Intelligent process construction and dynamic optimization method based on low codes

Abstract

The invention relates to the technical field of enterprise informatization and process automation, and discloses an intelligent process construction and dynamic optimization method based on low codes, and the method comprises the steps: selecting a process category based on preset process classification; an initial process template is generated based on the process category, the initial process template is output in a visual mode, and the initial process template comprises an initial task node, an initial process branch and an initial gateway; adjusting an initial task node, an initial process branch and an initial gateway for the initial process template through visual dragging to form a basic process; executing the basic process, recording the execution condition of the basic process, and generating an optimization suggestion according to a preset optimization model; the user selects an optimization scheme according to the optimization suggestion and outputs an optimization process; and adjusting the initial process template based on a preset process adjustment strategy. The method has the advantages of reducing a process construction technical threshold, realizing process dynamic optimization and improving compliance control capability.

Description

Technical Field

[0001] The present invention relates to the field of enterprise informatization and process automation technology, and more specifically to a low-code-based intelligent process construction and dynamic optimization method. Background Art

[0002] In the daily operations of modern enterprises, office automation (OA) systems have become the core infrastructure supporting various business processes, widely used in key scenarios such as procurement approval, expense reimbursement, and personnel changes. The efficiency and compliance of these processes are directly related to the company's operating costs and risk control capabilities. However, current mainstream enterprise OA systems generally have significant limitations in process construction and execution management, restricting their full value.

[0003] Traditional OA systems rely heavily on specialized technical personnel to build business processes. Whenever a new approval process needs to be added or modified, such as adjusting travel reimbursements or updating procurement specifications, companies often need to hire developers to write tedious code or configure complex rules. This process is not only time-consuming and labor-intensive, but also slows response. When business needs or regulations change, the system struggles to adapt in a timely manner, causing processes to become disconnected from actual operations. Update cycles often take weeks or even months, severely impacting the company's operational agility.

[0004] More significantly, the existing system lacks intelligent process optimization capabilities. Once a process is designed, its operational efficiency depends entirely on the quality of the initial design, with no ability to dynamically adjust based on actual operational data. When bottlenecks or anomalies arise in the process, managers often need to manually analyze logs and interview relevant personnel to identify the issue. This passive response model leads to delayed problem discovery and long optimization cycles. Furthermore, the lack of intelligent analysis capabilities for regulatory regulations makes process design prone to compliance loopholes, which often go undetected until the approval process, resulting in extensive duplication of work and approval delays.

[0005] Furthermore, traditional systems lack effective intelligent assistance during process execution. Approval personnel rely solely on personal experience to determine the compliance of completed forms, making it difficult to quickly identify potential risk points. When encountering complex approval scenarios, the system is unable to dynamically adjust the approval process based on real-time circumstances, resulting in low approval efficiency and insufficient risk control. Summary of the Invention

[0006] In response to the shortcomings of the existing technology, the purpose of the present invention is to provide a low-code-based intelligent process construction and dynamic optimization method and system, which has the advantages of lowering the technical threshold for process construction, realizing dynamic process optimization, and improving compliance control capabilities.

[0007] To achieve the above object, the present invention provides the following technical solutions: Low-code-based intelligent process construction and dynamic optimization methods include: Select a process category based on the preset process classification; Generate an initial process template based on the process category, the initial process template is output in a visual manner, and the initial process template includes an initial task node, an initial process branch, and an initial gateway; The basic process is formed by visually dragging and dropping the initial task node, initial process branch, and initial gateway of the initial process template; Execute basic processes and record process execution data; Analyze the process execution data based on the preset rule statistics module and generate several optimization suggestions; The user selects an optimization plan based on the optimization suggestions and outputs the optimization process; The initial process template is adjusted based on a preset process adjustment strategy.

[0008] Furthermore, a rule optimization auxiliary module is configured, including: Construct a process operation database, which records historical process operations and corresponding process execution data. The historical process operations include task nodes, process branches, and gateways of each historical process operation. The process execution data includes task node execution time, processing data of node configuration personnel, selection frequency of process branches, gateway decision basis and execution results, total process time, number of rollbacks, and rollback opinions; Count and output abnormal links based on the process operation database. The abnormal links reflect the task nodes, process branches or gateways that have an impact on the execution results during the process operation; The user inputs several optimization goals and formulates several optimization plans based on abnormal links, and generates multiple optimization suggestions based on the preset rule base; Each optimization plan is manually scored, and the optimization plan with the highest score is output as the optimization suggestion.

[0009] Furthermore, an optimization solution verification strategy is configured, and the optimization solution verification strategy is executed after the optimization process is actually run. The optimization solution verification strategy includes: Continuously collect actual operation data after the optimization process runs, Compare the actual operation data with the corresponding simulated operation data. If the actual operation data meets the expectations of the simulated operation data, the optimization plan and the actual operation data are written into the process operation database.

[0010] Furthermore, the process adjustment strategy includes: Get the user's historically selected optimization plan rating information, According to the user's scoring weights on different optimization goals, a comprehensive scoring index is formed. The task nodes and process path configurations in the initial process template are adjusted based on the comprehensive scoring index.

[0011] Furthermore, a rule-assisted construction step is configured, which is synchronously executed when the initial process template is visually dragged, including: Pre-extract structured process rules based on institutional documents to form a rule set; When users build a basic process by dragging and dropping, the current process structure is compared with the constraints in the rule set in real time; If it is detected that the current process structure does not meet the constraints in the rule set, a prompt message and corresponding correction suggestions will be output.

[0012] Furthermore, a process initiation pre-review step is configured, including: Get the historical running process associated with the current running process, use the historical running process as the index to search in the process running database and the corresponding process historical execution data, Obtain high-frequency error points in the current running process based on the process historical execution data. The high-frequency error points are reflected in the links in the historical running process where the error frequency exceeds the frequency threshold. Output high-frequency error points and provide corresponding institutional regulations.

[0013] Furthermore, a form compliance verification step is configured, which is executed synchronously during the execution of the basic process, including: Automatically scan form data and check the form's integrity and format compliance based on preset field validation rules; Obtain historical approval form data, compare the current form content with the constraints in the preset rule library, and identify potential risk points. The potential risk points are specifically the current form that violates the corresponding system documents when combined with the historical approval form; Output risk warning information based on potential risk points.

[0014] Furthermore, the form compliance verification step is configured with a dynamic path adjustment strategy. When a potential risk point is detected, a corresponding task node is automatically added based on the specific content of the potential risk point and the real-time rule matching result.

[0015] Furthermore, the initial process template is configured with an initial template auxiliary generation strategy, including: Receive the process template requirement instruction input by the user, extract keywords from the process template requirement instruction and compare the similarity with the historical running process, call the initial process template corresponding to the historical running process with the highest similarity and output it.

[0016] The low-code-based intelligent process construction and dynamic optimization method includes: Process modeling engine, which provides a visual drag-and-drop modeling interface for process components such as nodes, paths, and forms; The rule center module is used to maintain structured process rule sets and field compliance rules; The data service module is used to collect process execution data during process operation and synchronize the process execution data to the optimization suggestion generation module in real time; Optimization engine, which analyzes process data based on preset algorithms, generates optimization suggestions, and verifies the results; A rules management module, which converts regulatory texts into executable compliance rules based on pre-set structured standards; The intelligent assistance module is used to provide real-time assistance based on preset rules during process construction and approval. From the above, it can be seen that the low-code based intelligent process construction and dynamic optimization method and system provided by this application generates basic processes through visual drag and drop and dynamically adjusts the process structure based on the AI ​​optimization model, which solves the problem that traditional OA systems rely on professional and technical personnel and have a long optimization cycle. It has the advantages of lowering the technical threshold for process construction, realizing dynamic process optimization, and improving compliance control capabilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a flowchart of the low-code-based intelligent process construction and dynamic optimization method in the present invention; Figure 2 It is a flowchart of the optimization model training step in the present invention; Figure 3 It is a flow chart of the optimization scheme verification strategy in the present invention; Figure 4 It is an architectural diagram of the low-code-based intelligent process construction and dynamic optimization system in the present invention. DETAILED DESCRIPTION

[0018] 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 ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0019] It should be noted that when a component is referred to as being "fixed to" another component, it may be directly on the other component or there may also be a central component. When a component is considered to be "connected to" another component, it may be directly connected to the other component or there may also be a central component. When a component is considered to be "set on" another component, it may be directly set on the other component or there may also be a central component. The terms "vertical", "horizontal", "left", "right" and similar expressions used herein are for illustrative purposes only.

[0020] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one skilled in the art to which this invention pertains. The terms used in this specification of the present invention are for the purpose of describing specific embodiments only and are not intended to limit the present invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0021] Existing technologies have long faced technical bottlenecks in process development and optimization. Traditional systems rely on professional developers to perform code-level modifications, resulting in lengthy process adjustment cycles and difficulty adapting to rapidly changing business needs. When efficiency bottlenecks or compliance vulnerabilities arise in business processes, the lack of effective automated analysis tools forces managers to manually retrieve operation logs to identify the issues, often missing optimal optimization opportunities. During form approval, approvers rely entirely on personal experience to determine compliance, which can easily lead to overlooking potential risks.

[0022] To address these issues, the R&D team observed that inefficient process construction stemmed from high technical barriers to entry, while delayed optimization stemmed from a lack of a data-driven decision-making mechanism. By analyzing enterprise process management needs, they discovered that visual construction tools could reduce technical reliance, and that real-time data collection and analysis could provide a basis for optimization. Further research revealed that converting regulatory norms into structured rules enabled automated compliance verification. Based on these findings, they developed a technical approach that leveraged a low-code platform for visual construction, dynamic optimization using operational data, and a rules engine to ensure compliance.

[0023] Therefore, this application proposes a low-code-based intelligent process construction and dynamic optimization method, see Figure 1, including selecting a process category based on a preset process classification; generating an initial process template output in a visual manner based on the process category, the template including the initial task node, process branch and gateway; adjusting the template components by visual dragging and dropping to form a basic process; executing the process and recording the execution data; analyzing the data through the rule statistics module to generate optimization suggestions; outputting the optimized process according to the optimization plan selected by the user; and adjusting the initial template based on the process adjustment strategy. The preset process classification is a tree-like classification system divided by business areas, such as procurement, personnel, and finance, approval levels, such as department level / company level, and process complexity, such as single node / multiple branches. It includes specific categories such as procurement, expense reimbursement, and personnel changes. Users select matching categories through classification catalogs or tags.

[0024] Among them, process classification selection refers to the division of process types according to business scenarios, which can be implemented by a tree-like classification catalog or a label system to ensure the targeted generation of process templates. Initial process template generation refers to calling the preset template library based on the classification results, which can be implemented by a template matching algorithm, and displaying an editable process framework through a visual interface. Visual drag and drop adjustment refers to modifying the process structure through graphical operations, which can be implemented by the HTML5 drag and drop API, allowing non-technical personnel to directly modify node relationships and path logic. Process execution data recording refers to collecting operation logs during process runtime, which can be implemented by tracking technology to capture key indicators such as task duration and path selection frequency. The rule statistics module refers to the built-in data analysis engine, which can be implemented by an association rule mining algorithm to identify inefficient links and compliance risks in the process. The process adjustment strategy refers to the mechanism for dynamically updating templates, which can be implemented by version control technology to automatically iterate the template configuration based on the optimization results.

[0025] Specifically, when an enterprise needs to create a new procurement approval process, the operator first selects the "Procurement" process from the category catalog. The system automatically generates an initial template containing standard nodes such as purchase requisition, price comparison review, and contract approval. The supplier qualification review node is added by dragging and dropping, and the approval path is adjusted to parallel countersignature mode. After the process goes live, the system records an abnormal average time consumption at the contract approval node. Analysis by the rule statistics module reveals that this node has been repeatedly modified due to a lack of legal pre-audit. Based on this, an optimization suggestion is generated to add a legal review node. After the user confirms the recommendation, the system automatically updates the initial template configuration, and all subsequent new processes will include this optimized node by default.

[0026] Compared to existing technologies, traditional system modifications require developers to rewrite approval logic code. This solution, however, uses visual operations to adjust process structure, reducing modification cycles from weeks to hours. Existing technologies lack operational data analysis capabilities and are unable to proactively identify process bottlenecks. However, this solution, through point-of-care data collection and rule analysis, automatically locates inefficiencies and proposes optimization solutions. Regarding compliance assurance, traditional systems rely on manual verification of regulatory documents. This solution transforms regulations into structured rules, enabling real-time verification during form completion.

[0027] Through the above technical solution, this application enables non-technical personnel to independently complete process design and adjustment, significantly reducing reliance on professional development resources. By continuously collecting operational data and generating optimization suggestions, process performance is dynamically improved. Combined with a structured rules engine, compliance risks caused by human oversight are effectively reduced, ensuring that business processes always meet the latest regulatory requirements.

[0028] This application further proposes a solution for configuring rule optimization auxiliary modules, see Figure 2 , including building a process operation database, which records the historical operation process and the corresponding process execution data. The historical operation process includes the task nodes, process branches and gateways of each historical operation process. The process execution data includes the execution time of the task node, the processing data of the node configuration personnel, the selection frequency of the process branch, the gateway decision basis and execution results, the total process time, the number of rollbacks and rollback opinions; based on the process operation database, the abnormal links are counted and output, and the abnormal links reflect the task nodes, process branches or gateways that have an impact on the execution results during the process operation; the user inputs several optimization goals and formulates several optimization plans based on the abnormal links, and generates multiple optimization suggestions based on the preset rule base; each optimization plan is manually scored, and the optimization plan with the highest score is output as the optimization suggestion. The preset rule base is composed of structured rules. The rules are condition-action logic pairs, such as if-else statements, and are stored in a relational database. The sources include the results of enterprise system document analysis and historical optimization experience accumulation, and the latest system updates are automatically synchronized every week; the abnormal links reflect the task nodes, process branches or gateways that affect the execution results during the process operation; the impact types include efficiency, execution time exceeding the average of similar nodes by 20%, compliance, violation of the preset rule base constraints, and accuracy rollback times ≥3 times. If any type is met, it is determined to be an abnormality.

[0029] Among them, the process operation database refers to a database that stores historical process operation records. It can be implemented as a relational database or a time-series database, and is used to persist the structured data generated during the process execution. Abnormal links refer to links that affect process efficiency or results identified through data analysis. They can be implemented using statistical analysis algorithms or machine learning models to locate process bottlenecks or high-incidence points of errors. Optimization goals refer to the improvement directions that users expect to achieve. They can be implemented by receiving keywords or options entered by users through an interactive interface to guide the generation of optimization solutions. The preset rule base refers to a set of rules that includes compliance constraints and optimization logic. It can be implemented using a rule engine or a decision tree model to automatically generate optimization suggestions that meet business specifications. Manual scoring refers to the subjective evaluation of alternative solutions by users. It can be implemented by collecting user scoring results for different dimensions through a scoring interface to comprehensively evaluate the actual feasibility of optimization solutions.

[0030] Specifically, the process operation database collects data such as the execution time of task nodes in historical processes and the frequency of process branch selection to form an analyzable process execution data set. The abnormal link identification module automatically filters out links with abnormal execution time, excessive rollback frequency, or missing decision-making basis based on preset statistical thresholds or machine learning models. After the user enters the optimization goal through the interactive interface, the rule optimization auxiliary module combines the characteristics of the abnormal link with the compliance requirements in the rule base to generate multiple candidate optimization solutions. For example, when it is detected that the average processing time of an approval node exceeds the preset threshold, the system may recommend adding parallel approval paths or adjusting the node responsible person configuration. After all candidate solutions are scored by users from dimensions such as efficiency improvement, resource consumption, and compliance risks, the solution with the highest score is selected as the final optimization suggestion.

[0031] Compared with existing technologies, traditional OA systems lack the ability to systematically analyze historical operational data. Identifying abnormal links relies on manual judgment, resulting in a highly subjective and slow optimization solution generation process. This solution, by building a process operation database and automated analysis modules, can quickly identify process bottlenecks and generate data-driven optimization recommendations. It also incorporates a manual scoring mechanism to ensure that solutions meet actual business needs.

[0032] Through the above technical solutions, this application can effectively solve the problems of insufficient data analysis and lack of objective basis for optimization suggestions in traditional process optimization. By automatically identifying abnormal links and generating multi-dimensional optimization solutions, the process optimization cycle is significantly shortened. At the same time, the introduction of a manual scoring mechanism avoids the business adaptability issues that may arise from pure algorithmic recommendations, ensuring that the final selected optimization solution not only conforms to the data analysis conclusions but also meets the personalized needs of actual business scenarios.

[0033] This application further proposes to configure an optimization solution verification strategy, which is executed after the optimization process is actually run. Figure 3 ,The optimization scheme verification strategy includes continuously collecting the actual ,operation data after the optimization process runs, and comparing the actual ,operation data with the corresponding simulation operation data. If the actual ,operation data meets the expectations of the simulation operation data, the optimization scheme and ,actual operation data are written into the process operation database.

[0034] The optimization solution verification strategy refers to a mechanism for verifying the effectiveness of implemented process optimization solutions. This can be achieved through automated monitoring modules and data comparison algorithms to ensure that the optimization solution's actual results meet expectations. Actual operation data refers to the execution records generated by the optimized process in real business scenarios. This can be obtained through point collection and log analysis technologies to reflect the actual operating status of the optimization solution. Simulated operation data refers to predictive operation indicators generated based on historical data and algorithmic models. This can be generated using machine learning models or process simulation tools to establish an expected baseline for optimization results. The process operation database refers to a knowledge base that stores historical process data and optimization solutions. This can be implemented using a distributed database architecture to accumulate optimization experience and provide data support for subsequent process iterations.

[0035] Specifically, after the optimization process is put into operation, the system automatically starts the data collection module to capture the execution time of each node in the process, the frequency of path selection and other operating indicators in real time. The actual operation data collected is processed by the data cleaning module and compared at the field level with the simulated operation data generated in advance by the process simulation engine. When the deviation value between the key indicators in the actual operation data and the simulated data is within the preset threshold range, the optimization plan is judged to have passed the verification. The verified optimization plan and its corresponding operation data will be automatically archived to the process operation database as a reference for subsequent process optimization. The entire verification process is triggered by preset automated rules, and there is no need for manual intervention in the data comparison and result judgment links.

[0036] Compared with existing technologies, traditional process optimization solutions lack a systematic verification mechanism after implementation, often relying on manual sampling inspections or post-analysis statistical analysis. This leads to long verification cycles and incomplete sample coverage. This solution, by establishing an automated verification mechanism, enables real-time monitoring and objective evaluation of optimization results, avoiding subjective errors in human judgment and significantly shortening the verification feedback cycle.

[0037] Through the above technical solution, this application solves the problem of delayed verification and lack of data support after the implementation of traditional process optimization solutions, and realizes closed-loop verification of the optimization effect. In actual business scenarios, when an enterprise implements a new procurement approval process optimization solution, the system can automatically verify whether the optimized approval time has achieved the expected reduction target, and convert the verified optimization solution into a reusable process template, providing a reliable data basis for subsequent optimization of similar processes.

[0038] This application further proposes a process adjustment strategy including obtaining the user's historically selected optimization plan scoring information, forming a comprehensive scoring index based on the user's scoring weights on different optimization goals, and adjusting the task nodes and process path configurations in the initial process template based on the comprehensive scoring index.

[0039] Among them, the scoring information of the optimization schemes selected by users in the past refers to the data set of users' quantitative evaluation of the actual effects of the implemented optimization schemes. Specifically, it can be implemented by a five-level scoring system or a numerical interval scoring method, which is used to reflect the actual effects of the optimization schemes in dimensions such as efficiency improvement and compliance improvement. The scoring weight refers to the relative importance of different optimization goals in the overall evaluation system. Specifically, it can be calculated by the hierarchical analysis method or the entropy weight method, and is used to quantify the differences in the degree of user attention to goals such as cost control and approval timeliness. The comprehensive scoring index refers to a multi-dimensional evaluation parameter formed by weighted calculation. Specifically, it can be implemented by a linear weighted model or a fuzzy comprehensive evaluation algorithm, which is used to convert scattered scoring information into a quantifiable and comparable global optimization basis. Task node and process path configuration adjustment refers to the dynamic modification of elements such as approval links and branch conditions in the process template. Specifically, process mining technology can be used to identify inefficient nodes and reconfigure the approval path based on the scoring results.

[0040] Specifically, after the user completes the selection and implementation of the optimization plan, the system will continue to collect evaluation data of the plan in actual operation. For example, in the procurement approval scenario, the user may give a higher score to the optimization plan that shortens the approval cycle, and a lower score to the cost control plan. The scoring weight module automatically identifies that the user is more concerned about timeliness indicators by analyzing the user's historical operation data. The comprehensive scoring indicator generation module sets the timeliness weight to 0.7 and the cost control weight to 0.3 to form a personalized evaluation system. The process adjustment engine iteratively optimizes the process template based on this indicator, such as promoting the parallel approval nodes set in the high-scoring plan to similar processes, while reducing the frequency of budget reviews involved in low-scoring plans.

[0041] Compared to existing technologies, traditional process management systems lack the ability to dynamically learn user preferences, and process optimization relies primarily on manual judgment. Existing static scoring models are unable to differentiate between objective priorities in different business scenarios, resulting in poor alignment between optimization solutions and actual needs. This solution establishes a quantifiable, comprehensive scoring system that enables continuous improvement of process templates based on real user feedback, enabling personalized process configuration while ensuring regulatory compliance.

[0042] Through the above-mentioned technical solution, this application solves the technical problem of traditional process management systems' optimization strategies being out of sync with actual user needs. By quantitatively analyzing user focus on different optimization objectives, process adjustment strategies can be precisely aligned with business needs. For example, compliance can be prioritized in expense reimbursement processes, while timeliness can be emphasized in procurement approval processes. This dynamic adjustment mechanism effectively improves the adaptability of process templates and the effectiveness of optimization solutions, avoiding the subjective bias and lags caused by manual adjustments.

[0043] The present application further proposes a rule-assisted construction step, which is executed synchronously when the initial process template is visually dragged, including pre-extracting structured process rules based on system documents to form a rule set; when the user builds the basic process by dragging, the current process structure is compared with the constraints in the rule set in real time; if it is detected that the current process structure does not meet the constraints in the rule set, a prompt message and corresponding correction suggestions are output.

[0044] Among them, structured process rules refer to standardized process requirements parsed from corporate system documents. Specifically, natural language processing technology can be used to extract key process elements, such as approval levels, amount thresholds, departmental permissions, etc., to form a machine-recognizable rule base. Constraints refer to compliance requirements that must be met when building a process. For example, purchase applications for a specific amount must pass the review node of the finance department. Specifically, the text system can be converted into logical judgment conditions through the rule engine. Prompt information and correction suggestions refer to the automatically generated problem location description and adjustment plan when a violation of the process structure is detected. For example, when it is detected that a necessary approval link is missing, it can be recommended to insert an approval node for a specified role.

[0045] Specifically, when users drag and drop task nodes through the visual interface to build a process, the system background simultaneously runs a rule verification mechanism. For example, when building a procurement approval process, if the user does not set a financial review node in the process, and the system rules require a financial review when the procurement amount exceeds the set threshold, the system will immediately trigger an early warning prompt. The early warning information can specify the missing node type, the violated system clause number, and provide operational suggestions for inserting a financial review node. Users can directly click to confirm the addition of the node according to the prompt, or manually adjust the process structure until all constraints are met.

[0046] Compared to existing technologies, traditional OA systems lack real-time rule verification capabilities during the process design phase. Designers rely solely on memory or manual verification of system documents, which can easily lead to missing critical approval steps. For example, when building cross-departmental collaborative processes, process breakpoints often occur due to unfamiliarity with other departments' systems. These issues are often not discovered until the process is actually running, resulting in extensive rework. However, this solution, through automated rule parsing and real-time verification, can immediately intercept illegal operations during the process construction phase, preventing incorrect processes from entering the execution phase.

[0047] Through the above technical solutions, this application achieves real-time compliance assurance during the process design process, effectively reducing process defects caused by human negligence. For example, when building complex processes involving multi-department collaboration, the system can automatically identify missing signing nodes or unauthorized operations to ensure that the process structure fully matches the institutional requirements. This dynamic verification mechanism significantly reduces the workload of subsequent process adjustments and exception handling, making the processes built on the low-code platform both flexible and compliant.

[0048] This application further proposes a process initiation pre-review step, including obtaining the historical running process associated with the current running process, using the historical running process as an index to search in the process running database and the corresponding process historical execution data, and obtaining the high-frequency error points in the current running process based on the process historical execution data. The high-frequency error points are reflected in the links in the historical running process where the error frequency exceeds the frequency threshold, and the high-frequency error points are output and the corresponding system regulations are given.

[0049] The process execution database is a structured database that stores historical process execution records. It can be implemented as a relational database or a time-series database. It is used to persistently store task node status, approval records, time-consuming data, and other operational traces generated during process execution. By establishing a mapping between process versions and execution data, this database provides a data foundation for error pattern analysis.

[0050] Frequent error points are identified through statistical analysis of historical process execution data as recurring errors. Specifically, a sliding window algorithm is used to calculate the frequency of errors, combined with a preset frequency threshold for judgment. For example, if an approval node experiences five or more timeouts or rollbacks in the last 30 process instances, that node is marked as a frequent error point.

[0051] Institutional matching involves associating frequent errors with relevant institutional document clauses. Specifically, natural language processing techniques can be used to extract keywords from institutional texts and establish a database of mappings between error types and institutional clauses. When frequent errors are detected, the corresponding institutional basis is automatically retrieved by matching the error's characteristic keywords.

[0052] Specifically, during the process initiation phase, the system automatically retrieves historical run instances corresponding to the current process template. By analyzing data such as approval logs and rollback records stored in the process operation database, it identifies task nodes or path branches where error rates exceed a preset threshold. For example, in the procurement approval process, the system discovered that the contract review link had required supplementary materials more than 20% of the time in the historical process, thus determining that this link was a high-frequency error point. The system then retrieved the procurement contract review specification clauses from the system management module and pushed the specific material list requirements and review standards to the process initiator.

[0053] Compared with existing technologies, traditional OA systems lack proactive early warning mechanisms for historical error patterns when processes are initiated, often forcing reviewers to repeatedly handle the same form errors. This solution, by establishing an automated analysis mechanism for process operation data, proactively identifies potential risk areas during process initiation and provides targeted guidance based on regulatory compliance, preventing recurrence of errors in newly created processes.

[0054] Through the above technical solution, this application can identify error-prone links in advance and push compliance guidance at the process initiation stage, effectively reducing the probability of approval interruption due to form filling errors or process design defects, reducing the number of process rollbacks and communication costs, and at the same time strengthening the compliance guarantee of process design through the instant association of system terms.

[0055] This application further proposes to configure a form compliance verification step, and synchronously execute the form compliance verification step during the execution of the basic process, including automatically scanning the form data, checking the integrity and format compliance of the form based on the preset field verification rules; obtaining historical approval form data, comparing the current form filling content with the constraints in the preset rule library, and identifying potential risk points. The potential risk points are specifically the current form that violates the content of the corresponding system document after combining with the historical approval form; and outputting risk warning information based on the potential risk points.

[0056] Among them, the preset field verification rules refer to a pre-defined set of logical conditions for verifying the compliance of form fields. Specifically, they can be implemented by regular expression matching, data type verification, or business rule logical judgment to ensure that the required items of the form are complete and the data format complies with the specifications. Historical approval form data refers to the form instances of completed approval processes and their associated approval result data stored in the database. Specifically, they can be obtained through the database query interface to provide historical reference for the compliance judgment of the current form. Potential risk points refer to abnormal situations where the current form content conflicts with historical approval records and system requirements. Specifically, they can be identified through the rule engine through logical reasoning of the form field values ​​and the constraints in the preset rule library, which is used to discover compliance risks caused by errors in form filling or deviations in rule understanding.

[0057] Specifically, during the expense reimbursement process, when the applicant submits an electronic expense report, the system automatically scans the expense type, amount, invoice number and other fields in the form to verify whether required items are missing or the amount format is incorrect. For example, if it is detected that the invoice number is not filled in according to the "year-department-serial number" format, it will immediately prompt for correction. At the same time, the system retrieves the approval records of similar expenses of the applicant in the past three months. When it is found through comparison that the current reimbursement amount exceeds the historical average by a certain percentage, combined with the provisions in the expense management system that require additional explanations for abnormal amounts, the system identifies the potential risk of not filling in the situation description. At this time, the system generates a prompt message containing a specific risk description and a reference to the system terms, pushes it to the approver interface, and automatically adds a financial review node to the approval path.

[0058] Compared to existing technologies, traditional OA systems rely on reviewers to manually check form content, making it difficult to quickly identify formatting errors or historical data conflicts. This solution, however, automatically verifies basic form compliance through pre-set rules and dynamically identifies potential risks based on historical review data, enabling proactive risk warnings. While reviewers in existing technologies must manually review historical documents to identify issues like duplicate reimbursements, this solution uses a rule engine to automatically correlate historical data, significantly improving risk identification efficiency.

[0059] Through the above technical solution, this application solves the problems of delayed form compliance checks and reliance on manual experience in traditional approval processes. It enables real-time verification and risk warnings during the form filling phase, avoiding process rollbacks caused by form errors. At the same time, by linking historical approval data with institutional rules, it can identify complex risks that are difficult to detect through single form checks, such as duplicate applications and abnormal amount fluctuations, thereby improving the accuracy of approval decisions and risk control capabilities.

[0060] This application further proposes that the form compliance verification step is configured with a dynamic path adjustment strategy. When a potential risk point is detected, the corresponding task node is automatically added based on the specific content of the potential risk point and the real-time rule matching results.

[0061] Among them, potential risk points refer to the content of the current form that violates the corresponding system documents after being combined with the historical approval forms. Specifically, natural language processing technology can be used to perform semantic analysis on the form fields, combined with the rule engine for logical judgment to achieve this. This feature is used to identify form data combinations that may violate preset compliance requirements. Among them, real-time rule matching refers to the instant comparison of the current form data with the constraints in the preset rule library, which can be implemented specifically by regular expression matching or decision tree algorithm. This feature ensures that risk judgments are based on the latest compliance standards. Among them, the dynamic path adjustment strategy refers to the automatic generation of remedial action nodes based on the risk type, which can be achieved by dynamically inserting approval links or supplementary material nodes through the API interface of the process engine. This feature realizes the intelligent reconstruction of the process path to deal with sudden risks.

[0062] Specifically, during the procurement application process, if the system detects a correlation between a supplier's qualifications for a purchase order and a historical blacklisted supplier, the system automatically triggers a risk warning mechanism. By matching the supplier access rules in the procurement management system in real time, the system determines that the order has a qualification review flaw. At this point, the process engine automatically inserts a legal review node into the existing approval chain, requiring the legal department to conduct a special review of the supplier's background. This newly added node also generates a corresponding review form template and pushes it to the designated approver's workstation.

[0063] Compared to existing technologies, traditional systems can only interrupt the process or return for revisions when form issues are discovered, without automatically adjusting subsequent processing paths based on the nature of the risk. This solution, through a linkage mechanism of rule matching and process reconstruction, can accurately locate risk resolution links while maintaining process continuity, avoiding efficiency losses caused by process interruptions.

[0064] Through the above technical solution, this application effectively addresses the rigidity of traditional approval processes when faced with compliance risks. After detecting a potential risk, the system automatically adapts to the optimal response path, such as submitting supplementary materials, inserting a special approval process, or initiating a multi-departmental review and approval mechanism, seamlessly integrating risk management with business processes. This approach not only ensures the rigidity of system implementation but also avoids response delays caused by manual intervention, significantly improving risk control efficiency in complex business scenarios.

[0065] This application further proposes that the initial process template is configured with an initial template auxiliary generation strategy, including receiving process template requirement instructions input by the user, extracting keywords from the process template requirement instructions and comparing the similarity with historical running processes, calling the initial process template corresponding to the historical running process with the highest similarity and outputting it.

[0066] Among them, the process template requirement instructions refer to the process construction requirements entered by the user through natural language or structured forms. Specifically, they can be implemented by using a text input box combined with a drop-down selection component to clarify the core elements of the process to be built. Keyword extraction refers to the use of natural language processing technology to identify key business elements from instructions. Specifically, it can be implemented by using a word segmentation algorithm combined with a domain dictionary to accurately capture user needs. Similarity comparison refers to the use of an algorithm to calculate the degree of match between current requirements and historical processes. Specifically, it can be implemented by using a cosine similarity algorithm combined with a semantic vector model to quickly locate reusable templates. Historical running processes refer to verified process instances stored in the database. Specifically, they can be stored in the form of process version snapshots, which contain structured data such as node configurations and path rules, and are used to provide reliable templates that have been tested in practice.

[0067] Specifically, when a user needs to create a new business process, the system receives process building requirements, including business scenarios, approval steps, and other elements, through an interactive interface. After semantic parsing of the input text, key business elements are extracted as standardized tags. The system then performs a multi-dimensional match between these tags and historical process metadata stored in the process database. By calculating weighted semantic and structural similarity, the system selects the historical process template with the highest matching score. After compliance verification, this template is automatically loaded into the visual modeling interface as the initial construction basis.

[0068] In some implementations, when a user enters a "purchase order approval process" requirement, the system automatically matches a process template from the historical process library that meets similarity criteria in terms of purchase amount range, supplier type, and other dimensions. If multiple candidate templates are available, a visual comparison view can be provided to assist the user in selecting.

[0069] Compared to existing technologies, traditional systems require users to manually build processes from a blank canvas. This solution, however, rapidly initiates process building through intelligent matching of historical templates. While existing technologies rely on manual retrieval of historical processes, this solution uses algorithms to automatically recommend optimal templates, significantly shortening the process design cycle.

[0070] Through the above technical solution, this application effectively addresses the low template reuse rate and poor construction efficiency of traditional process construction systems. By automatically matching historical best practice templates, it reduces the workload of repeated design while ensuring that new processes comply with existing business specifications. Users can quickly generate a compliant initial process framework without professional process design experience, significantly lowering the technical threshold for process construction.

[0071] This application further proposes a low-code-based intelligent process construction and dynamic optimization system, see Figure 4 , including a process modeling engine, a rule center module, a data service module, an optimization engine, a rule management module, and an intelligent assistance module. The process modeling engine is used to provide a visual drag-and-drop modeling interface for process components such as nodes, paths, and forms; the rule center module is used to maintain structured process rule sets and field compliance rules; the data service module is used to collect process execution data during process operation and synchronize the process execution data to the optimization suggestion generation module in real time; the optimization engine is used to analyze process data based on preset algorithms, generate optimization suggestions, and verify the effects; the rule management module is used to convert institutional texts into executable compliance rules based on preset structured standards; and the intelligent assistance module is used to provide real-time assistance based on preset rules during process construction and approval.

[0072] The process modeling engine is the core component that supports process design through a graphical interface. This can be implemented using a browser-based drag-and-drop editor, lowering the technical barrier to entry by pre-setting standardized process elements. The rule center module is the database system that stores and manages process constraints. This can be implemented using a relational database combined with a rule engine, ensuring that compliance requirements are automatically met during process construction. The data service module is the middleware responsible for collecting and transmitting process operation data. This can be implemented using message queues combined with data pipeline technology, enabling real-time synchronization and persistent storage of operational data. The optimization engine is the computational module that generates process optimization recommendations based on data analysis. This can be implemented using machine learning algorithms combined with a business rule library, verifying the effectiveness of optimization solutions through simulation. The rule management module is the processing unit that converts natural language regulations into machine-executable rules. This can be implemented using natural language processing technology combined with knowledge graphs, enabling structured analysis of regulatory clauses. The intelligent assistance module is an interactive component that provides real-time guidance during process operations. This can be implemented using a rule matching engine combined with context-aware technology, proactively delivering compliance alerts at key points.

[0073] Specifically, during the process construction phase, the process modeling engine provides a visual interface for business personnel to directly drag and drop process elements, and the rule management module simultaneously converts relevant system documents into structured rules and stores them in the rule center module. When the user configures the form fields, the intelligent assistance module automatically calls the field compliance rules in the rule center module for real-time verification. During the process operation, the data service module continuously collects data such as task execution time and path selection frequency. The optimization engine identifies process bottlenecks based on this data and generates optimization suggestions. In the approval process, the intelligent assistance module dynamically marks potential risk points by comparing historical approval data with the current form content. When it is detected that the process path deviates from the preset rules, the rule center module triggers an alarm and recommends a correction plan. At the same time, the optimization engine simulates and verifies the adjusted process.

[0074] Compared to existing technologies, traditional OA systems require professional developers to write code to configure processes. This system, however, leverages a visual modeling engine to enable business personnel to directly participate in process design, significantly reducing technical reliance. Existing systems lack real-time rule verification capabilities, while this system utilizes a rule management module to automatically translate and execute regulatory regulations, ensuring compliance with requirements from the very beginning of process development. Traditional solutions struggle to promptly identify process defects during operation, while this system leverages a data service module and optimization engine to continuously analyze operational data and dynamically optimize process configurations.

[0075] Through the above technical solutions, this application solves the problems of traditional OA system process construction relying on coding, delayed compliance verification, and lack of data support for optimization. Business personnel can quickly build compliance processes without a technical background, and system specifications are directly integrated into the process design link through automated analysis, avoiding compliance risks caused by human omissions. The real-time collection and analysis of operating data provides an objective basis for process optimization, so that process efficiency can continue to improve with business development. In the approval process, the system actively identifies form filling anomalies and path deviations, effectively reducing approval delays and operational errors.

[0076] The above are merely preferred embodiments of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions based on the principles of the present invention are within the scope of protection of the present invention. It should be noted that improvements and modifications that do not depart from the principles of the present invention are within the scope of protection of the present invention.

Claims

1. Low-code-based intelligent process construction and dynamic optimization method, characterized by: include: Select a process category based on the preset process classification; Generate an initial process template based on the process category, the initial process template is output in a visual manner, and the initial process template includes an initial task node, an initial process branch, and an initial gateway; The basic process is formed by visually dragging and dropping the initial task node, initial process branch, and initial gateway of the initial process template; Execute basic processes and record process execution data; Analyze the process execution data based on the preset rule statistics module and generate several optimization suggestions; The user selects an optimization plan based on the optimization suggestions and outputs the optimization process; The initial process template is adjusted based on a preset process adjustment strategy.

2. The low-code-based intelligent process construction and dynamic optimization method according to claim 1 is characterized in that: Equipped with rule optimization auxiliary modules, including: Construct a process operation database, which records historical process operations and corresponding process execution data. The historical process operations include task nodes, process branches, and gateways of each historical process operation. The process execution data includes task node execution time, processing data of node configuration personnel, selection frequency of process branches, gateway decision basis and execution results, total process time, number of rollbacks, and rollback opinions; Count and output abnormal links based on the process operation database. The abnormal links reflect the task nodes, process branches or gateways that have an impact on the execution results during the process operation; The user inputs several optimization goals and formulates several optimization plans based on abnormal links, and generates multiple optimization suggestions based on the preset rule base; Each optimization plan is manually scored, and the optimization plan with the highest score is output as the optimization suggestion.

3. The low-code-based intelligent process construction and dynamic optimization method according to claim 2 is characterized in that: An optimization solution verification strategy is configured and executed when the optimization process is actually running. The optimization solution verification strategy includes: Continuously collect actual operation data after the optimization process runs, Compare the actual operation data with the corresponding simulated operation data. If the actual operation data meets the expectations of the simulated operation data, the optimization plan and the actual operation data are written into the process operation database.

4. The low-code-based intelligent process construction and dynamic optimization method according to claim 2 is characterized in that: The process adjustment strategy includes: Get the user's historically selected optimization plan rating information, According to the user's scoring weights on different optimization goals, a comprehensive scoring index is formed. The task nodes and process path configurations in the initial process template are adjusted based on the comprehensive scoring index.

5. The low-code-based intelligent process construction and dynamic optimization method according to claim 1 is characterized in that: A rule-assisted construction step is configured, and the rule-assisted construction step is executed synchronously when the initial process template is visually dragged, including: Pre-extract structured process rules based on institutional documents to form a rule set; When users build a basic process by dragging and dropping, the current process structure is compared with the constraints in the rule set in real time; If it is detected that the current process structure does not meet the constraints in the rule set, a prompt message and corresponding correction suggestions will be output.

6. The low-code-based intelligent process construction and dynamic optimization method according to claim 5 is characterized in that: The configuration includes the following steps for initiating a pre-review process: Get the historical running process associated with the current running process, use the historical running process as the index to search in the process running database and the corresponding process historical execution data, Obtain high-frequency error points in the current running process based on the process historical execution data. The high-frequency error points are reflected in the links in the historical running process where the error frequency exceeds the frequency threshold. Output high-frequency error points and provide corresponding institutional regulations.

7. The low-code-based intelligent process construction and dynamic optimization method according to claim 5 is characterized in that: A form compliance verification step is configured, which is executed synchronously during the execution of the basic process, including: Automatically scan form data and check the form's integrity and format compliance based on preset field validation rules; Obtain historical approval form data, compare the current form content with the constraints in the preset rule library, and identify potential risk points. The potential risk points are specifically the current form that violates the corresponding system documents when combined with the historical approval form; Output risk warning information based on potential risk points.

8. The low-code-based intelligent process construction and dynamic optimization method according to claim 7 is characterized in that: The form compliance verification step is configured with a dynamic path adjustment strategy. When a potential risk point is detected, the corresponding task node is automatically added based on the specific content of the potential risk point and the real-time rule matching results.

9. The low-code-based intelligent process construction and dynamic optimization method according to claim 1 is characterized in that: The initial process template is configured with an initial template auxiliary generation strategy, including: Receive the process template requirement instruction input by the user, extract keywords from the process template requirement instruction and compare the similarity with the historical running process, call the initial process template corresponding to the historical running process with the highest similarity and output it.

Images