A method for generating work order summaries based on low-code configuration

Abstract

The present invention discloses a method for generating a work order summary based on a low-code configuration, which relates to the technical field of work order summary generation and includes the following steps: based on a constructed directed graph structure, identifying all field combinations in which fields are configured to reference each other, including performing a path analysis operation in the directed graph structure to screen out all field combinations with direct or indirect bidirectional reference relationships; extracting the path depth of all identified field combinations, the number of times the fields are referenced in different paths, and the number of paths with loops in the constructed directed graph structure, and analyzing the degree of closure of the field reference paths of each identified field combination based on the extraction results; classifying each identified field combination according to the evaluation results, and determining whether to interrupt the summary generation process and execute the corresponding control strategy. The present invention solves the problem that circular references of fields in low-code configurations are difficult to detect, and realizes intelligent identification and risk control of field reference structures.

Description

Technical Field

[0001] The present invention relates to the technical field of work order summary generation, and in particular to a work order summary generation method based on low-code configuration. Background Art

[0002] Work order summary generation based on low-code configuration refers to an intelligent method that can automatically generate summaries of work order content through a graphical interface or visualization module without writing a lot of code. This method combines the summary writing process in the traditional work order management system with the low-code development platform, allowing users to define the summary generation logic by simply dragging controls, configuring rules, or selecting templates, such as setting extraction fields, summary rules, text compression algorithms, or trigger conditions, thereby achieving customized and automated work order summary output. Compared with the traditional method that relies on manual writing or hard-coded development, the solution based on low-code configuration greatly reduces the development and operation and maintenance threshold, allowing non-technical personnel to quickly deploy and adjust summary generation rules according to business needs, while having higher scalability and adaptability; especially in the changing and complex enterprise service scenarios, this method not only improves the efficiency and consistency of work order processing, but also significantly optimizes the intelligence level and user experience of the work order system, so it has broad application value and promotion prospects.

[0003] Existing low-code ticket summary generation technologies typically integrate a visual configuration platform with a text processing engine to implement a user-customizable summary generation process. The entire process begins with data source configuration. Through a low-code interface, users can select data fields from the ticket system, such as incident descriptions, handling procedures, and customer feedback, as the raw input for summary generation. Next, they enter the rule-setting phase. Without writing code, users simply drag and drop controls, set parameters, or select pre-set templates to define summary generation logic, such as keyword extraction, semantic classification, text compression strategies, and trigger conditions (such as state changes and time points). The system then intelligently processes the selected fields using an integrated NLP engine, extracting key points and organizing them into structured or natural language summaries. Finally, the summary output can be flexibly integrated into various interfaces or systems through a configurable display module, such as embedding into a CRM, pushing to a customer service notification platform, or generating PDF reports. This "low-code + configurable" approach allows non-technical users to flexibly build a ticket summary generation mechanism tailored to their business scenarios, avoiding complex custom development and deployment processes while facilitating future maintenance and expansion of the system.

[0004] The existing technology has the following deficiencies:

[0005] When using a low-code configuration platform to generate work order summaries, if the configuration personnel set the reference relationship between fields by means of "calculated fields", for example, configuring field A as "the summary of field B + field C" and setting field C as "the summary paragraph excerpt of field A", a circular reference relationship between fields will be inadvertently formed. In this case, because the low-code platform allows field values ​​to dynamically depend on the output results of other fields during summary generation, and the platform does not establish a directed dependency graph of the field reference path, nor does it set up a circular detection mechanism, the system cannot recognize that field A depends on field C, and field C in turn depends on field A, thus forming a logical closed loop. Because this closed loop has no intuitive identification in the visual configuration interface, configuration personnel usually find it difficult to detect the existence of this problem, which in turn causes the summary generation engine to enter an infinite recursive state during execution. The existing work order summary generation technology based on low-code configuration cannot determine whether the summary generation process should be interrupted based on the degree of closure of the field reference path when the fields are configured to reference each other. As a result, the system summary thread continues to run and cannot exit, causing computing resources to be occupied indefinitely and generation tasks to be blocked. Not only will the summary output fail, but it will also slow down or even interrupt the execution of other summary tasks, thereby affecting the automation efficiency and platform stability of the entire work order processing process.

[0006] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not form the prior art that is already known to a person of ordinary skill in the art. Summary of the Invention

[0007] The purpose of the present invention is to provide a work order summary generation method based on low-code configuration to solve the problems in the above-mentioned background technology.

[0008] To achieve the above objectives, the present invention provides the following technical solution: a method for generating a work order summary based on low-code configuration, specifically comprising the following steps:

[0009] Establish reference path relationships between fields, including recording the dependency path between each field and its referenced fields when configuring summary expressions to form a field reference chain set;

[0010] Constructing a directed graph structure of the field reference chain set, including converting each dependency path in the field reference chain set into a directed edge to generate a directed graph consisting of field nodes and reference edges, which is used to represent the reference order relationship between fields;

[0011] Identifying all combinations of fields configured to reference each other based on the constructed directed graph structure, including performing a path analysis operation in the directed graph structure to screen out all combinations of fields that have direct or indirect bidirectional reference relationships;

[0012] Extract the path depths, the number of times a field is referenced in different paths, and the number of paths with loops for all identified field combinations from the constructed directed graph structure. Analyze the degree of closure of the field reference paths for each identified field combination based on the extraction results.

[0013] Classify the identified field combinations based on the evaluation results, determine whether to interrupt the summary generation process, and execute the corresponding control strategy;

[0014] The identified field combination and its corresponding reference path structure are saved as a snapshot, and an identification tag of the field combination is generated based on the saved snapshot to prevent the field combination with the same reference relationship from being established again in the subsequent configuration process.

[0015] Preferably, a directed graph structure of a field reference chain set is constructed, specifically:

[0016] Resolve each dependency path in the field reference chain set into a one-way dependency relationship between the starting field node and the target field node, and create a node set and an edge set based on all dependencies;

[0017] Map each field name to a unique node in a directed graph structure, and convert each dependency path into a directed edge from the starting field node to the target field node;

[0018] By traversing the field reference chain set, a complete field reference graph topology is established to form a directed graph structure consisting of all field nodes and the reference relationships between them, which is used to represent the reference order and dependency path distribution between fields.

[0019] Preferably, identifying all combinations of fields configured to reference each other based on the constructed directed graph structure includes performing a path analysis operation in the directed graph structure to screen out all combinations of fields that have direct or indirect bidirectional reference relationships, specifically:

[0020] Based on the constructed directed graph structure, perform path analysis on the starting node corresponding to each field name, obtain all target node paths reachable from the starting node, and record the forward dependency path relationship between fields;

[0021] For each target node that appears in the forward path, a reverse path analysis operation is further performed to find whether there is a path from the target node to the original starting node, so as to determine whether the forward path and its corresponding reverse path form a direct or indirect bidirectional reference relationship;

[0022] For field path combinations that are determined to have a bidirectional reference relationship, all fields involved in the path are extracted, and multiple field paths with path intersections and shared fields are aggregated to form a complete field combination unit;

[0023] Through the aggregation operation, all field combinations that have direct or indirect bidirectional reference relationships in the constructed directed graph structure are screened out as the identification results of all field combinations in which the fields are configured to reference each other.

[0024] Preferably, the method of performing the reverse path analysis operation on each target node appearing in the forward path includes:

[0025] Starting from each target node, perform a reverse path search in the constructed directed graph structure, searching upwards level by level to see if there is a path pointing to the original starting node;

[0026] During the path search process, the intermediate field nodes passed on each path are recorded to identify whether the path forms a complete return chain;

[0027] When there is at least one path from the target node back to the starting node, determine whether the path constitutes a direct or indirect bidirectional reference relationship with the original path, and mark the target node and the original starting node together as a potential mutual reference field combination;

[0028] Through the reverse path analysis operation, the reachability of the target nodes in all forward paths is verified to determine whether there is a reference path loop between fields.

[0029] Preferably, the path depths of all identified field combinations, the number of times the fields are referenced in different paths, and the number of paths with loops are extracted from the constructed directed graph structure, and the degree of closure of the field reference paths of each identified field combination is analyzed based on the extraction results, specifically including the following steps:

[0030] Based on the constructed directed graph structure, the maximum path depth, the total number of times the field is referenced in different paths, and the number of closed loops detected in the path are extracted for each identified field combination. After extraction, unit normalization is performed to form a normalized feature vector.

[0031] A three-dimensional structure tensor is constructed based on the normalized feature vector. The three tensor dimensions are the maximum path depth, the total number of times a field is referenced in different paths, and the number of closed loops detected in the path. This tensor represents the feature distribution of each field combination in the reference structure space.

[0032] Performing a principal direction extraction operation on the constructed three-dimensional structure tensor to obtain the structural change direction corresponding to the principal component of the tensor, and performing a projection deviation rate calculation between the structural change direction and the preset standard direction;

[0033] The inverse of the projection deviation rate of each field combination is used as the reference path closure coefficient of each field combination, and the reference path closure coefficient of each field combination is compared with the pre-set reference path closure coefficient threshold interval. Based on the comparison results, the reference path closure degree of each field combination when configured to reference each other is evaluated.

[0034] Preferably, based on the constructed directed graph structure, the maximum path depth, the total number of times the field is referenced in different paths, and the number of closed loops detected in the path are extracted for the field reference paths contained in each identified field combination, and unit normalization is performed after extraction to form a normalized feature vector, specifically:

[0035] Obtain the maximum and minimum values ​​of all field combinations in three dimensions: maximum path depth, total number of times the field is referenced in different paths, and number of closed loops detected in the path, and construct the original three-dimensional structure vector for each field combination.

[0036] Based on the maximum and minimum values ​​corresponding to each dimension, the structure vector of each field combination is normalized to convert the original parameter values ​​of different dimensions into dimensionless values ​​in a unified range;

[0037] The normalized maximum path depth value, the total number of times the field is referenced in different paths, and the number of closed loops detected in the path are concatenated into a normalized feature vector in a fixed order.

[0038] Preferably, a principal direction extraction operation is performed on the constructed three-dimensional structure tensor to obtain the structural change direction corresponding to the principal component of the tensor, and the projection deviation rate of the structural change direction is calculated with the preset standard direction, specifically:

[0039] The constructed three-dimensional structure tensor is converted into a corresponding matrix form, and the tensor is flattened along the path depth dimension, the total number of times the field is referenced in different paths, and the number of closed loops detected in the path through the expansion operation;

[0040] Perform principal component analysis on the tensor based on the flattened matrix form, extract the principal component direction of each field combination that changes most significantly in the multidimensional structure space, and use the principal component direction vector as the main direction representing the trend of citation structure change;

[0041] Setting a preset standard direction as a reference direction vector indicating that the structure has no abnormal closing trend;

[0042] The extracted principal component direction vector is projected onto the preset standard direction to obtain the cosine value of the angle between the two, and the projection deviation rate between the two is calculated based on this.

[0043] Preferably, the inverse of the projection deviation rate of each field combination is used as the reference path closure coefficient of each field combination, and the reference path closure coefficient of each field combination is compared with a preset reference path closure coefficient threshold range. Based on the comparison results, the reference path closure degree of each field combination when configured to reference each other is evaluated, as follows:

[0044] When the reference path closure coefficient of the field combination is less than the minimum value of the preset threshold interval, the reference path closure degree of the field combination when configured to reference each other is low;

[0045] When the reference path closure coefficient of a field combination is greater than or equal to the minimum value of a preset threshold interval and less than or equal to the maximum value, the reference path closure degree of the field combination when configured to reference each other is medium;

[0046] When the reference path closure coefficient of the field combination is greater than the maximum value of the preset threshold interval, the reference path closure degree of the field combination when configured to reference each other is high.

[0047] Preferably, the identified field combinations are classified according to the evaluation results, and it is determined whether to interrupt the summary generation process and execute the corresponding control strategy, specifically:

[0048] Field combinations with low reference path closure levels are classified as high-risk reference groups. When an identified field combination is classified as a high-risk reference group, the summary generation process for that field combination is interrupted, preventing it from participating in the generation of subsequent summary expressions. A structural disconnection suggestion is generated to assist configuration personnel in adjusting field dependencies.

[0049] Field combinations with a moderate degree of reference path closure are classified as warning reference groups. When an identified field combination is classified as a warning reference group, it is temporarily allowed to participate in summary generation. However, before the summary is executed, structural integrity check logic is inserted to dynamically monitor the execution process and generate field reconstruction suggestions to indicate potential closure risks.

[0050] Field combinations with a high degree of citation path closure are divided into stable citation groups. When the identified field combinations are classified as stable citation groups, their normal participation in the summary generation process is maintained without interrupting execution and generating intervention suggestions.

[0051] In the above technical solution, the technical effects and advantages provided by the present invention are:

[0052] 1. The present invention systematically presents the dependency relationships between fields by establishing a reference path chain set between fields and constructing a directed graph structure. It also performs path analysis and bidirectional reference identification operations in the graph structure, accurately screening out all field combinations with direct or indirect bidirectional dependencies. Compared with the current state of the art in traditional low-code configuration methods that lack reference path visualization and detection mechanisms, this solution implements structured identification and active modeling of field reference path closure risks, effectively avoiding problems such as recursive dead loops and execution blocking in the summary engine caused by circular dependencies during the configuration process, and significantly improving the stability and configuration controllability of the work order summary generation process.

[0053] 2. This invention introduces a three-dimensional structural tensor and principal direction deviation analysis mechanism, normalizing the structural distribution of field combinations across the three dimensions of "path depth, number of citations, and number of closed loops" to model the structure. Tensor principal component analysis is then used to extract structural variation trends and assess the degree of closure of field reference paths. This approach is more granular and explanatory than the traditional single-dimensional "loop / no loop" judgment, achieving a technological leap from "structural risk" to "quantifiable structural risk," significantly improving the system's analytical accuracy and decision-making flexibility when dealing with complex configuration-dependent structures.

[0054] 3. This invention further introduces the evaluation results into the policy control process. By dividing field combinations into high-risk reference groups, alert reference groups, and stable reference groups, differentiated response measures such as interrupting summaries, inserting structure monitoring logic, and normal execution are adopted respectively. It also provides link breaking suggestions for high-risk combinations and structural reconstruction suggestions for alert combinations. At the same time, through the snapshot recording and identification tagging mechanism, the recurrence of the same structure is effectively prevented, forming a complete "detection-assessment-classification-intervention-blocking" closed-loop mechanism. This not only improves the automation continuity of the summary processing process, but also enhances the robustness and long-term maintainability of the low-code platform in dealing with dynamic configuration risk scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction to the drawings required for use in the embodiments will be given below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0056] Figure 1 This is a flow chart of a method for generating a work order summary based on low-code configuration according to the present invention. DETAILED DESCRIPTION

[0057] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, these example embodiments are provided so that the description of this disclosure will be thorough and complete and will fully convey the concepts of the example embodiments to those skilled in the art.

[0058] The present invention provides Figure 1 The method for generating a work order summary based on low-code configuration shown in the figure specifically includes the following steps:

[0059] Establish reference path relationships between fields, including recording the dependency path between each field and its referenced fields when configuring summary expressions to form a field reference chain set;

[0060] In the low-code configuration platform, automatic recording of field reference paths can be achieved by monitoring configuration expression input events. When the user sets an expression for a certain field when configuring the work order summary rule (for example, "field A = field B + field C summary"), the system can identify the referenced fields that appear in the expression in sequence during parsing, and record dependency relationships such as "field A depends on field B" and "field A depends on field C" in the form of key-value pairs, forming a set of one-way reference relationships between fields. For each recorded reference relationship, the platform can construct a path element from the current field to the referenced field, and append it to the field reference chain set in sequence. This set can be dynamically maintained and queried as a structured storage data structure (such as an adjacency list, path sequence dictionary, etc.), supporting subsequent analysis operations for graph structure construction and closed-loop detection.

[0061] Automatically recording reference paths between fields in this way can minimize dependency omissions or configuration ambiguities. This is especially true in low-code environments, where configuration personnel often visually drag fields or use template expressions for rapid combination, lacking a global understanding of underlying dependencies. Therefore, the platform needs to capture and archive field dependency paths during the expression parsing phase. By abstracting all reference paths into a chain collection, it not only provides a complete data foundation for the subsequent construction of directed graph structures and execution loop detection, but also enables automatic alignment between configuration behaviors and reference structures, effectively improving the robustness and controllability of the summary generation engine in dynamic reference scenarios, preventing difficult-to-locate logical errors caused by circular dependencies, and ensuring the stable operation of the work order summary system.

[0062] Constructing a directed graph structure of the field reference chain set, including converting each dependency path in the field reference chain set into a directed edge to generate a directed graph consisting of field nodes and reference edges, which is used to represent the reference order relationship between fields;

[0063] In this embodiment, a directed graph structure of a field reference chain set is constructed, specifically:

[0064] Resolve each dependency path in the field reference chain set into a one-way dependency relationship between the starting field node and the target field node, and create a node set and an edge set based on all dependencies;

[0065] In the low-code configuration platform, whenever a user completes the summary expression configuration for a field, the system automatically resolves the field into the current field (i.e., the starting field node) and identifies the referenced fields (i.e., the target field node) that appear in its expression, thereby treating the expression as a one-way path where the starting field depends on the target field. The system can extract the structure of each field reference chain through a syntax parser or expression analysis engine and abstract each path into a set of ordered field pairs. On this basis, two sets are established: one is a node set with field names as elements, which is used to uniquely represent all participating fields in the graph structure; the other is an edge set with field dependency pairs as elements, which is used to record directed reference relationships between fields. This approach not only ensures the uniqueness of field nodes in the graph and the directionality of reference edges, but also provides a rigorous data foundation for subsequent construction of directed graphs, analysis of dependency topology, and closed-loop detection. It is a key step in achieving structured reference path modeling and can effectively support the discovery and management of circular reference issues.

[0066] Map each field name to a unique node in a directed graph structure, and convert each dependency path into a directed edge from the starting field node to the target field node;

[0067] To map each field name to a unique node in a directed graph structure and convert each dependency path into a directed edge, a graph model data structure can be constructed. Specifically, the platform creates an in-memory mapping table that uniquely registers all field names as node identifiers in the graph structure, ensuring that each field corresponds to only one node instance. Then, within the field reference chain, each "field A depends on field B" path is parsed as a pair of ordered fields. Based on this, the system adds a directed edge from field A to field B in the graph structure, with the direction of the edge explicitly indicating the flow of the dependency. This approach not only constructs a directed graph of inter-field reference relationships but also provides the underlying graph structure for subsequent path traversal, loop detection, and closure analysis. This approach is necessary because in low-code configuration scenarios, references between fields are dynamic, mutable, and non-linear. Traditional configuration alone cannot identify potential circular dependencies or hierarchical depth. By mapping field reference relationships into a graph structure, structured management and path visualization analysis are enabled, ensuring the stability and controllability of the summary generation process.

[0068] By traversing the field reference chain set, a complete field reference graph topology is established to form a directed graph structure consisting of all field nodes and the reference relationships between them, which is used to represent the reference order and dependency path distribution between fields.

[0069] The topology of the field reference graph is dynamically constructed by traversing the set of field reference chains one by one. During the traversal, the system sequentially reads each field dependency path, locates the starting field node and the target field node using the previously established field-node mapping, and converts the dependency relationship between them into a directed edge in the graph structure. As the traversal continues, all field names are registered as nodes, and all reference relationships are converted into edges, gradually building a complete directed graph consisting of nodes and edges. This topology can be stored as a graph data model such as an adjacency list, adjacency matrix, or edge list, facilitating subsequent graph algorithm operations such as path search, dependency chain tracing, and loop detection. This graph topology is essential because field dependencies in low-code platforms can span multiple layers and sources, and the configuration logic is highly flexible. Without a complete graph, it is impossible to understand the dependency order and path distribution between fields from a holistic perspective, making it difficult to promptly identify hidden closed loops or redundant paths, which can easily cause the summary engine to enter an infinite loop or redundant calculation. Therefore, establishing a topological structure through graph traversal is the core foundation for achieving structured modeling and controllable generation of global field references.

[0070] Identifying all combinations of fields configured to reference each other based on the constructed directed graph structure, including performing a path analysis operation in the directed graph structure to screen out all combinations of fields that have direct or indirect bidirectional reference relationships;

[0071] In this embodiment, all combinations of fields configured to reference each other are identified based on the constructed directed graph structure, including performing a path analysis operation in the directed graph structure to screen out all combinations of fields that have direct or indirect bidirectional reference relationships, specifically:

[0072] Based on the constructed directed graph structure, perform path analysis on the starting node corresponding to each field name, obtain all target node paths reachable from the starting node, and record the forward dependency path relationship between fields;

[0073] Path analysis is accomplished by performing a node-by-node traversal of the constructed directed graph structure. Specifically, the system uses a depth-first search (DFS) or breadth-first search (BFS) algorithm for each starting node corresponding to a field name, recursively or hierarchically traversing all outgoing edges in the graph. During the traversal, the system continuously records the field nodes traversed along the current path and dynamically stores this path information as an ordered list of field nodes. Whenever a path traversal reaches an end node or is interrupted, the system records the path as a valid "field forward dependency path" in the path collection. By performing this traversal on each field node in the graph, the platform fully captures all downstream dependent field paths that can be reached from the starting node, forming a complete forward dependency path structure. This is necessary because inter-field references in low-code configuration platforms are often hierarchical and chain-like. Field A may depend on field B, which in turn indirectly depends on field C. The complete dependency chain cannot be identified using static configuration information alone. By performing systematic path analysis on directed graphs, we can not only clarify the actual dependency order between fields but also provide quantifiable and traceable foundational data for subsequent identification of inter-reference relationships, determination of closed-loop structures, and implementation of link-breaking strategies. This process is highly automated and suitable for integration into the pre-analysis process before abstract generation, enabling early warning and control of potential logical closed loops.

[0074] For each target node that appears in the forward path, a reverse path analysis operation is further performed to find whether there is a path from the target node to the original starting node, so as to determine whether the forward path and its corresponding reverse path form a direct or indirect bidirectional reference relationship;

[0075] For field path combinations that are determined to have a bidirectional reference relationship, all fields involved in the path are extracted, and multiple field paths with path intersections and shared fields are aggregated to form a complete field combination unit;

[0076] For field path combinations that have been determined to have bidirectional reference relationships, the system can complete the construction of field combination units by performing aggregation analysis on the path result set. Specifically, the system first extracts path node information from all identified bidirectional reference paths and extracts the set of field names involved in the path; then, the system further identifies the existence of path cross-references or field duplicate references in this set, that is, the same field nodes or partial field sequences are nested in two or more paths. Once such cross-references or field overlaps are detected, the system merges the field sets corresponding to these paths to form a larger field combination unit, and ensures that the reference relationships of all fields in each combination unit in the graph structure constitute a potential closed loop or complex dependency block. The system can automate this aggregation process with the help of graph clustering algorithms or merging logic based on the field node co-occurrence matrix, and mark each combination unit as an independent entity to be evaluated. The goal is to unify multiple seemingly independent but actually cross-dependent paths into a single analysis unit, enabling more accurate identification of multi-field circular reference structures and avoiding missing potential dependency loops. This also establishes clear data boundaries and target scopes for subsequent operations such as field path structure parameter extraction, risk level assessment, and broken link control. This approach allows the system to identify complex dependency issues within configuration logic at a higher granularity before summarization is generated, effectively improving the comprehensiveness and accuracy of closed-loop identification.

[0077] Through the aggregation operation, all field combinations that have direct or indirect bidirectional reference relationships in the constructed directed graph structure are screened out as the identification results of all field combinations in which the fields are configured to reference each other.

[0078] In order to filter out all field combinations that have direct or indirect bidirectional reference relationships in the constructed directed graph structure through aggregation operations, an aggregate mapping relationship between fields can be constructed based on the aforementioned field path intersection and shared field analysis results. The specific implementation method is as follows: After the system completes the bidirectional reference identification of all field paths, it regards the field names involved in each set of bidirectional paths as the initial field set, and merges multiple sets with path intersections, shared fields, or intersections between fields. It uses a union-find algorithm or a connected subgraph identification technology based on graph partitioning to classify all fields into several "closed reference domains." Each reference domain represents a field combination, which the system saves as the final identified mutually referencing field combination. The system can also uniquely identify each combination to facilitate accurate reference during subsequent processing. The reason for adopting this aggregation method is that in low-code configuration scenarios, field reference paths do not always exist in isolation, and often manifest as complex cross-dependencies between multiple fields. Relying solely on pairwise judgment of paths will result in structural omissions or redundant judgments. By merging and clustering all field paths that can form reference loops, we can form closed, indivisible field combination units, which serve as the basis for subsequent risk assessment, closure analysis, and broken link handling. This not only improves identification efficiency but also avoids repeated analysis of the same path structure, thereby enhancing the overall stability and execution performance of the system when handling circular dependencies.

[0079] In this embodiment, the method of performing the reverse path analysis operation on each target node appearing in the forward path includes:

[0080] Starting from each target node, perform a reverse path search in the constructed directed graph structure, searching upwards level by level to see if there is a path pointing to the original starting node;

[0081] Starting from each target node, a reverse path search is performed within the constructed directed graph structure, searching upwards level by level to determine whether there is a path back to the original starting node. This is achieved by constructing a reverse index structure within the graph. Specifically, after the directed graph is constructed, the system simultaneously generates a "backward reference table" for each field node, recording which upstream field nodes it depends on. When performing a reverse path search, the system starts from the current target field node and recursively calls the corresponding upstream node list in the reverse reference table, tracing back through the field reference chain layer by layer. Each time a field node is accessed, it is added to the current reverse path set and checked to see if it is equal to the field of the starting node. If, at a certain level, a path from the target node to the original starting field is found, the system determines that a back-path exists and forms part of a closed reference loop. At the implementation level, a depth-first search (DFS) approach can be used to limit the maximum recursion depth to prevent stack overflow, or a queue-based breadth-first search (BFS) can be used to expand the path for improved efficiency. The reason for performing this step-by-step reverse path search is that the dependencies between fields in low-code platforms are often not one-way linear. Although field A directly references field B, field B may indirectly reference field A through other fields. This closed loop cannot be identified through a one-way path. Through systematic reverse path analysis, it is possible to comprehensively determine whether the target field has the possibility of structurally returning to the starting field, and then identify indirect or implicit circular reference problems in the configuration logic, providing an accurate structural basis for subsequent field combination aggregation, risk level assessment, and broken chain control strategies. This process is highly automated and controllable, and is suitable for integration into dependency integrity analysis before summary generation.

[0082] During the path search process, the intermediate field nodes passed on each path are recorded to identify whether the path forms a complete return chain;

[0083] To record the intermediate field nodes traversed along each path during the path search process and identify whether the path forms a complete return chain, the system dynamically maintains a "path tracking stack" or "node access sequence list" during the path search (whether forward or reverse). This stack stores the names of the field nodes visited sequentially during the search. Whenever the path algorithm jumps from one field node to the next, the system adds the field to the current path tracking list and saves the complete path sequence as a field path instance when the search terminates. By fully recording all intermediate field nodes in the path, the system not only determines the structural order of the path but also determines whether the path truly forms a valid chain from the target node back to the starting node. In scenarios with multiple intersecting paths or branching nodes, it can be difficult to determine whether a path conforms to closed-loop logic based solely on the start and end points. Recording intermediate node information helps the system verify whether the path contains anomalies such as interruptions, duplications, or asymmetric jumps, thereby ensuring the logical integrity of the identified reference chain. Furthermore, this intermediate node recording mechanism also provides a data foundation for subsequent modules such as path aggregation, field combination merging, and citation strength analysis, enabling the system to understand the dependency network between fields at a more granular level, improving the accuracy and controllability of overall structural analysis. Therefore, this step is not only necessary for structural tracking but also directly impacts the reliability of identifying inter-reference relationships and the stability of system responses.

[0084] When there is at least one path from the target node back to the starting node, determine whether the path constitutes a direct or indirect bidirectional reference relationship with the original path, and mark the target node and the original starting node together as a potential mutual reference field combination;

[0085] When the system detects during a reverse path search that a target field node has at least one valid path back to the original starting field node, forming a complete closed path loop from the starting point to the target and back to the starting point, the system can determine whether this path forms a direct or indirect bidirectional reference relationship with the forward path. To achieve this, the system establishes a unique identifier for each set of field paths during the path search, uniformly storing the path's starting node, target node, and complete sequence of intermediate nodes. When a closed loop is detected, the forward path is compared with its corresponding reverse closed-loop path. If the endpoint of the reverse path matches the starting node of the forward path, and there is a valid field dependency between the intermediate nodes, the system deems the two paths to form a closed reference chain. At this point, the system adds the target node and the starting node to a set of potential mutually referencing fields and marks them as "suspected mutually referencing" using a field name index. This mark serves as an identification anchor in the internal data structure for subsequent field aggregation, risk analysis, and link-breaking strategy execution. The core purpose of this operation is to separate the field nodes that actually constitute the dependent closed loop in the complex path structure from the graph structure, and to build a clear two-way reference combination from a logical level to support the subsequent independent evaluation and control processing of the combination. Especially in low-code configuration scenarios with multi-path and multi-field cross-references, if the path closed loop is not verified and the field combination mark is not generated, it will be difficult for the system to accurately define "whether the closed loop actually exists" and "which fields are affected by the closed loop", which may cause closed loop identification misjudgment or processing omissions, affecting the stability and correctness of summary generation. Therefore, this step is a key decision point for identifying logical closed loop field combinations, and has a high degree of structural abstraction capability and practical operational value.

[0086] Through the reverse path analysis operation, the reachability of the target nodes in all forward paths is verified to determine whether there is a reference path loop between fields.

[0087] To verify the reachability of all target nodes in forward paths through reverse path analysis, the system employs a "path backtracking detection" strategy from graph theory, combining a traversal mechanism with a path mapping table to implement a closed-loop verification process. Specifically, during the forward path traversal, the system establishes a one-to-one correspondence between the starting node of each path and all its target nodes, storing this information in a path mapping table. Subsequently, during the reverse path analysis phase, starting from each target node, the system uses a reverse reference table (i.e., the inverse adjacency table of the graph structure) to trace the path layer by layer, determining whether a path connects back to the corresponding original starting node. During the path analysis process, the system records the node paths traversed during each backtracking attempt and prunes them using path pointers and access status to ensure that duplicate judgments are not made. When a target node has at least one return path, the system considers it and the original starting node to form a node pair that forms a closed reference path loop, and archives this structure as a potential bidirectional reference base combination. The significance of this reachability verification operation is that the dependencies between fields in a low-code configuration environment are often nonlinear and indirect, and potential structural loops cannot be discovered through a one-way reference path alone. In particular, hidden closed loop problems are more likely to occur when there are long path chains, path merging, or multi-level dependencies between fields. Through comprehensive reachability verification, the system can not only ensure that all field reference relationships are accurately resolved, but also significantly improve the detection accuracy of circular references, reduce configuration logic risks, and enhance the stability and controllability of the work order summary generation process. Therefore, this step is not only a necessary condition for judging closed-loop relationships, but also a structural prerequisite for the subsequent field combination classification and the execution of broken-link strategies.

[0088] Extract the path depths, the number of times a field is referenced in different paths, and the number of paths with loops for all identified field combinations from the constructed directed graph structure. Analyze the degree of closure of the field reference paths for each identified field combination based on the extraction results.

[0089] In this embodiment, the path depths of all identified field combinations, the number of times a field is referenced in different paths, and the number of paths with loops are extracted from the constructed directed graph structure. Based on the extraction results, the degree of closure of the field reference paths of each identified field combination is analyzed. Specifically, the following steps are included:

[0090] Based on the constructed directed graph structure, the maximum path depth, the total number of times the field is referenced in different paths, and the number of closed loops detected in the path are extracted for each identified field combination. After extraction, unit normalization is performed to form a normalized feature vector.

[0091] A three-dimensional structure tensor is constructed based on the normalized feature vector. The three tensor dimensions are the maximum path depth, the total number of times a field is referenced in different paths, and the number of closed loops detected in the path. This tensor represents the feature distribution of each field combination in the reference structure space.

[0092] In order to construct a three-dimensional structural tensor based on normalized feature vectors, the normalized feature vectors of multiple field combinations can be organized into a multidimensional array with a three-dimensional structure through software, where the three dimensions of the tensor correspond to the maximum path depth, the total number of times the field is referenced in different paths, and the number of closed loops detected in the path. The specific implementation method includes: first, after the software generates normalized feature vectors for all field combinations, these vectors are used as the basic units of the tensor and arranged in sequence according to the order of the field combinations; then, according to the normalized feature values ​​of each field combination in the three structural dimensions, it is mapped to a three-dimensional coordinate system, and the system automatically assigns the position of the combination in the tensor in the three-dimensional coordinate system, thereby expressing all combinations in a unified three-dimensional space. The position and value of each point in the tensor represent the structural feature distribution state of a field combination in terms of the reference path structure. The core value of this approach lies in that it not only preserves the local feature information of each field combination but also constructs an overall structural feature space. This enables complex operations such as high-dimensional trend modeling, structural anomaly identification, and principal component direction extraction in this multidimensional space. This enhances macroscopic insight into the closure trends of reference paths between fields, providing higher-dimensional and more precise support for subsequent assessment of circular reference risks. The entire construction process can be efficiently implemented in software through matrix operations, tensor data structures, and spatial mapping algorithms, offering excellent scalability and automation capabilities.

[0093] Performing a principal direction extraction operation on the constructed three-dimensional structure tensor to obtain the structural change direction corresponding to the principal component of the tensor, and performing a projection deviation rate calculation between the structural change direction and the preset standard direction;

[0094] The inverse of the projection deviation rate of each field combination is used as the reference path closure coefficient of each field combination, and the reference path closure coefficient of each field combination is compared with the pre-set reference path closure coefficient threshold interval. Based on the comparison results, the reference path closure degree of each field combination when configured to reference each other is evaluated.

[0095] In this embodiment, based on the constructed directed graph structure, the maximum path depth, the total number of times the field is referenced in different paths, and the number of closed loops detected in the path are extracted for each identified field reference path contained in the field combination. After extraction, unit normalization is performed to form a normalized feature vector, specifically:

[0096] Obtain the maximum and minimum values ​​of all field combinations in three dimensions: maximum path depth, total number of times the field is referenced in different paths, and number of closed loops detected in the path, and construct the original three-dimensional structure vector for each field combination.

[0097] To obtain the maximum and minimum values ​​for all field combinations across the three dimensions of "maximum path depth, total number of field references across different paths, and number of closed loops detected in a path" and construct the original three-dimensional structure vector for each field combination based on these values, the following approach can be used: First, the system traverses all identified field combinations in the constructed directed graph structure and analyzes their corresponding field reference paths one by one. For each field combination, the maximum path depth, total number of field references across multiple paths, and number of closed loops in the corresponding path are extracted. To obtain the global maximum and minimum values ​​for each dimension, these three parameters are recorded as lists. Statistical aggregation is then performed to sequentially extract the global maximum and minimum values ​​for each of these three parameter categories, which serve as the basis for subsequent normalization. Based on this, for each field combination, the original parameter values ​​for the three categories are organized into a structure vector in a fixed order to construct the complete original three-dimensional structure vector. In this way, it can be ensured that each field combination has a unified structural data representation, providing standardized input for subsequent normalization processing and structural modeling analysis, eliminating dimensional differences between field combinations caused by different scales or path complexity, and thus laying the foundation for further path closure trend assessment.

[0098] Based on the maximum and minimum values ​​corresponding to each dimension, the structure vector of each field combination is normalized to convert the original parameter values ​​of different dimensions into dimensionless values ​​in a unified range;

[0099] In order to achieve the standardization of parameters of different dimensions in the field combination structure vector, linear unit normalization can be used, that is, in each dimension, the original parameter value is converted into a dimensionless value within a unified interval (usually 0, 1), so that different field combinations are comparable in the path structure parameter dimension. In the specific implementation process, firstly, based on the maximum value of the three dimensions of the maximum path depth of all field combinations extracted in the previous step, the total number of times the field is referenced in different paths, and the number of closed loops detected in the path and minimum value For each component of the original structure vector of each field combination Perform normalization calculation. The formula used for normalization is: in It represents the normalized value, which represents the standardized performance of the field combination in this dimension. This formula will linearly map all the original data to between 0 and 1, retaining the relative proportional differences between the original data, while removing the unit dimension, so that the structural vector will not be affected by the dimensional difference during subsequent tensor modeling, deviation calculation or cluster analysis. This process can be executed in batches through the normalization function set in the software, combined with vectorization processing to improve efficiency. It is a widely used and reliable standard method in data preprocessing and pattern analysis. This method can ensure that the path structure features are mathematically consistent in subsequent comparisons, providing an accurate input basis for identifying differences in path closure trends between field combinations.

[0100] The normalized maximum path depth value, the total number of times the field is referenced in different paths, and the number of closed loops detected in the path are concatenated into a normalized feature vector in a fixed order for subsequent analysis and evaluation of the field reference path structure.

[0101] In order to splice the normalized maximum path depth value, the total number of times the field is referenced in different paths, and the number of closed loops detected in the path into a normalized feature vector in a fixed order, the software can introduce a structured vector generation operation after the normalization process is completed. Specifically, after the system completes the normalization of the three-dimensional structure parameters of each field combination, it will splice these three normalized results in a pre-defined field order, for example, always putting the maximum path depth first, the total number of references second, and the number of closed loops third, thereby generating a vector form with fixed dimensions and in order, such as . This operation can usually be completed in memory through the construction of lists, arrays or vector structures, and can be uniformly stored as feature fields or matrices in a data table for further structural modeling, cluster analysis or multidimensional space mapping operations. By splicing in a fixed order, the semantic expression consistency of the feature vectors of all field combinations can be ensured, so that subsequent algorithms can accurately understand the structural characteristics represented by each dimension when processing these vectors, avoiding evaluation errors caused by dimensional mismatch. In addition, this vector form also facilitates the subsequent use of tensor modeling methods to uniformly model multi-field combinations, thereby performing trend identification and risk assessment on the overall distribution status of the reference path structure, which is a key intermediate step in supporting complex structural pattern recognition.

[0102] In this embodiment, a principal direction extraction operation is performed on the constructed three-dimensional structure tensor to obtain the structural change direction corresponding to the principal component of the tensor, and the projection deviation rate of the structural change direction is calculated with the preset standard direction, specifically:

[0103] The constructed three-dimensional structure tensor is converted into a corresponding matrix form, and the tensor is flattened along the path depth dimension, the total number of times the field is referenced in different paths, and the number of closed loops detected in the path through the expansion operation;

[0104] To convert the constructed three-dimensional structural tensor into a matrix form suitable for subsequent analysis, tensor unfolding or flattening can be used. This process is typically accomplished in software through a combination of data structure transformations and linear algebra operations. Specifically, each dimension of the three-dimensional tensor—namely, the maximum path depth, the total number of times a field is referenced across different paths, and the number of closed loops detected in a path—is first used as an unfolding dimension. The unfolding operation, centered around a particular dimension, retains all slices along that dimension as rows or columns, while the combination of the remaining two dimensions is flattened into a single dimension, forming a two-dimensional matrix. For example, when flattening based on path depth, the system retains slices at each path depth and flattens the corresponding number of references and number of closed loops into a row vector, transforming the three-dimensional tensor into a two-dimensional matrix. This unfolding process can be repeated along each dimension to obtain matrix representations from three different perspectives, capturing the varying trends of structural features across different dimensions. Through the expanded matrix structure, the software can more efficiently perform linear dimensionality reduction techniques such as principal component analysis (PCA), thereby extracting directional features in the structural space that best express the differences between field combinations. This is a key basic step in realizing structural trend analysis and projection deviation calculation.

[0105] Perform principal component analysis on the tensor based on the flattened matrix form, extract the principal component direction of each field combination that changes most significantly in the multidimensional structure space, and use the principal component direction vector as the main direction representing the trend of citation structure change;

[0106] To perform principal component analysis (PCA) on a flattened matrix tensor to extract the principal component directions where field combinations vary most significantly in the multidimensional structural space, linear algebraic methods such as eigenvalue decomposition (Eigenvalue Decomposition) or singular value decomposition (SVD) can be used in the software. Specifically, after the tensor is flattened into matrices according to different dimensions, the normalized features of each field combination are represented as a row or column in the matrix. The system first performs mean centering on the matrix to eliminate data skew. The covariance matrix is ​​then calculated to measure the joint variation across various structural dimensions (such as path depth, citation count, and closed loop count). Eigenvalue decomposition is then performed on the covariance matrix to extract the principal component vector with the largest eigenvalue, which represents the principal direction of structural variation. Alternatively, SVD can be used to directly decompose the original matrix, where the principal direction is determined by the eigenvector corresponding to the first principal component. This vector geometrically represents the direction where the field combination is most widely distributed and varies most strongly in the multidimensional structural space. Extracting this direction helps reduce high-dimensional structural features to the most discriminative single dimension, thereby providing the optimal direction for subsequent risk identification, closure trend analysis, and projection deviation rate calculation. This processing method is not only automated by the data analysis module but also supports batch and high-performance computing, making it suitable for real-time operation on large-scale low-code configuration platforms.

[0107] Setting a preset standard direction as a reference direction vector indicating that the structure has no abnormal closing trend;

[0108] To achieve the goal of "setting a preset standard direction as a reference direction vector indicating a structurally normal, non-anomalous closure trend," the software can combine statistical modeling of historical normal configuration data with empirical rule-setting. Specifically, the system first constructs a reference dataset consisting of known field combinations without circular dependencies and archives the normalized feature vectors generated by these combinations. The software then performs principal component analysis or mean clustering on this set of feature vectors to extract the main distribution trends of these "normal field reference structures" in three-dimensional structural space, ultimately determining a representative direction vector as the standard direction. This direction vector typically exhibits a characteristic distribution direction with a shallow maximum path depth, a balanced distribution of field references, and a near-zero number of closed loops. During system initialization or policy updates, this reference direction can be manually confirmed through human intervention or automatically generated based on statistical indicators (such as high-frequency region centers and feature mean vectors). It is stored as a fixed vector in the configuration parameter table for subsequent projection deviation rate calculations. The purpose of setting this standard direction is to provide a unified structural "health benchmark" for all field combinations to be evaluated, so that the system can quantify the degree to which the field reference relationship deviates from the normal configuration trend in a vector space manner, thereby supporting structural stability assessment and risk identification. It is one of the core scales in the entire closure judgment mechanism.

[0109] The extracted principal component direction vector is projected onto the preset standard direction to obtain the cosine value of the angle between the two. Based on this, the projection deviation rate between the two is calculated to indicate the degree of deviation of the structural trend.

[0110] To project the extracted principal component direction vectors onto a pre-set standard direction, obtain the cosine of the angle between them, and calculate the projection deviation rate, the software automatically performs this process based on cosine similarity analysis using the vector space operation module. The system first takes the principal component direction vectors extracted from the 3D structure tensor and the pre-set standard direction vector as input data. Both vectors are normalized to ensure their modulus is 1, eliminating scale artifacts. The cosine of the two vectors is then calculated using the vector dot product formula. This is the dot product of the principal component direction vector and the standard direction vector divided by the product of their moduli (due to the normalization, the denominator is 1). This yields the cosine of the angle, which reflects the degree of alignment between the two directions. The system then calculates the absolute difference between this cosine value and the ideal alignment (cosine of 1) to generate a "projection deviation rate" between 0 and 1, which quantifies the difference in structural trends between the principal and standard directions. Values ​​closer to 0 indicate more stable structural features; values ​​closer to 1 indicate that field reference relationships deviate significantly from expected directions, posing potential risks such as circular dependencies. This method is not only computationally efficient and suitable for automated analysis processes in large-scale configuration environments, but also provides the system with an accurate risk measurement foundation in an intuitive numerical form, facilitating subsequent classification and control strategy execution.

[0111] In this embodiment, the inverse of the projection deviation rate of each field combination is used as the reference path closure coefficient of each field combination, and the reference path closure coefficient of each field combination is compared with a preset reference path closure coefficient threshold range. Based on the comparison results, the reference path closure degree of each field combination when configured to reference each other is evaluated, as follows:

[0112] When the reference path closure coefficient of the field combination is less than the minimum value of the preset threshold interval, the reference path closure degree of the field combination when configured to reference each other is low;

[0113] When the reference path closure coefficient of a field combination falls below the preset minimum threshold, it indicates a clear closed-loop tendency in its reference structure, and a strong bidirectional reference relationship between fields, often manifesting as multiple field cross-references, frequent repeated references, and unreasonable path depth. This situation can easily lead to recursive call dead loops during summary generation, causing the summary engine to fall into an infinite reference chain when parsing these fields, resulting in thread freezes, abnormal CPU resource consumption, and even the downfall of the entire platform's summary service. Therefore, such field combinations are assessed as high-risk combinations and should generally be immediately interrupted from participating in the summary generation process, triggering the configuration-level error warning mechanism.

[0114] When the reference path closure coefficient of a field combination is greater than or equal to the minimum value of a preset threshold interval and less than or equal to the maximum value, the reference path closure degree of the field combination when configured to reference each other is medium;

[0115] When the reference path closure coefficient of a field combination falls within the preset threshold range, it indicates that its field reference structure has a certain degree of controllability, but potential structural risks still exist, such as long reference chains, closed loops that do not form strong recursion, or fields that appear multiple times in different paths but do not yet constitute a core conflict. In most cases, such combinations will not immediately cause systemic errors, but in environments with intensive summary generation tasks or dynamic changes in field references, they may gradually evolve into an uncontrollable state. Therefore, field combinations at this level are generally considered medium-risk combinations and need to be recorded and recommended for structural optimization or subsequent monitoring to prevent risk accumulation.

[0116] When the reference path closure coefficient of the field combination is greater than the maximum value of the preset threshold interval, the reference path closure degree of the field combination when configured to reference each other is high.

[0117] When the reference path closure coefficient of a field combination exceeds the preset maximum threshold, it indicates that its reference structure has good topological stability and a high degree of non-closure. There are no obvious bidirectional circular references between fields, the reference path is clear and hierarchical, the number of field references is reasonable, and the path depth is moderate, making it unlikely to cause structural dependency conflicts or computational blockages. This type of field combination can be considered low-risk or safe, allowing safe participation in summary generation configuration. It also exhibits good compatibility and stability during dynamic summary execution, helping to ensure high system availability and automated processing efficiency.

[0118] The "pre-set reference path closure coefficient threshold range" can be determined through statistical modeling based on historical summary configuration data. Its implementation process includes the following key steps: First, a large amount of historical ticket summary configuration data must be collected from the existing ticket system. This data should include information such as field reference structure, generation anomalies, and generation efficiency. Next, for each historical field combination, the corresponding reference path closure coefficient is calculated using a method consistent with the current evaluation logic. Based on their execution results during the actual summary generation process, these field combinations are classified into categories such as "safe execution," "has caused anomalies," and "requires manual intervention." Then, using this labeled data as a sample, cluster analysis (such as K-means) or distribution fitting (such as Gaussian mixture models) is used to statistically summarize the distribution characteristics of the closure coefficient values ​​within each category, thereby extracting the boundary range between each category. Finally, based on the cluster boundaries or the confidence interval of the fitted distribution, a lower and upper limit of the closure coefficient is set as the evaluation boundary between "low risk" and "high risk," forming the final closure coefficient threshold range. The threshold intervals set in this way accurately reflect the relationship between field structure stability and system operational risk, and can be dynamically updated and optimized as data accumulates. This method can be implemented entirely through software logic, through data modeling and algorithm execution, and is suitable for automated risk modeling scenarios within low-code platforms.

[0119] Classify the identified field combinations based on the evaluation results, determine whether to interrupt the summary generation process, and execute the corresponding control strategy;

[0120] In this embodiment, the identified field combinations are classified according to the evaluation results, and a decision is made as to whether to interrupt the summary generation process and execute the corresponding control strategy, specifically:

[0121] Field combinations with low reference path closure levels are classified as high-risk reference groups. When an identified field combination is classified as a high-risk reference group, the summary generation process for that field combination is interrupted, preventing it from participating in the generation of subsequent summary expressions. A structural disconnection suggestion is generated to assist configuration personnel in adjusting field dependencies.

[0122] When the evaluation module determines that the reference path closure coefficient of a field combination falls below the minimum of a preset threshold, the system automatically marks the combination as a "high-risk reference group" and inserts a rule-based interception mechanism before the configuration expression execution engine initiates summary generation. This mechanism triggers an interrupt instruction upon detecting a high-risk reference group involved in field concatenation or summary segment generation, terminating the expression calculation and preventing recursive calls from entering an infinite loop. Furthermore, the summary configuration analysis engine analyzes the reference path structure of the field combination within the directed graph to identify critical loop paths between fields and automatically recommends one or more breakpoints. For example, it prioritizes disconnecting secondary dependency paths while preserving the primary expression field chain, generating structural breakpoint recommendations. These recommendations are graphically displayed in the low-code configuration interface to help configuration personnel understand the breakpoints and adjustable areas of field dependencies, reducing the probability of configuration errors. This approach not only ensures the stable operation of the summary generation engine but also enables preventive configuration-level fixes without manual log review, providing robust system reliability and automated diagnostic value.

[0123] Field combinations with a moderate degree of reference path closure are classified as warning reference groups. When an identified field combination is classified as a warning reference group, it is temporarily allowed to participate in summary generation. However, before the summary is executed, structural integrity check logic is inserted to dynamically monitor the execution process and generate field reconstruction suggestions to indicate potential closure risks.

[0124] When a field combination's reference path closure coefficient falls within the middle range of a preset threshold, resulting in a "moderate" closure, the system places the combination in the "warning reference group." Before this group participates in the summary generation process, the configuration logic automatically inserts a structural integrity checkpoint. This is achieved by traversing the field dependency path in real time as the summary execution engine loads the expression tree, verifying whether new loops or abnormal increases in path depth exist. Furthermore, during execution, the call sequence and depth of the field value backfill chain are continuously recorded. If local field values ​​remain unreturned for extended periods or the call stack experiences backtracking, a warning signal is triggered, marking the current expression chain as presenting a structural risk. The system also automatically generates field refactoring suggestions based on the combination's historical configuration performance and key nodes in the current execution path. For example, recommendations include extracting frequently referenced fields as intermediate cache fields and merging parallel reference paths, prompting configuration personnel to proactively mitigate potential path closure trends. This process, accomplished through the insertion of auxiliary analysis logic, dynamic dependency tracking, and intelligent notification mechanisms, offers predictive and preventative value, enabling risk mitigation and configuration optimization without interrupting the execution process.

[0125] Field combinations with a high degree of citation path closure are divided into stable citation groups. When the identified field combinations are classified as stable citation groups, they maintain their normal participation in the summary generation process without interrupting execution and generating intervention suggestions, thereby improving the automation continuity and configuration efficiency of summary processing.

[0126] When a field combination is configured to reference each other and its reference path closure is assessed as high, the system classifies the combination as a "stable reference group." To implement this strategy, a software-based mechanism for marking field combination status and selecting processing paths is implemented. After the field closure assessment module calculates the reference path closure coefficient, if the assessment value exceeds a set maximum threshold, the system assigns the field combination a "stable reference group" designation. During the subsequent summary generation process, the scheduling engine differentiates field combinations based on their group information. Field combinations in the stable reference group automatically skip structural integrity checks and runtime monitoring logic, directly incorporating them into the expression tree construction process and allowing them to fully participate in value splicing and summary content generation. This is achieved by constructing a field combination-group mapping table and referencing it during the expression parsing phase to determine whether a field group is stable. If the reference group is stable, path loop detection, dependency chain monitoring, and suggestion generation are not performed, and the standard execution channel is directly entered. Furthermore, the risk warning component associated with this field combination is hidden from the configuration interface to reduce interface clutter, increase configurator confidence in reliable fields, and enhance efficiency. This not only reduces unnecessary system resource consumption but also ensures the consistency and automation of structurally stable fields during summary generation. It is one of the key strategies to ensure execution efficiency and stability in low-risk scenarios.

[0127] The identified field combination and its corresponding reference path structure are saved as a snapshot, and an identification tag of the field combination is generated based on the saved snapshot to prevent the field combination with the same reference relationship from being established again in the subsequent configuration process.

[0128] In order to achieve the goal of saving the identified field combinations and their corresponding reference path structures as snapshots, this can be achieved through persistent data structures and versioned configuration strategies in the software system. After the field reference path analysis is completed, the reference path corresponding to each identified field combination (i.e., the dependency edge relationship between the fields that constitute the combination) is abstracted as directed graph structure data and encoded in a structural serialized manner, such as using an adjacency list or adjacency matrix, to encapsulate all reference path logical expressions corresponding to the field combination into a persistent graph snapshot data object. These graph snapshot objects are then stored in a structured database table, and a unique combination identifier is generated for each field combination as the primary key, thereby achieving structural archiving of the field reference logic in the current configuration state.

[0129] After completing the saving of the field combination snapshot, it is also necessary to use software to generate an identification mark for the field combination based on the structural features in the snapshot. Specifically, a distinguishing structural signature can be generated by hashing or vectorizing the field names and their reference relationships in the field combination. To improve detection efficiency, a set of fast-search data indexing mechanisms can be constructed, such as inverted indexes or Bloom filters, which are specifically used to match and compare the structural signatures of newly created field combinations during subsequent configuration. Once a new combination that is consistent with a combination structure in a historical snapshot is identified, the system can trigger the early warning logic and prompt the configuration personnel that the combination has a potential circular reference risk, thereby blocking the repeated construction of such reference structures in advance.

[0130] The purpose of generating snapshots of field combination reference paths and identification tags is to enhance the low-code configuration platform's ability to memorize field reference structures. Because the mutual reference relationships between field combinations are difficult to visually identify in complex structures and with a large number of fields, the likelihood of repeatedly building high-risk structures is high. By automatically saving structure snapshots and generating identification tags, the system can detect structural conflicts and mitigate risks early in the configuration process, significantly improving the security and intelligence of the configuration process and preventing the recurrence of identified issues, thereby enhancing the maintainability and stability of the overall platform.

[0131] The above formulas are all dimensionless and numerical calculations. The formulas are obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formulas are set by technicians in this field according to actual conditions.

[0132] The above embodiments can be implemented in whole or in part via software, hardware, firmware, or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. A computer program product comprises one or more computer instructions or computer programs. When the computer instructions or computer program are loaded or executed on a computer, the processes or functions according to the embodiments of the present application are fully or partially generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. Computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless means (e.g., infrared, wireless, microwave, etc.). A computer-readable storage medium can be any available medium accessible by a computer or a data storage device such as a server or data center that contains a collection of one or more available media. Available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media. Semiconductor media can be solid-state drives.

[0133] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0134] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0135] In the several embodiments provided in this application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the embodiments described above are merely illustrative. For example, the division of units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interface, device or unit, which can be electrical, mechanical or other forms.

[0136] Units described as separate components may or may not be physically separate, and components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0137] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0138] The above are only specific embodiments of the present application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

Claims

1. A method for generating work order summaries based on low-code configuration, characterized in that: The specific steps include: Establish reference path relationships between fields, including recording the dependency path between each field and its referenced fields when configuring summary expressions to form a field reference chain set; Constructing a directed graph structure of the field reference chain set, including converting each dependency path in the field reference chain set into a directed edge to generate a directed graph consisting of field nodes and reference edges, which is used to represent the reference order relationship between fields; Identifying all combinations of fields configured to reference each other based on the constructed directed graph structure, including performing a path analysis operation in the directed graph structure to screen out all combinations of fields that have direct or indirect bidirectional reference relationships; Extract the path depths, the number of times a field is referenced in different paths, and the number of paths with loops for all identified field combinations from the constructed directed graph structure. Analyze the degree of closure of the field reference paths for each identified field combination based on the extraction results. Classify the identified field combinations based on the evaluation results, determine whether to interrupt the summary generation process, and execute the corresponding control strategy; The identified field combination and its corresponding reference path structure are saved as a snapshot, and an identification tag of the field combination is generated based on the saved snapshot to prevent the field combination with the same reference relationship from being established again in the subsequent configuration process.

2. A method for generating a work order summary based on low-code configuration according to claim 1, characterized in that: Construct a directed graph structure of the field reference chain set, specifically: Resolve each dependency path in the field reference chain set into a one-way dependency relationship between the starting field node and the target field node, and create a node set and an edge set based on all dependencies; Map each field name to a unique node in a directed graph structure, and convert each dependency path into a directed edge from the starting field node to the target field node; By traversing the field reference chain set, a complete field reference graph topology is established to form a directed graph structure consisting of all field nodes and the reference relationships between them, which is used to represent the reference order and dependency path distribution between fields.

3. A method for generating a work order summary based on low-code configuration according to claim 2, characterized in that: Based on the constructed directed graph structure, all combinations of fields that are configured to reference each other are identified, including performing path analysis operations in the directed graph structure to filter out all combinations of fields that have direct or indirect bidirectional reference relationships, specifically: Based on the constructed directed graph structure, perform path analysis on the starting node corresponding to each field name, obtain all target node paths reachable from the starting node, and record the forward dependency path relationship between fields; For each target node that appears in the forward path, a reverse path analysis operation is further performed to find whether there is a path from the target node to the original starting node, so as to determine whether the forward path and its corresponding reverse path form a direct or indirect bidirectional reference relationship; For field path combinations that are determined to have a bidirectional reference relationship, all fields involved in the path are extracted, and multiple field paths with path intersections and shared fields are aggregated to form a complete field combination unit; Through the aggregation operation, all field combinations that have direct or indirect bidirectional reference relationships in the constructed directed graph structure are screened out as the identification results of all field combinations in which the fields are configured to reference each other.

4. A method for generating a work order summary based on low-code configuration according to claim 3, characterized in that: The reverse path analysis operation is performed on each target node that appears in the forward path in the following ways: Starting from each target node, perform a reverse path search in the constructed directed graph structure, searching upwards level by level to see if there is a path pointing to the original starting node; During the path search process, the intermediate field nodes passed on each path are recorded to identify whether the path forms a complete return chain; When there is at least one path from the target node back to the starting node, determine whether the path constitutes a direct or indirect bidirectional reference relationship with the original path, and mark the target node and the original starting node together as a potential mutual reference field combination; Through the reverse path analysis operation, the reachability of the target nodes in all forward paths is verified to determine whether there is a reference path loop between fields.

5. A method for generating a work order summary based on low-code configuration according to claim 4, characterized in that: Extract the path depths, the number of times a field is referenced in different paths, and the number of paths with loops for all identified field combinations from the constructed directed graph structure. Analyze the degree of closure of the field reference paths for each identified field combination based on the extraction results. This specifically includes the following steps: Based on the constructed directed graph structure, the maximum path depth, the total number of times the field is referenced in different paths, and the number of closed loops detected in the path are extracted for each identified field combination. After extraction, unit normalization is performed to form a normalized feature vector. A three-dimensional structure tensor is constructed based on the normalized feature vector. The three tensor dimensions are the maximum path depth, the total number of times a field is referenced in different paths, and the number of closed loops detected in the path. This tensor represents the feature distribution of each field combination in the reference structure space. Performing a principal direction extraction operation on the constructed three-dimensional structure tensor to obtain the structural change direction corresponding to the principal component of the tensor, and performing a projection deviation rate calculation between the structural change direction and the preset standard direction; The inverse of the projection deviation rate of each field combination is used as the reference path closure coefficient of each field combination, and the reference path closure coefficient of each field combination is compared with the pre-set reference path closure coefficient threshold interval. Based on the comparison results, the reference path closure degree of each field combination when configured to reference each other is evaluated.

6. A method for generating a work order summary based on low-code configuration according to claim 5, characterized in that: Based on the constructed directed graph structure, the maximum path depth, the total number of times the field is referenced in different paths, and the number of closed loops detected in the path are extracted for each identified field combination. After extraction, unit normalization is performed to form a normalized feature vector, which is specifically: Obtain the maximum and minimum values ​​of all field combinations in three dimensions: maximum path depth, total number of times the field is referenced in different paths, and number of closed loops detected in the path, and construct the original three-dimensional structure vector for each field combination. Based on the maximum and minimum values ​​corresponding to each dimension, the structure vector of each field combination is normalized to convert the original parameter values ​​of different dimensions into dimensionless values ​​in a unified range; The normalized maximum path depth value, the total number of times the field is referenced in different paths, and the number of closed loops detected in the path are concatenated into a normalized feature vector in a fixed order.

7. A method for generating a work order summary based on low-code configuration according to claim 6, characterized in that: The main direction extraction operation is performed on the constructed three-dimensional structure tensor to obtain the structural change direction corresponding to the principal component of the tensor, and the projection deviation rate of the structural change direction is calculated with the preset standard direction, specifically: The constructed three-dimensional structure tensor is converted into a corresponding matrix form, and the tensor is flattened along the path depth dimension, the total number of times the field is referenced in different paths, and the number of closed loops detected in the path through the expansion operation; Perform principal component analysis on the tensor based on the flattened matrix form, extract the principal component direction of each field combination that changes most significantly in the multidimensional structure space, and use the principal component direction vector as the main direction representing the trend of citation structure change; Setting a preset standard direction as a reference direction vector indicating that the structure has no abnormal closing trend; The extracted principal component direction vector is projected onto the preset standard direction to obtain the cosine value of the angle between the two, and the projection deviation rate between the two is calculated based on this.

8. A method for generating a work order summary based on low-code configuration according to claim 7, characterized in that: The inverse of the projection deviation rate of each field combination is used as the reference path closure coefficient of each field combination. The reference path closure coefficient of each field combination is compared with the pre-set reference path closure coefficient threshold range. Based on the comparison results, the reference path closure degree of each field combination when configured to reference each other is evaluated, as follows: When the reference path closure coefficient of the field combination is less than the minimum value of the preset threshold interval, the reference path closure degree of the field combination when configured to reference each other is low; When the reference path closure coefficient of a field combination is greater than or equal to the minimum value of a preset threshold interval and less than or equal to the maximum value, the reference path closure degree of the field combination when configured to reference each other is medium; When the reference path closure coefficient of the field combination is greater than the maximum value of the preset threshold interval, the reference path closure degree of the field combination when configured to reference each other is high.

9. A method for generating a work order summary based on low-code configuration according to claim 8, characterized in that: Based on the evaluation results, the identified field combinations are classified and a decision is made as to whether to interrupt the summary generation process and execute the corresponding control strategy, specifically: Field combinations with low reference path closure levels are classified as high-risk reference groups. When an identified field combination is classified as a high-risk reference group, the summary generation process for that field combination is interrupted, preventing it from participating in the generation of subsequent summary expressions. A structural disconnection suggestion is generated to assist configuration personnel in adjusting field dependencies. Field combinations with a moderate degree of reference path closure are classified as warning reference groups. When an identified field combination is classified as a warning reference group, it is temporarily allowed to participate in summary generation. However, before the summary is executed, structural integrity check logic is inserted to dynamically monitor the execution process and generate field reconstruction suggestions to indicate potential closure risks. Field combinations with a high degree of citation path closure are divided into stable citation groups. When the identified field combinations are classified as stable citation groups, their normal participation in the summary generation process is maintained without interrupting execution and generating intervention suggestions.

Images