Intelligent approval rule modeling method for process automation

Abstract

The application discloses a process automation-oriented intelligent approval rule modeling method and relates to the technical field of business process management, and comprises the following steps: S100, in the process of constructing a rule candidate set, extracting the scene features, field semantic levels and participant role information of each historical approval data, generating a context semantic label set, and using the context semantic label set for subsequent rule difference modeling. The application enhances the semantic boundary recognition capability of the rule by introducing the alignment of the context semantic label and the ternary feature; constructs a rule feature matrix and a differentiation candidate set, realizes the accurate classification and processing of ambiguous rules; in combination with expression sensitivity reinforcement and simulation verification, identifies approval deviation and risks in advance; finally, realizes the dynamic optimization of the rule model through backtracking correction, improves the stability and accuracy of the rule model in multiple scenarios, and constructs a closed-loop reliable intelligent approval rule system.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of business process management, in particular to a process automation-oriented intelligent approval rule modeling method. BACKGROUND

[0002] The "process automation-oriented intelligent approval rule modeling" refers to structuring modeling and intelligent rule extraction of the approval process in an enterprise or organization by fusing process modeling technology and artificial intelligence methods, so as to realize automatic generation, optimization and application of the approval rules. In the traditional approval process, the approval rules often depend on manual setting, and the rules are fixed, the response is lagging and it is difficult to adapt to complex business changes. This method automatically identifies the approval conditions, judgment logic and constraint relationships by intelligently analyzing historical approval data, business process nodes, user operation behaviors and other information, and constructs a formal and executable rule model, so that it can be embedded in the process automation system to drive the automatic judgment and execution of the subsequent process, improve the efficiency, accuracy and flexibility of the approval, and realize the closed-loop intelligent process governance system from "rule design - execution - feedback - optimization".

[0003] The prior art has the following disadvantages:

[0004] In the process automation-oriented intelligent approval rule modeling process, it is usually necessary to extract rules and structure modeling based on a large amount of historical approval data to realize automatic induction and formal expression of approval conditions, judgment logic and execution path. However, in the actual modeling process, due to the existence of a large number of samples with similar structure, similar semantics or partially overlapping condition combinations in historical approval records, the current intelligent rule extraction algorithm often misjudges the rules with key detail differences as equivalent redundant items due to insufficient feature recognition granularity or limited context perception ability when performing redundant compression or semantic fusion processing, and then performs improper merging or deletion operations.

[0005] Such misjudgment problems mainly manifest as: ignoring the approval strategy differences brought by additional restrictions, boundary conditions or context features under the same main condition, thereby misaggregating multiple independent rules with fine-grained differences into a single logical path. Such erroneous modeling behavior will directly lead to the loss of differentiated constraints in the original approval rule system, and the decline of the approval path precision, especially in process scenarios with multi-role collaboration, sensitive element judgment or high-risk constraint conditions, which will cause serious consequences such as authority delegation, process bypassing or failure of risk control mechanisms, seriously affecting the execution accuracy of the intelligent approval system and the compliance stability of organizational governance.

[0006] For example, taking the enterprise fund approval process as an example, there are the following two rules in the historical data:

[0007] Rule A: When the purchase amount is greater than or equal to 1 million yuan, and the supplier is a domestic enterprise, the general manager needs to approve;

[0008] Rule B: When the purchase amount is greater than or equal to 1 million yuan, and the supplier is an overseas enterprise, the board of directors needs to approve.

[0009] The two rules are highly consistent in the main condition "amount ≥ 100 million", but there are essential differences in the additional condition "supplier type" and the corresponding approval level requirements. The existing rule modeling algorithm may only capture the common logic of the same amount threshold in the two rules in the extraction process, ignoring the additional constraint difference of "domestic / overseas", and then mistakenly merging it into a general rule: "amount ≥ 100 million needs to be approved by the general manager". This erroneous merger downgrades the overseas high-value procurement process that should be approved by the board of directors to the general manager level, resulting in a mismatch in the approval path and an overstepping of organizational responsibilities, which may cause serious compliance risks, foreign exchange capital loss, international supply chain management failure, and ineffective internal audit accountability, and other serious consequences.

[0010] The above information disclosed in the background section is only used to strengthen the understanding of the background of the present disclosure, and therefore it can include information that does not constitute prior art known to those of ordinary skill in the art. SUMMARY

[0011] The purpose of the present application is to provide an intelligent approval rule modeling method for process automation, which enhances the semantic boundary recognition ability of the rules by introducing context semantic tags and ternary feature alignment; constructs a rule feature matrix and a differentiated candidate set to realize accurate classification and processing of ambiguous rules; combines expression sensitivity enhancement and simulation verification to identify approval deviation and risks in advance; finally realizes dynamic optimization of the rule model through backtracking correction, improves its stability and accuracy in multiple scenarios, and builds a closed-loop reliable intelligent approval rule system to solve the problems in the above background technology.

[0012] In order to achieve the above purpose, the present application provides the following technical scheme: an intelligent approval rule modeling method for process automation, comprising the following steps:

[0013] S100, in the process of constructing the rule candidate set, the scene features, field semantic level and participant role information of each historical approval data are extracted, a context semantic tag set is generated, and is used for subsequent rule difference modeling;

[0014] S200, based on the context semantic tag set, the feature alignment of the rule expression is performed, the main condition field, the context additional attribute and the decision action are extracted, the rule feature matrix is constructed, and is used as the basic structure for rule comparison and screening;

[0015] S300, based on the rule feature matrix, identify rule combinations with the same main condition field but different expression methods, extract ambiguous additional attributes, and generate a rule differentiation candidate set;

[0016] S400, based on the rule differentiation candidate set, embed key additional attributes into the original rule structure, perform expression sensitivity enhancement processing, explicitly mark details that are easily ignored by compression algorithms, and generate an enhanced rule model set;

[0017] S500, based on the enhanced rule model set, perform multi-path simulation combined with historical approval process data, simulate process paths under different rule configurations, identify approval deviations and logical mismatches caused by rule expression differences, and output simulation comparison results and risk warning information;

[0018] S600, according to the simulation comparison results and risk warning information, determine the execution stability of the rule model in various context scenarios, start consistency backtracking and structure correction for detected deviations and conflicts, trace the context labels, adjust the rule expression and semantic label configuration, and complete the closed-loop optimization of the rule model.

[0019] Preferably, step S100 includes:

[0020] In the process of extracting each piece of historical approval data, the scene features, field semantic levels and participant role information corresponding to the approval behavior are collected;

[0021] Based on approval logs, business process definitions, organizational structures, business metadata and unstructured text, perform context data enhancement extraction to obtain timestamp, process stage, post level, business line and note semantic information;

[0022] Classify the collected fields according to semantic levels, and divide field types and expression levels according to the preset business logic knowledge base;

[0023] Generate a structured context semantic label set in the form of a "label dimension, label key, label value" triple, and bind it one-to-one with the historical approval data.

[0024] Preferably, step S200 includes:

[0025] Based on the context semantic label set, structurally analyze the logic expression of each rule, identify judgment fields, state fields and role fields;

[0026] Extract three types of key elements: main condition fields, context additional attributes and decision actions, and construct a unified triple expression format;

[0027] Align the three types of elements according to the preset field template to construct a rule feature matrix;

[0028] The rule feature matrix is normalized and coded and semantic weight is marked to improve semantic resolution in rule comparison.

[0029] Preferably, the step S300 comprises:

[0030] The semantic element recognition and structure segmentation processing are performed on each historical approval data, the main condition field, the context additional attribute and the decision action of the rule are identified based on the context semantic tag set;

[0031] The three types of key elements identified are mapped to the pre-defined field site, the rule feature matrix is constructed, and the missing field is processed with null value or default unconditional processing;

[0032] Expression normalization processing is performed on the rule feature matrix to standardize the expression form of the semantic equivalent field;

[0033] According to the role strength of the field in the decision, semantic weight adjustment is performed, the rule combination with the same main condition field but different context additional attributes is identified, the additional attribute with ambiguity is extracted, and a rule differentiation candidate set is generated.

[0034] Preferably, the step S400 comprises:

[0035] Based on the rule differentiation candidate set, the rule combination with the same main condition field but different semantic differences in the context additional attribute is identified, and the key additional attribute easily ignored by the compression algorithm is extracted;

[0036] The extracted key additional attribute is embedded into the corresponding original rule structure, the logic is kept unchanged and the expression content is explicitly expanded;

[0037] Expression sensitivity enhancement processing is performed on the updated rule structure to mark the semantic weight and saliency of the key detail element;

[0038] The processing results are summarized to generate an enhanced rule model set.

[0039] Preferably, the step S500 comprises:

[0040] The representative historical approval process data are screened, and the input field, process node sequence, historical decision path and context label of each data are extracted;

[0041] For the same approval input, the original rule model set and the enhanced rule model set are respectively applied to perform process simulation to deduce the respective approval trend and decision result;

[0042] The process path difference under two kinds of rule configurations is compared to identify the approval deviation and logic mismatch behavior, and whether there is authority decentralization, role overreach or key node omission is analyzed;

[0043] The simulation comparison report containing path difference, offset type and risk level is output, and corresponding risk warning information is generated.

[0044] Preferably, the step S600 comprises:

[0045] The approval offset path and logic conflict node in the simulation output are analyzed in combination to identify the problem type that repeatedly occurs in multiple context combination scenarios;

[0046] Based on the context semantic label and semantic weight label embedded in the rule model, the key field causing the offset and its context source are traced back to determine the expression defect or modeling deficiency reason;

[0047] The rule model with structural problems is modified, including supplementing context additional attributes, optimizing field nested structure, and refining semantic label content;

[0048] The modified rule model is simulated and verified again to confirm that it has semantic stability and execution consistency under the same context condition.

[0049] In the above technical solution, the technical effects and advantages provided by the present application are as follows:

[0050] The present application introduces the context semantic label mechanism and ternary structure feature alignment strategy to ensure that the rule has clear semantic boundary and structure expression in the early modeling stage, significantly enhancing the identification ability of fine-grained differences. By constructing the rule feature matrix and rule differentiation candidate set, the rule combination with expression ambiguity and context conflict is further accurately classified and processed, avoiding the situation that semantic overlap is misjudged as equivalent rules. Combined with the expression sensitivity enhancement mechanism and simulation verification process, potential path offset and decision conflict can be identified before rule execution, and approval risk can be warned in advance. Finally, through the consistency backtracking and structure modification process, the dynamic self-optimization of the rule model is realized, ensuring its execution stability and expression accuracy in different context scenarios. Overall, this scheme not only improves the accuracy and reliability of rule modeling, but also builds a closed-loop system for rule generation, verification and optimization, providing a highly controllable and reliable rule basis for intelligent decision-making in the process automation environment. BRIEF DESCRIPTION OF DRAWINGS

[0051] In order to more clearly illustrate the technical solutions in the embodiments or prior art, the drawings needed in the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments described in the present application, and other drawings can also be obtained by those skilled in the art based on these drawings.

[0052] Figure 1 The method flowchart of the intelligent approval rule modeling method for process automation of the present application. Detailed Implementation

[0053] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the description of this disclosure will be more complete and fully convey the concept of the exemplary embodiments to those skilled in the art.

[0054] This invention provides, for example Figure 1 The intelligent approval rule modeling method for process automation shown includes the following steps:

[0055] S100, during the process of constructing the rule candidate set, when extracting each piece of historical approval data, the scene features, field semantic level and participating role information corresponding to the data are recorded, and the corresponding context semantic label set is generated for subsequent structured modeling of rule differences;

[0056] To achieve comprehensive semantic feature perception of historical approval data and avoid misjudgments in rule merging due to missing contextual information during rule modeling, a method for generating a set of contextual semantic tags is proposed. This method is used to construct a candidate set of rules while simultaneously enhancing the context of each historical approval data entry. Specifically, it includes the following steps:

[0057] While collecting historical approval record data, enhanced contextual data extraction is performed. Specifically, when parsing each approval record, not only are its core logical fields (such as conditional parameters, judgment values, and operational actions) extracted, but contextual data directly related to the approval behavior is also extracted simultaneously. This includes, but is not limited to, the business line identifier to which the approval business belongs, the current stage label of the approval process, the business timestamp of the data generation, the organizational level of the initiator and approver, and the region or area identifier where the business was initiated. This method ensures that each approval data entry not only contains executable logical elements but also carries contextual semantic clues that reflect the actual application background. The key to this step is achieving a two-way binding between approval behavior data and its contextual information, ensuring reproducibility and contextual reproducibility during subsequent rule extraction.

[0058] The contextual data enhancement extraction can be implemented in various ways, including: extracting the timestamp, geographic location and device information when the approval behavior occurs based on the log system; obtaining the process stage and link type to which the node belongs in combination with the business process definition file; parsing the post level and responsibility range of the initiator and approver from the organizational structure; extracting the business line, approval matter classification and applicable rule version, etc. from the business metadata system; and using natural language processing methods to perform semantic analysis on unstructured text such as approval notes and operation instructions to identify potential contextual elements. These methods can enhance the contextual expression of the approval data from multiple dimensions and improve the accuracy and discrimination ability of rule modeling.

[0059] On the basis of completing the extraction of contextual information, the fields involved in the approval data are subjected to semantic level classification processing. According to the preset business logic knowledge base, the fields are subjected to semantic type division (such as numerical judgment field, enumeration type state field, role identity field, etc.), and are grouped and identified according to the semantic accuracy level. For example, “amount” is a first-level quantitative field, “amount>100 million yuan” is a second-level threshold expression field, and “amount>100 million yuan and overseas supplier” is a third-level semantic linkage field. This semantic level division helps to accurately identify the expression granularity difference between rules subsequently, especially in the case of slight differences in field combination expression, the level mark can significantly improve the accuracy of semantic recognition, and avoid mistaking the fine-grained difference as expression redundancy.

[0060] Based on the extracted contextual information and semantic level classification results, a unified format contextual semantic tag set is constructed for each historical approval data. The set is organized in a structured manner, and each tag is composed of a “tag dimension, tag key, tag value” triple, wherein the tag dimension includes scene dimension (such as business line, time period), field dimension (such as field level, field role type), role dimension (such as initiator role, approval role) and the like. For example, the contextual semantic tag set in a piece of approval data can include the following entries: “scene dimension: business line = overseas procurement”, “field dimension: amount field level = threshold judgment”, “role dimension: approval person identity = general manager” and the like. The structured tag set not only retains the original decision information of the approval data, but also embeds the semantic background of the decision formation, so that the subsequent rule extraction can be based on semantic reference to perform difference modeling and comparative analysis.

[0061] Each piece of historical approval data and its generated context semantic tag set are bound one by one, and are uniformly stored and managed as the basic unit of the rule candidate set. On this basis, subsequent modeling steps such as rule feature comparison, difference identification and combination optimization can call the corresponding semantic tag set as an auxiliary information source to determine whether there is an actual context difference between the rules, thereby avoiding the decision path deviation caused by relying only on the main condition field for merging processing. In addition, the context semantic tag set also supports dynamic expansion and feedback optimization. For example, when rule mismatch problems are found in simulation verification, the existing tag set can be updated in reverse to further improve the comprehensiveness and discrimination of the tag representation, forming a semantic-aware modeling mechanism with evolution ability.

[0062] In the process of constructing the rule candidate set, when extracting each piece of historical approval data, the scene features, field semantic levels and participating role information associated with the data are recorded, and the corresponding context semantic tag set is generated. Its core role is to provide fine-grained and structured context support for subsequent rule modeling. Traditional rule extraction methods focus more on the explicit logic of condition fields and result actions, ignoring the real business environment, process stage and organizational role differences behind the approval behavior, which can easily cause mismerging of structurally similar but semantically different rules. By introducing context semantic tags, a reproducible and analyzable context reference system can be established for each approval behavior, so that subsequent judgment of whether two rules have substantial differences can no longer rely only on value comparison at the field level, but can combine their business application scenarios, semantic expression levels and role participation relationships for multi-dimensional comparison. This tagged context modeling approach improves the semantic granularity of rule expression and enhances the discriminability of rule merging and clustering processing, fundamentally avoiding the problem of rule abstraction distortion caused by lack of context awareness, ensuring that the approval rules constructed in the process automation scenario are executable and retain necessary scene constraints and policy differences, significantly improving the accuracy and stability of rule modeling.

[0063] S200, based on the generated context semantic tag set, performing feature alignment processing on the rule expression, extracting three types of key elements of the main condition field, the context additional attribute and the decision action in each rule, constructing a rule feature matrix as a unified basic structure for rule comparison and combination screening;

[0064] To solve the problem of inconsistent rule expression structure, inconsistent semantic granularity, and difficulty in accurate difference judgment and combination screening in the prior art, a rule expression feature alignment processing method based on a context semantic tag set is proposed. By extracting three types of key elements of the main condition field, the context additional attribute and the decision action, a unified format rule feature matrix is constructed for subsequent rule comparison, aggregation and difference analysis. The method specifically includes the following steps:

[0065] Based on the set of context semantic labels generated for each piece of historical approval data, the logical expression contained in the rule is structurally parsed. During the parsing process, all explicit condition fields in the rule are identified and categorized according to a predefined field classification dictionary. The identification includes judging fields (such as amount, time, project level, etc.), status fields (such as supplier type, approval priority, etc.), role fields (such as approver job level, initiator department, etc.), and the like. Through this parsing process, the basic classification of all semantic constituent elements in the rule is established, laying a logical foundation for subsequent feature element division.

[0066] After completing the field classification and parsing, the rule expression is extracted into three core feature elements. Specifically, the fields reflecting the pre-trigger logic of the rule are extracted as main condition fields, which usually include core numerical conditions or Boolean conditions for triggering judgment; the conditions that accompany the main conditions and are used to refine or limit the triggering scenarios are defined as context additional attributes, such as role requirements, regional restrictions, supplier background, etc.; the approval behavior, flow action, or result output determined by the rule is classified as decision action. Through this element classification and extraction method, a structured semantic skeleton can be constructed from the original rule text, so that each type of rule is represented in a unified ternary expression format, effectively improving the comparison and screening efficiency between rules.

[0067] The "unified ternary expression format" refers to the representation of each rule as a structured expression consisting of main condition fields, context additional attributes, and decision actions. The core of this format is that, regardless of the complexity or differences in the original approval rule expression, it must be abstracted and summarized into these three core elements: the main condition field is used to describe the main judgment logic on which the rule is based (such as whether the amount exceeds a certain threshold); the context additional attribute is used to capture the scene semantics that limit the scope of the rule (such as supplier type, approver position, business region, etc.); the decision action is used to explicitly define the execution behavior after the rule is triggered (such as being approved by a certain role or transferred to a certain node). The use of the unified ternary expression format provides a unified semantic structure for subsequent alignment, comparison, clustering, and difference detection between rules, enabling the rules to be transformed from "uncertain expressions at the natural language level" to "standard entities at the structured semantic level", thereby achieving efficient rule modeling, similarity calculation, and mis-merging avoidance. It not only improves the automation and comparability of rule modeling, but also significantly enhances the model's ability to perceive semantic differences, which is a key foundation for building a high-precision, context-sensitive approval rule system.

[0068] After the extraction of the three key elements, the main condition field, the context additional attribute, and the decision action are aligned according to the preset structure template to construct a rule feature matrix. The feature matrix is a two-dimensional table structure, each row corresponds to a rule, and each column corresponds to a standardized field site, including the numerical judgment area, the enumeration state area, the semantic constraint area, and the execution action area, etc. In the matrix construction process, if a rule is missing a certain type of field, use "null" or "undefined" to occupy the place to maintain the consistency of the matrix structure. The structured matrix not only improves the comparison efficiency of rule data, but also supports subsequent vectorization processing, similarity calculation, and combined rule screening, and is the basic structure for realizing rule semantic clustering and difference detection.

[0069] After the feature matrix is completed, the three types of elements of each rule are normalized and coded and the semantic weight is labeled to improve the response ability of the rule to fine-grained differences in the comparison process. For example, for the same field "amount", if the judgment values of two rules are similar but not completely consistent, the distance can be quantified after normalization to assist in identifying "approximately but not equivalent" judgment logic. For the details in the context additional attribute that are easily overlooked, such as approval role and project level, the weighted labeling improves their discrimination weight in rule comparison to prevent them from being incorrectly merged. Finally, the constructed rule feature matrix not only has good readability and consistency, but also has efficient comparability, providing a unified structural basis and semantic guarantee for subsequent rule clustering, ambiguous rule screening, and combined judgment operations.

[0070] The main role of this step is to realize the structured reconstruction and semantic alignment of historical approval rules, so as to provide a unified expression basis for accurate comparison, semantic difference identification and combined screening between rules. In the actual process automation scene, the approval rules are often derived from natural language description, manual setting or semi-structured records. These rules have high heterogeneity in expression, field structure and semantic granularity. If direct comparison or combination is performed, it is easy to cause misidentification, misaggregation and other problems due to inconsistent structure or ambiguous semantics. By performing feature alignment processing of rule expression based on the context semantic tag set, three types of core semantic elements can be accurately extracted from each rule: the main condition field is used to mark the core logic basis for triggering the rule; the context additional attribute is used to describe the limited environment and business context applicable to the rule; and the decision action defines the execution path and processing behavior after the rule is triggered. These three elements constitute the skeleton of the rule semantics. By uniformly extracting and mapping them into a rule feature matrix, the standardized representation of rules in structure and the accurate comparison of rules in semantics can be realized, avoiding the problems of covering actual differences due to expression differences or misjudging equivalence due to structural coincidence. As the underlying structure for subsequent rule clustering, difference identification, and optimization, the feature matrix not only improves the accuracy and operability of rule modeling, but also enhances the ability to identify fine-grained semantic differences. It is a key step to realize the improvement of intelligent approval rule modeling accuracy and risk control.

[0071] S300, based on the constructed rule feature matrix, identifying a rule combination with semantic overlap but different expression under the premise of the same main condition field, extracting the ambiguous additional attributes in the rule combination, generating a rule differentiation candidate set for distinguishing rules with potential decision logic differences;

[0072] To solve the problems of comparison difficulty, miscombination and semantic ambiguity caused by non-uniform rule expression form and mixed semantic elements in existing rule modeling, a rule expression feature alignment processing method based on a set of context semantic tags is proposed. By standardizing the structure of rule expression and aligning it, the key semantic elements in the rule are extracted as three types of core components, and a rule feature matrix with clear structure and consistent semantics is constructed as the basic structure for rule comparison and combination screening. The method includes the following steps:

[0073] Based on the set of context semantic labels generated from each historical approval data, the rule text is processed for semantic element recognition and structural segmentation. The core of this step is to divide the expression content in the rule into identifiable and classifiable independent elements through semantic parsing strategy. Among them, the preset semantic template and context label matching mechanism are used to anchor the semantic of the condition judgment, business constraint, trigger statement and behavior action in the rule text. For example, for the rule "when the procurement amount is greater than or equal to 100 million yuan and the supplier is an overseas enterprise, the board of directors will approve", through the guidance of context label, "amount greater than or equal to 100 million yuan" is identified as the main condition, "supplier is an overseas enterprise" is identified as the additional constraint, and "board of directors will approve" is identified as the decision action. This step realizes the first conversion from free expression text to structured semantic unit, and lays the semantic foundation for subsequent standardized expression.

[0074] On the basis of completing semantic segmentation, key element extraction operation is performed to classify the semantic fragments in each rule into three types of core components: main condition field, context additional attribute and decision action. Among them, the main condition field refers to the core judgment condition that triggers the rule to take effect, which is usually a quantitative or qualitative logical judgment such as amount interval, time range, approval category, etc.; the context additional attribute includes semantic elements closely related to the business context, such as applicable organization range, participant identity, regional label, risk level, etc., to depict the limited scenario of rule application; the decision action is the specific behavior after the rule is triggered, including transfer object, approval method, processing path, etc. The extraction of the three types of elements depends on the semantic anchor point and label correspondence table generated in the previous step to ensure that all extracted content has clear semantic boundaries and business meaning.

[0075] After completing the extraction of the three types of elements, in order to ensure the consistency of semantic alignment between different rules, a pre-defined field site is used to build a unified structure of rule feature matrix. The matrix takes rule as row and standard field position as column, respectively corresponding to the main condition field set, the context additional attribute set and the decision action set. Each field site defines clear semantic dimension and standardized expression format, such as "amount judgment field", "regional label field", "approval post field", "final approval path field", etc. In the rule mapping process, if a rule is missing a type of field, "null" or "default unconditional" is used for placeholder processing to maintain the integrity and alignment consistency of the matrix structure. The matrix not only supports structured storage, but also facilitates semantic similarity calculation, feature weight analysis and rule aggregation processing, which is the core basis for efficient and controllable rule management.

[0076] To enhance the sensitivity and accuracy of the rule feature matrix to semantic differences, uniform expression normalization and semantic weight adjustment processing are performed on the constructed matrix. Expression normalization is mainly used to standardize fields with equivalent semantics but different expressions, such as normalizing "amount greater than or equal to 1 million" and "amount greater than or equal to 1 million" to the standard field "amount greater than or equal to 1 million"; semantic weight adjustment assigns different weights to the fields according to their influence on the actual decision-making, such as setting a higher recognition weight for the "supplier attribute" field in high-risk approval to avoid being misjudged as irrelevant conditions in subsequent rule screening. Through this step, the rule feature matrix not only has the ability to align the form, but also has the ability to respond sensitively to semantic changes, significantly improving the accuracy and recognition ability of rule comparison.

[0077] The role of this step is to provide a unified expression framework for the structured modeling and semantic accurate comparison of approval rules, thereby effectively solving the inconsistency problems of existing approval rules in expression format, field level and context information. Since historical approval rules often come from manual configuration, free text description or multi-source heterogeneous systems, these rules may have similar intentions in logical expression, but show obvious differences in structure, making it difficult to directly judge or identify differences. To solve this problem, this step is based on the generated context semantic tag set to deeply analyze and align the rule expression, and clearly divide and extract three key elements: main condition field, context additional attribute and decision action. The main condition field reflects the core basis for the logical establishment of the rule and is the main axis for judging the rule trigger; the context additional attribute is used to describe the restrictive environment applicable to the rule, covering information such as process node, role identity, regional identifier, business scenario, etc.; the decision action defines the operation path to be performed after the rule is triggered, such as approval role, flow direction, processing method, etc. Mapping these three elements into the rule feature matrix achieves the consistency of rules in the structure level and the explicitness in the semantic level, so that rules with different expression forms can be stored and compared in a homogeneous way in the standard matrix. This processing method not only improves the accuracy of rule comparison, reduces the risk of misjudgment due to structural inconsistency or semantic invisibility, but also provides a foundation for subsequent rule screening, combination optimization, clustering and merging, and ambiguity avoidance. It is an indispensable key step in the transition from data to model in intelligent approval rule modeling.

[0078] S400, based on the rule differentiation candidate set, embedding the extracted key additional attributes into the corresponding original rule structure, performing expression sensitivity enhancement processing on the updated rule structure, explicitly labeling the detailed elements that are easily ignored by the rule compression algorithm in the rule model, and generating an enhanced rule model set;

[0079] To solve the problem of fine-grained semantic misjudgment in rule extraction, especially the situation that additional conditions are not complete or the semantic weight is low and compressed algorithm misrecognizes it as redundant, a rule enhancement method based on rule differentiation candidate set is proposed. This method re-embeds the key additional attributes into the rule structure and performs expression sensitivity enhancement processing, so that the details that are easy to be ignored are explicitly presented in the rule structure, thus forming an enhanced rule model set, ensuring the completeness of rule semantics and the accuracy of decision expression. It includes the following steps:

[0080] Based on the rule differentiation candidate set constructed in the early stage, identify each group of rule combinations with the same main condition field but different context attributes, and extract the key additional attributes that are easy to be ignored by the merging algorithm. This step uses the previously constructed rule feature matrix to locate rules with the same main condition field but different context attributes or decision actions, and pays special attention to the additional elements with weakened semantic expression, ambiguous field content or low weight, such as role range restriction, geographical application label, organizational level boundary, etc. Through semantic distance calculation and semantic conflict labeling mechanism, these potential details that are easy to be lost are extracted to form the additional attribute set associated with the original rule.

[0081] Embed the extracted key additional attributes into the corresponding original rule structure to form the updated rule expression. The embedding process needs to keep the original logical framework of the rule unchanged. Under the premise of keeping the main condition field unchanged, the additional attributes are explicitly presented in the rule logic expression. Specific methods include condition statement level nesting, constraint field insertion or semantic annotation extension, etc. to revise the structure, so that the rule remains consistent in logical structure, but is more clear, specific and complete in expression level. For example, the original rule only states "amount ≥ 100 million yuan", and the revised rule adds "and the supplier is an overseas enterprise, and the contract period exceeds 12 months" as explicit additional conditions.

[0082] Perform expression sensitivity enhancement processing on the updated rule structure to improve the retention ability of key details in the extraction and compression stages. This processing marks the additional attributes that are easy to be ignored in the rule by adding semantic weight prompts, field saliency labels and context prompt labels, etc. in the rule expression, to strengthen their presence and incompressibility in the rule modeling algorithm. For example, add weight labels to conditions such as "the approver must be a board member" to avoid being down-weighted in the parsing or aggregation process; or explicitly classify expressions such as "limited to overseas procurement processes" to prevent them from being misrecognized as similar conditions in domestic scene rules. This processing mechanism ensures that the key semantics of the rule remain stable during multiple rounds of modeling, reconstruction and compression.

[0083] The enhanced rule model set is generated by aggregating the rule set that has been embedded with key attributes and subjected to expression sensitivity enhancement processing. The set retains the main logical structure in the original rules and enhances the context restoration ability, comparison discrimination ability and compression anti-interference ability of the rules through explicit presentation of additional attributes and strengthening of semantic expression. The enhanced rule model not only provides higher discrimination accuracy in subsequent process simulation, path simulation and difference detection, but also significantly improves the adaptability and reliability of the rules in dynamic business scenarios. At the same time, the set can be used as input for the rule feedback optimization link to achieve stable inheritance and accurate reconstruction of rule expression in multiple rounds of strategy adjustment.

[0084] The purpose of this step is to solve the problem of insufficient semantic detail recognition in the rule extraction and compression process. By re-embedding key additional attributes into the original rule structure and performing sensitivity enhancement processing on the rule expression, an enhanced rule model set with more complete expression and clearer semantics is generated. In existing rule modeling practices, automatic extraction algorithms often aggregate and compress based on the commonality of condition fields, lacking awareness of semantic micro-differences and contextual additional restrictions. Especially in cases where the expression form is not uniform, the additional condition boundary is fuzzy, or the semantic weight is low, it is easy to misjudge rules with essential differences as redundant and merge them into a single rule, leading to path mismatch in the approval process and risk control failure. This step reconstructs the original rules by identifying key additional attributes in the rule differentiation candidate set, embedding the hidden or marginalized additional restrictions into the original rule expression in an explicit and standardized structured form. At the same time, an expression sensitivity enhancement mechanism is introduced to make the details that are easily ignored by modeling algorithms have a higher presence and non-compressibility in rule expression, thereby improving the semantic stability and recognition accuracy of rule models in subsequent comparison, clustering and simulation. The enhanced rule model set constructed finally has higher expression rigor and difference retention ability, which can effectively avoid false judgments and decision deviations caused by rule abstraction distortion in the process automation system, and is a key step to achieve accurate rule modeling and reliable execution of risk control logic.

[0085] S500, based on the enhanced rule model set, multi-path process simulation is performed in combination with historical approval process data to simulate process decision paths under different rule configurations, identify approval deviation and logic mismatch behaviors caused by rule expression differences, and output simulation comparison results and corresponding risk warning information;

[0086] To further verify the decision accuracy and behavior consistency of the constructed enhanced rule model set in actual process application, a multi-path process simulation method based on historical approval process data is proposed. By simulating the process flow generated by different rule configuration schemes under the same process input conditions, the approval deviation and logic mismatch problems caused by rule expression differences are identified, and simulation comparison results and risk warning information are generated, providing a basis for subsequent rule correction and model optimization. The method includes the following steps:

[0087] Prepare the input data set required for multi-path simulation. This step is based on the archived historical approval process data for screening, selecting representative approval samples as simulation inputs. These samples need to cover various business types, role structures, amount intervals, and geographic regions, etc. key dimensions to ensure the comprehensiveness and coverage of the simulation results. For each sample process data, extract its original input fields, process node sequence, historical decision path, and corresponding context label information as the basis for subsequent simulation. The goal of this step is to build a real and comparable data environment, so that the enhanced rule model can accurately receive diversified approval input conditions in the simulation process.

[0088] Representative approval samples refer to historical approval records that can reflect key business scenarios, diversified condition combinations, and typical decision patterns in the approval process. Such samples usually include: high-amount or high-risk transaction approval processes, approval processes involving cross-department or multi-role coordination, processes containing special approval conditions (such as regional restrictions, supplier classification, contract duration, etc.), and processes that have experienced approval conflicts, abnormal paths, or rule adjustment records. These samples cover a wide range of businesses, have high condition complexity, and have different decision paths, which help to comprehensively test the adaptability and accuracy of the rule model in real business in the simulation, and are the key input basis for evaluating the effectiveness of rule modeling.

[0089] In the simulation process, two rule configuration schemes are applied to simulate the process for the same set of historical approval inputs: one is the original rule model set before construction, and the other is the enhanced rule model set containing expression reinforcement. In the execution process, according to the structure content, condition judgment, and decision path of the two rule sets, the approval path of each process under the two rule configurations is derived, including the number of approval loops, changes in participating roles, final decision results, and time used, etc. By comparing the process results of the same input under different rule driving, the process path changes caused by rule expression differences are identified, especially focusing on the approval decision change points caused by additional condition expression reinforcement.

[0090] The approval deviation identification and logic mismatch analysis are performed on the process path differences found in the simulation process. The deviation identification mainly focuses on the level change of the approval node, the adjustment of the approval person's identity, and the process jump phenomenon; the logic mismatch analysis mainly evaluates whether there are phenomena such as authority delegation, role overreach, repeated approval, or missing of key approval nodes caused by incomplete rule expression. For example, in the original rule model, a procurement process with an amount exceeding 100 million yuan is approved by the general manager without distinguishing the supplier type, while in the enhanced rule model, it is approved by the board of directors according to the additional condition “the supplier is an overseas enterprise”, then the path difference is considered as a key approval deviation point. This analysis process combines context labels to assist in determining whether the deviation behavior constitutes a process logic risk.

[0091] The approval deviations and logic mismatch behaviors found in the above simulation comparison are sorted and archived to generate a simulation comparison report containing process number, input condition, approval path difference under two types of rule configuration, deviation type, potential risk level, and other elements, and output corresponding risk warning information. The risk warning information can not only be used to prompt business personnel to audit the rationality of the configuration of the relevant process rules, but also can be fed back to the rule modeling mechanism to trigger further optimization and precision revision of the corresponding enhanced rule model. This step realizes the verification closed loop after rule modeling, so that the modeling, expression, simulation, and correction of the approval process form a dynamic linkage, and improves the business adaptability, semantic consistency, and application reliability of the rule model.

[0092] The role of this step is to verify the behavior of the enhanced rule model in the actual process through multi-path flow simulation, so as to identify the approval deviation and logic mismatch problems that may be caused by differences in rule expression, and ensure the consistency, accuracy and compliance of the approval rules at the execution level. Although the completeness of the rules has been improved through context semantic annotation and expression enhancement during the rule modeling process, the actual running effect still needs to be verified in the process scenario because the rule modeling itself is still an abstract semantic structure. This step matches the enhanced rule model set with the real historical approval process data, simulates the approval path under different rule configurations under the same input conditions, and compares the differences in the execution order, participating roles, processing time and path changes of the approval nodes under the original rule model and the enhanced model. Through comparison and analysis, the specific influence of the expression changes after the strengthening of additional conditions, context semantics and other factors on the approval logic can be identified, especially the positioning of high-risk behaviors such as delegation of authority, process bypass, repeated processing or skipping of key nodes caused by improper rule expression. In addition, this step also outputs simulation comparison reports and risk warning information, clearly indicating the process number, trigger condition, risk level and suggested handling method of the existing deviation, providing a basis for subsequent rule optimization and process adjustment. Therefore, this step not only provides a verification closed-loop mechanism for the rule model, but also ensures the logical accuracy and governance specification of the process through simulation means, which is a key link to realize the effective conversion of rules from modeling to application.

[0093] S600, according to the simulation comparison results and risk warning information, jointly determine the execution stability of the rule model in various context scenarios, and start the consistency backtracking and structure correction process for the detected path deviation and decision conflict, trace the context label corresponding to the risk point, adjust the rule structure expression and semantic label configuration, ensure the decision consistency and expression accuracy of the rule model in the real process, and complete the dynamic closed-loop optimization of the rule modeling process;

[0094] To ensure that the approval logic based on the enhanced rule model has high consistency and semantic expression accuracy in diversified context scenarios, a rule consistency backtracking and structure correction method based on simulation comparison results and risk warning information is proposed. It can be used to analyze, trace, adjust and update the structure and expression of rules for approval deviation and decision conflict problems, so as to realize the dynamic closed-loop optimization of the rule model from extraction to application. The steps include:

[0095] Based on the comparison results and risk warning information output by the multi-path process simulation, the execution performance of the rule model under different context conditions is jointly judged. This step comprehensively analyzes the approval deviation paths, logical conflict nodes, decision imbalance behaviors and their corresponding trigger conditions existing in each simulation process, and determines whether the problem has stability characteristics, i.e., whether it occurs repeatedly in multiple context combination scenarios, through statistical and clustering analysis. If a certain type of deviation problem occurs in different business types, regional tags, and approval role structures, it is determined that the rule has expression defects or context modeling deficiencies, and needs to be included in the consistency backtracking process.

[0096] The rule model that has been determined to be inconsistent or have semantic conflicts is subjected to backtracking processing. This processing is based on the context semantic tags and semantic weight identification information embedded in the enhanced rule model to trace the key fields that cause problems and their context sources. For example, when the problem of "approval role level deviation" occurs, by analyzing the semantic tags corresponding to the "role limitation" field in the rule model, it is identified whether the algorithm ignores this element due to missing original tags, inaccurate attribute division, or low weight configuration. At the same time, other rules with similar expressions in the related rule group are compared horizontally to determine whether the deviation is caused by structural problems such as expression misalignment or inconsistent additional conditions.

[0097] Based on the traceability results, structural correction processing is performed on the rule model with problems. Structural correction includes: first, supplementing missing context additional attributes and incorporating them into the rule structure expression in an explicit manner to ensure expression completeness; second, adjusting the nesting structure of the main condition field and additional attributes to optimize the judgment order and combination of rule logic and improve the discrimination ability of the decision path; third, updating and refining the semantic tag set to enhance its recognition and non-ignorability in the modeling algorithm. For example, the original tag labeled "approval role = high-level manager" is refined to "approval role = general manager" or "approval role = board member" to avoid recognition errors caused by ambiguous expression. All structural correction actions should maintain the original meaning of the rule unchanged, only enhancing the semantic recognizability and context consistency of the expression.

[0098] The corrected rule model is re-integrated into the enhanced rule model set and matched with the original process sample for simulation verification again to confirm that the corrected model no longer produces approval deviation or logical conflict in the same context scenario. If there are still some abnormal processes, the backtracking and iterative correction can continue until the rule model remains semantically stable, path accurate and execution consistent in multiple scenario simulations. This step ensures that the rule modeling process has dynamic feedback capability and self-adaptive optimization mechanism, so that the rule structure not only meets the formal logical correctness, but also meets the real business decision needs, thereby realizing the closed-loop self-correction and continuous evolution of intelligent approval rule models in actual processes.

[0099] The role of this step is to build a dynamic closed-loop feedback mechanism in the rule modeling process, ensuring that the rule model has stable, consistent, and reliable execution capabilities under different context conditions, and can locate the risk source and correct the expression defects in a timely manner when problems are found, thereby improving the semantic accuracy and decision consistency of the rule model in real business processes. In the approval process modeling, even if the context perception and expression enhancement has formed a relatively complete rule structure, it may still lead to rule mismatch, path deviation, approval result conflict, and other problems in some specific scenarios due to semantic ambiguity, context conflict, or expression boundary ambiguity. Based on the simulation comparison results and the generated risk warning information of the previous stage, this step comprehensively analyzes the behavior performance of the rule model under diversified context scenarios, and judges whether the model has stability defects by comparing whether the decision path is stable, the approval role is consistent, and the trigger condition is accurate. Once deviation paths or conflict behaviors are found, the consistency backtracking and structure correction process is started, the context semantic tags corresponding to the risk rules are extracted from the model, and whether it is caused by tag missing, attribute ambiguity, or semantic coverage error is confirmed through tag tracing. Further, by adjusting the rule structure expression, reconstructing the logic order, and optimizing the semantic tag configuration, the rules are more suitable for their real application situation and can prevent algorithm errors from aggregating or misidentifying in structure. Finally, the corrected rules are re-verified to form a closed-loop optimization chain, ensuring the continuous evolution and robustness control of the rules from construction, verification to application process. This mechanism not only improves the adaptability of the model to complex business differences, but also strengthens the behavior stability of the rule execution stage, which is a key step to ensure the logical correctness, compliance, and controllability of the intelligent approval system.

[0100] By the intelligent approval rule modeling method for process automation, the problems of rule mismerging and approval logic mismatch caused by low semantic recognition granularity and lack of context awareness in existing rule extraction technologies can be effectively solved. The method introduces a context semantic tag mechanism and a triple structure feature alignment strategy to ensure that the rule has a clear semantic boundary and structural expression at the initial modeling stage, significantly enhancing the recognition ability of fine-grained differences. By constructing a rule feature matrix and a rule differentiation candidate set, the rule combination of expression ambiguity and context conflict is further accurately classified and processed, avoiding the situation of misjudging as equivalent rules due to semantic overlap. Combined with the expression sensitivity enhancement mechanism and the simulation verification process, potential path deviation and decision conflict can be identified before rule execution, and approval risks can be warned in advance. Finally, through the consistency backtracking and structure modification process, the dynamic self-optimization of the rule model is realized, ensuring its execution stability and expression accuracy in different context scenarios. Overall, the scheme not only improves the accuracy and reliability of rule modeling, but also builds a closed-loop system of rule generation, verification and optimization, providing a highly controllable and reliable rule basis for intelligent decision-making in the process automation environment.

[0101] The above formulas are dimensionless numerical calculations, and the formulas are obtained by software simulation of a large amount of data to obtain a formula of the latest real situation. The preset parameters in the formula are set by a person skilled in the art according to the actual situation.

[0102] The above only describes certain exemplary embodiments of the present application by way of illustration, and it is self-evident that those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present application. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present application.

[0103] It should be noted that in this text, if there are relationship terms such as first and second, etc., they are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between the entities or operations. Moreover, the terms "include", "contain" or any other variants thereof are intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method, article or device. Without more limitations, the element defined by the sentence "includes a" does not exclude the presence of another identical element in the process, method, article or device including the element.

[0104] It should be understood that the size of the sequence number of the above processes does not mean the order of execution in various embodiments of the present application, and the execution order of the processes should be determined according to its function and inherent logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0105] Those skilled in the art can realize that the units and algorithm steps of the examples described in combination with the embodiments disclosed herein can be realized in electronic hardware or a combination of computer software and electronic hardware. Whether the functions are realized in hardware or software depends on the specific application and design constraints of the technical solution. The skilled person can use different methods to realize the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0106] Those skilled in the art can clearly understand that, for the convenience and brevity of the description, the specific working process of the system, device and unit described above can refer to the corresponding process in the foregoing method embodiments, which will not be repeated here.

[0107] The units described as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, i.e. they can be located in one place or distributed on multiple network units. Part or all of the units can be selected to achieve the purpose of the embodiment according to actual needs.

[0108] In addition, each functional unit in each embodiment of the present application can be integrated into one processing unit, or each unit can exist physically, or two or more units can be integrated into one unit.

[0109] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto, and any skilled person in the art can easily think of changes or replacements within the technical scope disclosed in the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

[0110] The above only describes some exemplary embodiments of the present application by way of illustration, and it is needless to say that those skilled in the art can modify the described embodiments in various ways without deviating from the spirit and scope of the present application. Therefore, the above figures and descriptions are illustrative in nature and should not be understood as limiting the scope of protection of the claims of the present application.

Claims

1. A method for modeling intelligent approval rules for process automation, characterized in that, Includes the following steps: S100, in the process of constructing the rule candidate set, extract the scene features, field semantic hierarchy and participating role information of each historical approval data to generate a set of context semantic tags for subsequent rule difference modeling; S200, based on the contextual semantic label set, performs feature alignment on rule expression, extracts main condition fields, contextual additional attributes and decision actions, and constructs a rule feature matrix as the basic structure for rule comparison and filtering; include: Based on the contextual semantic tag set, the logical expression of each rule is structured and parsed to identify judgment fields, status fields and role fields; Extract three key elements: main condition field, contextual additional attributes, and decision action, and construct a unified ternary expression format; The three key elements are aligned according to a preset field template to construct a rule feature matrix; Normalize the feature matrix of the rules and label it with semantic weights to improve the semantic discrimination ability in rule comparison; S300, based on the rule feature matrix, identifies rule combinations with the same main condition field but different expression methods, extracts ambiguous additional attributes, and generates a rule differentiation candidate set; S400, based on the rule-differentiated candidate set, embeds key additional attributes into the original rule structure, performs expression-sensitive enhancement processing, explicitly labels detailed elements that are easily ignored by compression algorithms, and generates an enhanced rule model set; S500, based on an enhanced set of rule models, performs multi-path simulations by combining historical approval process data. It simulates process paths under different rule configurations, identifies approval deviations and logical mismatches caused by differences in rule expression, and outputs simulation comparison results and risk warning information. Based on simulation comparison results and risk warning information, S600 determines the execution stability of the rule model in various context scenarios, initiates consistency backtracking and structural correction for detected offsets and conflicts, verifies context labels, adjusts rule expressions and semantic tag configurations, and completes closed-loop optimization of the rule model.

2. The intelligent approval rule modeling method for process automation according to claim 1, characterized in that, Step S100 includes: During the extraction of each historical approval data, the scene characteristics, field semantic hierarchy, and participating role information corresponding to the approval behavior are collected. Based on approval logs, business process definitions, organizational structure, business metadata and unstructured text, perform enhanced extraction of execution context data to obtain timestamps, process stages, job levels, business lines and semantic information of remarks; The collected fields are semantically categorized, and the field types and expression levels are divided according to the preset business logic knowledge base; Generate a structured set of contextual semantic tags, organized in the form of "tag dimension, tag key, tag value" triples, and bind them one by one with historical approval data.

3. The intelligent approval rule modeling method for process automation according to claim 1, characterized in that, Step S300 includes: Semantic element recognition and structural segmentation are performed on each historical approval data, and the main condition fields, contextual additional attributes, and decision actions are identified based on the contextual semantic tag set. The three types of key elements identified are mapped to predefined field positions to construct a rule feature matrix, and missing fields are handled with null values ​​or default unconditional handling. Perform expression normalization processing on the rule feature matrix to standardize the expression form of semantically equivalent fields; Based on the strength of the field's role in decision-making, semantic weight adjustment is performed to identify rule combinations that have the same main condition fields but different contextual additional attributes, extract ambiguous additional attributes, and generate a rule differentiation candidate set.

4. The intelligent approval rule modeling method for process automation according to claim 1, characterized in that, Step S400 includes: Based on the rule-differentiated candidate set, we identify rule combinations that have the same main condition field but semantic differences in contextual additional attributes, and extract key additional attributes that are easily ignored by compression algorithms. The extracted key additional attributes are embedded into the corresponding original rule structure, keeping the logic unchanged and explicitly expanding the content of expression; The updated rule structure is subjected to expression-sensitive enhancement processing, and the semantic weights and saliency markers of key detail elements are annotated. The processing results are summarized to generate an enhanced set of rule models.

5. The intelligent approval rule modeling method for process automation according to claim 1, characterized in that, Step S500 includes: Filter representative historical approval process data and extract the input fields, process node sequence, historical decision path and context labels for each data point; For the same approval input, the original rule model set and the enhanced rule model set are applied to perform process simulations to deduce their respective approval paths and decision results; Compare the differences in process paths under the two rule configurations, identify approval deviations and logical mismatches, and analyze whether there is delegation of authority, overstepping of role authority, or omission of key nodes. Output a simulation comparison report that includes path differences, offset types, and risk levels, and generate corresponding risk warning information.

6. The intelligent approval rule modeling method for process automation according to claim 1, characterized in that, Step S600 includes: The joint analysis process simulation output identifies approval offset paths and logical conflict nodes, and identifies problem types that repeatedly occur in multiple context combination scenarios. Based on the contextual semantic tags and semantic weight identifiers embedded in the rule model, we trace back the key fields that caused the offset and their contextual sources to determine the reasons for expression defects or insufficient modeling. Corrections are made to rule models with structural problems, including adding contextual attributes, optimizing field nesting structures, and refining semantic tag content; The revised rule model was re-simulated and verified to confirm that it has semantic stability and execution consistency under the same context.

Images