A dynamic enterprise portrait generation method based on multi-source heterogeneous data

Abstract

The present invention relates to a method for generating a dynamic enterprise portrait based on multi-source heterogeneous data, which relates to the field of data processing. The method comprises obtaining a business process topology diagram of a target enterprise, wherein the business process topology diagram includes multiple business nodes and the node relationships between the business nodes; collecting node association data of the business nodes in real time to generate a current state portrait of the target enterprise; obtaining the pending business operations of the business nodes to form a business operation sequence; dynamically generating an enterprise state portrait based on the current state portrait and in combination with the business operation sequence to determine a predicted enterprise state portrait sequence; determining risk sequence points, risk business nodes and risk enterprise state portraits according to the predicted enterprise state portrait sequence, and displaying them in a management cockpit, thereby solving the technical problems of being unable to dynamically predict the enterprise state and being difficult to adapt to complex business processes for strategic management and operational monitoring.

Description

Technical Field

[0001] The present invention relates to the field of data processing, and in particular to a method for generating dynamic enterprise portraits based on multi-source heterogeneous data. Background Art

[0002] In today's highly competitive business environment, enterprises face complex and volatile market conditions and internal operational challenges. To better address these challenges, companies need timely and accurate understanding of their operational status and market environment to make informed decisions. However, traditional enterprise data management and analysis methods often suffer from data silos, data inconsistencies, and inefficient data processing, making them unable to meet enterprises' needs for dynamic, comprehensive, and accurate business portraits.

[0003] With the development of data integration and processing technologies, data-driven intelligent enterprise portrait generation methods have emerged. These methods integrate multi-source, heterogeneous data from different systems and formats to construct a comprehensive enterprise portrait, thereby helping enterprises better understand their own operating status and market environment. However, existing enterprise portrait generation methods mostly remain at a static level, that is, they can only reflect the state of the enterprise at a certain moment and cannot predict the enterprise's future operating status and risk points. In addition, as enterprises expand in size and business scope, their business processes become increasingly complex. Existing enterprise portrait generation methods often struggle to adapt to this complexity and cannot accurately reflect the relationships between the various business nodes of an enterprise and their impact on the enterprise's overall operating status. Therefore, a method that can dynamically predict the enterprise status is needed to help enterprises better cope with the complex and changing business environment. Summary of the Invention

[0004] The present invention aims to solve the technical problems in the existing technology that the enterprise status cannot be dynamically predicted and it is difficult to adapt to complex business processes for strategic management and operation monitoring, and provides a dynamic enterprise portrait generation method based on multi-source heterogeneous data to solve the problem.

[0005] The technical solution of the present invention to solve the above technical problems is as follows:

[0006] The present invention provides a method for generating a dynamic enterprise portrait based on multi-source heterogeneous data, the method comprising: obtaining a business process topology diagram of a target enterprise, the business process topology diagram comprising a plurality of business nodes and node relationships between the business nodes; collecting node association data of each of the business nodes in real time to generate a current state portrait of the target enterprise; obtaining pending business operations of each of the business nodes to form a business operation sequence; dynamically generating an enterprise state portrait based on the current state portrait and in combination with the business operation sequence to determine a predicted enterprise state portrait sequence; determining risk sequence points, risk business nodes, and risk enterprise state portraits based on the predicted enterprise state portrait sequence, and displaying them in a management cockpit;

[0007] The node association data of each business node is collected in real time to generate a current status portrait of the target enterprise, including:

[0008] Performing data collection configuration on each of the service nodes, and determining the data source and data type of each of the service nodes;

[0009] Extracting real-time data related to each of the business nodes from a multi-source heterogeneous system based on the data source and data type of each of the business nodes, the multi-source heterogeneous system including an enterprise resource planning system, a financial management system, a supply chain management system, and a customer relationship management system;

[0010] Clean, convert and standardize the real-time data of each business node to form node-related data in a unified format;

[0011] Generating a current status portrait of the target enterprise based on the node association data includes:

[0012] Acquire a multi-dimensional indicator system, wherein the multi-dimensional indicator system includes multiple evaluation dimensions set by a user terminal;

[0013] Determining core indicator values of multiple evaluation dimensions respectively according to the node association data;

[0014] Based on the core indicator values of each evaluation dimension, a multidimensional feature vector representing the current status of the enterprise is constructed;

[0015] The multi-dimensional feature vector is mapped into a current state portrait.

[0016] In one embodiment, the multiple evaluation dimensions include at least a financial dimension, an operational dimension, and a market dimension.

[0017] In one embodiment, based on the current state portrait and in combination with the business operation time sequence, the enterprise state portrait is dynamically generated to determine the enterprise state portrait prediction time sequence, including:

[0018] Constructing a target enterprise status prediction model, wherein the target enterprise status prediction model is used to predict the enterprise status profile after the business operation is executed;

[0019] Based on the business operation sequence, obtaining a first business operation to be executed and a corresponding first business node, wherein the business operation sequence includes N business operations to be executed, and each of the business operations to be executed has a corresponding business node;

[0020] Based on the current state profile, the first to-be-performed business operation, and the first business node, predicting the enterprise state profile using the target enterprise state prediction model to obtain a first predicted enterprise state profile;

[0021] Continuing to obtain a second to-be-executed business operation and a corresponding second business node based on the business operation time sequence, and obtaining a second predicted enterprise status portrait based on the first predicted enterprise status portrait, the second to-be-executed business operation, and the second business node;

[0022] Iterate and execute until the Nth predicted enterprise status portrait is obtained;

[0023] The first predicted enterprise status portrait, the second predicted enterprise status portrait, and the Nth predicted enterprise status portrait are arranged and stored in time sequence to obtain the enterprise status portrait prediction time sequence.

[0024] In one embodiment, building a target enterprise status prediction model includes:

[0025] Uploading the business process topology diagram to an enterprise management data sharing platform, and receiving a plurality of matching enterprise status prediction models fed back by the enterprise management data sharing platform, each matching enterprise status prediction model having a matching degree identifier;

[0026] Performing integrated fitting on multiple matching enterprise status prediction models according to the matching degree identifier to obtain an initial enterprise status prediction model;

[0027] Obtain historical business operation execution data of the target enterprise and generate multiple enterprise historical portrait samples, each enterprise historical portrait sample including a sample initial state portrait, a sample pending business operation, a sample business node, and a sample result state portrait;

[0028] Based on the multiple enterprise historical portrait samples, the initial enterprise status prediction model is trained and optimized to obtain the target enterprise status prediction model.

[0029] In one embodiment, determining risk time sequence points, risk business nodes, and risky enterprise status profiles based on the enterprise status profile prediction time sequence includes:

[0030] Acquire multiple evaluation dimension thresholds, where the multiple evaluation dimension thresholds are set by a user terminal and correspond one-to-one to the multiple evaluation dimensions;

[0031] Based on the multiple evaluation dimension thresholds, a risk judgment is made on the enterprise status portrait prediction time series to determine the risk time series points, risk business nodes and risk enterprise status portraits.

[0032] In one embodiment, the enterprise status portrait prediction time series includes N predicted enterprise status portraits; based on the multiple evaluation dimension thresholds, risk judgment is performed on the enterprise status portrait prediction time series to determine risk time series points, risk business nodes, and risk enterprise status portraits, including:

[0033] Traversing N predicted enterprise status portraits in the enterprise status portrait prediction time series to obtain a first predicted enterprise status portrait;

[0034] Extracting the core indicator value of each evaluation dimension in the first predicted enterprise status portrait, and comparing the core indicator value of each evaluation dimension with the corresponding evaluation dimension threshold;

[0035] When the core indicator value of any evaluation dimension exceeds the corresponding evaluation dimension threshold, the time position corresponding to the first predicted enterprise status portrait is determined as the risk time point, the business node corresponding to the to-be-executed business operation that causes the risk time point to appear is determined as the risk business node, and the first predicted enterprise status portrait is determined as the risk enterprise status portrait.

[0036] The beneficial effects of the present invention are: by obtaining the business process topology diagram of the target enterprise, collecting business node data in real time to generate a current status portrait, and combining it with the business operations to be executed for dynamic prediction, the future status portrait and potential risks of the enterprise can be determined, and finally displayed in the management cockpit, which can comprehensively and dynamically reflect the operating status of the enterprise, effectively identify risk points, and provide strong support for the enterprise's strategic management and operation monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 A flowchart of a method for generating a dynamic enterprise portrait based on multi-source heterogeneous data provided by the present invention.

[0038] Figure 2 A schematic diagram of the process of determining the timing of enterprise status portrait prediction in a dynamic enterprise portrait generation method based on multi-source heterogeneous data provided by the present invention. DETAILED DESCRIPTION

[0039] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making any creative efforts shall fall within the scope of protection of the present invention.

[0040] In the description of the present invention, the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of the technical features indicated. Therefore, a feature specified as "first" or "second" may explicitly or implicitly include one or more of the specified features. In the description of the present invention, "plurality" means two or more, unless otherwise specifically defined.

[0041] In the description of the present invention, the term "for example" is used to mean "used as an example, illustration or illustration". Any embodiment of the present invention described as "for example" is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is given to enable any person skilled in the art to implement and use the present invention. In the following description, details are listed for the purpose of explanation. It should be understood that a person of ordinary skill in the art can recognize that the present invention can be implemented without using these specific details. In other examples, well-known structures and processes are not elaborated in detail to avoid obscuring the description of the present invention with unnecessary details. Therefore, the present invention is not intended to be limited to the embodiments shown, but is consistent with the widest scope consistent with the principles and features disclosed herein. Example

[0042] like Figure 1 As shown, an embodiment of the present invention provides a method for generating a dynamic enterprise portrait based on multi-source heterogeneous data, the method comprising:

[0043] S10: Obtain a business process topology diagram of the target enterprise, where the business process topology diagram includes multiple business nodes and node relationships between the business nodes.

[0044] For example, we first obtain the target enterprise's business process topology map, which serves as the cornerstone for building a business profile. This map details the multiple business nodes within the enterprise, representing various aspects of its operations, such as procurement, production, and sales. It also clearly illustrates the relationships between these nodes, specifically how they connect and influence each other. These relationships reflect the logic and processes of the enterprise's operations and are key to understanding its operational status.

[0045] For example, a manufacturing company may include multiple business nodes, including raw material procurement, production and processing, and finished product sales. In a business process topology diagram, these business nodes are clearly labeled, and the relationships between them, such as logistics, information flow, and capital flow, are also displayed. The topology diagram allows for a visual representation of how raw materials flow from procurement to production, then through processing into finished products, and ultimately to the market through sales.

[0046] Acquiring a business process topology map not only helps comprehensively understand the operational status of an enterprise but also provides a foundation for subsequent data collection, profile generation, and risk prediction. Only by accurately capturing an enterprise's business nodes and the relationships between them can targeted data be collected, leading to the generation of an accurate enterprise profile and the prediction of potential risk points. Therefore, acquiring the target enterprise's business process topology map is the first and crucial step in the entire dynamic enterprise profile generation method.

[0047] S20: Collect node association data of each of the business nodes in real time to generate a current status portrait of the target enterprise.

[0048] Furthermore, real-time collection of node-related data from each business node is a key step in generating a portrait of the target enterprise's current state. This involves real-time capture and processing of the massive amounts of data generated during the enterprise's operations. This data comes from various business nodes of the enterprise, such as production, sales, and finance, and is closely related and mutually influential. Real-time collection emphasizes the timeliness and dynamism of data, meaning the need to capture the latest developments in the enterprise's operations at all times to ensure that the generated enterprise portrait truly reflects the enterprise's current state. Node-related data refers to various types of data closely related to each business node. These data may include production data, sales data, financial data, and so on. These data are interconnected through business logic and together form a complete picture of the enterprise's operations.

[0049] For example, consider a retail business operating underway. Data from its sales department will reflect real-time sales activity, such as sales volume, sales revenue, and inventory levels. Simultaneously, data from its finance department will also be updated in real time, reflecting the company's cash flow and profit status. These data components are all crucial components of node-linked data, interconnected through business logic and collectively revealing the company's operational status. By collecting this data in real time and then cleaning, converting, and standardizing it, we can generate a corporate profile that accurately reflects the company's current status, providing strong support for strategic management and operational monitoring.

[0050] S30: Acquire the to-be-executed service operations of each service node to form a service operation sequence.

[0051] Subsequently, obtaining pending business operations for each business node aims to comprehensively analyze the company's future planned business operations, providing a foundation for subsequent dynamic forecasting and risk assessment. These pending business operations refer to the company's upcoming or planned business activities, distributed across various business nodes, such as production plans, sales plans, and procurement plans. These operations not only reflect the company's operational strategy and market demand but also directly impact its operational status and future development.

[0052] In order to form a complete business operation sequence, it is necessary to traverse each business node of the enterprise, collect and record the business operations to be executed on each node. These operations will be sorted according to their logical order and time schedule in the enterprise operation, thus forming an orderly business operation sequence.

[0053] For example, a manufacturing company might develop a production plan for the coming week, including specific production tasks for each production line, raw material procurement plans, and product sales plans. These plans are broken down into specific business operations to be executed and sorted according to the timelines of the production process. By capturing this operational information, a clear and organized timeline of business operations can be formed, providing strong support for the subsequent dynamic generation of a company status profile. Through this step, companies can gain a clearer understanding of their future operational plans, enabling them to make more informed decisions, optimize resource allocation, and improve operational efficiency.

[0054] S40: Based on the current status portrait, the enterprise status portrait is dynamically generated in combination with the business operation sequence, and the enterprise status portrait prediction sequence is determined.

[0055] Optionally, dynamic generation of an enterprise state profile based on the current state profile, combined with a business operation time series, aims to dynamically display the enterprise's future operational status by predicting the impact of future business operations. The current state profile represents the enterprise's current operational status and provides a comprehensive, accurate, and real-time reflection. The business operation time series displays the enterprise's planned future business operations and their schedules. The combination of these two elements simulates the enterprise's future operational processes, generating a series of enterprise state profiles, known as the enterprise state profile prediction time series.

[0056] Specifically, starting with the current state profile, the company's future operational state is gradually simulated based on the pending business operations listed in the business operation timeline. Each time a business operation is executed, the company's state profile is updated, reflecting the impact of that operation on the company's operational state. This process continues until all operations in the business operation timeline have been simulated, forming a complete enterprise state profile prediction timeline.

[0057] For example, a manufacturing company might use its current production and sales plans to forecast production and sales for the next few months. First, a current state profile is generated based on current production and sales data. Then, based on future production plans (such as adding production lines or adjusting production plans) and sales plans (such as expanding into new markets or hosting promotional events) within the business operation timeline, the company gradually simulates the impact of these operations on the company's operations, generating a series of predicted business state profiles. These profiles are arranged chronologically to form a clear timeline of business state profile forecasts, helping companies better understand future operational trends and make more informed decisions.

[0058] S50: Determine risk time sequence points, risk business nodes and risk enterprise status portraits based on the enterprise status portrait prediction time sequence, and display them in the management cockpit.

[0059] Specifically, the ultimate goal and value lies in determining risk time series points, risky business nodes, and risky enterprise status profiles based on the enterprise status profile forecast time series, and visually displaying this information in the management cockpit. The management cockpit is used for strategic management and operational monitoring of the enterprise, with the primary purpose of centrally displaying key data and indicators. The enterprise status profile forecast time series represents a series of chronologically arranged future enterprise status profiles. These are generated by simulating the enterprise's future business operations and can comprehensively reflect the enterprise's future operational status and development trends. Risk time series points, risky business nodes, and risky enterprise status profiles are potential risk points identified in these forecast profiles, and they are of great significance to the enterprise's strategic management and operational monitoring.

[0060] Specifically, the system traverses each predicted profile in the enterprise status profile prediction time series and performs a risk assessment on each profile based on preset evaluation dimensions and thresholds. When the core indicator value of a certain evaluation dimension in a profile exceeds the corresponding threshold, the corresponding time series position of the profile is identified as a risky time series point. The business node corresponding to the pending business operation that caused the risk is identified as a risky business node, and the profile is then designated as a risky enterprise status profile.

[0061] For example, a retail company might use dynamic enterprise profile generation to forecast sales for the coming months. The forecast results might reveal that the inventory turnover rate, a core metric in the sales forecast profile for a particular month, falls significantly below a preset threshold. In this case, that month is designated as a risk time series point. Business nodes corresponding to related operations that led to a decrease in inventory turnover (such as inappropriate procurement plans or untimely adjustments to sales strategies) are identified as risky business nodes, and the sales forecast profile for that month is designated as a risky enterprise status profile. This information is intuitively displayed in the management cockpit for decision makers to reference, enabling them to take timely action, adjust operational strategies, and mitigate potential risks.

[0062] In a preferred embodiment, node-related data of each of the business nodes is collected in real time to generate a current status portrait of the target enterprise, including: configuring data collection for each of the business nodes to determine the collection data source and data type of each of the business nodes; extracting real-time data related to each of the business nodes from multi-source heterogeneous systems based on the collection data source and data type of each of the business nodes, the multi-source heterogeneous systems including an enterprise resource planning system, a financial management system, a supply chain management system, and a customer relationship management system; cleaning, converting, and standardizing the real-time data of each of the business nodes to form node-related data in a unified format; and generating a current status portrait of the target enterprise based on the node-related data.

[0063] Optimally, real-time collection of node-related data from each business node to generate a profile of the target enterprise's current status is a key step. This process first requires configuring data collection for each business node, clarifying the data source and data type for each node to ensure data accuracy and relevance. For example, a sales business node might be configured to collect customer order data from a customer relationship management system (CRM), with data types including order amounts and customer information.

[0064] Next, based on the configured data sources and data types, real-time data related to each business node is extracted from multiple heterogeneous systems. These systems include enterprise resource planning (ERP), financial management systems, and supply chain management systems, each storing data from different aspects of the enterprise. For example, production progress data can be extracted from the ERP system, while cash flow data can be obtained from the financial management system.

[0065] However, extracted real-time data often suffers from inconsistent formats and varying quality. Therefore, it requires cleaning, conversion, and standardization to eliminate noise and errors and convert the data into a unified format for subsequent analysis and processing. This processed data forms node-related data in a unified format that accurately reflects the relationships between business nodes.

[0066] Finally, based on this node-related data, a current status profile of the target enterprise can be generated. This profile provides a comprehensive, accurate, and real-time reflection of the enterprise's current operational status, including the status of each business node and the impact between them. For example, through the current status profile, an enterprise can clearly see the order status of the sales business node, the production progress of the production business node, and the funding status of the finance business node, providing strong support for enterprise decision-making.

[0067] In a preferred embodiment, generating a current status profile of the target enterprise based on the node-related data includes: obtaining a multi-dimensional indicator system, the multi-dimensional indicator system including multiple evaluation dimensions set by a user; determining core indicator values for each of the multiple evaluation dimensions based on the node-related data; constructing a multi-dimensional feature vector representing the current status of the enterprise based on the core indicator values of each evaluation dimension; and mapping the multi-dimensional feature vector into a current status profile. The multiple evaluation dimensions include at least a financial dimension, an operational dimension, and a market dimension.

[0068] Specifically, the multi-dimensional indicator system obtained covers multiple evaluation dimensions set by the user side, such as financial dimension, operational dimension, market dimension and risk dimension, etc. These dimensions comprehensively reflect the operating status of the enterprise.

[0069] Based on the node-linked data, the core indicator values for each assessment dimension are determined. Specifically, for the financial dimension, the current ratio indicator is calculated, which is obtained by dividing the current assets data obtained from the financial system node by the current liabilities data to assess the company's short-term debt repayment ability. For the operational dimension, the inventory turnover rate indicator is calculated by dividing the cost of goods sold data obtained from the supply chain system node by the average inventory data to measure the efficiency of the company's inventory management. For the market dimension, the customer satisfaction indicator is calculated by taking a weighted average of the customer rating data obtained from the customer relationship management system node to understand the company's reputation in the market. For the risk dimension, the accounts receivable delinquency rate indicator is calculated by dividing the overdue accounts receivable data obtained from the financial system node by the total accounts receivable data to assess the company's capital recovery risk.

[0070] After determining the core indicator values for each evaluation dimension, a multidimensional feature vector representing the company's current state is constructed based on these values. This vector is a mathematical representation that integrates information from each dimension and can comprehensively and accurately reflect the company's current state. Finally, this multidimensional feature vector is mapped into a current state portrait, forming an intuitive and visual display of the company's operating status. For example, through the current state portrait, a company can clearly see its performance in terms of finance, operations, market, and risk, thereby promptly identifying potential problems and adjusting operational strategies. For example, if the current ratio in the financial dimension is low, the company may need to focus on improving its short-term debt repayment ability; if the inventory turnover rate in the operational dimension is low, the company may need to optimize its inventory management process; if customer satisfaction in the market dimension is low, the company needs to strengthen customer service quality, etc.

[0071] In a preferred embodiment, Figure 2 As shown, based on the current state portrait, the enterprise state portrait is dynamically generated in combination with the business operation sequence, and the enterprise state portrait prediction sequence is determined, including: constructing a target enterprise state prediction model, the target enterprise state prediction model is used to predict the enterprise state portrait after the business operation is executed; based on the business operation sequence, a first business operation to be executed and a corresponding first business node are obtained, wherein the business operation sequence includes N business operations to be executed, and each business operation to be executed has a corresponding business node; based on the current state portrait, the first business operation to be executed and the first business node, the enterprise state portrait is predicted by the target enterprise state prediction model to obtain a first predicted enterprise state portrait; continue to obtain a second business operation to be executed and a corresponding second business node based on the business operation sequence, and obtain a second predicted enterprise state portrait based on the first predicted enterprise state portrait, the second business to be executed and the second business node; iterative execution until the Nth predicted enterprise state portrait is obtained; the first predicted enterprise state portrait, the second predicted enterprise state portrait, and the second predicted enterprise state portrait are stored in chronological order until the Nth predicted enterprise state portrait is obtained.

[0072] Exemplarily, based on the current state portrait, dynamically generating the enterprise state portrait in combination with the business operation sequence and determining the prediction sequence involves building a target enterprise state prediction model, which can predict the future enterprise state portrait based on the current enterprise state and business operations.

[0073] Furthermore, based on the business operation sequence, pending business operations and their corresponding business nodes are gradually obtained. The business operation sequence contains N pending business operations, each of which is associated with a specific business node. For example, a manufacturing company's business operation sequence may include operations such as production plan adjustment, raw material procurement, and production line startup. Each operation corresponds to a different business node, such as production, procurement, and equipment management.

[0074] First, the first pending business operation and its corresponding first business node are obtained from the business operation time series. Then, based on the current state profile, the first pending business operation, and the first business node, the target enterprise state prediction model is used to predict the enterprise state profile, resulting in a first predicted enterprise state profile. This predicted profile reflects the expected state of the enterprise after executing the first pending business operation.

[0075] Subsequently, based on the business operation time sequence, the second pending business operation and its corresponding second business node are obtained. Based on the first predicted enterprise status profile, the second pending business operation, and the second business node, a second predicted enterprise status profile is obtained using the prediction model. This process continues iteratively, with each iteration generating a new predicted profile based on the previous predicted profile and the next pending business operation, until the Nth predicted enterprise status profile is obtained.

[0076] Finally, these predicted enterprise status profiles are stored in a time series, resulting in a predicted time series of enterprise status profiles. This predicted time series demonstrates the dynamic evolution of the enterprise's status profile over the coming period as various business operations are executed. For example, through this predicted time series, an enterprise can clearly see how its production status and inventory status change after executing business operations such as production plan adjustments and raw material procurement, providing strong support for decision-making.

[0077] In a preferred embodiment, a target enterprise status prediction model is constructed, including: uploading the business process topology diagram to an enterprise management data sharing platform, receiving multiple matching enterprise status prediction models fed back by the enterprise management data sharing platform, each matching enterprise status prediction model having a matching degree identifier; integrating and fitting the multiple matching enterprise status prediction models according to the matching degree identifier to obtain an initial enterprise status prediction model; obtaining the historical business operation execution data of the target enterprise, and producing multiple enterprise history portrait samples, each enterprise history portrait sample including a sample initial status portrait, a sample business operation to be executed, a sample business node and a sample result status portrait; based on the multiple enterprise history portrait samples, training and optimizing the initial enterprise status prediction model to obtain the target enterprise status prediction model.

[0078] Specifically, when building a target enterprise status prediction model, the business process topology diagram must be uploaded to the enterprise management data sharing platform. This step aims to leverage the platform's resources to obtain multiple enterprise status prediction models that match the target enterprise's business processes. These matching models are desensitized, containing only the model architecture and training parameters, ensuring that the original enterprise's business data is not leaked, thereby ensuring data security and privacy. Each matching model is also assigned a matching degree indicator to measure its degree of alignment with the target enterprise's business processes.

[0079] Next, based on the matching degree indicators, multiple matching enterprise status prediction models are integrated and fitted to form an initial enterprise status prediction model. The integrated fitting process aims to combine the strengths of each matching model to improve the accuracy and generalization of the prediction model. For example, if one matching model performs well in predicting status changes in the production stage, while another model is more accurate in predicting status changes in the sales stage, integrated fitting can combine the strengths of these two models to form a more comprehensive initial prediction model.

[0080] To further optimize the initial enterprise status prediction model, we need to obtain the target enterprise's historical business operation execution data and generate multiple enterprise historical profile samples based on this data. Each sample includes the sample's initial state profile, the sample pending business operation, the sample business node, and the sample result state profile. These historical profile samples provide rich data support for model training, enabling the model to learn the patterns of enterprise status changes as business operations change.

[0081] Finally, based on these historical enterprise profile samples, the initial enterprise status prediction model is trained and optimized. The training and optimization process continuously adjusts the model's parameters to enable it to more accurately predict changes in enterprise status. After training and optimization, the resulting target enterprise status prediction model will be able to more accurately predict the enterprise status profile after future business operations are executed, providing strong support for the company's strategic management and operational monitoring. For example, in retail companies, this model can predict the impact of future sales activities on inventory, cash flow, and other aspects of the company's status, helping the company make more informed decisions.

[0082] In a preferred embodiment, risk time series points, risk business nodes and risk enterprise status portraits are determined based on the enterprise status portrait prediction time series, including: obtaining multiple evaluation dimension thresholds, multiple evaluation dimension thresholds are set by the user end, and correspond one-to-one to the multiple evaluation dimensions; based on the multiple evaluation dimension thresholds, risk judgment is made on the enterprise status portrait prediction time series to determine risk time series points, risk business nodes and risk enterprise status portraits.

[0083] Furthermore, multiple assessment dimension thresholds are obtained. These thresholds are set by the user based on the company's actual situation and risk appetite, and correspond one-to-one with the multiple assessment dimensions mentioned above (such as financial, operational, and market dimensions). For example, a user might set the current ratio threshold in the financial dimension to 1.5, meaning that when the current ratio falls below this value, the company is considered to have financial risk.

[0084] Next, based on these assessment dimension thresholds, a risk assessment is performed on the enterprise status profile forecast time series. Specifically, each enterprise status profile in the forecast time series is traversed, and the core indicator values for each assessment dimension are checked to see if they exceed the corresponding thresholds. If the core indicator value for a particular assessment dimension in a profile exceeds the threshold, the time point corresponding to that profile is identified as a risky time series point. The business node corresponding to the pending business operation that caused the risk is identified as a risky business node, and the profile is then designated as a risky enterprise status profile.

[0085] For example, in a manufacturing enterprise's forecast time series, if the core indicator value of inventory turnover, an assessment dimension, in the enterprise status profile at a certain point in time is found to be significantly lower than the set threshold, this point in time will be identified as a risk time series point. The business nodes corresponding to the relevant business operations that led to the decline in inventory turnover (such as unreasonable procurement plans and untimely production plan adjustments) will be identified as risky business nodes, and the enterprise status profile at that point in time will be designated as a risky enterprise status profile. This allows the enterprise to identify potential risk points in advance and take appropriate measures to reduce risks, ensuring stable operations.

[0086] In a preferred embodiment, the enterprise status portrait prediction time series includes N predicted enterprise status portraits; based on the multiple evaluation dimension thresholds, the enterprise status portrait prediction time series is subjected to risk judgment, and risk time series points, risk business nodes and risk enterprise status portraits are determined, including: traversing the N predicted enterprise status portraits in the enterprise status portrait prediction time series to obtain a first predicted enterprise status portrait; extracting the core indicator values of each evaluation dimension in the first predicted enterprise status portrait, and comparing the core indicator values of each evaluation dimension with the corresponding evaluation dimension threshold; when the core indicator value of any evaluation dimension exceeds the corresponding evaluation dimension threshold, the time series position corresponding to the first predicted enterprise status portrait is determined as a risk time series point, the business node corresponding to the to-be-executed business operation that causes the risk time series point to appear is determined as a risk business node, and the first predicted enterprise status portrait is determined as a risk enterprise status portrait.

[0087] Specifically, during the risk assessment process for predictive enterprise status profiles, a sequence of N predicted enterprise status profiles is presented. This sequence is dynamically generated based on multi-source heterogeneous data combined with the enterprise's business processes and operational time series, aiming to reflect the enterprise's operational status at different points in the future. To identify potential risks, these predicted profiles are individually reviewed based on multiple assessment thresholds set by the user.

[0088] Specifically, starting with the first predicted enterprise status profile, the core indicator values for each evaluation dimension are extracted. These evaluation dimensions may include finance, operations, marketing, and other aspects. Each dimension has specific core indicators, such as current ratio, inventory turnover rate, and customer satisfaction. Next, these core indicator values are compared with the corresponding evaluation dimension thresholds. Thresholds are set by the user based on the company's actual situation and risk appetite to determine whether the indicator values are within a safe range.

[0089] During the comparison process, if the core indicator value of any assessment dimension exceeds the corresponding threshold, the corresponding time series position in the predicted enterprise status profile is immediately identified as a risk time series point. This indicates that the enterprise may face potential risk at this time series point. Simultaneously, the pending business operations that led to this risk time series point are traced, and the corresponding business node is identified as a risk business node. This allows the source of the risk to be clearly identified, providing strong support for subsequent risk response.

[0090] Finally, predicted enterprise status profiles that exceed the threshold are identified as risky enterprise status profiles, allowing enterprises to intuitively understand their specific operational status at risk time series points. For example, in a retail enterprise's forecast time series, if the inventory turnover rate in the forecast profile at a certain point in time is significantly below the threshold, this point in time is identified as a risky time series point. The business nodes corresponding to the relevant business operations that lead to inventory backlogs (such as over-purchasing) are also identified as risky business nodes, and the forecast profile at that point in time is also designated as a risky enterprise status profile. This allows the enterprise to take timely measures, such as adjusting procurement plans and strengthening inventory management, to reduce risks and ensure stable operations.

[0091] The embodiment of the present invention provides a method for generating a dynamic enterprise portrait based on multi-source heterogeneous data, which has at least the following technical effects:

[0092] 1. By collecting node-related data from multi-source heterogeneous systems in real time and combining it with a multi-dimensional indicator system, a comprehensive and accurate portrait of the company's current status can be generated. This portrait not only reflects the company's operating status at a certain moment, but also reveals the inherent connections between the various dimensions of the company through multi-dimensional feature vectors, providing the company with an in-depth and detailed self-awareness tool.

[0093] 2. Utilizing the constructed target enterprise status prediction model, combined with business operation time series, a dynamic enterprise status profile prediction time series can be generated. This allows enterprises to foresee the impact of future business operations on their status, allowing them to adjust strategies and mitigate potential risks. Furthermore, by setting assessment dimension thresholds, risky time series points, risky business nodes, and risky enterprise status profiles can be automatically identified, providing enterprises with timely risk warnings.

[0094] 3. In terms of model construction, by integrating and fitting multiple matching enterprise status prediction models and combining them with the historical business operation execution data of the target enterprise for training and optimization, a target enterprise status prediction model that is highly adapted to the characteristics of the enterprise is obtained. This integration and customization approach not only improves the model's prediction accuracy, but also enables the model to better adapt to the unique needs of different enterprises, providing enterprises with more efficient and personalized decision-making support.

[0095] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present application. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application is not limited to the embodiments shown herein, but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for generating dynamic enterprise portraits based on multi-source heterogeneous data, characterized in that: The method comprises: Obtaining a business process topology diagram of a target enterprise, wherein the business process topology diagram includes a plurality of business nodes and a node relationship between the business nodes; Collect node association data of each business node in real time to generate a current status portrait of the target enterprise; Obtaining the business operations to be executed of each of the business nodes to form a business operation time sequence; Based on the current state portrait, the enterprise state portrait is dynamically generated in combination with the business operation time sequence, and the enterprise state portrait prediction time sequence is determined; Determine risk time series points, risk business nodes, and risk enterprise status profiles based on the enterprise status profile prediction time series, and display them in the management cockpit; The node association data of each business node is collected in real time to generate a current status portrait of the target enterprise, including: Performing data collection configuration on each of the service nodes, and determining the data source and data type of each of the service nodes; Extracting real-time data related to each of the business nodes from a multi-source heterogeneous system based on the data source and data type of each of the business nodes, the multi-source heterogeneous system including an enterprise resource planning system, a financial management system, a supply chain management system, and a customer relationship management system; Clean, convert and standardize the real-time data of each business node to form node-related data in a unified format; Generating a current status portrait of the target enterprise based on the node association data includes: Acquire a multi-dimensional indicator system, wherein the multi-dimensional indicator system includes multiple evaluation dimensions set by a user terminal; Determining core indicator values of multiple evaluation dimensions respectively according to the node association data; Based on the core indicator values of each evaluation dimension, a multidimensional feature vector representing the current status of the enterprise is constructed; Mapping the multidimensional feature vector into a current state portrait; The process of dynamically generating an enterprise status portrait based on the current status portrait and combining the business operation sequence to determine the enterprise status portrait prediction sequence includes: Constructing a target enterprise status prediction model, wherein the target enterprise status prediction model is used to predict the enterprise status profile after the business operation is executed; Based on the business operation sequence, obtaining a first business operation to be executed and a corresponding first business node, wherein the business operation sequence includes N business operations to be executed, and each of the business operations to be executed has a corresponding business node; Based on the current state profile, the first to-be-performed business operation, and the first business node, predicting the enterprise state profile using the target enterprise state prediction model to obtain a first predicted enterprise state profile; Continuing to obtain a second to-be-executed business operation and a corresponding second business node based on the business operation time sequence, and obtaining a second predicted enterprise status portrait based on the first predicted enterprise status portrait, the second to-be-executed business operation, and the second business node; Iterate and execute until the Nth predicted enterprise status portrait is obtained; The first predicted enterprise status portrait, the second predicted enterprise status portrait, and the Nth predicted enterprise status portrait are arranged and stored in time sequence to obtain the enterprise status portrait prediction time sequence.

2. The method according to claim 1, characterized in that The multiple evaluation dimensions include at least a financial dimension, an operational dimension, and a market dimension.

3. The method according to claim 1, characterized in that Construct a target enterprise status prediction model, including: Uploading the business process topology diagram to an enterprise management data sharing platform, and receiving a plurality of matching enterprise status prediction models fed back by the enterprise management data sharing platform, each matching enterprise status prediction model having a matching degree identifier; Performing integrated fitting on multiple matching enterprise status prediction models according to the matching degree identifier to obtain an initial enterprise status prediction model; Obtain historical business operation execution data of the target enterprise and generate multiple enterprise historical portrait samples, each enterprise historical portrait sample including a sample initial state portrait, a sample pending business operation, a sample business node, and a sample result state portrait; Based on the multiple enterprise historical portrait samples, the initial enterprise status prediction model is trained and optimized to obtain the target enterprise status prediction model.

4. The method according to claim 1, wherein Determining risk time sequence points, risk business nodes, and risk enterprise status profiles based on the enterprise status profile prediction time sequence includes: Acquire multiple evaluation dimension thresholds, where the multiple evaluation dimension thresholds are set by a user terminal and correspond one-to-one to the multiple evaluation dimensions; Based on the multiple evaluation dimension thresholds, a risk judgment is made on the enterprise status portrait prediction time series to determine the risk time series points, risk business nodes and risk enterprise status portraits.

5. The method according to claim 4, characterized in that The enterprise status portrait prediction time series includes N predicted enterprise status portraits; Based on the multiple evaluation dimension thresholds, risk judgment is performed on the enterprise status profile prediction time series to determine risk time series points, risk business nodes, and risk enterprise status profiles, including: Traversing N predicted enterprise status portraits in the enterprise status portrait prediction time series to obtain a first predicted enterprise status portrait; Extracting the core indicator value of each evaluation dimension in the first predicted enterprise status portrait, and comparing the core indicator value of each evaluation dimension with the corresponding evaluation dimension threshold; When the core indicator value of any evaluation dimension exceeds the corresponding evaluation dimension threshold, the time position corresponding to the first predicted enterprise status portrait is determined as the risk time point, the business node corresponding to the to-be-executed business operation that causes the risk time point to appear is determined as the risk business node, and the first predicted enterprise status portrait is determined as the risk enterprise status portrait.

Images