method of claim 1, wherein, The S4 comprises the following steps: S4a. According to the comprehensive rating results, matching a preset resource allocation strategy library, automatically configuring differentiated policy support schemes for enterprises of different levels, and generating an initial resource allocation list; S4b. Based on a dynamic monitoring mechanism, periodically re-evaluating the enterprises in the grading management list: when a major risk event or an abnormal fluctuation of indicators is detected, real-time rating update is triggered, and the policy support scheme is adjusted synchronously; S4c. Bind the policy support scheme and dynamic reevaluation results to the spatial geographic information library, visualize the enterprise distribution heat and policy coverage intensity in the electronic map, and output the regional investment optimization atlas.

6. The method of claim 1, wherein, Also includes a pre-direct grading step: Before S2, detect whether the structured data set contains a national industrial fund investment record of ≥100 million yuan or an international standard leading formulation certificate; if so, skip steps S2-S3, directly generate an S-level comprehensive evaluation result, and write it to the grading management list.

7. The method of claim 2, wherein, In S1b, perform data heterogeneous hierarchical storage: Import structured data in the initial heterogeneous data set into a relational database storage, semi-structured data into a document database storage, and unstructured data into a distributed file system storage, and establish a unified metadata index to associate the three types of storage carriers.

8. The method of claim 3, wherein, The execution frequency of S2a and S2b is dynamically decoupled: S2a. Call the organizational structure change record, core team stability data, and short-term capital flow indicators in the structured data set monthly, calculate the enterprise growth score, and update the dynamic growth quantitative parameter; S2b. Call the investment and financing event data, intellectual property output, and brand communication volume indicators in the structured data set quarterly, calculate the enterprise activity score, and update the dynamic activity quantitative parameter.

9. The method of claim 3, wherein, In S2b, perform industry weight dynamic allocation: According to the national economic industry classification code, retrieve the preset weight configuration rule, increase the production capacity qualification indicator weight of manufacturing enterprises to 25%-35%, increase the intellectual property indicator weight of technology enterprises to 30%-40%, and simultaneously reduce the proportion of non-core dimension weight.

10. The method of claim 3, wherein, In the innovation capability model, perform innovation energy efficiency conversion evaluation: Extract the technology transaction contract registration information in the structured data set, add 10 points to enterprises with annual technology procurement or transfer amount ≥100 million yuan, and add 5 points / 3 points to state-level / provincial-level research cooperation projects, respectively, and superimposed to the innovation capability basic score.

11. The method of claim 4, wherein, In S3a: When a major legal risk event or continuous two years of financial loss is detected, ignore the scores of other dimensions in the multi-dimensional quantitative parameter, directly lower the comprehensive evaluation result to D level, and freeze the policy support resource allocation.

12. An electronic device for a dynamic assessment method of enterprise-wide capabilities, comprising: Display; One or more processors; Memory; And one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including instructions for executing any of the methods according to claims 1-11.

Images