Intelligent workflow guidance methods and systems for government processes

Abstract

This invention discloses an intelligent workflow guidance method and system for government affairs processes. Relating to the field of e-government technology, this invention first integrates a government affairs dictionary, historical form data, and process-related data. After cleaning and standardizing these data using ETL tools, the Apriori algorithm is employed to mine high-frequency field relationships, thereby constructing a tree-shaped form field filling chain and calculating its weight. When a user fills out a form, the system automatically identifies high-priority triggering fields, uses an inverted index to recall candidate filling chains, and combines context similarity to determine the optimal chain for automatic filling. Finally, the system performs multi-dimensional compliance checks on the form's format, required fields, and business rules. If the checks fail, a correction prompt is generated and guidance is provided for correction. This invention improves the standardization and accuracy of government affairs data filling, effectively increasing the efficiency of government affairs process processing.

Description

Technical Field

[0001] This invention relates to the field of e-government technology, specifically to an intelligent workflow guidance method and system for e-government processes. Background Technology

[0002] Workflow guidelines for government processes need to meet the requirements of standardization, efficiency, and compliance. However, existing technologies have the following shortcomings in practical applications: Firstly, government processes involve various types of data, such as government dictionary databases, historical form data, and process-related data. However, existing technologies lack a unified cleaning and standardization mechanism, resulting in chaotic data formats, redundancy, and repetition, which can easily lead to errors in form filling. Secondly, existing technologies rely heavily on manual sorting of field relationships, which not only makes it difficult to discover implicit relationships but also fails to quantify the strength of the relationships. Furthermore, they do not adapt the filling chain to contextual information such as government process nodes and user positions, resulting in low usability of the filling results. Third, existing validations are mostly limited to simple format validations, failing to cover the validation of mandatory fields and business rules necessary for government scenarios. Furthermore, when validation fails, the error messages are vague and lack precise location and correction guidance. Therefore, there is an urgent need for an intelligent workflow guidance method and system for government affairs processes. Summary of the Invention

[0003] To address the shortcomings of existing technologies, this invention provides an intelligent workflow guidance method and system for government affairs processes, which solves the problems of non-standard data, poor data matching, and incomplete verification in existing government affairs processes.

[0004] To achieve the above objectives, the present invention provides the following technical solution: an intelligent workflow guidance method for government affairs processes, comprising: Step 1: Integrate the government affairs dictionary, historical form data, and process association data to construct form field filling chains. When a user fills out a form, the system triggers a filling chain match based on the filled fields. The specific steps for constructing the form field filling chain are as follows: Use an ETL tool to clean the government affairs dictionary, historical form data, and process association data, and classify and store them according to business type and field category; Use the Apriori association rule mining algorithm to analyze the high-frequency association relationships between fields in the historical form data, and organize these association relationships into rule entries of trigger field-related field-association probability, and store them in the association rule library; With the trigger field as the root node, and in descending order of association probability, use related fields as child nodes to construct a tree-structured filling chain, and label each filling chain with the corresponding applicable process node and position; Assign weights to the obtained filling chains based on historical data. Step 2: Perform multi-dimensional compliance checks on the filled form fields, specifically including format validation, required field validation, and business rule validation; Step 3: For fields that fail validation, generate correction prompts in order of priority, and guide users to make real-time corrections and re-validate.

[0005] As a further aspect of the present invention, the specific steps for assigning weights to the obtained filling chain based on historical data are as follows: Calculate the effective application rate of each fill chain from the system's historical operation logs, and extract four-dimensional data that is strongly correlated with the actual application effect of the fill chain. The four-dimensional data includes fill success rate, user adoption rate, business compliance rate, and data freshness. The Spearman rank correlation coefficient was used to calculate the correlation strength between the four-dimensional data and the effective application rate, and the rank contribution coefficient was obtained. The contribution coefficients of each dimension are converted into weights that sum to 1. The specific formula is as follows: ,in, This represents the ranking correlation strength between the data in the j-th dimension and the effective application rate, where j∈[1,4]; Using Min-Max normalization, the four dimensions of the original data for each padding chain are mapped to the [0,1] interval, resulting in... Calculate the weight of each fill chain as follows: .

[0006] As a further aspect of the present invention, the Spearman rank correlation coefficient is used to calculate the correlation strength between four-dimensional data and the effective application rate, and the specific operation to obtain the rank contribution coefficient is as follows: For all population chains, sort them in descending order by population success rate, user adoption rate, business compliance rate, and data freshness, and also sort them in descending order by effective application rate. According to the formula The ranking correlation strength between four-dimensional data and effective application rate was calculated, where, Let be the difference between the dimensional ranking and the effective application rate ranking of the i-th filling chain, and n be the total number of filling chains; right Take the absolute value to obtain the level contribution coefficients corresponding to the four dimensions.

[0007] As a further aspect of the present invention, the specific steps for the system to trigger fill chain matching based on the filled fields are as follows: By listening to front-end events, the form fields filled in by users are captured in real time. Combined with the priority table of core business fields, the first high-priority field to be filled in is identified as the trigger field, and its field value is extracted. Based on the trigger field and its corresponding field value, the trigger field and field value are mapped to the corresponding fill chain ID through inverted index technology. All candidate fill chains that contain the trigger field and field value and are applicable to the current process node and user position are retrieved from the fill chain library. Calculate the similarity between the current context information and the candidate filling chain, and correct the similarity with the weights corresponding to the candidate filling chain to obtain the final similarity = weights corresponding to the candidate filling chain × similarity; The system employs a quick sorting algorithm to rank candidate fill chains from highest to lowest based on their final similarity. The top-ranked chain is selected as the optimal fill chain. If multiple fill chains have the same similarity, they are sorted a second time from highest to lowest based on their association probability. Once the optimal fill chain is determined, the system automatically fills the associated field values ​​in the chain into the corresponding positions in the form.

[0008] As a further aspect of the present invention, the step of calculating the similarity between the current context information and the candidate filling chain is as follows: Define organizational, business, and behavioral keywords: Organizational keywords must come from the government organizational structure dictionary, including the full name of the user's department, the unique code of the current process node, and the user's job title and level, and must completely match the dictionary entries; Business-related keywords must come from the government business element dictionary, including business type codes and form field values, and must include a unique business code; Behavioral keywords are selected from system operation logs and must meet the following requirement: operation frequency ≥ 3 months. Next, among them This is the operation frequency threshold; Match the three types of keywords mentioned above from the current context information and the candidate filling chain to obtain the organizational keyword set, business keyword set, and behavioral keyword set, respectively. Calculate the bidirectional overlap rate for organizational categories, business categories, and behavioral categories respectively, and obtain the similarity = bidirectional overlap rate for organizational categories × bidirectional overlap rate for business categories × bidirectional overlap rate for behavioral categories; The specific method for calculating the three types of bidirectional overlap rate is as follows: Find the intersection of the current context's keyword set and the candidate fill chain corresponding to the same keyword set. Count the number of the current context's keyword set, the candidate fill chain corresponding to the same keyword set, and the intersection as num1, Num1, and Num2, respectively. Calculate the intersection ratios num1 / Num1 and num1 / Num2, and then add the two intersection ratios together to obtain the bidirectional overlap rate corresponding to that category.

[0009] As a further aspect of the present invention, the specific steps for format verification are as follows: In the rule configuration library, a unique format validation rule is bound to each field type, forming a field type-rule ID-validation logic mapping table; The format validation rules for the corresponding field can be invoked through the oninput or onblur events on the front end. If the format validation fails, the system adds a red border to the field and displays the specific error reason when the mouse hovers over it. If the validation passes, the red border disappears automatically.

[0010] As a further aspect of the present invention, the specific steps for validating required fields are as follows: In the process node-required field mapping table, according to the government business specifications, a unique required field is configured for each process node, and the blocking level of the field is marked; When a user clicks submit, the system uses a traversal algorithm to check whether each field in the list of required fields for the current process node has been filled in: if a blocking field is not filled in, the submission operation is immediately blocked and the name of the unfilled field is recorded; if a prompt field is not filled in, a confirmation pop-up window appears, allowing the user to choose whether to continue. For required fields that are not filled in, the system displays a summary prompt at the top of the form, adds a yellow border to the unfilled fields, and automatically scrolls the page to the first unfilled field using the page scrolling API.

[0011] As a further aspect of the present invention, the specific steps for business rule verification are as follows: We sort out the business logic rules in government scenarios, use a rule engine to encode the rules into executable logic, and store them in the business rule library, and mark the corresponding applicable process nodes for each rule; Before the form is submitted, the system extracts the values ​​of the related fields that need to be validated, and matches the applicable rules from the business rule library according to the current process node, and performs rule validation one by one; If the business rule validation fails, the system will display an orange border next to the corresponding field in the form and provide a text message explaining the specific reason for the conflict. If it is a cross-field conflict, both related fields will be marked, and the conflict relationship will be clearly stated.

[0012] As a further aspect of the present invention, the specific steps for guiding users to perform real-time correction and re-verification are as follows: Errors that fail validation are categorized into blocking errors and informational errors, with blocking errors taking precedence over informational errors. For errors of the same type, they are then sorted a second time according to their position in the form, and an error handling list is generated. In the error handling list, add a location field button to each error entry. When the user clicks the button, the system can smoothly scroll the page to the corresponding error field. For errors that can be automatically corrected, the system will display an auto-correct button, which the user can click to automatically correct the field values; for errors that require manual correction, the system will provide reference suggestions based on historical data. After the user corrects the error, the system re-validates the field in real time via the onblur event: if the correction is compliant, the red / orange border of the erroneous field disappears and the corresponding entry in the error list is deleted; if the correction is still non-compliant, the error message is updated and the border warning is maintained.

[0013] Intelligent workflow guidance systems for government processes include: The data integration module is used to integrate government dictionary databases, historical form data, and process-related data, and calls ETL tools to complete data cleaning, classification, and storage. The fill chain construction module is used to mine association rules through the Apriori algorithm, construct a tree-shaped fill chain and label the applicable scenarios, and assign values ​​to the fill chain weights based on historical data. The compliance verification module is used to perform multi-dimensional compliance verification on the filled form fields, specifically including format verification, required field verification, and business rule verification. The correction guidance module is used to generate correction prompts for fields that fail validation, sort them by priority, and guide users to make real-time corrections and re-validate.

[0014] This invention provides an intelligent workflow guidance method and system for government affairs processes, which has the following advantages compared with the prior art: (1) This invention uses ETL tools to uniformly clean and classify government dictionary databases, historical form data, and process-related data, effectively solving the problems of chaotic and redundant formats of multi-source government data in the prior art. At the same time, relying on the government dictionary database, it lays a standardized data foundation for intelligent filling and ensures that the filling fields conform to government business specifications. (2) This invention uses the Apriori algorithm to automatically mine the high-frequency association between fields in historical forms, solving the problem that it is difficult to find implicit associations by manual sorting. In this way, a tree structure filling chain is constructed and the applicable process nodes and positions are marked. Combined with weight calculation based on Spearman coefficient and context similarity matching, the filling chain is accurately adapted to the current business scenario and user needs. (3) This invention constructs a multi-dimensional compliance verification system for format, required fields, and business rules, covering the core verification needs of government scenarios. It also generates an error list by prioritizing fields that fail verification, providing field location, automatic correction, and reference suggestions to guide users to make efficient corrections. Attached Figure Description

[0015] Figure 1 This is a flowchart of the steps of the present invention; Figure 2 This is a flowchart illustrating the steps for triggering fill chain matching in this invention. Figure 3 This is the system principle block diagram of the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] Example 1 like Figure 1 This invention provides an intelligent workflow guidance method for government affairs processes, including: Step 1: Integrate government dictionary database, historical form data, and process association data to build a form field filling chain; The government affairs dictionary database is an authoritative source of standardized data in the government affairs field. It contains basic data that has been officially standardized, such as administrative divisions, job levels, and case types. It can ensure that the basic values ​​of the fields to be filled in comply with government affairs business specifications, avoid errors in form filling due to inconsistent data expressions, and provide a compliance baseline for intelligent filling. It is the basic support for standardized filling of form fields. Historical form data records users' actual filling habits in real business situations, containing a large number of non-standardized but frequently occurring relationships. This allows the filling results to better match actual business needs and makes up for the lack of flexibility in the government affairs dictionary database. The process-related data is closely linked to the current stage of the business and the user's role. This ensures that the fields to be filled are accurately matched with the process progress and user permissions, avoiding data filling that is out of step with the current process stage and making the filling results real-time and scenario-specific. The specific steps for constructing the form field population chain are as follows: Use ETL tools to clean the government dictionary database, historical form data, and process-related data, including removing duplicate data from historical forms, correcting data with format errors, and classifying and storing them according to business type and field category; The raw data in government affairs scenarios, namely the aforementioned government affairs dictionary database, historical forms, and process-related data, generally suffer from problems such as chaotic formatting, redundancy, and exposure of sensitive information. For example, in historical forms, "job level" may include descriptions such as deputy section chief, deputy section chief level, and deputy section chief level, while the date formats in the process-related data are mixed, such as YYYY.MM.DD and YYYY / MM / DD. Direct use of these formats will lead to deviations in the filling results. For such large and redundant data, manual processing alone would take several days and may miss sensitive information. ETL tools (such as Apache NiFi and Kettle) can automate the cleaning and standardization process, reducing manual processing costs while ensuring data format consistency. The Apriori association rule mining algorithm is used to analyze the high-frequency association relationships between fields in historical form data. These association relationships are organized into rule entries of trigger field - associated field - association probability and stored in the association rule library. For example, rule 1: trigger field is "department = case management office" → associated field is "department head = Zhang XX", and the association probability is 85%. Manually sorting out the relationships between government data fields has two major pain points: ① Implicit relationships are difficult to discover, and manual analysis of historical data is required, which can easily lead to omissions; ② The strength of the relationship is difficult to quantify, and can only be set based on experience, resulting in poor targeting. The Apriori algorithm can automatically mine high-frequency relationships between fields from massive amounts of historical data and quantify the strength of the relationship with the probability of the relationship. Using the trigger field as the root node, and in descending order of association probability, the associated fields are used as child nodes to construct a tree-structured filling chain, and each filling chain is labeled with the corresponding applicable process node and position. If only the flat relationship between trigger fields and associated fields is stored, problems such as hierarchical confusion and scenario mismatch will occur. A tree structure can clearly present the hierarchical relationship between triggers and associations, avoiding field filling chaos. At the same time, marking the applicable scenarios (process nodes, positions) can narrow the matching range and improve the filling accuracy. The specific steps for assigning weights to the obtained fill chain based on historical data are as follows: Calculate the effective application rate of each fill chain from the system's historical operation logs, and extract four-dimensional data that are strongly correlated with the actual application effect of the fill chain, including fill success rate, user adoption rate, business compliance rate, and data freshness. Effective application rate, the proportion of times a certain fill chain is triggered that simultaneously satisfies user acceptance (the fill data is not modified and the form is submitted) and business compliance (the subsequent process is not rejected due to the fill data) out of the total number of triggers, directly anchors the core needs of government affairs scenarios; The success rate of filling is the proportion of times when the user did not manually modify the value of the filling field after a certain filling chain was triggered, out of the total number of triggers. It reflects the degree of matching between the filling chain data and the user's needs. User adoption rate, which is the proportion of times the populated chain's data is retained when a user finally submits the form after a certain populated chain has been populated in the past, out of the total number of triggers, reflects the applicability of the populated data to the business. Business compliance rate, which is the percentage of times that data filled in a certain filling chain was not returned due to data non-compliance in subsequent approval processes out of the total number of triggers, reflects the degree to which the filled data complies with government regulations; Data freshness is the reciprocal of the number of days since the last update of related data in a certain populated chain, reflecting the timeliness of the populated data; The Spearman rank correlation coefficient was used to calculate the correlation strength between the four-dimensional data and the effective application rate. The specific operation was as follows: For all fill chains, sort them in descending order by fill success rate, user adoption rate, business compliance rate, and data freshness, and also sort them in descending order by effective application rate to obtain the ranking number of each fill chain in each dimension and target indicator. According to the formula The ranking correlation strength between four-dimensional data and effective application rate was calculated, where, Let be the difference between the dimensional ranking and the effective application rate ranking of the i-th filling chain, and n be the total number of filling chains; If a certain dimension If the value is negative, its absolute value is taken to ensure that the dimension's contribution to the objective is positive. That is, the higher the dimension value, the higher the effective application rate, thus obtaining the level contribution coefficient. The contribution coefficients of each dimension are converted into weights that sum to 1. The specific formula is as follows: ,in, This represents the ranking correlation strength between the data in the j-th dimension and the effective application rate, where j∈[1,4]; Using Min-Max normalization, the four dimensions of the original data for each padding chain are mapped to the [0,1] interval, resulting in... And calculate the weight of each filling chain i. ; When a user fills out a form, the system triggers a fill chain matching based on the fields already filled in.

[0018] Step 2: Perform multi-dimensional compliance checks on the filled form fields, specifically including format validation, required field validation, and business rule validation; The specific steps for format verification are as follows: In the rule configuration library, a unique format validation rule is bound to each field type, forming a field type-rule ID-validation logic mapping table; Binding field types to rules one by one can avoid confusion in validation logic. When adding new fields in the future, you only need to associate them with existing rules, without the need for repeated development. For example, the "Promotion Time" field is of type "Date" and is bound to rule ID201 (YYYY-MM-DD format). When the user enters "2020 / 01 / 01", the system calls rule ID201 for validation. The format validation rules for the corresponding field can be invoked through the front-end oninput event (triggered in real time during user input) or onblur event (triggered when the field loses focus); The oninput event can detect formatting errors in real time, avoiding prompts only after the user has filled in all the fields, while the onblur event ensures that final validation is performed after the fields are filled in. For example, when a user enters an ID number “110101****0101123” (17 digits), the oninput event will prompt “ID number needs to be 18 digits, please add the last digit”. If the user adds “110101****01011234” (18 digits, the last digit is 4), the onblur event will trigger a full validation to confirm that the format is compliant. If the format validation fails, the system will add a red border to the field and display the specific error reason when the mouse hovers over it. If the validation passes, the red border will disappear automatically. The specific steps for validating the required fields are as follows: In the process node-required field mapping table, according to the government business specifications, each process node is configured with a unique required field. For example, the required fields for the clue registration node are the source of the clue, the name of the person being reported, and the content of the clue. The blocking level of the field is marked (blocking: if not filled, submission is not possible; prompting: if not filled, submission is possible but confirmation is required). When a user clicks submit, the system uses a traversal algorithm to check whether each field in the list of required fields for the current process node has been filled in: if a blocking field is not filled in, the submission operation is immediately blocked and the name of the unfilled field is recorded; if a prompt field is not filled in, a confirmation pop-up window appears, allowing the user to choose whether to continue. For required fields that are not filled, the system displays a summary prompt at the top of the form, adds a yellow border to the unfilled fields, and automatically scrolls the page to the first unfilled field using the page scrolling API, making it easy for users to quickly locate the field. The specific steps for verifying the business rules are as follows: We sort out the business logic rules in government scenarios, use a rule engine to encode the rules into executable logic, and store them in the business rule library. Each rule is marked with the corresponding applicable process node. The rules engine can decouple rules from code. When adding or modifying rules, you only need to adjust them in the rule library, without modifying the front-end code. For example, coding rule R4: if promotion time > start date then validation fails, applicable nodes are clue registration and personnel information entry; Before the form is submitted, the system extracts the values ​​of the related fields that need to be validated. For example, when validating R4, it extracts the field values ​​of "promotion time" and "work start time" and matches the applicable rules from the business rule library according to the current process node, and performs rule validation one by one. Extracting related fields ensures the completeness of the data required for rule validation, avoiding validation failures due to missing data, while matching rules by nodes reduces invalid validations and improves validation efficiency. If the business rule validation fails, the system will display an orange border next to the corresponding field in the form and provide a text message indicating the specific reason for the conflict, such as "Promotion date (2020-01-01) is later than the start date (2019-01-01), which does not comply with government regulations." If it is a cross-field conflict, both related fields will be marked, and the conflict relationship will be clearly stated. The orange border makes it easy to distinguish between "business logic error" and "formatting error" to avoid user confusion, while the specific reasons and related field annotations help users understand the nature of the error and improve the accuracy of correction.

[0019] Step 3: Generate specific correction prompts for fields that fail validation and guide the user to make corrections. The specific steps are as follows: Errors that fail validation are categorized into blocking errors and informational errors, with blocking errors taking precedence over informational errors. For errors of the same type, they are then sorted a second time according to their position in the form, and an error handling list is generated. Prioritization can guide users to resolve "core errors that affect submission" first, avoiding wasting time on non-core errors, while position sorting allows users to correct errors according to the order in which the form was filled in, reducing the cost of redirection; In the error handling list, add a location field button to each error entry. When the user clicks the button, the system can smoothly scroll the page to the corresponding error field. For errors that can be automatically corrected, such as date formats "2024 / 05 / 20" and "2024-5-20" being converted to "2024-05-20", the system will display an auto-correct button, which the user can click to automatically correct the field value. For errors that require manual correction, such as negative case amounts or promotion dates that are later than the start date of employment, the system provides reference suggestions based on historical data. For example, "The historical average case amount is 50,000 yuan, so it is recommended to enter a non-negative number; the start date of employment has been filled in as 2019-01-01, so it is recommended that the promotion date be ≤2019-01-01." After the user corrects the error, the system re-validates the field in real time via the onblur event: if the correction is compliant, the red / orange border of the erroneous field disappears and the corresponding entry in the error list is deleted; if the correction is still non-compliant, the error message is updated and the border warning is maintained. For example, if the phone number is changed from "138**13800" to "138**13800a", the message will be "The phone number must be 11 digits, currently containing letters".

[0020] Example 2 This embodiment further discloses a method for triggering fill chain matching based on filled fields, based on embodiment 1. The specific content includes: By listening to front-end events, the form fields filled in by users are captured in real time. Combined with the priority table of core business fields, such as clue type, department, and identity of the person being reported as high priority, and remarks and attachment description as low priority, the first high priority field filled in is identified as the trigger field and its field value is extracted. For example, if the user fills in "clue type = funds" first, then this field is the trigger field and the value is "funds". When users fill out forms, the order and type of the input fields are chaotic. If all fields are blindly used as trigger conditions, it will lead to a chaotic filling chain matching. Front-end event listeners can capture user input behavior in real time, avoiding waiting for the user to fill in all fields before triggering the matching. Meanwhile, the core business field priority table can filter low-value fields, ensuring that the field that best represents the business intent is used as the trigger condition, thereby improving the accuracy of matching. Based on the trigger field and its corresponding field value, the trigger field and field value are mapped to the corresponding fill chain ID through inverted index technology. All candidate fill chains that contain the trigger field and field value and are applicable to the current process node and user position are retrieved from the fill chain library. The populate chain library may store hundreds of chains. If the candidate chains are retrieved by traversing the entire library, the query efficiency will be extremely low. Especially in scenarios where users input data in real time, the delay will seriously affect the user experience. However, the inverted index technology can realize fast querying of user input values ​​to populate chains through the mapping relationship of "trigger field - field value → populate chain ID". Calculate the similarity between the current context information and the candidate filling chain, and correct the similarity with the weights corresponding to the candidate filling chain to obtain the final similarity = weights corresponding to the candidate filling chain × similarity; The specific steps for calculating the similarity between the current context information and the candidate filling chain are as follows: Define the organizational, business, and behavioral keywords to be extracted: organizational keywords are from the government organizational structure dictionary, including the full name of the user's department, the unique code of the current process node, and the user's job title and level, and must completely match the dictionary entries; Business-related keywords are derived from the government business element dictionary, including business type codes and form field values, and must include a unique business code. Behavioral keywords are filtered from system operation logs and must meet the following criteria: operation frequency ≥ 3 months. The format is fixed as: "Government operational verb + business object," where the verb is limited to government operational terms from the dictionary. The operation frequency threshold needs to be set according to the actual situation; Match the three types of keywords mentioned above from the current context information and the candidate filling chain to obtain the organizational keyword set, business keyword set, and behavioral keyword set, respectively. Calculate the bidirectional overlap rate for organizational categories, business categories, and behavioral categories respectively, and obtain the similarity = bidirectional overlap rate for organizational categories × bidirectional overlap rate for business categories × bidirectional overlap rate for behavioral categories; The specific method for calculating the three types of bidirectional overlap rate is as follows: Find the intersection of a certain set of keywords in the current context and the corresponding set of keywords in the candidate filling chain. Count the number of the current set of keywords, the corresponding set of keywords in the candidate filling chain, and the number of intersections as num1, Num1, and Num2. Calculate the intersection ratios num1 / Num1 and num1 / Num2, and add the two intersection ratios together to obtain the bidirectional overlap rate corresponding to this category. The system employs a quick sorting algorithm to rank candidate fill chains from highest to lowest based on their final similarity. The top-ranked chain is selected as the optimal fill chain. If multiple fill chains have the same similarity, they are sorted a second time from highest to lowest based on their association probability. Once the optimal fill chain is determined, the system automatically fills the associated field values ​​in the chain into the corresponding positions in the form.

[0021] Example 3 like Figure 3 An intelligent workflow guidance system for government processes includes: The data integration module is used to integrate government dictionary databases, historical form data, and process-related data, and calls ETL tools to complete data cleaning, classification, and storage. The fill chain construction module is used to mine association rules through the Apriori algorithm, construct a tree-shaped fill chain and label the applicable scenarios, and assign values ​​to the fill chain weights based on historical data. The compliance verification module is used to perform multi-dimensional compliance verification on the filled form fields, specifically including format verification, required field verification, and business rule verification. The correction guidance module is used to generate correction prompts for fields that fail validation, sort them by priority, and guide users to make real-time corrections and re-validate.

[0022] Some of the data in the above formulas are numerical calculations with dimensions removed, and the contents not described in detail in this specification are all prior art known to those skilled in the art.

[0023] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. An intelligent workflow guidance method for government affairs processes, characterized in that: include: Step 1: Integrate the government affairs dictionary, historical form data, and process association data to construct a form field filling chain. When a user fills out a form, the system triggers the filling chain matching based on the filled fields. The specific steps for constructing the form field filling chain are as follows: Use an ETL tool to clean the government affairs dictionary, historical form data, and process association data, and classify and store them according to business type and field category; Use the Apriori association rule mining algorithm to analyze the high-frequency association relationships between fields in the historical form data, and organize these association relationships into rule entries of trigger field-related field-association probability, and store them in the association rule library. Using the trigger field as the root node, and ranking the associated fields as child nodes according to their association probability from high to low, a tree-structured filling chain is constructed. Each filling chain is labeled with its corresponding applicable process node and job position. Weights are assigned to the resulting filling chains based on historical data. Specifically, the effective application rate of each filling chain is calculated from the system's historical operation logs, and four-dimensional data strongly correlated with the actual application effect of that filling chain is extracted. This four-dimensional data includes filling success rate, user adoption rate, business compliance rate, and data freshness. The Spearman rank correlation coefficient is used to calculate the correlation strength between the four-dimensional data and the effective application rate, obtaining the rank contribution coefficient. The rank contribution coefficients of each dimension are converted into weights that sum to 1, using the following formula: ,in, Let j represent the ranking correlation strength between the j-th dimension data and the effective application rate, where j∈[1,4]; Min-Max normalization is used to map the original data of the four dimensions of each filling chain to the interval [0,1], resulting in... Calculate the weight of each fill chain as follows: ; Step 2: Perform multi-dimensional compliance checks on the filled form fields, specifically including format validation, required field validation, and business rule validation; Step 3: For fields that fail validation, generate correction prompts in order of priority, and guide users to make real-time corrections and re-validate.

2. The intelligent workflow guidance method for government affairs processes according to claim 1, characterized in that, The Spearman rank correlation coefficient is used to calculate the correlation strength between four-dimensional data and the effective application rate, and the specific operation to obtain the rank contribution coefficient is as follows: For all population chains, sort them in descending order by population success rate, user adoption rate, business compliance rate, and data freshness, and also sort them in descending order by effective application rate. According to the formula The ranking correlation strength between four-dimensional data and effective application rate was calculated, where, Let be the difference between the dimensional ranking and the effective application rate ranking of the i-th filling chain, and n be the total number of filling chains; right Take the absolute value to obtain the level contribution coefficients corresponding to the four dimensions.

3. The intelligent workflow guidance method for government affairs processes according to claim 1, characterized in that, The specific steps by which the system triggers fill chain matching based on the filled fields are as follows: By listening to front-end events, the form fields filled in by users are captured in real time. Combined with the priority table of core business fields, the first high-priority field to be filled in is identified as the trigger field, and its field value is extracted. Based on the trigger field and its corresponding field value, the trigger field and field value are mapped to the corresponding fill chain ID through inverted index technology. All candidate fill chains that contain the trigger field and field value and are applicable to the current process node and user position are retrieved from the fill chain library. Calculate the similarity between the current context information and the candidate filling chain, and correct the similarity with the weights corresponding to the candidate filling chain to obtain the final similarity = weights corresponding to the candidate filling chain × similarity; The system employs a quick sorting algorithm to rank candidate fill chains from highest to lowest based on their final similarity. The top-ranked chain is selected as the optimal fill chain. If multiple fill chains have the same similarity, they are sorted a second time from highest to lowest based on their association probability. Once the optimal fill chain is determined, the system automatically fills the associated field values ​​in the chain into the corresponding positions in the form.

4. The intelligent workflow guidance method for government affairs processes according to claim 3, characterized in that, The steps for calculating the similarity between the current context information and the candidate padding chain are as follows: Define organizational, business, and behavioral keywords: Organizational keywords must come from the government organizational structure dictionary, including the full name of the user's department, the unique code of the current process node, and the user's job title and level, and must completely match the dictionary entries; Business-related keywords must come from the government business element dictionary, including business type codes and form field values, and must include a unique business code; Behavioral keywords are selected from system operation logs and must meet the following requirement: operation frequency ≥ 3 months. Next, among them This is the operation frequency threshold; Match three categories of keywords—organizational, business, and behavioral—from the current context information and candidate fill chain to obtain sets of organizational keywords, business keywords, and behavioral keywords, respectively. Calculate the bidirectional overlap rate for organizational, business, and behavioral categories respectively, and obtain the similarity = bidirectional overlap rate for organizational category × bidirectional overlap rate for business category × bidirectional overlap rate for behavioral category; The specific method for calculating the three types of bidirectional overlap rates is as follows: Find the intersection of the current context's keyword set and the candidate fill chain corresponding to the same keyword set. Count the number of the current context's keyword set, the candidate fill chain corresponding to the same keyword set, and the intersection as num1, Num1, and Num2, respectively. Calculate the intersection ratios num1 / Num1 and num1 / Num2, and then add the two intersection ratios together to obtain the bidirectional overlap rate corresponding to that category.

5. The intelligent workflow guidance method for government affairs processes according to claim 1, characterized in that, The specific steps for format verification are as follows: In the rule configuration library, a unique format validation rule is bound to each field type, forming a field type-rule ID-validation logic mapping table; The format validation rules for the corresponding field can be invoked through the oninput or onblur events on the front end. If the format validation fails, the system adds a red border to the field and displays the specific error reason when the mouse hovers over it. If the validation passes, the red border disappears automatically.

6. The intelligent workflow guidance method for government affairs processes according to claim 1, characterized in that, The specific steps for validating required fields are as follows: In the process node-required field mapping table, according to the government business specifications, a unique required field is configured for each process node, and the blocking level of the field is marked; When a user clicks submit, the system uses a traversal algorithm to check whether the fields in the list of required fields for the current process node have been filled in. If a blocking field is not filled in, the submission operation is immediately blocked and the name of the unfilled field is recorded. If a prompt field is left blank, a confirmation pop-up will appear, allowing the user to choose whether to continue. For required fields that are not filled in, the system displays a summary prompt at the top of the form, adds a yellow border to the unfilled fields, and automatically scrolls the page to the first unfilled field using the page scrolling API.

7. The intelligent workflow guidance method for government affairs processes according to claim 1, characterized in that, The specific steps for business rule verification are as follows: We sort out the business logic rules in government scenarios, use a rule engine to encode the rules into executable logic, and store them in the business rule library, and mark the corresponding applicable process nodes for each rule; Before the form is submitted, the system extracts the values ​​of the related fields that need to be validated, and matches the applicable rules from the business rule library according to the current process node, and performs rule validation one by one; If the business rule validation fails, the system will display an orange border next to the corresponding field in the form and provide a text message explaining the specific reason for the conflict. If it is a cross-field conflict, both related fields will be marked, and the conflict relationship will be clearly stated.

8. The intelligent workflow guidance method for government affairs processes according to claim 1, characterized in that, The specific steps to guide users to perform real-time corrections and re-verification are as follows: Errors that fail validation are categorized into blocking errors and informational errors, with blocking errors taking precedence over informational errors. For errors of the same type, they are then sorted a second time according to their position in the form, and an error handling list is generated. In the error handling list, add a location field button to each error entry. When the user clicks the button, the system can smoothly scroll the page to the corresponding error field. For errors that can be automatically corrected, the system will display an autocorrect button, which the user can click to automatically correct the field values. For errors that require manual correction, the system provides reference suggestions based on historical data; After the user corrects the error, the system re-validates the field in real time via the onblur event: if the correction is compliant, the red / orange border of the erroneous field disappears and the corresponding entry in the error list is deleted; if the correction is still non-compliant, the error message is updated and the border warning is maintained.

9. An intelligent workflow guidance system for government processes, used to execute the intelligent workflow guidance method for government processes as described in any one of claims 1-8, characterized in that, include: The data integration module is used to integrate government dictionary databases, historical form data, and process-related data, and calls ETL tools to complete data cleaning, classification, and storage. The fill chain construction module is used to mine association rules through the Apriori algorithm, construct a tree-shaped fill chain and label the applicable scenarios, and assign values ​​to the fill chain weights based on historical data. The compliance verification module is used to perform multi-dimensional compliance verification on the filled form fields, specifically including format verification, required field verification, and business rule verification. The correction guidance module is used to generate correction prompts for fields that fail validation, sort them by priority, and guide users to make real-time corrections and re-validate.

Images