Total-factor intelligent business modeling method, device and equipment

By acquiring multi-source heterogeneous data for data completion and element mining, business process models are constructed and decision predictions are made, solving the problems of data missingness and element bias in traditional modeling, and improving the accuracy of modeling and the scientific nature of decision-making.

CN120975668APending Publication Date: 2025-11-18BEIJING BASIC POINT ORIGIN INFORMATION TECHNOLOGY CO LTD
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
CN202511072497.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-07-31
Publication Date
2025-11-18

AI Technical Summary

Technical Problem

When faced with complex and ever-changing real-world business scenarios, existing technologies rely on static, manual processes and cannot automatically handle missing data and element deviations, resulting in low accuracy in modeling and processing.

Method used

By acquiring heterogeneous data from multiple sources, data completion processing is performed to extract entity, environment, and tool elements, a business process model is constructed, and decision prediction processing is carried out. The business decision results are determined by combining predictive information and business optimization goals.

Benefits of technology

It significantly improves the accuracy of modeling and the scientific nature of business decisions, solves the problems of data loss and element bias in traditional modeling, and achieves more comprehensive data support and dynamic and accurate element representation.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure BDA0005529704500000073
    Figure BDA0005529704500000073
  • Figure BDA0005529704500000102
    Figure BDA0005529704500000102
  • Figure BDA0005529704500000111
    Figure BDA0005529704500000111
Patent Text Reader

Abstract

The invention discloses a total-factor intelligent business modeling method, device and equipment, and relates to the technical field of computers.The method comprises the steps that multi-source heterogeneous data and a business optimization target in a business scene are obtained, data completion processing is conducted on the multi-source heterogeneous data, and a structured data set is obtained; performing element mining processing on the structured data set to obtain a structured element set; the structured element set comprises entity elements, environment elements and tool elements; performing process digital processing based on the structured element set, and constructing a business process model; the business process model is used for representing a process relationship among elements in the structured element set; performing decision prediction processing on the business process model to obtain predictive information; and performing service target optimization processing based on the predictive information and the service optimization target, and determining a service decision result. According to the invention, the obtained modeling elements are more comprehensive, and the accuracy of modeling processing is improved.
Need to check novelty before this filing date? Find Prior Art

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of computer, in particular to a full-factor intelligent business modeling method, device and equipment. BACKGROUND

[0002] With the deepening of the process of industrial digitization, business modeling and intelligent decision-making technology has gradually become the core driving force for breaking through the efficiency bottleneck in various industries. Whether it is precise production control in manufacturing, global collaborative management in the supply chain, risk prevention and control in the financial field, precise diagnosis in the medical industry, or efficient scheduling of energy systems, the accuracy, response speed and adaptive ability of modeling and decision-making have never been higher. Therefore, in order to improve business efficiency, optimize decision-making quality, and strengthen risk prevention, it is particularly important to study how to intelligently model business activities.

[0003] At present, the related technology adopts a traditional modeling method for business modeling, which includes statistical modeling, machine learning, business process modeling, etc. However, when facing complex and variable actual business scenarios, this scheme needs to rely on static manual driving and cannot automatically handle data missing and factor deviation, resulting in low modeling processing accuracy. SUMMARY

[0004] The purpose of the present application is to provide a full-factor intelligent business modeling method, device and equipment.

[0005] To achieve the above purpose, the present application provides the following scheme:

[0006] In a first aspect, the present application provides a full-factor intelligent business modeling method, comprising:

[0007] Obtaining multi-source heterogeneous data in a business scenario and a business optimization target, performing data padding processing on the multi-source heterogeneous data to obtain a structured data set;

[0008] Performing factor mining processing on the structured data set to obtain a structured factor set; the structured factor set includes entity factors, environmental factors and tool factors;

[0009] Performing process digitization processing based on the structured factor set to construct a business process model; the business process model is used to represent the process relationship between each factor in the structured factor set;

[0010] Performing decision prediction processing on the business process model to obtain predictive information;

[0011] Performing business target optimization processing based on the predictive information and the business optimization target to determine a business decision result.

[0012] In a second aspect, the present application provides a full-element intelligent business modeling device, the device comprising:

[0013] a data padding module configured to acquire multi-source heterogeneous data in a business scenario and a business optimization target, perform data padding processing on the multi-source heterogeneous data, and obtain a structured data set;

[0014] an element mining module configured to perform element mining processing on the structured data set, and obtain a structured element set; the structured element set comprises an entity element, an environment element, and a tool element;

[0015] a process digitalization module configured to perform process digitalization processing based on the structured element set, and construct a business process model; the business process model is configured to represent a process relationship between elements in the structured element set;

[0016] a decision prediction module configured to perform decision prediction processing on the business process model, and obtain predictive information;

[0017] a target optimization module configured to perform business target optimization processing based on the predictive information and the business optimization target, and determine a business decision result.

[0018] In a third aspect, the present application provides a computer device, comprising a memory, a processor, a computer program stored in the memory and executable on the processor, and the processor executes the computer program to implement the steps of the full-element intelligent business modeling method according to any one of the above aspects.

[0019] According to the embodiments of the present application, the following technical effects are disclosed:

[0020] The application provides a full-element intelligent business modeling method, device and equipment, which comprises the following steps: acquiring multi-source heterogeneous data in a business scene and a business optimization target, performing data supplementing processing on the multi-source heterogeneous data to obtain a structured data set, performing element mining processing on the structured data set to obtain a structured element set, the structured element set comprising entity elements, environment elements and tool elements, performing process digitization processing based on the structured element set to construct a business process model, the business process model being used to represent the process relationship between the elements in the structured element set, performing decision prediction processing on the business process model to obtain predictive information, and performing business target optimization processing based on the predictive information and the business optimization target to determine a business decision result. Compared with the prior art, the application solves the problem of weak modeling basis caused by data loss in traditional modeling by acquiring multi-source heterogeneous data and performing data supplementing processing, and provides complete and reliable structured data support for subsequent modeling. The entity elements, environment elements and tool elements are extracted by performing element mining on the structured data set, which overcomes the defects of single and biased element data in traditional modeling, and ensures that the obtained modeling elements are more comprehensive and targeted. The business process model is constructed based on the structured element set, which breaks the limitation of traditional business process modeling which relies on static artificial driving, and dynamically and accurately represents the relationship between elements through process digitization. The decision prediction processing is performed on the business process model, which changes the status quo that traditional modeling is difficult to cope with dynamic changes in complex scenes, and can generate actual predictive information. The business decision result is determined in combination with the predictive information and the business optimization target, which avoids the problem that the traditional modeling decision is disconnected with the target, and finally significantly improves the accuracy of modeling processing and the scientificity and effectiveness of business decision. BRIEF DESCRIPTION OF DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiments will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0022] Figure 1 The structural schematic diagram of the application environment of the full-element intelligent business modeling method in an embodiment of the present application;

[0023] Figure 2 The flowchart of the full-element intelligent business modeling method provided by an embodiment of the present application;

[0024] Figure 3 The structural schematic diagram of the structured element set provided by an embodiment of the present application;

[0025] Figure 4A flowchart of a full-factor intelligent business modeling method provided for another embodiment of the present application is shown in the figure;

[0026] Figure 5 A functional module diagram of a full-factor intelligent business modeling device provided for an embodiment of the present application is shown in the figure;

[0027] Figure 6 A structural diagram of a computer device provided for an embodiment of the present application is shown in the figure. DETAILED DESCRIPTION

[0028] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of protection of the present application.

[0029] The above purposes, features and advantages of the present application can be more obvious and easy to understand. The present application will be further described in detail below with reference to the drawings and specific embodiments.

[0030] In the related art, a traditional modeling method is used for business modeling, which includes statistical modeling, machine learning, business process modeling, etc. However, when facing complex and variable actual business scenarios, this scheme needs to rely on static manual driving and cannot automatically process data accuracy and factor deviation, resulting in low modeling processing accuracy.

[0031] Based on the above defects, the present application provides a full-factor intelligent business modeling method. Compared with the prior art, in the present scheme, by acquiring multi-source heterogeneous data and performing data completion processing, the problem of weak modeling foundation caused by data missing in traditional modeling is solved, providing complete and reliable structured data support for subsequent modeling; the structured data set is mined for factors and the entity, environment and tool factors are extracted, overcoming the defects of single and large deviation of factor data in traditional modeling, ensuring that the obtained modeling factors are more comprehensive and more targeted; and a business process model is constructed based on the structured factor set, breaking the limitation of traditional business process modeling relying on static manual driving, and realizing dynamic and accurate representation of the relationship between factors through process digitization; the business process model is processed for decision prediction, changing the status quo that traditional modeling is difficult to cope with complex scene dynamic changes, and being able to generate predictive information that fits the actual situation; the business decision result is determined in combination with the predictive information and the business optimization target, avoiding the problem that the traditional modeling decision is out of touch with the target, and finally significantly improving the accuracy of modeling processing and the scientificity and effectiveness of business decision.

[0032] The full-factor intelligent business modeling method provided in the embodiments of the present application can be applied to, for example,Figure 1 An application environment of the full-element intelligent business modeling method is shown. The application environment includes a terminal 102, a server 104, and a data storage system. The terminal 102 communicates with the server 104 through a network. The data storage system can store multi-source heterogeneous data in a business scenario obtained by the server 104. The data storage system can be separately arranged, integrated on the server 104, or placed on a cloud or other servers. The terminal 102 can send the obtained test vehicle information of a closed site to the server 104. After the server 104 obtains the multi-source heterogeneous data, the server 104 determines a business decision result through data completion, element mining, process digitization, decision prediction, and business target optimization processing. In addition, in some embodiments, the intelligent business modeling method can also be implemented by the server 104 or the terminal 102 alone, for example, the terminal 102 can directly obtain multi-source heterogeneous data in a business scenario to construct an initial population, and perform data completion, element mining, process digitization, decision prediction, and business target optimization processing to determine a business decision result.

[0033] The terminal 102 can be, but is not limited to, various desktop computers, notebook computers, smart phones, tablet computers, Internet of Things devices, and portable wearable devices. The Internet of Things device can be a smart speaker, a smart television, a smart air conditioner, a smart vehicle device, etc. The portable wearable device can be a smart watch, a smart bracelet, a head-mounted device, etc. The server 104 can be implemented by an independent server or a server cluster composed of multiple servers, and can also be a cloud server.

[0034] In an exemplary embodiment, as shown in Figure 2 A full-element intelligent business modeling method is provided, which is executed by a computer device, specifically, can be executed by a terminal or a server, or can be executed by a terminal and a server together. In the embodiments of the present application, the method is applied to the server 104 in Figure 1 The method includes the following steps S201 to S205. Specifically,

[0035] In step S201, multi-source heterogeneous data in a business scenario and a business optimization target are obtained, and the multi-source heterogeneous data is processed by data completion to obtain a structured data set.

[0036] It should be noted that the above business scenarios can include a variety of scenarios, for example, precise production regulation of manufacturing industry, global collaborative management of supply chain, risk prevention and control in the financial field, precise diagnosis in the medical industry, efficient scheduling of energy systems, etc. Among them, multi-source refers to the diversity of data sources, and heterogeneous refers to the difference in data format. The multi-source and heterogeneous data in the business scenario refers to data of different structures from different data sources. The multi-source and heterogeneous data corresponding to different business scenarios is different. The multi-source and heterogeneous data can be obtained through a multi-level storage architecture.

[0037] Optionally, the multi-source and heterogeneous data in the above business scenario can be obtained from an external device, or can be obtained from a blockchain or a database, or can be obtained by real-time analysis of the entire business process in the business scenario. The embodiment does not limit the acquisition method of the multi-source and heterogeneous data in the business scenario.

[0038] Exemplarily, taking the business scenario of e-commerce marketing as an example, the multi-source and heterogeneous data can come from user behavior logs, transaction systems, customer service records, external third parties, etc. The data format can be structured data, semi-structured data, and unstructured data. Among them, the structured data can be transaction data in an excel table, the semi-structured data can be user behavior logs in json format, and the unstructured data can be customer service voice transcription text, product evaluation pictures, etc.

[0039] The above business optimization target refers to the target of data processing. For example, for retail business, the business optimization target can be to improve the repeat purchase rate, and for supply chain business, the business optimization target can be to reduce the inventory turnover days.

[0040] It can be understood that since the multi-source and heterogeneous data can have data missing, in order to make the obtained data more comprehensive, data completion processing is needed to fill in the missing values. The data completion processing can be performed by a rule-based completion method, an algorithm-based completion method, or a cross-source complementary completion method, so as to obtain a structured data set. The structured data set refers to comprehensive structured data after data completion processing.

[0041] In this step, by obtaining the multi-source and heterogeneous data of the business scenario, the obtained data can cover the full-dimensional features of the business, solving the problem of weak modeling basis caused by single data source and serious missing in traditional modeling, and ensuring the relevance of data and business targets, providing higher quality data information for subsequent factor mining and process modeling.

[0042] In one of the embodiments, the application also provides a specific implementation method of performing data completion processing on the multi-source and heterogeneous data to obtain a structured data set, and the method comprises:

[0043] The first specified operation is executed in a loop until the comprehensive quality score of the current dataset is not less than the quality score threshold, obtaining a structured data set;

[0044] The first specified operation includes:

[0045] The current dataset is comprehensively evaluated to obtain a data completeness score and a data consistency score; when the first specified operation is executed for the first time, the current dataset is multi-source heterogeneous data, and when the first specified operation is not executed for the first time, the current dataset is an updated dataset obtained when the first specified operation is executed last time.

[0046] Based on the data completeness score and the data consistency score, a comprehensive quality score is obtained, and it is determined whether the comprehensive quality score is less than the quality score threshold.

[0047] When the comprehensive quality score is less than the quality score threshold, a data supplement source is obtained, and based on the data supplement source and the multi-source heterogeneous data, an updated dataset is determined. And control enters the next first specified operation.

[0048] When the comprehensive quality score is not less than the quality score threshold, control does not enter the next first specified operation.

[0049] It can be understood that any effective modeling needs to be based on high-quality, complete data. The original multi-source heterogeneous data D raw cannot directly meet this requirement. Therefore, a heterogeneous function module is needed, which can automatically process the multi-level storage architecture multi-source heterogeneous data D raw into a high-quality structured data set D quality In this application, the self- emergence stage is set to realize the supplement processing of high-quality data, and the self- emergence stage adopts a self- emergence data supplement algorithm.

[0050] Specifically, after obtaining the multi-source heterogeneous data D raw , the self- emergence method can be used to combine the preset knowledge set K expert to perform data supplement processing on the multi-source heterogeneous data, obtaining a structured data set D quality . The function D qua quality=f emerge (D raw , K expert ) can be used to represent. Wherein, the function improves the data quality in an iterative manner through the built-in self- emergence data supplement algorithm until the preset data completeness evaluation threshold is met. The preset knowledge data set can be obtained through expert knowledge and business objectives.

[0051] The process of obtaining the structured data set in the embodiment is modeled as an iterative optimization problem, and the algorithm iterative process can be represented by the following formula:

[0052]

[0053] Wherein, D (k) is the current data set, D (k+1) is the updated data set, is the optimal data point.

[0054] First, the current data set D (k) is obtained, which represents the data set at the kth iteration. The comprehensive evaluation of the current data set can include checking the completeness of the current data set, obtaining the completeness score S c (D (k) ), the completeness state includes: the proportion of missing fields, the record completeness, etc., and detecting the data conflict rate, the format uniformity, etc., obtaining the consistency score S a (D (k) ), and then combining according to the preset weight to obtain the comprehensive quality score Q(D (k) ), which can be represented by the following formula:

[0055] Q(D)=w c ·S c (D)+w a ·S a (D);

[0056] Wherein, S c (D) represents the completeness score, S a (D) represents the data consistency score, w c is the weight corresponding to the completeness score, w a is the weight corresponding to the consistency score. The weights w c , w a can be determined by expert knowledge setting, historical modeling effect feedback adjustment, or importance analysis based on specific business scenarios, and support dynamic adjustment to adapt to different application requirements, which can be customized according to the importance of the completeness score and the consistency score.

[0057] The comprehensive quality score is compared with the quality score threshold Q threshold , when Q(D (k) )<Q threshold , the data filling process is triggered, forming a quality-driven iterative mechanism; when Q(D (k) )≥Q thresholdIn the iteration process, the current data set is taken as the structured data set. In the first iteration, the current data set is the initial multi-source heterogeneous data, and in the subsequent iteration, the current data set is the multi-source heterogeneous data obtained after the previous processing. In the first execution of the iteration process, the originally obtained multi-source heterogeneous data is taken as the initial current data set, and a comprehensive evaluation process is performed to obtain a comprehensive quality score, so as to determine whether the quality score is less than a quality score threshold; in the non-first execution of the iteration process, the updated data set obtained through the previous iteration operation is taken as the current data set in the next iteration, and a comprehensive evaluation process is performed to execute the iteration process.

[0058] According to the data source and the multi-source heterogeneous data, the updated data set is determined, including: obtaining a candidate data subset from the data completion source; obtaining a preset knowledge data set, determining the posterior probability of each data point in the candidate data subset under the condition of the multi-source heterogeneous data and the knowledge data set; selecting a data point with the maximum posterior probability from the candidate data subset as an optimal data point; and supplementing the optimal data point to the multi-source heterogeneous data to obtain the updated data set.

[0059] In the execution of the data completion process, the system can automatically locate a potential data completion source, which can include a history library and a knowledge library, and obtain a candidate data subset from the data completion source Then, core completion and expert knowledge enhancement operations are performed, and the core is to select an optimal data point from the candidate data subset The posterior probability of each data point in the candidate data subset under the condition of the multi-source heterogeneous data and the knowledge data set can be obtained, and then a data point with the maximum posterior probability is selected as an optimal data point, and the optimal data point is represented by the following formula:

[0060]

[0061] Wherein, is the selected optimal data point, d is a single data point in the candidate data subset, is the candidate data subset selected from the potential source such as the history library and the knowledge library in the kth iteration, P(d|D (k) ,K expert is the posterior probability of the data point d under the condition of the given current data set D (k) and the knowledge data set K expert .

[0062] After the optimal data point is determined, the optimal data point is supplemented to the current data set D (k) to obtain the updated data set D (k+1)Then, the updated dataset is used as the current dataset for the next iteration, and a comprehensive evaluation is performed on it to obtain the overall quality score Q(D) of the updated dataset. (k+1) ), and compare it with the quality score threshold Q. threshold Perform a comparison, when Q(D) (k+1) )<Q threshold When Q(D) is reached, the next iteration is executed; when Q(D) is reached... (k+1) )≥Q threshold When the loop terminates, the final high-quality structured data set D is output. quality .

[0063] In this embodiment, by executing the first specified operation, the current dataset (initially multi-source heterogeneous data, and subsequently the results of the previous update) is comprehensively evaluated to obtain data completeness and consistency scores. Then, the comprehensive quality score is calculated and compared with the threshold. If it does not meet the standard, the updated dataset is determined based on the data completion source and multi-source heterogeneous data, and the loop continues until the comprehensive quality score meets the standard and a structured data set is obtained. This realizes dynamic iterative optimization of data completion. With continuous evaluation and targeted completion, it effectively solves the problem of weak foundation caused by data missing and inconsistency in traditional modeling, and provides more complete, accurate and high-quality structured data support for subsequent business modeling, improving the accuracy of modeling processing from the data source.

[0064] Step S202: Perform feature mining processing on the structured data set to obtain a structured feature set; the structured feature set includes: entity features, environmental features, and tool features.

[0065] After obtaining the structured dataset, a set of structured elements A for business activities can be obtained through a self-reinforcing approach. This process involves complex element definition and mining; therefore, a functional module is needed to automatically mine and structurally define structured elements from the data. These structured elements include entity elements, environmental elements, and tool elements, with tool elements being the most complex. This application implements element mining processing by setting a self-reinforcing stage, which employs a self-reinforcing element mining algorithm.

[0066] In one embodiment, a specific implementation method for performing feature mining on a structured data set to obtain a structured feature set is also provided, the method including:

[0067] The second specified operation is executed repeatedly until the current structured dataset meets the iteration termination condition, resulting in a set of structured features. The iteration termination condition includes reaching the required number of iterations or the integrity score being no less than the integrity score threshold. The second specified operation includes:

[0068] The integrity of each element in the current structured data set is evaluated to obtain an integrity metric value; when the second specified operation is executed for the first time, the current structured data set is the structured data set, and when the second specified operation is not executed for the first time, the current structured data set is the updated structured data set obtained when the second specified operation is executed for the last time.

[0069] Based on the integrity metric value and the integrity weight value, an integrity score is obtained, and it is determined whether the current structured data set meets the iteration end condition.

[0070] If the current structured data set does not meet the iteration end condition, candidate generation and mining processing are performed based on the structured data set to obtain an updated structured data set, and control is entered into the next second specified operation.

[0071] If the current structured data set meets the iteration end condition, control is not entered into the next second specified operation.

[0072] Specifically, after obtaining the structured data set D quality , the structured element definition can be mined and completed in a self-reinforcing manner until the preset element integrity evaluation standard is met, thereby obtaining the structured element set A, which can be represented by the function A=f enhance (D quality ). Among them, the self-reinforcing element mining algorithm is configured inside the function to mine and complete all element definitions in an iterative manner until the preset element integrity evaluation standard is met.

[0073] In this embodiment, the element mining process is modeled as a structured knowledge discovery and enhancement process, and the goal is to maximize the integrity score C(A) of the structured element set A, which is used to comprehensively evaluate the internal structural integrity of all elements in the structured data set. The integrity score can be represented by the following formula:

[0074]

[0075] Among them, a i is a single element in the structured data set, V(a i ) is the integrity metric value, w iThe importance weight is a weight. The integrity measurement value is used to represent the attribute completeness, (whether the necessary attribute is complete), the relationship integrity (whether the association with other elements is complete), the structural consistency (whether the internal structure conforms to the standard), and the like. The importance weight ranges from 0 to 1, and the integrity score is obtained by weighted summation. The importance weight can be determined based on the business impact degree, the use frequency, the criticality degree, the dependency relationship complexity, and the like of the element, and can be obtained and dynamically adjusted by expert evaluation, historical data analysis, business value contribution calculation, and the like.

[0076] The updated structured data set is obtained based on the structured data set, and includes the following steps:

[0077] A preset standard knowledge base is obtained, and a candidate incremental update option is generated based on the standard knowledge base. The candidate incremental update option includes at least one of the following: complete data of missing attributes, repair information of key connections, enhanced components of functional modules, and version upgrade configuration information of tool elements. The confidence of each update item in the candidate incremental update option is calculated. An update set is selected from the candidate incremental update option, and the update set is applied to the structured data set to obtain the updated structured data set.

[0078] The self-reinforcing algorithm in the embodiment continuously performs incremental updates on the element set in a spiral iterative reinforcement manner, and the iterative process can be represented by the following formula:

[0079] A (k+1) =A (k) ⊕ΔA * ;

[0080] A (k+1) is the updated structured data set, A (k) is the structured data set, and ΔA * is the update set.

[0081] Specifically, in the iterative process, the missing detection and candidate generation operation is performed first. The integrity of each element in the current structured data set is evaluated, the necessary attributes, relationship links, and functional completeness of each element are analyzed by structured analysis, and the standard element template or domain knowledge base is compared to identify specific missing links such as attribute missing, relationship broken, and incomplete function, and the influence degree is quantified to obtain the integrity measurement value of each element. The integrity score is obtained by weighted summation of the corresponding weight. Subsequently, candidate incremental update options are generated based on knowledge graph reasoning, similar element matching, expert knowledge base query, and the like The candidate incremental update option can be understood as a candidate incremental update scheme, including: data completion for missing attributes, repair information for key connections, enhanced components for functional modules, and version upgrade configuration information for tool elements, etc., which are structured update contents.

[0082] After missing detection and candidate generation, deep mining and inference completion are performed. The core decision of this algorithm lies in selecting an optimal, high-confidence update set ΔA from the candidate solutions. * This screening process is based on confidence assessment, meaning that only those samples from the pre-defined standard knowledge base MathCalcA are considered. (k) Under the given conditions, only updates with a confidence level higher than a preset threshold will be adopted. The update set, consisting of updates with a confidence level greater than the preset threshold, can be represented by the following formula:

[0083]

[0084] Where, ΔA * Let δ be the update set, and δ be the update term. A is a candidate incremental update option. (k) For structured data collections, Conf threshold This is the confidence threshold.

[0085] After obtaining the updated set, the selected high-confidence updated sets are applied to the current structured dataset to obtain the updated structured dataset A. (k+1) Then, a comprehensive evaluation is performed on the updated structured dataset to obtain an integrity score. To determine whether the integrity score has converged, the iteration termination condition is met when the integrity score satisfies the integrity score convergence (i.e., |C(A)|). (k+1) )-C(A (k) When the integrity score threshold is reached or the maximum number of iterations is reached, the loop terminates, and the current structured dataset is used as the final structured feature set A. If the integrity score does not meet the iteration conditions, the next iteration is executed.

[0086] This embodiment systematically proposes for the first time a unified modeling framework for the three elements of entity, environment, and tool, especially the versioned structure design of the tool element, which provides new data guidance for the reusable and evolving modeling of business processes.

[0087] For example, please see Figure 3 As shown, Figure 3 This is a schematic diagram of the structured element set provided in the embodiments of this application. Taking manufacturing production scheduling as a scenario, multi-source heterogeneous data in this business scenario is obtained, and data completion processing and element mining processing are performed on it to obtain a structured element set A = {E}. ent ity,Eenv ,T set The framework includes: entity elements, environmental elements, and tool elements. Entity elements include physical objects such as production equipment, workers, raw materials, and product orders. Each entity element has a set of attributes describing its characteristics. This set of characteristic attributes can be accessed via E... entity E indicates entity ={a1, a2, ..., a n}, where a n This represents the nth characteristic attribute of the entity. Environmental Element: Used to characterize external conditions and background information affecting business activities, including production time, market demand, policies and regulations, etc. Each environmental element has a set of attributes describing its characteristics. The set of status attributes corresponding to this environmental element can be obtained through E... env E indicates env ={e1, e2, ..., e m}, where e m This represents the m-th status attribute of the environment. Tool elements are used to represent the smallest granularity of abstract information about operational nodes in a business process, and can be represented by T... set This means that production scheduling algorithms, quality inspection processes, and equipment maintenance strategies can all be abstracted into tool elements with versioned structures.

[0088] For entity elements, personnel include skill attributes, role definitions, and performance records; equipment includes technical parameters, operating status, and maintenance history; materials include physical characteristics, quality standards, and supply information; and products include specifications, quality indicators, and market positioning. For environmental elements, time includes seasonal cycles, business cycles, and unexpected events; location includes geographical location, infrastructure, and resource distribution; market includes demand changes, competitive landscape, and price fluctuations; and policy includes regulatory requirements, industry standards, and compliance constraints. Tool elements are represented using a versioned structure, including business data (historical execution, training samples, and performance records); model parameters (algorithm logic, model application, and parameter configuration); and metadata (input / output contracts, execution constraints, and performance indicators). A full-element model is established, comprising an ordered combination of multiple tool elements, represented as a Directed Acyclic Graph (DAG) business process diagram, possessing interpretable, traceable, and executable characteristics.

[0089] The above-mentioned tool elements can be represented by the following versioned structure:

[0090]

[0091]

[0092] Optionally, in addition to the meta-model-data three-element structure, the versioned structure of the tool element can also use other structured definition methods, such as the configuration-algorithm-history organization form, but the core idea is still the abstraction of the versioned executable unit.

[0093] In this embodiment, a self-reinforcing element mining algorithm is used to continuously improve the internal structure definition of the tool element through spiral iteration, and dynamically update the element knowledge graph. Specifically, by repeatedly executing the second specified operation, taking the structured data set as the initial, performing integrity assessment on its elements for the first time, continuously assessing based on the previous update result, combining the integrity measurement value and the weight value to obtain the integrity score, and determining whether to end the iteration according to the score and the number of iterations, if not, updating the data set through candidate generation and mining processing and continuing the loop, until the conditions are met to obtain the structured element set. This process can dynamically and continuously optimize the integrity of the elements in the structured data set, solve the problem of omission and deviation in traditional modeling element mining, make the refined structured element set more comprehensive and accurate, and lay a solid element foundation for subsequent business process model construction, and improve the adaptability and accuracy of business modeling for complex scenarios.

[0094] Step S203, performing process digitization based on the structured element set to construct a business process model; the business process model is used to represent the process relationship between each element in the structured element set.

[0095] It can be understood that after obtaining the structured element set A, according to the mathematical model relationship between the elements, through self-arrangement, according to the mathematical model of the relationship between the elements, an executable business process model G DAG is assembled. Therefore, it is necessary to have a function module to automatically arrange the tool elements into a complete business process according to the interface contract and dependency relationship of each tool element.

[0096] In one embodiment, performing process digitization based on the structured element set to construct a business process model includes: performing tool instantiation processing on the structured element set to obtain an instantiated tool set; based on the instantiated tool set and the dependency relationship between the elements, using a topological sorting rule to construct an initial business process; obtaining an interface rule, and according to the interface rule and the updated data set, the initial business process, obtaining an interface-compatible intermediate business process; performing logic recognition and completion processing on the intermediate business process to obtain a logically complete complete business process; performing global verification processing on the complete business process to obtain an executable business process model.

[0097] After obtaining the structured element set A, through the function G DAG = f orchestrate(A) performing processing to obtain a business process model G DAG The function assembles the scattered tool elements into a topologically ordered, interface compatible and globally complete executable business process through the built-in intelligent DAG arrangement and global integrity verification algorithm. The structure of the business process model can be a directed acyclic graph (DAG).

[0098] Specifically, after obtaining the structured element set, a tool instantiation operation is performed, the optimal version and configuration are selected by inputting the structured element set A, and the instantiated tool set T is output instances , and an intelligent arrangement operation is performed, the initial business process G is constructed based on topological sorting by inputting the instantiated tool set T instances and the dependency relationship initial , and interface matching is performed, the initial business process G is input initial and the interface specification, and the interface compatible intermediate business process G is output compatible , then breakpoint supplement is performed, the intermediate business process G is input compatible , logical missing is identified and completed, and the logically complete complete business process G is output complete , and global verification is performed, the complete business process G is input complete , global verification processing is performed, and the final executable business process model G is obtained DAG .

[0099] Optionally, in addition to being based on a knowledge graph, the modeling method of the element relationship in the embodiment can also model the element relationship in a relational database, a vector database, etc., but the core idea is still the structured expression and reasoning application of the relationship between elements, and the modeling method of the element relationship in the embodiment is not limited in any way.

[0100] In the embodiment, the self-arrangement mechanism is used to realize intelligent conversion from scattered tool elements to complete business processes, has complete capabilities of automatic arrangement, intelligent optimization and global verification, breaks the limitations of traditional business process modeling relying on static manual driving, and realizes dynamic and accurate representation of the relationship between elements through process digitization.

[0101] The mathematical model between the above elements includes an objective relationship matrix R obj ∈R N×N , a process dependency relationship R flow , and an input-output mapping relationship f. The objective relationship matrix is used to represent the inherent relationship between the entity elements and the environmental elements, the process dependency relationship is used to represent the relationship between the tool elements, which can be represented by a DAG structure, for example, a directed acyclic graph G=(V,E), where V is a tool element set and E is a dependency edge set. The input-output mapping relationship f:Input→Output is used to represent the processing transformation relationship between the tool elements and the entity elements.

[0102] The digital representation and storage architecture of the element adopts a multi-level storage architecture, including a vectorization representation layer, a relationship graph layer, and a model storage layer. The vectorization representation layer refers to the vectorization of element attributes v i ∈R d The relationship graph layer refers to the element knowledge graph KG=(Entities, Relations, Attributes), and the model storage layer refers to the versioned model storage and management of tool elements.

[0103] Step S204, decision prediction processing is performed on the business process model to obtain predictive information.

[0104] It can be understood that any static business process model will be invalid due to changes in the external environment. Therefore, the system must have a functional module to run the model in a simulated environment and continuously and automatically optimize the model according to the difference between the simulation results and the reality feedback, so that it has adaptability to environmental changes. After obtaining the business process model G DAG , a self-evolution method can be used. In the self-evolution stage, a self-evolution prediction optimization algorithm is used to obtain predictive information I predict , which can be represented by the function I predict =f evolve (G DAG ). This function simulates and learns the business process model in the digital twin environment through the built-in self-evolution prediction optimization algorithm to continuously optimize the prediction accuracy of the model in an iterative manner. The predictive information can be understood as predictive information insight and decision support information.

[0105] In an embodiment of the present application, the decision prediction processing is performed on the business process model to obtain predictive information, including:

[0106] The business process model is used as an agent in the reinforcement learning framework to perform business simulation actions in the digital environment and construct an objective function. The objective function aims to maximize business value. The business process model is executed to obtain interaction trajectories, and key deviation points between simulation prediction results and real results are identified. The interaction trajectories are analyzed to evaluate the contribution of each element in the business process model to the key deviation points. According to the contribution, the policy gradient method is used to optimize the internal model parameters until the prediction performance indicators of the business process model converge or reach the maximum training rounds, thereby obtaining the predictive information.

[0107] Specifically, after obtaining the business process model, the prediction optimization process in the twin universe is modeled as a reinforcement learning (Reinforcement Learning) problem. In this framework, the entire business process model G DAGIt is regarded as an agent, which in the dynamic environment of digital twin, through the interaction with the environment trial and error, learn how to adjust its own strategy (i.e. the model parameters of internal tool elements) to maximize the long-term business value.

[0108] The core goal of the above algorithm is to find the optimal business process model The objective function is constructed to maximize the long-term expected reward that can be obtained by performing a series of actions in the digital twin environment, i.e. business simulation, which can be represented by the following formula:

[0109]

[0110] Where s t is the environment state at time t, a t is the action performed by G DAG , and R(s t ,a t ) is the immediate reward calculated by comparing the predicted results with the actual business value according to the action, and γ is the discount factor for future rewards. In this objective function, the known parameters include the upper limit of the time range, the discount factor, T is the upper limit of the time range, γ is the discount factor, and R(s t ,a t ) is the reward function of the environment state and the agent action, the parameters to be optimized are the business process model G DAG and the model parameters θ i of its internal tool elements, and the dynamic parameters are the environment state and the action sequence s t ,a t and the expected value E[·].

[0111] In this embodiment, in order to achieve the above expected value maximization, the following core mechanisms are used to achieve self-evolution reinforcement: environment interaction and deviation detection: execute the business process model G DAG in the digital twin universe to obtain a series of interaction trajectories (s0, a0, r1, s1, …), and identify the key deviation points between simulation prediction and real results. Credit allocation and impact positioning operation: by analyzing the interaction trajectory, the algorithm evaluates the contribution of each tool element T i to the final overall reward (or deviation), thereby locating the key factors affecting the prediction accuracy. Strategy optimization and accurate update operation: based on the contribution of each tool, the algorithm uses methods such as policy gradient (Policy Gradient) to accurately optimize the internal model parameters θ i (i.e. part of the agent's strategy π θ ). Its strategy update rule can be represented as:

[0112]

[0113] wherein, is the policy performance metric of the tool, a is the learning rate, is the internal model parameter at the kth iteration, is the updated internal model parameter. This process continues for iterations until the prediction performance indicator (e.g., average reward) of the overall business process model converges or is terminated when the maximum training epoch is reached, resulting in the predictive information I predict .

[0114] In this embodiment, the self-evolving predictive optimization algorithm is used to continuously optimize the performance of tool elements through reinforcement learning, thereby improving the prediction accuracy and reliability of the twin universe.

[0115] Step S205, based on the predictive information and business optimization target, business target optimization processing is performed to determine the business decision result.

[0116] It can be understood that after obtaining the predictive information I predict , it needs to be converted into an executable business decision to create value, that is, a functional module is used to combine the predictive insight with the business target to generate the optimal decision, and according to the execution effect, a closed-loop feedback is performed to realize the continuous improvement of business value. In this application, through self-optimization, it can be represented by the function S optimal =f optimize (I predict ,K expert ), which combines the predictive insight with the business optimization target through the built-in closed-loop iteration mechanism based on business value feedback to generate a quantifiable and sustainable optimized business decision scheme. Determine the final business decision result S optimal .

[0117] wherein, based on the predictive information and the business optimization target, the business target optimization processing is performed to determine the business decision result, comprising:

[0118] Obtain the initial business decision options, obtain the actual execution effect information through business monitoring; calculate the deviation amount according to the actual execution effect information and the predictive information; adjust the business decision logic and process parameters based on the deviation amount and the preset knowledge base to obtain the adjusted business decision options; optimize the key elements in the adjusted business decision options to obtain the optimized element set; when the optimized element set meets the evaluation standard, determine the business decision result based on the optimized element set.

[0119] Specifically, the initial business decision options S optimal are obtained, the actual execution effect information is collected through business monitoring, the actual execution effect information is output, which is represented by the data set D actual , and deviation analysis is performed: input Dactual and the expected target I predict , calculate the deviation amount of the actual execution effect information and the predictive information and identify the cause, output the deviation analysis report Δ analysis . Then perform strategy adjustment: input the deviation analysis report Δ analysis and the preset knowledge dataset K expert , adjust the decision logic and process parameters, output the adjusted business decision options S adjusted , and perform element optimization: input S adjusted , identify the key elements therein and perform targeted optimization according to the key elements, output the optimized element set Then perform closed-loop iterative processing to determine whether the optimized element set meets the evaluation standard, such as the degree of business value improvement, and when it meets the evaluation standard, output the final optimized business decision result When the optimized element set does not meet the evaluation standard, trigger a new round of iteration operation until the business decision result is obtained.

[0120] Optionally, the above algorithm can also be customized according to actual needs, for example, in addition to self- emergence, self-strengthening, and self-evolution algorithms, different mathematical optimization algorithms such as genetic algorithm, particle swarm optimization, and reinforcement learning can also be used, but the core mechanism is still iterative improvement based on quality evaluation. In the implementation mode of the above digital twin environment, other different simulation technologies and visualization technologies can also be used to build the twin environment, but the core idea is still virtual-real mapping and prediction deduction based on the full-element model.

[0121] The intelligent business modeling method provided in the present application, through the "five-self" collaborative mechanism (self- emergence, self- strengthening, self- arrangement, self- evolution, and self- optimization), supported by innovative core algorithms such as self- emergence data completion, self- strengthening element mining, and self- evolution prediction optimization algorithm, constructs a system-level modeling architecture from raw data to executable business model, realizes prediction optimization closed loop with the help of digital twin, and directly converts the modeling result into business value. At the same time, relying on the built-in feedback and closed-loop optimization mechanism, the system has the ability of continuous self- optimization, forms an end-to-end intelligent modeling closed loop, significantly improves the automation and intelligence level of the modeling process, guarantees the continuous evolution of modeling quality, efficiently adapts to complex and variable business scenarios, and promotes the business modeling to cross from static artificial driving to dynamic intelligent iteration.

[0122] Please refer to Figure 4As shown, the acquired multi-source heterogeneous data, professional knowledge and business target are first input into the input layer, and are processed through the ATOMS core system, are processed through the self-organizing data completion algorithm, high-quality structured data sets (high-quality data elements) are obtained, and are processed through the self-reinforcing element mining algorithm, structured element sets (structured full element sets) are obtained, executable business process models are obtained through the intelligent DAG arrangement algorithm, predictive information (i.e., predictive insight information) is obtained through the self-evolution prediction optimization algorithm, and the final business decision scheme (business decision result) is obtained through the closed-loop iterative optimization.

[0123] The present application provides a full-element intelligent business modeling method, which comprises: acquiring multi-source heterogeneous data and business optimization targets in a business scenario, performing data completion processing on the multi-source heterogeneous data to obtain a structured data set; performing element mining processing on the structured data set to obtain a structured element set; the structured element set comprises: entity elements, environmental elements and tool elements; performing process digitization processing based on the structured element set to construct a business process model; the business process model is used to represent the process relationship between each element in the structured element set; performing decision prediction processing on the business process model to obtain predictive information; performing business target optimization processing based on the predictive information and the business optimization target to determine a business decision result. Compared with the prior art, in the present scheme, the multi-source heterogeneous data is acquired and data completion processing is performed, which solves the problem of weak modeling foundation caused by data missing in traditional modeling, and provides complete and reliable structured data support for subsequent modeling; the structured data set is mined for elements and entity, environmental and tool elements are extracted, which overcomes the defects of single and biased element data in traditional modeling, and ensures that the obtained modeling elements are more comprehensive and more targeted; and a business process model is constructed based on the structured element set, which breaks the limitation of traditional business process modeling relying on static artificial driving, and dynamically and accurately represents the relationship between elements through process digitization; decision prediction processing is performed on the business process model, which changes the status quo that traditional modeling is difficult to cope with dynamic changes in complex scenarios, and can generate predictive information that fits the actual situation; the business decision result is determined in combination with the predictive information and the business optimization target, which avoids the problem that traditional modeling decision is disconnected from the target, and finally significantly improves the accuracy of modeling processing and the scientificity and effectiveness of business decision.

[0124] Based on the same inventive concept, the present application also provides an intelligent business modeling device for implementing the above-mentioned intelligent business modeling device. The implementation scheme of the problem solving provided by the device is similar to the implementation scheme described in the above method, so the specific limitations in one or more full-element intelligent business modeling device embodiments provided below can be referred to the limitations of the full-element intelligent business modeling method in the above text, which will not be described here again.

[0125] In one example embodiment, as shown in Figure 5 A full-factor intelligent business modeling apparatus is provided, which comprises:

[0126] A data complementing module 510 is configured to obtain multi-source heterogeneous data in a business scenario and a business optimization target, and perform data complementing processing on the multi-source heterogeneous data to obtain a structured data set.

[0127] An element mining module 520 is configured to perform element mining processing on the structured data set to obtain a structured element set, wherein the structured element set comprises entity elements, environment elements, and tool elements.

[0128] A process digitalization module 530 is configured to perform process digitalization processing based on the structured element set to construct a business process model, wherein the business process model is configured to represent process relationships between the elements in the structured element set.

[0129] A decision prediction module 540 is configured to perform decision prediction processing on the business process model to obtain predictive information.

[0130] A target optimization module 550 is configured to perform business target optimization processing based on the predictive information and the business optimization target to determine a business decision result.

[0131] As an optional implementation, the data complementing module 510 is specifically configured to:

[0132] perform a first specified operation in a loop until a comprehensive quality score of a current data set is not less than a quality score threshold to obtain the structured data set.

[0133] The first specified operation comprises:

[0134] performing comprehensive evaluation processing on the current data set to obtain a data completeness score and a data consistency score; when the first specified operation is performed for the first time, the current data set is the multi-source heterogeneous data, and when the first specified operation is not performed for the first time, the current data set is an updated data set obtained when the first specified operation is performed last time.

[0135] obtaining the comprehensive quality score based on the data completeness score and the data consistency score, and determining whether the comprehensive quality score is less than the quality score threshold.

[0136] when the comprehensive quality score is less than the quality score threshold, obtaining a data complementing source, determining the updated data set based on the data complementing source and the multi-source heterogeneous data, and controlling to enter the first specified operation next time.

[0137] when the comprehensive quality score is not less than the quality score threshold, controlling not to enter the first specified operation next time.

[0138] As an optional implementation, the data padding module 510 is further configured to:

[0139] obtain a candidate data subset from the data padding source;

[0140] obtain a preset knowledge data set, and determine a posterior probability of each data point in the candidate data subset under conditions of the multi-source heterogeneous data and the knowledge data set;

[0141] select a data point with the maximum posterior probability from the candidate data subset as an optimal data point;

[0142] supplement the optimal data point to the multi-source heterogeneous data to obtain an updated data set.

[0143] As an optional implementation, the element mining module 520 is specifically configured to:

[0144] perform the second specified operation in a loop until the current structured data set meets an iteration end condition, to obtain a structured element set, the iteration end condition including reaching an iteration number or an integrity score being not less than an integrity score threshold;

[0145] the second specified operation includes:

[0146] perform integrity evaluation on each element in the current structured data set to obtain an integrity measurement value; when the second specified operation is performed for the first time, the current structured data set is the structured data set, and when the second specified operation is not performed for the first time, the current structured data set is an updated structured data set obtained when the second specified operation is performed last time;

[0147] obtain an integrity score based on the integrity measurement value and an integrity weight value, and determine whether the current structured data set meets the iteration end condition;

[0148] if the current structured data set does not meet the iteration end condition, perform candidate generation and mining processing based on the structured data set to obtain an updated structured data set, and control to enter the second specified operation next time;

[0149] if the current structured data set meets the iteration end condition, control not to enter the second specified operation next time.

[0150] As an optional implementation, the element mining module 520 is further configured to:

[0151] obtain a preset standard knowledge base, and generate a candidate incremental update option based on the standard knowledge base; the candidate incremental update option includes at least one of the following: complete data of a missing attribute, repair information of a key connection, an enhanced component of a functional module, and version upgrade configuration information of a tool element;

[0152] Calculate the confidence of each update item in the candidate incremental update options;

[0153] Select the update item with a confidence greater than a preset confidence threshold from the candidate incremental update options as an update set;

[0154] Apply the update set to the structured data set to obtain an updated structured data set.

[0155] As an optional implementation, the process digitization module 530 is specifically configured to:

[0156] Perform tool instantiation processing on the structured element set to obtain an instantiated tool set;

[0157] Based on the instantiated tool set and the dependency relationship between elements, an initial business process is constructed using a topological sorting rule;

[0158] Obtain interface rules, and according to the interface rules and the initial business process, an interface-compatible intermediate business process is obtained;

[0159] Perform logical recognition and completion processing on the intermediate business process to obtain a logically complete complete business process;

[0160] Perform global verification processing on the complete business process to obtain an executable business process model.

[0161] As an optional implementation, the decision prediction module 540 is specifically configured to:

[0162] The business process model is used as an agent in a reinforcement learning framework to perform business simulation actions in a digital environment and construct a target function; the target function aims to maximize the expected value;

[0163] Execute the business process model, obtain an interaction trajectory, and identify key deviation points between simulation prediction results and real results;

[0164] Analyze the interaction trajectory to evaluate the contribution of each element in the business process model to the key deviation points;

[0165] According to the contribution, the internal model parameters are optimized using a policy gradient method until the prediction performance indicators of the business process model converge or reach the maximum training round, to obtain predictive information.

[0166] As an optional implementation, the target optimization module 550 is specifically configured to:

[0167] Obtain initial business decision options and obtain actual execution effect information through business monitoring;

[0168] According to the actual execution effect information and the predictive information, calculate the deviation amount;

[0169] adjust the business decision logic and process parameters based on the deviation amount and the preset knowledge base to obtain an adjusted business decision option;

[0170] optimizing a key element in the adjusted business decision option to obtain an optimized element set;

[0171] when the optimized element set meets the evaluation standard, determining a business decision result based on the optimized element set.

[0172] In the device, multi-source heterogeneous data is acquired and data is supplemented to solve the problem of weak modeling foundation caused by data loss in traditional modeling, and to provide complete and reliable structured data support for subsequent modeling; structured data sets are mined for elements and entity, environment and tool elements are extracted to overcome the defects of single and biased element data in traditional modeling, and to ensure that the acquired modeling elements are more comprehensive and more targeted; and a business process model is constructed based on the structured element set to break the limitation of traditional business process modeling relying on static artificial driving, and to realize dynamic and accurate representation of the relationship between elements through process digitization; the business process model is processed for decision prediction to change the status quo that traditional modeling is difficult to cope with dynamic changes in complex scenarios, and to generate predictive information that fits the actual situation; the business decision result is determined in combination with the predictive information and the business optimization target to avoid the problem of disconnection between traditional modeling decision and target, and finally to significantly improve the accuracy of modeling processing and the scientificity and effectiveness of business decision.

[0173] In an exemplary embodiment, a computer device, which can be a server or a terminal, has an internal structure as shown in Figure 6 The computer device includes a processor, a memory, an input / output interface (I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store video tag processing data. The input / output interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used to communicate with external terminals through network connection. The computer program is executed by the processor to implement a full-element intelligent business modeling method.

[0174] Those skilled in the art can understand that,Figure 6 The structure shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. The specific computer device can include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0175] In an exemplary embodiment, a computer device is also provided, including a memory and a processor, the memory storing a computer program, and the processor implementing the steps in the above method embodiments when executing the computer program.

[0176] In an exemplary embodiment, a computer readable storage medium is provided, storing a computer program, which, when executed by a processor, implements the steps in the above method embodiments.

[0177] In an exemplary embodiment, a computer program product is provided, including a computer program, which, when executed by a processor, implements the steps in the above method embodiments.

[0178] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present application are all information and data authorized by the user or authorized by all parties, and the collection, use and processing of related data need to comply with relevant regulations.

[0179] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer readable storage medium, and when the computer program is executed, the processes of the above-mentioned embodiments of the methods can be included. Any reference to memory, databases or other media used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical storage, high-density embedded non-volatile memory, resistive memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration but not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.

[0180] The database involved in the embodiments provided in the present application can include at least one of a relational database and a non-relational database. The non-relational database can include a distributed database based on a blockchain, etc., without being limited thereto. The processor involved in the embodiments provided in the present application can be a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., without being limited thereto.

[0181] The technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, all possible combinations of the technical features in the above embodiments are not described, however, as long as the combinations of the technical features do not exist contradictory, they should be considered as the scope of the present application.

[0182] The principles and implementation modes of the present application are described by applying specific examples in the present application. The above-mentioned embodiments are only used to help understand the method and its core idea of the present application; at the same time, for those skilled in the art, according to the idea of the present application, the specific implementation mode and application range will be changed. In conclusion, the content of the present application should not be understood as a limitation.

Claims

1. A method for modeling all-element intelligent business processes, characterized in that, The comprehensive intelligent business modeling method includes: Obtain multi-source heterogeneous data and business optimization goals from the business scenario, perform data completion processing on the multi-source heterogeneous data, and obtain a structured data set; The structured dataset is subjected to element mining processing to obtain a structured element set; the structured element set includes: entity elements, environmental elements, and tool elements; Based on the structured element set, process digitization is performed to construct a business process model; the business process model is used to characterize the process relationships between the various elements in the structured element set. The business process model is subjected to decision prediction processing to obtain predictive information; Based on the predictive information and the business optimization objectives, business objectives are optimized to determine the business decision results.

2. The full-element intelligent business modeling method according to claim 1, characterized in that, The multi-source heterogeneous data is subjected to data completion processing to obtain a structured data set, including: The first specified operation is executed repeatedly until the overall quality score of the current dataset is not less than the quality score threshold, thus obtaining the structured data set; The first specified operation includes: The current dataset is comprehensively evaluated to obtain a data completeness score and a data consistency score; when the first specified operation is executed for the first time, the current dataset is the multi-source heterogeneous data; when the first specified operation is not executed for the first time, the current dataset is the updated dataset obtained when the first specified operation was executed previously. Based on the data completeness score and the data consistency score, a comprehensive quality score is obtained, and it is determined whether the comprehensive quality score is less than the quality score threshold. When the overall quality score is less than the quality score threshold, the data completion source is obtained. Based on the data completion source and the multi-source heterogeneous data, the updated dataset is determined, and the system is controlled to proceed to the next first specified operation. When the overall quality score is not less than the quality score threshold, the control will not proceed to the next specified operation.

3. The full-element intelligent business modeling method according to claim 2, characterized in that, Based on the data source and the multi-source heterogeneous data, the updated dataset is determined, including: Obtain a subset of candidate data from the data completion source; Obtain a preset knowledge data set, and determine the posterior probability of each data point in the candidate data subset under the conditions of the multi-source heterogeneous data and the knowledge data set; Select the data point with the highest posterior probability from the candidate data subset as the optimal data point; The optimal data points are added to the multi-source heterogeneous data to obtain the updated dataset.

4. The full-element intelligent business modeling method according to claim 1, characterized in that, The structured dataset is subjected to feature mining processing to obtain a structured feature set, including: The second specified operation is executed repeatedly until the current structured dataset meets the iteration termination condition, and the structured feature set is obtained. The iteration termination condition includes reaching the number of iterations or the integrity score is not less than the integrity score threshold. The second specified operation includes: Integrity assessment is performed on each element in the current structured dataset to obtain an integrity metric value; when the second specified operation is executed for the first time, the current structured dataset is the structured data set; when the second specified operation is not executed for the first time, the current structured dataset is the updated structured data set obtained in the previous execution of the second specified operation. Based on the integrity metric and integrity weight, an integrity score is obtained, and it is determined whether the current structured dataset meets the iteration termination condition. If the current structured dataset does not meet the iteration termination condition, candidate generation and mining are performed based on the structured dataset to obtain an updated structured dataset, and then the process is controlled to enter the next second specified operation. If the current structured dataset meets the iteration termination condition, control will not proceed to the next specified second operation.

5. The full-element intelligent business modeling method according to claim 4, characterized in that, Based on the structured dataset, candidate generation and mining processes are performed to obtain an updated structured dataset, including: Obtain a preset standard knowledge base, and generate candidate incremental update options based on the standard knowledge base; the candidate incremental update options include at least one of the following: data to complete missing attributes, information to repair key connections, enhanced components of functional modules, and version upgrade configuration information of tool elements; Calculate the confidence level of each update item in the candidate incremental update options; From the candidate incremental update options, select update items with a confidence level greater than a preset confidence threshold as the update set; The updated set is applied to the structured data set to obtain the updated structured data set.

6. The full-element intelligent business modeling method according to claim 1, characterized in that, Based on the aforementioned set of structured elements, a business process model is constructed through digital processing, including: The structured element set is instantiated to obtain an instantiated tool set; Based on the instantiated toolset and the dependencies between elements, an initial business process is constructed using topological sorting rules. Obtain the interface rules, and based on the interface rules and the initial business process, obtain the interface-compatible intermediate business process; The intermediate business processes are logically identified and completed to obtain a logically complete business process. The complete business process is then subjected to global verification to obtain an executable business process model.

7. The full-element intelligent business modeling method according to claim 1, characterized in that, The business process model is subjected to decision prediction processing to obtain predictive information, including: The business process model is used as an agent in a reinforcement learning framework to perform business simulation actions in a digital environment, and an objective function is constructed; the objective function aims to maximize the expected value. Execute the business process model, obtain the interaction trajectory, and identify the key deviation points between the simulated prediction results and the actual results; Analyze the interaction trajectory and evaluate the contribution of each element in the business process model to the key deviation point; Based on the contribution, the internal model parameters are optimized using the policy gradient method until the predictive performance index of the business process model converges or reaches the maximum number of training rounds, thereby obtaining the predictive information.

8. The full-element intelligent business modeling method according to claim 1, characterized in that, Based on the predictive information and the business optimization objectives, business objective optimization processing is performed to determine the business decision results, including: Obtain initial business decision options and obtain actual execution effect information through business monitoring; The deviation is calculated based on the actual performance information and the predictive information; Based on the deviation and the preset knowledge base, the business decision-making logic and process parameters are adjusted to obtain the adjusted business decision options. The key elements in the adjusted business decision options are optimized to obtain an optimized set of elements. When the optimized set of elements meets the evaluation criteria, the business decision result is determined based on the optimized set of elements.

9. A full-element intelligent business modeling device, characterized in that, The all-element intelligent business modeling device includes: The data completion module is used to acquire multi-source heterogeneous data and business optimization goals in the business scenario, and to perform data completion processing on the multi-source heterogeneous data to obtain a structured data set. The feature mining module is used to perform feature mining processing on the structured data set to obtain a structured feature set; the structured feature set includes: entity features, environmental features, and tool features; The process digitization module is used to perform process digitization processing based on the structured element set and construct a business process model; the business process model is used to represent the process relationships between the various elements in the structured element set. The decision prediction module is used to perform decision prediction processing on the business process model to obtain predictive information; The target optimization module is used to perform business target optimization processing based on the predictive information and the business optimization target, and determine the business decision result.

10. A computer device, comprising: The memory and processor contain a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the steps of the full-element intelligent business modeling method according to any one of claims 1-8.

Citation Information

Patent Citations

  • Modeling and collaborative optimization method

    CN111079997A

  • Business process guided self-learning optimization algorithm

    CN112364220A

  • Automatically building business intelligence models

    US20240346427A1

  • Statistical machine learning and business process models systems and methods

    US9092802B1