An automatic flow transfer method and system for cross-departmental collaborative examination and approval of government affairs services

Abstract

The application discloses a kind of automatic flow method and system of government service cross-departmental collaborative examination and approval, it is related to government information technology field, the method reads and executes semantic rule analysis to joint matter text material, generates rule candidate set, and calculates the bit sequence discrete quantity S of each candidate sentence section;Rule candidate set is carried out segment division and structured disassembly processing, forms rule structure chain, and the segment that is not hung to existing rule structure chain is recorded as closed gap segment, and the structure closed value is calculated;Rule structure chain is classified, and target department benchmark rule chain is constructed, and the cross-departmental mapping value G and conclusion acceptance value E of rule structure chain to target department are calculated;Current examination and approval file is transcribed into file structure segment chain and is compared with target department rule unit slot by slot, and the number of aggregated closed gaps B and the number of cross conflict points K are counted, and comprehensive flow determination value T is constructed to determine examination and approval flow.

Description

Technical Field

[0001] This invention relates to the field of e-government technology, specifically to an automated workflow method and system for cross-departmental collaborative approval of e-government services. Background Technology

[0002] In practice, government services typically involve joint processing of matters such as business establishment, project approval, qualification certification, engineering construction, and public services. These services often involve multiple approval departments reviewing applications sequentially according to their established responsibilities. This has led to a gradual evolution from single-window acceptance to cross-departmental collaborative approval. In cross-departmental collaborative approval scenarios, the review opinions, correction records, material verification results, and processing conclusions generated by preceding departments all need to serve as the basis for judgment in subsequent departments. Therefore, the government service processing is no longer limited to single-node internal review but has shifted to a collaborative process involving the connection of rules across multiple departments, the acceptance of materials, and the continuation of conclusions. As the number of joint processing matters, the types of rules, and the inter-departmental relationships continue to increase in complexity, the unified analysis and breakdown of joint processing documents, historical approval guidelines, and real-time case file content, along with the automatic flow between nodes, has become a fundamental processing requirement in cross-departmental collaborative approval of government services.

[0003] Existing cross-departmental collaborative approval processes for government services mostly rely on matching item names, comparing document catalogs, or manually reviewing prior opinions before forwarding. They often lack a unified, structured mechanism for breaking down approval conditions, supporting documents, exceptions, and conclusions in documents such as policy documents, service guides, correction notices, and approval templates. They also lack a chain-like expression method for the sequential relationships between different departments. Because prior and target departments differ in their descriptions of the subject matter, the definition of conditions, the writing of document names, and exception handling paths, existing processes often struggle to determine whether prior rules can be continued by the target department, and also fail to identify the attachment status of case file content in the target department's rule slots. This results in joint processing items frequently relying on manual review to decide whether to push, correct, or return them when entering subsequent stages, making it difficult to meet the continuous processing needs of cross-departmental collaborative approvals. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides an automated workflow method and system for cross-departmental collaborative approval of government services, thus solving the problems mentioned in the background technology.

[0005] To achieve the above objectives, the present invention provides the following technical solution: an automated workflow method for cross-departmental collaborative approval of government services, comprising the following steps: S1. Read the text data of the joint handling matters and perform semantic rule parsing to generate a set of rule candidates and calculate the positional discreteness S of each candidate sentence segment; S2. Perform segmentation and structured decomposition on the candidate rule set to form a rule structure chain. Segments not attached to the existing rule structure chain are recorded as closed gap segments, and the structure closure value is calculated. S3. Classify the rule structure chain, construct the target department benchmark rule chain, and calculate the cross-departmental mapping value G and conclusion acceptance value E from the rule structure chain to the target department. S4. Translate the current approval case file into a case file structure fragment chain and align it with the target department's rule unit slot by slot. Calculate the number of aggregated closure gaps B and the number of cross-conflict points K, and construct a comprehensive circulation judgment value T to determine the approval circulation.

[0006] Preferably, S1 includes S11; S11. Read the text data of the current joint handling matter, perform semantic rule parsing on the text data in sequence, filter out the rule candidate sentence segments containing approval condition semantics, and generate a rule candidate set; The semantic rule parsing includes title level recognition, rule trigger word scanning, sentence segmentation and sentence dependency relationship extraction. The text materials include institutional documents, service guides, review caliber records, historical correction notices and historical approval opinion templates. The rule trigger word scanning is performed around the trigger words of the matter object, the trigger words of the limiting conditions, the trigger words of the material basis, the trigger words of the exception situation, and the trigger words of the processing conclusion.

[0007] Preferably, S1 further includes S12; S12. For each rule candidate segment in the rule candidate set, locate the order of appearance of the object segment u, the limiting condition segment l, the material basis segment m, the exception segment x, and the processing conclusion segment c in the original text, and calculate the positional discreteness S of the rule candidate segment based on the order position. This is used to describe the discrete state of the arrangement of the five types of segments within the original rule text, as follows: S j =|u j -l j |+|l j -m j |+|m j -x j |+|x j -c j |; Among them, S j Let u represent the positional discrete quantity of the candidate sentence segment for the j-th rule. j l j m j x j and cj These represent the subject matter segment, limiting condition segment, material basis segment, exception segment, and processing conclusion segment of the j-th rule candidate sentence segment, respectively.

[0008] Preferably, S2 includes S21; S21. Divide the candidate rule set into segments according to the item object slot, limiting condition slot, material basis slot, exception case slot and processing conclusion slot, and extract the number of item object segments O, limiting condition segments L, material basis segments M, exception case segments X and processing conclusion segments C in each candidate rule segment.

[0009] Preferably, S2 further includes S22 and S23; S22. After segmentation, perform structured decomposition processing on each structural segment, specifically: attach the limiting condition segment to the corresponding item object segment, attach the material basis segment to the corresponding limiting condition segment, attach the exception segment to the limiting condition segment corresponding to the current exception, when the current limiting condition segment is already associated with the material basis segment, attach the exception segment to its corresponding limiting condition segment and the material basis segment connected to the current limiting condition segment, and attach the processing conclusion segment to the end of the structural chain formed by the sequential connection of the item object segment, limiting condition segment, material basis segment and exception segment to form a rule structure chain; S23. Segments that are not attached to existing rule structure chains are recorded as closed gap segments, and the number of closed gaps B is counted. For two or more parallel structure chains derived from the same item object segment in the same rule, they are recorded as parallel branches, and the number of parallel branches P is counted. Then, the number of parallel branches P is counted and compared with the positional discreteness S to calculate the structural closure value R of the rule candidate sentence segment, as follows. ; Among them, O j L j M j X j C j B j and P j These represent the number of event object segments, limiting condition segments, supporting material segments, exception segments, processing conclusion segments, closing gaps, and parallel bifurcations in the j-th rule candidate sentence segment, respectively. j +L j +M j +X j +C j R represents the total number of segments in the candidate segment of rule j that have entered the five structural slots and been included in the linking chain. jThis represents the structural closure value of the j-th rule candidate segment, used to indicate the degree of closure of the j-th rule candidate segment in the five types of structural slots.

[0010] Preferably, S3 includes S31 and S32; S31. Read the rule structure chain of each department and classify the rule structure chain according to the name of the joint handling item. In the classified rule structure chain, extract the rule structure chain set corresponding to the target department. Arrange the item object slot, limiting condition slot, material basis slot, exception situation slot and processing conclusion slot with the highest frequency in the rule structure chain set corresponding to the target department according to the attachment order to generate the target department's benchmark rule chain. S32. Calculate the number of slots A for the alignment of the target department's benchmark rule chain with the following parameters: number of slots H for the inheritance of the constraint condition, number of material alias merged logs N, number of slot type misalignment Q, and number of caliber drifts D. Based on the statistical results, calculate the cross-departmental mapping value G from the rule structure chain to the target department, as follows. ; Among them, G j,k A represents the cross-departmental mapping value of the j-th rule structure chain in target department k. j,k H represents the number of slots in the j-th rule structure chain that can be directly aligned with the baseline rule chain of the target department k in the item object slot. j,k N represents the number of constraint inheritance slots in the j-th rule structure chain that are continued by the target department k. j,k Q represents the merge logarithm of material aliases in the j-th rule structure chain, where material names are merged with material names in the target department k. j,k D represents the number of slot misalignments between the j-th rule structure chain and the baseline rule chain of the target department k. j,k This represents the caliber drift number between the j-th rule structure chain and the target department k.

[0011] Preferably, S3 further includes S33; S33. After obtaining the cross-departmental mapping value G corresponding to all rule structure chains of the target department, count the number of rules n mapped to the current target department, and perform an exception slot comparison between the rule structure chain of the preceding department and the rule receiving structure of the target department to screen out exception fragments that are not attached to the corresponding limiting condition chain or material basis chain, and count the number of exception fragments to obtain the number of exceptions that are still not closed. Then, perform a material basis slot comparison between the rule structure chain of the preceding department and the rule receiving structure of the target department to screen out material branches that have set material basis receiving slots in the rule receiving structure of the target department but have not formed corresponding material basis fragments in the rule structure chain of the preceding department, and count the number of material branches to obtain the number of material branches that are still to be added. The conclusion acceptance value E of the target department k is calculated based on the number of unresolved exceptions Y and the number of material branches F to be added. k The details are as follows; ; Where, n k Y represents the number of rules that are mapped to the current target department k, with the target department as the unit. k This indicates the number of exceptions for target department k that have not yet been explicitly assigned to a specific condition chain, F. k This indicates the number of material branches that still need to be added to the current rule receiving structure of the target department k.

[0012] Preferably, S4 includes S41; S41. Call the item name, applicant information, electronic material catalog, supplementary explanation, and processing results formed by the preceding department corresponding to the current approval case file. Perform structured decomposition and transcribing through S2 into a case file structure fragment chain. Then, perform slot-by-slot alignment between the case file structure fragment chain and the target department rule unit. Identify the item object slots, limiting condition slots, material basis slots, exception situation slots, and processing conclusion slots that have been satisfied in the case file. Calculate the number of aggregation closure gaps B and the number of cross-conflict points K of the target department. Among them, the rule unit of the target department is a slotted continuation unit composed of the rule structure chain mapped to the target department in S3 and its corresponding item object receiving slot, limiting condition receiving slot, material basis receiving slot, exception situation receiving slot and processing conclusion receiving slot. The rule unit is used to represent the object type, condition chain type, material support type, exception attribution type and preceding conclusion status type that the target department is allowed to receive at the current approval node. The number of aggregated closure gaps B represents the sum of the number of unfilled item object slots, limiting condition slots, material basis slots, exception situation slots and processing conclusion slots. The number of cross conflict points K represents the sum of the number of object attachment conflicts, condition correspondence conflicts and path status conflicts.

[0013] Preferably, S4 further includes S42 and S43; S42. After obtaining the number of aggregation closure gaps B and the number of cross-conflict points K, combine the structural closure value R of each rule structure chain, the cross-departmental mapping value G of each rule structure chain to the target department, and the conclusion acceptance value E of the target department to calculate the comprehensive flow judgment value T of the target department, as follows; ; Among them, T k B represents the comprehensive turnover judgment value T for target department k. k and K k These represent the number of convergence closure gaps and the number of cross-conflict points in target department k, respectively. S43. Read the comprehensive transfer judgment value sequence of similar historical case files in the target department, and extract the median value of the comprehensive transfer judgment value sequence using the percentile method as the initial entry value U, extract the 75th percentile value of the comprehensive transfer judgment value sequence as the direct transfer value V, and make approval and transfer judgment with the real-time acquired comprehensive transfer judgment value T, as follows; When the comprehensive transfer judgment value T < the initial entry value U, it means that the current case file has not yet entered the basic transfer range in the rule continuation state of the target department, and the case file structure chain cannot be connected to the standard review chain of the target department. At this time, the push action of the current case file to the target department is frozen, and a gap list or conflict list is generated and written into the transfer record of the current case file. When the initial entry value U ≤ comprehensive circulation judgment value T < direct circulation value V, it means that the current case file has entered the circulation range in the target department, but has not yet entered the direct circulation range. At this time, the execution node push will attach the current case file to the pending queue of the target department, and then the review task will be assigned to the corresponding review port or manual review seat. When the comprehensive transfer judgment value T ≥ the direct transfer value V, it means that the current case file has entered the direct transfer range in the target department. At this time, the node receiving log, transfer time identifier and rule hit record are directly written, and the approval action of the target department is triggered.

[0014] An automated workflow system for cross-departmental collaborative approval of government services includes a rule source screening module, a disassembly and attachment module, a mapping and acceptance module, and a case file alignment and branching module. The rule source screening module is used to read the text data of the joint handling matters and perform semantic rule parsing to generate a set of rule candidates and calculate the positional discreteness S of each candidate sentence segment; The disassembly and attachment module is used to divide and disassemble the rule candidate set into segments to form a rule structure chain. Segments that are not attached to the existing rule structure chain are recorded as closed gap segments, and the structure closure value is calculated. The mapping and acceptance module is used to classify the rule structure chain, construct the target department benchmark rule chain, and calculate the cross-departmental mapping value G and conclusion acceptance value E from the rule structure chain to the target department. The case file alignment branch module is used to transcribe the current approval case file into a case file structure fragment chain and align it with the target department rule unit slot by slot, count the number of aggregated closure gaps B and the number of cross-conflict points K, and construct a comprehensive circulation judgment value T for approval circulation judgment.

[0015] This invention provides an automated workflow method and system for cross-departmental collaborative approval of government services. It offers the following advantages: (1) This method addresses the problems of fragmented rule expression, difficulty in directly connecting rule semantics, and inconsistent interpretations between different stages in the cross-departmental approval process for jointly handled matters. First, semantic rule parsing is performed on textual materials such as institutional documents, service guides, historical correction notices, and historical approval opinion templates. Then, based on five categories of fragments—object, limiting conditions, material basis, exceptions, and processing conclusions—positional identification and structural decomposition are performed. This transforms the rule content, which originally relied on manual reading and experience-based judgment, into a chain of rules that can be linked and compared. Compared with the existing method of searching through the original text segment by segment and interpreting the rules by each department, this method places the object relationship, condition relationship, exception relationship, and conclusion relationship in the approval rules within the same structural framework. This facilitates the identification of situations such as overlapping, inversion, insertion, and missing links in rule fragments, transforming the approval rules from a textual expression form into a slotted, chain-like, and connectable form.

[0016] (2) After the rule structure chain is formed, this method continues to classify the rule structure chains of each department, construct the target department's benchmark rule chain, and combine the cross-departmental mapping value and conclusion acceptance value to uniformly measure the object acceptance, condition inheritance, material alias merging, slot misalignment, and caliber drift between the rules of the preceding department and the rules received by the target department. Compared with the common "materials complete, push" or "preceding conclusions directly transferred" flow methods in the prior art, this invention no longer relies solely on a single field or a single conclusion for node transmission, but judges whether the preceding rules can be continued to be received by the target department based on the overall continuity status of the rule chain. It can identify problems such as unresolved exception clauses, material branches to be added, and rule calibers not connected, so that the rule connection relationship between cross departments has a clear structural basis, and also makes the rule basis before the target department receives it clearer.

[0017] (3) When the case file enters the specific circulation stage, this invention rewrites the case file information into a case file structure fragment chain and aligns it with the target department's rule unit slot by slot. It constructs a comprehensive circulation judgment value by combining the number of aggregated closure gaps, the number of cross-conflict points, the structural closure value, the cross-departmental mapping value, and the conclusion acceptance value. Then, it executes the freezing, pending review, or direct circulation judgment based on the statistical interval of similar historical case files. Compared with the existing methods that rely on fixed thresholds, manual supplementation, or serial transfer, this method can identify whether the case file has the basic continuation conditions, whether it is suitable to enter the review queue, or whether it has reached the direct reception state before entering the next approval node. It also writes the gap list, conflict list, rule hit record, and node reception log simultaneously, so that the case file circulation path, correction direction, and node reception basis all have corresponding data support. Overall, this method focuses on solving the problems of "difficult rule acceptance, difficult conclusion acceptance, and difficult case file judgment" in cross-departmental collaborative approval. Attached Figure Description

[0018] Figure 1This is a schematic diagram illustrating the steps of an automated workflow method for cross-departmental collaborative approval of government services according to the present invention. Figure 2 This is a schematic diagram of the workflow of an automated workflow system for cross-departmental collaborative approval of government services according to the present invention. Figure 3 This is a logic block diagram for an automatic workflow method for cross-departmental collaborative approval of government services according to the present invention. Detailed Implementation

[0019] 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.

[0020] Example 1 Please see Figure 1 This invention provides an automated workflow method for cross-departmental collaborative approval of government services. To achieve the above objectives, this invention employs the following technical solution, comprising the following steps: S1. Read the text data of the joint handling matters and perform semantic rule parsing to generate a set of rule candidates and calculate the positional discreteness S of each candidate sentence segment; S2. Perform segmentation and structured decomposition on the candidate rule set to form a rule structure chain. Segments not attached to the existing rule structure chain are recorded as closed gap segments, and the structure closure value is calculated. S3. Classify the rule structure chain, construct the target department benchmark rule chain, and calculate the cross-departmental mapping value G and conclusion acceptance value E from the rule structure chain to the target department. S4. Translate the current approval case file into a case file structure fragment chain and align it with the target department's rule unit slot by slot. Calculate the number of aggregated closure gaps B and the number of cross-conflict points K, and construct a comprehensive circulation judgment value T to determine the approval circulation.

[0021] In this embodiment, through steps S1 to S2, semantic rule parsing is first performed on the textual materials of the joint-processing matters. Then, the candidate rule segments are broken down into structural fragments such as the matter object, limiting conditions, material basis, exceptions, and processing conclusions, and further, a rule structure chain is formed. At the same time, closed gap fragments are identified and structural closure values ​​are calculated. Compared with the existing approval process, which mainly relies on manual reading of text and extraction of conditions based on experience, the above processing path transforms the rule content, which was originally scattered in institutional documents, service guides, historical correction notices, and approval opinion templates, into a structured expression that can be decomposed, linked, and compared. This allows the rule content to no longer remain at the level of natural language description, but to form a unified structural foundation for subsequent flow judgment. Thus, the tasks of pre-parsing, fragment consolidation, and chain expression of approval rules are completed, achieving the goal of transforming static text rules into contiguous rule units. Through S3, the rule structure chain is categorized and a baseline rule chain for the target department is constructed. Then, combined with the cross-departmental mapping value G and the conclusion acceptance value E, a unified analysis is performed on the object acceptance relationship, condition inheritance relationship, material name merging relationship, and caliber difference status between the preceding department rules and the target department rules. Compared with the existing technology's rough flow processing method based solely on item name, material catalog, or preceding conclusion, this processing path can determine whether there is a compatible relationship between cross-departments at the rule structure level. It can not only identify whether the rule content can be aligned, but also whether exception clauses have been closed, whether material branches still need to be added, and whether the caliber of the preceding and following department rules is consistent. Thus, the tasks of cross-departmental rule categorization, rule continuation foundation construction, and rule acceptance status calculation are completed, achieving the goal of establishing a clear acceptance basis for subsequent approval nodes. Through S4, the current approval file is transcribed into a file structure fragment chain and aligned slot by slot with the target department rule unit. Combined with the aggregation closure gap number B, the cross-conflict point number K, and the comprehensive flow judgment value T, it is determined whether the current file meets the conditions for entering the target department's approval stage. Compared to the linear processing methods commonly found in existing technologies, such as "processing upon completion of materials" or "direct transfer after submission of previous nodes," this processing path checks the structural integrity, rule conflict status, and cross-departmental acceptance status of the case file before it enters the next node. It then distinguishes between frozen, pending review, or direct processing, ensuring that the case file flow corresponds to rule matching, gaps, and conflicts. This completes the tasks of structured case file transcription, rule slot alignment, flow condition determination, and node branch processing, achieving the goal of making case file flow in cross-departmental collaborative approval more targeted, the judgment basis clearer, and the direction of correction more explicit.

[0022] Example 2 Please refer to Figure 3 Specifically: S1 includes S11; S11. After completing the qualification verification for reading authorized data from each department, read the text of the current joint handling matter, perform semantic rule parsing on the text in sequence, filter out the candidate rule segments containing approval condition semantics, and generate a set of candidate rules. The semantic rule parsing includes title level recognition, rule trigger word scanning, sentence segmentation and sentence dependency relationship extraction. The text materials include institutional documents, service guides, review caliber records, historical correction notices and historical approval opinion templates. The rule trigger word scanning is performed around the trigger words of the matter object, the trigger words of the limiting conditions, the trigger words of the material basis, the trigger words of the exception situation, and the trigger words of the processing conclusion.

[0023] S1 further includes S12; S12. For each rule candidate segment in the rule candidate set, locate the order of appearance of the object segment u, the limiting condition segment l, the material basis segment m, the exception segment x, and the processing conclusion segment c in the original text, and calculate the positional discreteness S of the rule candidate segment based on the order position. This is used to describe the discrete state of the arrangement of the five types of segments within the original rule text, as follows: S j =|u j -l j |+|l j -m j |+|m j -x j |+|x j -c j |; Among them, S j Let u represent the positional discrete quantity of the candidate sentence segment for the j-th rule. j l j m j x j and c j These represent the subject matter segment, limiting condition segment, supporting evidence segment, exception segment, and processing conclusion segment of the j-th rule candidate sentence segment, respectively. j -l j | indicates the positional interval between the item object fragment and the limiting condition fragment, |l j -m j | indicates the positional interval between the constraint fragment and the material basis fragment, |m j -x j | indicates the positional interval between material segments and exception segments, |x j -c j | indicates the positional interval between the exception segment and the processing conclusion segment.

[0024] In this embodiment, after the authorization data reading qualification verification of each department is passed, the system documents, service guides, review caliber records, historical correction notices, and historical approval opinion templates related to the current joint handling matter are read first. The text materials are then subjected to title level recognition, rule trigger word scanning, sentence segmentation, and sentence segment dependency extraction in sequence to identify the sentences carrying approval condition semantics and form a set of rule candidates. Subsequently, the order position of the matter object segment, limiting condition segment, material basis segment, exception segment, and processing conclusion segment in each rule candidate sentence segment is located, and the positional dispersion S is calculated based on the positional interval of each segment in the original text to characterize the arrangement of the five types of segments in the original text. A smaller positional dispersion S indicates that the internal structure of the current candidate sentence segment is closer to the chain arrangement required for subsequent rule decomposition; a larger positional dispersion S indicates that there are interleaved, inserted, or inverted segments in the current candidate sentence segment. Through the above processing, the semantic screening of the joint-process documents was first completed, followed by the quantitative representation of the internal structural order of the candidate rule segments. This transformed the natural language rules, originally scattered across different approval documents, into identifiable, sortable, and subsequently decomposable candidate structural units. Compared to existing technologies that primarily rely on manual text reading, extracting conditional statements based on experience, or initial screening solely based on keyword matching, this process first establishes text segment indexes for institutional documents, service guides, review records, historical correction notices, and historical approval opinion templates according to document source. Then, the clause hierarchy is identified based on title number, paragraph header identifier, and line break boundaries. Subsequently, within each clause hierarchy, a preset trigger word list is invoked to scan segments containing the matter name, qualification requirements, material name, exception description, and processing opinion. For the scanned segments, the dependency relationship is extracted, and adjacent segments with explanatory, limiting, or successive relationships with the currently matched segment are included in the same candidate rule segment. After obtaining the candidate rule segments, the five types of structural fragments within each segment are sequentially located. The position u of the object fragment, the position l of the limiting condition fragment, the position m of the material basis fragment, the position x of the exception fragment, and the position c of the processing conclusion fragment are recorded. Then, the positional discrepancy S of the candidate rule segment is calculated using the positional discrepancy formula. The positional discrepancy indicates whether the current candidate rule segment is suitable for direct entry into the next step of structural slot decomposition: when the positional intervals between the five types of fragments are relatively compact, the current candidate rule segment is given priority in the subsequent structured decomposition queue; when the positional discrepancy is large, it is treated as a key object for rearrangement in subsequent steps. The final output is a set of candidate rules containing the candidate rule segment text, source document identifier, the positional positions of the five types of fragments, and the positional discrepancy value.The above steps place the extraction of candidate segments and the determination of their internal order in the same processing chain. This not only identifies whether the rule content exists, but also whether the arrangement of the rule content in the original text is suitable for subsequent attachment. This provides a basis for subsequent segment division, structured decomposition, and rule chain construction, making the text screening criteria more consistent. The order in which candidate rules enter subsequent structural processing is clearer, and the decomposition deviation caused by differences in expression habits in the original text of the rules is also easier to distinguish.

[0025] Example 3 Please refer to Figure 3 Specifically: S2 includes S21; S21. Divide the candidate rule set into segments according to the item object slot, limiting condition slot, material basis slot, exception case slot and processing conclusion slot, and extract the number of item object segments O, limiting condition segments L, material basis segments M, exception case segments X and processing conclusion segments C in each candidate rule segment.

[0026] S2 also includes S22 and S23; S22. After segmentation, perform structured decomposition processing on each structural segment, specifically: attach the limiting condition segment to the corresponding item object segment, attach the material basis segment to the corresponding limiting condition segment, attach the exception segment to the limiting condition segment corresponding to the current exception, when the current limiting condition segment is already associated with the material basis segment, attach the exception segment to its corresponding limiting condition segment and the material basis segment connected to the current limiting condition segment, and attach the processing conclusion segment to the end of the structural chain formed by the sequential connection of the item object segment, limiting condition segment, material basis segment and exception segment to form a rule structure chain; S23. Segments that are not attached to existing rule structure chains are recorded as closed gap segments, and the number of closed gaps B is counted. For two or more parallel structure chains derived from the same item object segment in the same rule, they are recorded as parallel branches, and the number of parallel branches P is counted. Then, the number of parallel branches P is counted and compared with the positional discreteness S to calculate the structural closure value R of the rule candidate sentence segment, as follows. ; Among them, O j L j M j X j C j B j and P j These represent the number of event object segments, limiting condition segments, supporting material segments, exception segments, processing conclusion segments, closing gaps, and parallel bifurcations in the j-th rule candidate sentence segment, respectively. j +Lj +M j +X j +C j R represents the total number of segments in the candidate segment of rule j that have entered the five structural slots and been included in the linking chain. j This represents the structural closure value of the j-th rule candidate segment, used to indicate the degree of closure of the j-th rule candidate segment in the five types of structural slots.

[0027] In this embodiment, for the institutional documents, service guides, review records, historical correction notices, and historical approval opinion templates corresponding to the joint handling matters, the rule candidate sentence segments are first divided into segments based on the subject object slot, limiting condition slot, material basis slot, exception situation slot, and processing conclusion slot. The number of subject object segments (O), limiting condition segments (L), material basis segments (M), exception situation segments (X), and processing conclusion segments (C) are extracted respectively. Based on this, the following connection order is followed: "Subject Object—Limiting Conditions—Material Basis—Exception Situation—Processing Conclusion". The process involves structured decomposition, linking conditional segments to corresponding object segments, material basis segments to corresponding conditional segments, exception segment segments to corresponding condition chains or material chains, and processing conclusion segments to the end of the structural chain, forming a rule structure chain. Segments that fail to enter existing links are recorded as closed gap segments. The number of parallel branches P derived from the same object segment is counted, and the structural closure value R is calculated by combining this with the positional dispersion S. This value characterizes the chain-like convergence state of the current rule candidate segments in the five slot types. Through this implementation, approval rules, originally scattered in natural language, are transformed into structural chains with slot boundaries and linking orders. This eliminates the limitations of manual reading, experience extraction, or keyword matching. Instead, it identifies which segments within the rule have formed sequential relationships, which segments remain in a free state, and which segments have parallel branches or positional shifts, giving the internal organization of the rule text a statistically comparable and comparable structural expression. In the specific processing, each rule candidate segment output by S1 is first slotted. Semantic fragments representing the approval object are placed in the "matter object" slot; semantic fragments representing eligibility requirements, scope of application, preconditions, or status constraints are placed in the "limiting condition" slot; semantic fragments representing the names of supporting materials, evidence, attachments, or verification basis are placed in the "material basis" slot; semantic fragments representing exemption, incomplete submission, alternative submission, supplementary verification, or special application are placed in the "exceptional situation" slot; and semantic fragments representing acceptance, return for correction, suspension of review, transfer to manual verification, or termination of processing are placed in the "processing conclusion" slot. After the five slot categories are defined, the fragments within each slot are then linked accordingly: using the "matter object" segment as the head of the chain, limiting condition fragments related to that "matter object" are linked after the "matter object" segment; then, material basis fragments supporting these limiting conditions are linked after the corresponding limiting condition fragment; for exception situation fragments, they are linked to the chain segment containing the corresponding limiting condition fragment or material basis fragment according to their constraint object; and the processing conclusion fragment is linked to the end of the already formed complete condition chain. If a segment cannot find a corresponding attachment position in the current rule, then the segment is recorded as a closed gap segment and included in B; if multiple parallel constraint condition chains are attached to a certain item object, then the structure is recorded as a parallel fork and included in P. Based on this, the j-th rule is calculated using the structure closure value formula.The numerator of the structural closure value corresponds to the total number of segments identified and included in the structural chain, while the denominator corresponds to the structural discreteness formed by closure gaps, parallel bifurcations, and positional swings. This calculation result allows for the differentiation of rules: rules with higher structural closure values ​​are prioritized for the next step of cross-departmental mapping; rules with lower structural closure values ​​retain their source and gap identifiers during subsequent mapping processes. The final result is a set of rule chains containing five types of structural segment chains, their connection relationships, closure gap locations, parallel bifurcation locations, and structural closure values. Based on this processing path, on the one hand, a unified rule foundation can be laid for the subsequent construction of target department benchmark rule chains, cross-departmental mapping value calculation, and conclusion acceptance value analysis; on the other hand, it can make the conditional acceptance relationships, material support relationships, exception attribution relationships, and conclusion landing point relationships in the approval rules clearer, providing a clear structural basis for subsequent case file transcription, slot-by-slot alignment, and circulation determination.

[0028] In actual implementation, calculating the structural closure value R based on the comparison between "the total number of segments that have entered the structural slot and formed the mounting foundation" and "the abnormal quantity affecting the structural closure state" is more in line with engineering processing logic. Specifically, the number of object segments O, limiting condition segments L, material basis segments M, exception segment segments X, and processing conclusion segments C are obtained directly by performing slot identification on the same rule candidate segment. The sum of these five items reflects the total amount of structural content in the segment that has been identified and can participate in the organization of the object chain, condition chain, material chain, exception chain, and conclusion chain. Therefore, placing it at the beginning of the formula is to characterize how many basic segments the rule segment has that can be used to form chains. The number of closure gaps B comes from the number of free segments that failed to be incorporated into the existing rule structural chain during the mounting process. The larger this value, the more fragments are in the segment. The more content that cannot fall into the established structural relationship, the more complex the subsequent connection relationship becomes. The number of parallel branches P comes from the number of parallel branches derived from the same object fragment. The larger this value is, the more parallel condition chains or parallel conclusion chains exist under the same object. The positional dispersion S comes from the positional interval statistics of the object, limiting conditions, material basis, exceptions and processing conclusions in the original text. The larger this value is, the more inversions, insertions or interleavings exist in the original text. The more irregular the connection order after disassembly is. Therefore, the positional dispersion S, the number of parallel branches P and the positional dispersion S are accumulated as constraints in the subsequent terms because these three types of quantities correspond to the three sources of structural non-convergence: no connection, multiple branches and positional disturbance. The constant 1 is set in the denominator to avoid the denominator from having a zero value when there are no closed gaps, parallel branches and positional dispersion. Calculated in this way, the larger the structural closure value R, the more structural segments that can be attached to the unit anomaly, indicating that the candidate sentence segment of the rule is closer to the state that can be directly entered into the rule structure chain; the smaller the structural closure value R, the more structural segments there may be in the sentence segment, but these segments are more affected by gaps, forks and positional disturbances, and are more suitable for entering the process of disassembly, attachment or manual review.

[0029] Example 4 Please refer to Figure 3 Specifically: S3 includes S31 and S32; S31. Read the rule structure chain of each department and classify the rule structure chain according to the name of the joint handling item. In the classified rule structure chain, extract the rule structure chain set corresponding to the target department. Arrange the item object slot, limiting condition slot, material basis slot, exception situation slot and processing conclusion slot with the highest frequency in the rule structure chain set corresponding to the target department according to the attachment order to generate the target department's benchmark rule chain. S32. Calculate the number of slots A for the alignment of the target department's benchmark rule chain with the following parameters: number of slots H for the inheritance of the constraint condition, number of material alias merged logs N, number of slot type misalignment Q, and number of caliber drifts D. Based on the statistical results, calculate the cross-departmental mapping value G from the rule structure chain to the target department, as follows. ; Among them, G j,k A represents the cross-departmental mapping value of the j-th rule structure chain in target department k. j,k H represents the number of slots in the j-th rule structure chain that can be directly aligned with the baseline rule chain of the target department k in the item object slot. j,k N represents the number of constraint inheritance slots in the j-th rule structure chain that are continued by the target department k. j,k Q represents the merge logarithm of material aliases in the j-th rule structure chain, where material names are merged with material names in the target department k. j,k D represents the number of slot misalignments between the j-th rule structure chain and the baseline rule chain of the target department k. j,k This represents the caliber drift number between the j-th rule structure chain and the target department k.

[0030] S3 also includes S33; S33. After obtaining the cross-departmental mapping value G corresponding to all rule structure chains of the target department, count the number of rules n mapped to the current target department, and perform an exception slot comparison between the rule structure chain of the preceding department and the rule receiving structure of the target department to screen out exception fragments that are not attached to the corresponding limiting condition chain or material basis chain, and count the number of exception fragments to obtain the number of exceptions that are still not closed. Then, perform a material basis slot comparison between the rule structure chain of the preceding department and the rule receiving structure of the target department to screen out material branches that have set material basis receiving slots in the rule receiving structure of the target department but have not formed corresponding material basis fragments in the rule structure chain of the preceding department, and count the number of material branches to obtain the number of material branches that are still to be added. The conclusion acceptance value E of the target department k is calculated based on the number of unresolved exceptions Y and the number of material branches F to be added. k The details are as follows; ; Where, n k Y represents the number of rules that are mapped to the current target department k, with the target department as the unit. k This indicates the number of exceptions for target department k that have not yet been explicitly assigned to a specific condition chain, F. k This indicates the number of material branches that still need to be added to the current rule receiving structure of the target department k.

[0031] In this embodiment, the rule structure chains of each department output by S2 are first read, and then compared line by line based on the name of the joint handling item, the semantics of the item object, the material pointing relationship, and the conclusion attribution relationship. During the comparison, it first identifies whether the item object names in different departments belong to the same object expression, and then identifies whether the preceding department's limiting conditions can be used as a precondition for the target department. Next, materials in different departments pointing to the same proof content but with different names are merged using aliasing. Then, it checks whether exceptions can be passed to the target department along the original condition chain. Finally, it determines whether the processing conclusion formed by the preceding department has a basis for continuing to attach the condition chain in the target department. In the above comparison process, item object slots that can be directly aligned are counted as A. j,k For slots with limited conditions that can continue to be accepted, H is included. j,k For material alias correspondences that can be merged, they are included in N. j,k Cases where the rule structure chain and the target department's receiving structure are inconsistent in slot type are included in Q. j,k For the same matter, if there are differences in the description of conditions, the terminology of materials, or the wording of conclusions, it will be counted as D. j,k The cross-departmental mapping value G was then calculated using the cross-departmental mapping value formula. j,k This reflects the continuity of the j-th rule structure chain in the target department k. When the cross-department mapping value G is in a higher range, it indicates that the rule structure chain formed by the preceding department is relatively close to the receiving structure of the target department k in terms of the matter object, limiting conditions, and material basis; when the cross-department mapping value G is in a lower range, it indicates that there are still many slot misalignments or scope deviations in the target department k. After completing the mapping calculation of all rule structure chains to the target department k, the number C of the processing conclusion fragments formed in each rule structure chain mapped to that department is then calculated. j With the corresponding cross-departmental mapping value G j,k Summarize the data and combine it with the number of outstanding exceptions (Y) in the target department. k And the number of material branches F that still need to be added k The conclusion acceptance value E of the target department k is calculated. kThe conclusion acceptance value E represents the target department's acceptance status of the preceding approval rule chain: when the processing conclusion can enter the target department along the existing condition chain, and the gaps between the exception clause and the material branch are small, the conclusion acceptance value E is in a higher range; when the exception clause is still outside the condition chain, or the material branch has not yet been completed, the conclusion acceptance value E is in a lower range. The final output is the cross-departmental mapping value set G corresponding to the target department and the conclusion acceptance value E. Through the above implementation methods, the rule content, which originally remained at the level of departmental textual descriptions, is transformed into a categorizable, comparable, and inheritable slot chain structure. This not only completes tasks such as constructing the target department's baseline rule chain, analyzing cross-departmental rule continuity, identifying unresolved exceptions, and identifying missing material branches, but also enables the target department to form detailed judgment criteria based on the object's acceptance, conditional continuation, material merging, slot misconnection, and scope deviation when receiving previous approval results, rather than relying solely on the item name, material catalog, or manual experience. Compared to the current common method of directly transferring based on whether materials are complete or whether previous conclusions exist, the above processing path provides a corresponding structural basis for cross-departmental rule continuity relationships, unresolved exception positions, and supplementary material branch positions. The target department has a clearer understanding of the acceptance boundaries, acceptance conditions, and correction directions of previous rules, and the flow and connection relationships between departments are also easier to unify and organize.

[0032] In practice, the cross-departmental mapping value G is written in the form of "continuous items above, obstructive items below" because when approval rules flow from the preceding department to the target department, what truly determines whether a connection can be established is not simply looking at a single piece of material or a single conclusion, but rather how much content in the rule structure chain can be directly attached to the target department's baseline rule chain: Specifically, the number of slots for item alignment (A) indicates whether the preceding and following departments are on the same main line of acceptance for the same approval item; the number of slots for constraint inheritance (H) indicates whether conditions already judged by the preceding department can be continued by the target department; and the number of material alias merging logarithms (N) indicates whether, although the material... Although the naming conventions differ, the same material based slot can still be incorporated into the receiving rules. Therefore, the item object alignment slot number A, the constraint condition inheritance slot number H, and the material alias merging logarithm N are all continuable content and are placed in the numerator to represent the continuation basis when the rule enters the target department. The slot misalignment number Q indicates that although the segment exists, the attachment position is incorrect, and the caliber drift number D indicates that there is a discrepancy in the judgment caliber of the preceding and following departments regarding the same item. These two items will interfere with the continued attachment of the rule, so they are placed in the denominator to represent the resistance to rule continuation. The "1" in the denominator is used to retain the baseline division term to prevent errors when the slot misalignment number Q and the caliber drift number D are zero. Currently, there are no defined conditions. This calculation method aligns with the continuation judgment logic in engineering: the more content that can be directly accepted, the larger the cross-departmental mapping value G; the more slot misconnections and caliber offsets, the smaller the cross-departmental mapping value G. Similarly, the conclusion acceptance value E is calculated by dividing the sum of the previous conclusion's basis and the mapping state by the number of unresolved issues. This is because whether the target department can accept the previous conclusion depends not only on the number of processing conclusion fragments C and the mapping state G between these conclusions and the target department, but also on whether there are still exceptions Y in the current rule chain that have not been included in the specific condition chain or material chain, and whether the target department has reserved... The more material branches F that have not yet been completed in the preceding rules, the more exceptions Y that have not yet been closed, and the more material branches F that are currently waiting to be added, the more it indicates that although the preceding conclusion has been formed, there are still gaps in its acceptance boundary and support chain. Therefore, these are placed in the denominator to reduce E. In other words, both formulas essentially put "acceptable content" and "problems that block acceptance" into the same judgment framework. The purpose is to express with structured quantitative relationships whether cross-departmental rules can be accepted, to what extent they can be accepted, and where they are currently stuck, rather than directly determining whether they can be transferred based solely on the consistency of the item name or the similarity of the material catalog.

[0033] Example 5 Please refer to Figure 3 Specifically: S4 includes S41; S41. Call the item name, applicant information, electronic material catalog, supplementary explanation, and processing results formed by the preceding department corresponding to the current approval case file. Perform structured decomposition and transcribing through S2 into a case file structure fragment chain. Then, perform slot-by-slot alignment between the case file structure fragment chain and the target department rule unit. Identify the item object slots, limiting condition slots, material basis slots, exception situation slots, and processing conclusion slots that have been satisfied in the case file. Calculate the number of aggregation closure gaps B and the number of cross-conflict points K of the target department. Among them, the rule unit of the target department is a slotted continuation unit composed of the rule structure chain mapped to the target department in S3 and its corresponding item object receiving slot, limiting condition receiving slot, material basis receiving slot, exception situation receiving slot and processing conclusion receiving slot. The rule unit is used to represent the object type, condition chain type, material support type, exception attribution type and preceding conclusion status type that the target department is allowed to receive at the current approval node. The number of aggregated closure gaps B represents the sum of the number of unfilled item object slots, limiting condition slots, material basis slots, exception situation slots and processing conclusion slots. The number of cross conflict points K represents the sum of the number of object attachment conflicts, condition correspondence conflicts and path status conflicts.

[0034] S4 further includes S42 and S43; S42. After obtaining the number of aggregation closure gaps B and the number of cross-conflict points K, combine the structural closure value R of each rule structure chain, the cross-departmental mapping value G of each rule structure chain to the target department, and the conclusion acceptance value E of the target department to calculate the comprehensive flow judgment value T of the target department, as follows; ; Among them, T k B represents the comprehensive turnover judgment value T for target department k. k and K k These represent the number of aggregation closure gaps and the number of cross-conflict points for target department k, respectively. This reflects the overall closed-loop state of the rules mapped to the target department. It reflects the overall mapping status of each rule structure chain across departments; S43. Read the comprehensive transfer judgment value sequence of similar historical case files in the target department, and extract the median value of the comprehensive transfer judgment value sequence using the percentile method as the initial entry value U, extract the 75th percentile value of the comprehensive transfer judgment value sequence as the direct transfer value V, and make approval and transfer judgment with the real-time acquired comprehensive transfer judgment value T, as follows; When the comprehensive transfer judgment value T < the initial entry value U, it means that the current case file has not yet entered the basic transfer range in the rule continuation state of the target department, and the case file structure chain cannot be connected to the standard review chain of the target department. At this time, the push action of the current case file to the target department is frozen, and a gap list or conflict list is generated and written into the transfer record of the current case file. When the initial entry value U ≤ comprehensive circulation judgment value T < direct circulation value V, it means that the current case file has entered the circulation range in the target department, but has not yet entered the direct circulation range. At this time, the execution node push will attach the current case file to the pending queue of the target department, and then the review task will be assigned to the corresponding review port or manual review seat. When the comprehensive transfer judgment value T ≥ the direct transfer value V, it means that the current case file has entered the direct transfer range in the target department. At this time, the node receiving log, transfer time identifier and rule hit record are directly written, and the approval action of the target department is triggered.

[0035] In this embodiment, firstly, based on the formed rule structure chain, the target department's baseline rule chain, the structural closure value R, the cross-departmental mapping value G, and the conclusion acceptance value E, the item name, applicant information, electronic material catalog, supplementary explanation, and previous department processing results in the current approval file are read. Following the segment division and structured decomposition method in S2, the file content is transcribed into a file structure fragment chain. Subsequently, the file structure fragment chain is aligned slot-by-slot with the target department's rule units, checking the filling status of the item object slot, limiting condition slot, material basis slot, exception situation slot, and processing conclusion slot. The number of unfilled aggregate closure gaps B and the number of cross-conflict points K formed by object attachment conflicts, condition correspondence conflicts, and path status conflicts are counted. Then, combined with the structural closure value R of each rule structure chain, the cross-departmental mapping value G of each rule structure chain to the target department, and the conclusion acceptance value E of the target department, the comprehensive flow judgment value T is calculated, and historical similar files in the target department are read. The comprehensive flow judgment value sequence is used to extract the initial entry value U and the direct flow value V using the percentile method. The current case file is then frozen, pushed for review, or directly flowed. After adopting the above implementation method, whether a case file meets the conditions for entering the target department's approval node is no longer judged solely based on whether the materials have been submitted or whether the previous conclusion exists. Instead, it is judged uniformly based on the rule continuity status, structural gap status, conflict status, and cross-departmental acceptance status. This allows the missing slot content, the structural conflict, and the receiving level to be identified before the case file is flowed. At the same time, the gap list, conflict list, rule hit record, node receiving log, and flow time identifier are written into the flow record. This ensures that the node receiving basis, correction direction, and flow path in cross-departmental approval all have a corresponding structural basis. The need for repeated manual verification of the original text and repeated returns for correction is also eliminated. The rule continuity relationship, case file linkage relationship, and approval branch relationship between departments are also clearer.

[0036] In practice, the comprehensive transfer judgment value T formula breaks down whether the target department k has the conditions to receive the case file at the current moment into three parts for calculation: "whether the rule foundation is formed," "whether cross-departmental rules can be connected," and "whether there are still gaps and conflicts in the current case file." The total is obtained by statistically analyzing the structural closure value of each rule structure chain mapped to the target department. This calculation is because in actual approval, the focus is not on whether a single rule is complete, but on whether the entire set of rules that the department is currently responsible for has formed a usable chain structure. This involves accumulating the cross-departmental mapping values ​​from all rule structures to the target department. It represents how many of the objects, conditions, materials, exceptions, and conclusions formed by preceding departments can be received by the target department according to its existing slots. This total calculation aligns with the engineering scenario where multiple preceding rules converge on a single target department for joint processing matters. Multiplying these two totals is necessary because the rule's "self-closure" and "ability to map to the target department" must both be true simultaneously. If one is low, even if the other is high, the current node should not directly accept it. Therefore, a product is used to express the continuation state where both constraints are simultaneously satisfied. Adding the conclusion acceptance value E is because it reflects the target department's absorption of the preceding rule conclusions, especially the impact of unclosed exceptions and supplementary material branches on the continuation end. Therefore, it is added to the numerator as an acceptance correction to the aforementioned product result, indicating that it's necessary to consider not only whether the rule's front end can be attached but also whether the conclusion's end can be received. The denominator 1+B... k +K k This corresponds to the number of aggregation closure gaps and cross-conflict points that still exist when the current case file enters the target department. The reason for adopting the form of "1 + gap + conflict" is that the more gaps and conflicts there are, the less suitable the current case file is for entering the target department's standard review chain. The constant 1 is used to ensure that the denominator can still be calculated when there are no gaps or conflicts. Therefore, the comprehensive circulation judgment value T obtained by the whole formula essentially represents "the degree of node access after the current case file gaps and conflicts are reduced on the basis of the existing rule structure, cross-department mapping, and conclusion acceptance". The larger the value, the closer the continuity state of the case file and the target department's rule unit is to the direct circulation interval.

[0037] Example 6 Please refer to Figure 2 An automated workflow system for cross-departmental collaborative approval of government services includes a rule source screening module, a disassembly and attachment module, a mapping and acceptance module, and a case file alignment and branching module. The rule source screening module is used to read the text data of the joint handling matters and perform semantic rule parsing to generate a set of rule candidates and calculate the positional discreteness S of each candidate sentence segment; The disassembly and attachment module is used to divide and disassemble the rule candidate set into segments to form a rule structure chain. Segments that are not attached to the existing rule structure chain are recorded as closed gap segments, and the structure closure value is calculated. The mapping and acceptance module is used to classify the rule structure chain, construct the target department benchmark rule chain, and calculate the cross-departmental mapping value G and conclusion acceptance value E from the rule structure chain to the target department. The case file alignment branch module is used to transcribe the current approval case file into a case file structure fragment chain and align it with the target department rule unit slot by slot, count the number of aggregated closure gaps B and the number of cross-conflict points K, and construct a comprehensive circulation judgment value T for approval circulation judgment.

[0038] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended technical solutions and their equivalents.

Claims

1. An automated workflow method for cross-departmental collaborative approval of government services, characterized in that: Includes the following steps: S1. Read the text data of the joint handling matters and perform semantic rule parsing to generate a set of rule candidates and calculate the positional discreteness S of each candidate sentence segment; S2. Perform segmentation and structured decomposition on the candidate rule set to form a rule structure chain. Segments not attached to the existing rule structure chain are recorded as closed gap segments, and the structure closure value is calculated. S3. Classify the rule structure chain, construct the target department benchmark rule chain, and calculate the cross-departmental mapping value G and conclusion acceptance value E from the rule structure chain to the target department. S4. Translate the current approval case file into a case file structure fragment chain and align it with the target department's rule unit slot by slot. Calculate the number of aggregated closure gaps B and the number of cross-conflict points K, and construct a comprehensive circulation judgment value T to determine the approval circulation.

2. The automatic workflow method for cross-departmental collaborative approval of government services according to claim 1, characterized in that: S1 includes S11; S11. Read the text data of the current joint handling matter, perform semantic rule parsing on the text data in sequence, filter out the rule candidate sentence segments containing approval condition semantics, and generate a rule candidate set; The semantic rule parsing includes title level recognition, rule trigger word scanning, sentence segmentation and sentence dependency relationship extraction. The text materials include institutional documents, service guides, review caliber records, historical correction notices and historical approval opinion templates. The rule trigger word scanning is performed around the trigger words of the matter object, the trigger words of the limiting conditions, the trigger words of the material basis, the trigger words of the exception situation, and the trigger words of the processing conclusion.

3. The automatic workflow method for cross-departmental collaborative approval of government services according to claim 2, characterized in that: S1 further includes S12; S12. For each rule candidate segment in the rule candidate set, locate the order of appearance of the object segment u, the limiting condition segment l, the material basis segment m, the exception segment x, and the processing conclusion segment c in the original text, and calculate the positional discreteness S of the rule candidate segment based on the order position. This is used to describe the discrete state of the arrangement of the five types of segments within the original rule text, as follows: S j =|u j -l j |+|l j -m j |+|m j -x j |+|x j -c j |; Among them, S j Let u represent the positional discrete quantity of the candidate sentence segment for the j-th rule. j l j m j x j and c j These represent the subject matter segment, limiting condition segment, material basis segment, exception segment, and processing conclusion segment of the candidate sentence segment for rule j, respectively.

4. The automatic workflow method for cross-departmental collaborative approval of government services according to claim 3, characterized in that: S2 includes S21; S21. Divide the candidate rule set into segments according to the item object slot, limiting condition slot, material basis slot, exception case slot and processing conclusion slot, and extract the number of item object segments O, limiting condition segments L, material basis segments M, exception case segments X and processing conclusion segments C in each candidate rule segment.

5. The automatic workflow method for cross-departmental collaborative approval of government services according to claim 4, characterized in that: S2 also includes S22 and S23; S22. After segmentation, perform structured decomposition processing on each structural segment, specifically: attach the limiting condition segment to the corresponding item object segment, attach the material basis segment to the corresponding limiting condition segment, attach the exception segment to the limiting condition segment corresponding to the current exception, when the current limiting condition segment is already associated with the material basis segment, attach the exception segment to its corresponding limiting condition segment and the material basis segment connected to the current limiting condition segment, and attach the processing conclusion segment to the end of the structural chain formed by the sequential connection of the item object segment, limiting condition segment, material basis segment and exception segment to form a rule structure chain; S23. Segments that are not attached to existing rule structure chains are recorded as closed gap segments, and the number of closed gaps B is counted. For two or more parallel structure chains derived from the same item object segment in the same rule, they are recorded as parallel branches, and the number of parallel branches P is counted. Then, the number of parallel branches P is counted and compared with the positional discreteness S to calculate the structural closure value R of the rule candidate sentence segment, as follows. ； Among them, O j L j M j X j C j B j and P j These represent the number of event object segments, limiting condition segments, supporting material segments, exception segments, processing conclusion segments, closing gaps, and parallel bifurcations in the j-th rule candidate sentence segment, respectively. j +L j +M j +X j +C j R represents the total number of segments in the candidate sentence segment of rule j that have entered the five structural slots and been included in the linking chain. j This represents the structural closure value of the j-th rule candidate segment, used to indicate the degree of closure of the j-th rule candidate segment in the five types of structural slots.

6. The automatic workflow method for cross-departmental collaborative approval of government services according to claim 5, characterized in that: S3 includes S31 and S32; S31. Read the rule structure chain of each department and classify the rule structure chain according to the name of the joint handling item. In the classified rule structure chain, extract the rule structure chain set corresponding to the target department. Arrange the item object slot, limiting condition slot, material basis slot, exception situation slot and processing conclusion slot with the highest frequency in the rule structure chain set corresponding to the target department according to the attachment order to generate the target department's benchmark rule chain. S32. Calculate the number of slots A for the alignment of the target department's benchmark rule chain with the following parameters: number of slots H for the inheritance of the constraint condition, number of material alias merged logs N, number of slot type misalignment Q, and number of caliber drifts D. Based on the statistical results, calculate the cross-departmental mapping value G from the rule structure chain to the target department, as follows. ； Among them, G j,k A represents the cross-departmental mapping value of the j-th rule structure chain in target department k. j,k H represents the number of slots in the j-th rule structure chain that can be directly aligned with the baseline rule chain of the target department k in the item object slot. j,k N represents the number of constraint inheritance slots in the j-th rule structure chain that are continued by the target department k. j,k Q represents the merge logarithm of material aliases in the j-th rule structure chain, where material names are merged with material names in the target department k. j,k D represents the number of slot misalignments between the j-th rule structure chain and the baseline rule chain of the target department k. j,k This represents the caliber drift number between the j-th rule structure chain and the target department k.

7. The automatic workflow method for cross-departmental collaborative approval of government services according to claim 6, characterized in that: S3 also includes S33; S33. After obtaining the cross-departmental mapping value G corresponding to all rule structure chains of the target department, count the number of rules n mapped to the current target department, and perform an exception slot comparison between the rule structure chain of the preceding department and the rule receiving structure of the target department to screen out exception fragments that are not attached to the corresponding limiting condition chain or material basis chain, and count the number of exception fragments to obtain the number of exceptions that are still not closed. Then, perform a material basis slot comparison between the rule structure chain of the preceding department and the rule receiving structure of the target department to screen out material branches that have set material basis receiving slots in the rule receiving structure of the target department but have not formed corresponding material basis fragments in the rule structure chain of the preceding department, and count the number of material branches to obtain the number of material branches that are still to be added. The conclusion acceptance value E of the target department k is calculated based on the number of unresolved exceptions Y and the number of material branches F to be added. k The details are as follows; ； Where, n k Y represents the number of rules that are mapped to the current target department k, with the target department as the unit. k This indicates the number of exceptions for target department k that have not yet been explicitly assigned to a specific condition chain, F. k This indicates the number of material branches that still need to be added to the current rule receiving structure of the target department k.

8. The automatic workflow method for cross-departmental collaborative approval of government services according to claim 7, characterized in that: S4 includes S41; S41. Call the item name, applicant information, electronic material catalog, supplementary explanation, and processing results formed by the preceding department corresponding to the current approval case file. Perform structured decomposition and transcribing through S2 into a case file structure fragment chain. Then, perform slot-by-slot alignment between the case file structure fragment chain and the target department rule unit. Identify the item object slots, limiting condition slots, material basis slots, exception situation slots, and processing conclusion slots that have been satisfied in the case file. Calculate the number of aggregation closure gaps B and the number of cross-conflict points K of the target department. Among them, the rule unit of the target department is a slotted continuation unit composed of the rule structure chain mapped to the target department in S3 and its corresponding item object receiving slot, limiting condition receiving slot, material basis receiving slot, exception situation receiving slot and processing conclusion receiving slot. The rule unit is used to represent the object type, condition chain type, material support type, exception attribution type and preceding conclusion status type that the target department is allowed to receive at the current approval node. The number of aggregated closure gaps B represents the sum of the number of unfilled item object slots, limiting condition slots, material basis slots, exception situation slots and processing conclusion slots. The number of cross conflict points K represents the sum of the number of object attachment conflicts, condition correspondence conflicts and path status conflicts.

9. The automatic workflow method for cross-departmental collaborative approval of government services according to claim 8, characterized in that: S4 further includes S42 and S43; S42. After obtaining the number of aggregation closure gaps B and the number of cross-conflict points K, combine the structural closure value R of each rule structure chain, the cross-departmental mapping value G of each rule structure chain to the target department, and the conclusion acceptance value E of the target department to calculate the comprehensive flow judgment value T of the target department, as follows; ； Among them, T k B represents the comprehensive turnover judgment value T for target department k. k and K k These represent the number of convergence closure gaps and the number of cross-conflict points in target department k, respectively. S43. Read the comprehensive transfer judgment value sequence of similar historical case files in the target department, and extract the median value of the comprehensive transfer judgment value sequence using the percentile method as the initial entry value U, extract the 75th percentile value of the comprehensive transfer judgment value sequence as the direct transfer value V, and make approval and transfer judgment with the real-time acquired comprehensive transfer judgment value T, as follows; When the comprehensive transfer judgment value T < the initial entry value U, it means that the current case file has not yet entered the basic transfer range in the rule continuation state of the target department, and the case file structure chain cannot be connected to the standard review chain of the target department. At this time, the push action of the current case file to the target department is frozen, and a gap list or conflict list is generated and written into the transfer record of the current case file. When the initial entry value U ≤ comprehensive circulation judgment value T < direct circulation value V, it means that the current case file has entered the circulation range in the target department, but has not yet entered the direct circulation range. At this time, the execution node push will attach the current case file to the pending queue of the target department, and then the review task will be assigned to the corresponding review port or manual review seat. When the comprehensive transfer judgment value T ≥ the direct transfer value V, it means that the current case file has entered the direct transfer range in the target department. At this time, the node receiving log, transfer time identifier and rule hit record are directly written, and the approval action of the target department is triggered.

10. An automated workflow system for cross-departmental collaborative approval of government services, comprising the automated workflow method for cross-departmental collaborative approval of government services as described in any one of claims 1-9, characterized in that: This includes a rule source filtering module, a disassembly and attachment module, a mapping and acceptance module, and a case file alignment and branching module; The rule source screening module is used to read the text data of the joint handling matters and perform semantic rule parsing to generate a set of rule candidates and calculate the positional discreteness S of each candidate sentence segment; The disassembly and attachment module is used to divide and disassemble the rule candidate set into segments to form a rule structure chain. Segments that are not attached to the existing rule structure chain are recorded as closed gap segments, and the structure closure value is calculated. The mapping and acceptance module is used to classify the rule structure chain, construct the target department benchmark rule chain, and calculate the cross-departmental mapping value G and conclusion acceptance value E from the rule structure chain to the target department. The case file alignment branch module is used to transcribe the current approval case file into a case file structure fragment chain and align it with the target department rule unit slot by slot, count the number of aggregated closure gaps B and the number of cross-conflict points K, and construct a comprehensive circulation judgment value T for approval circulation judgment.

Images