Visual process automatic layout method based on dependency graph optimization

Abstract

The invention discloses a visual process automatic layout method based on dependency graph optimization, and relates to the technical field of visual layout optimization, and the method comprises the following steps: S1, carrying out the global scanning of an overall dependency graph, employing a self-adaptive structure entropy algorithm, extracting the connectivity and path span information of each node, and constructing a structure bit order vector of the connection strength between the nodes; according to the method, low coupling dependence is identified through structure bit sequence vectors and semantic confidence scores, spatial distribution of weak connection subgraphs is optimized by combining a field model and repulsive force, path smoothness and layout stability are ensured through dynamic path adjustment and local simulation optimization, and visual hierarchy definition is improved through hierarchy consistency verification and global correction. According to the scheme, the readability and interactivity of the flow chart and the stability and adaptability in dynamic updating and industrial application are remarkably enhanced.

Description

Technical Field

[0001] This invention relates to the field of visual layout optimization technology, specifically to a visual process automatic layout method based on dependency graph optimization. Background Technology

[0002] Automatic layout of visualization processes based on dependency graph optimization refers to the use of dependency graphs (which abstract the nodes in the process and their interdependencies into a directed graph) as the basic data structure when drawing and displaying complex flowcharts such as business processes, data flows, or control logic. By analyzing and calculating the dependencies, hierarchical structure, path lengths, and constraints between nodes, a spatial layout scheme is automatically generated, making the arrangement of nodes in the flowchart clearer, more hierarchical, with the shortest dependency paths and the fewest intersections. This avoids the inefficiency and chaos caused by manual dragging and arrangement, while visually presenting and optimizing the understanding of the process, improving users' understanding and operational efficiency of complex processes. It is widely used in scenarios that require process visualization, such as business modeling, software engineering, data processing, and automated management and control.

[0003] The existing technology has the following shortcomings:

[0004] In existing automated layout processes for visualized flowcharts based on dependency graph optimization, there are shortcomings in the layout identification and spatial arrangement of weakly connected subgraphs. Specifically, when a subgraph in a dependency graph is connected to the main flowchart only through a few virtual connections or long-distance edges spanning multiple levels, the layout algorithm, when performing dependency strength calculation and graph structure optimization, is prone to misclassifying such weakly dependent subgraphs as locally clustered blocks with strong coupling to the main flowchart based on the singularity of the connection and the physical existence of the connection path. Furthermore, during layout optimization, due to an imbalance in the distribution of attraction weights between nodes, this subgraph is untargetedly compressed to the spatial edges or corners of the flowchart, creating a visually apparent local high-density clustering phenomenon.

[0005] The existence of this problem directly undermines the overall hierarchical clarity and accurate representation of dependencies in flowcharts. It not only affects users' intuitive perception and semantic interpretation of the process logic, easily leading to misunderstandings of the process structure, but also may cause key auxiliary sub-processes to be overlooked or mistakenly included in the main process path during process verification, simulation, and deployment, resulting in configuration defects in the operational process or confusion in business logic. This problem is particularly amplified when dealing with large-scale complex processes or scenarios with dynamic dependency updates, and in severe cases, it can even lead to visualization rendering errors, graphical interface crashes, or incorrect topology presentation of dependency graphs, thus limiting the application stability and industrial deployment value of visual process layout methods.

[0006] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0007] The purpose of this invention is to provide an automatic layout method for visualized flowcharts based on dependency graph optimization. This method identifies low-coupling dependencies through structural positional vectors and semantic confidence scores, optimizes the spatial distribution of weakly connected subgraphs using force field models and repulsive forces, ensures smooth paths and stable layouts through dynamic path adjustment and local simulation optimization, and enhances visual hierarchy clarity through hierarchical consistency verification and global correction. This solution significantly improves the readability, interactivity, stability, and adaptability of flowcharts in dynamic updates and industrial applications, thereby addressing the problems mentioned in the background art.

[0008] To achieve the above objectives, the present invention provides the following technical solution: an automatic layout method for a visual process based on dependency graph optimization, comprising the following steps:

[0009] S1, perform a global scan of the overall dependency graph, extract the connectivity and path span information of each node, and construct a structural positional vector of the connection strength between nodes;

[0010] S2, based on the structural positional vector, normalizes the weights of the connecting edges and introduces semantic confidence scores in combination with contextual dependencies, dynamically adjusts the dependency weights of the weakly connected subgraphs, and forms a set of low-coupling dependency weights.

[0011] S3, based on a set of low-coupling dependent weights, constructs a force field model to apply a repulsive adjustment force to the weakly connected subgraphs, controlling their visual interference distance from the core area of ​​the main process.

[0012] S4, when applying a repulsive adjustment force, dynamically adjust the physical length of the connection path and the direction of the curved path, and simultaneously control the path curvature and turning point position according to the path span and node position to avoid visual intersection and path overlap.

[0013] S5. After the path is dynamically adjusted, local incremental simulation optimization is initiated to optimize the spatial position and path shape of weakly connected subgraphs and associated paths in a fine-grained manner, thereby improving the stability and boundary clarity of the overall layout.

[0014] S6, after local incremental simulation optimization, performs hierarchical consistency verification, verifies the overall layout based on the prominence of the critical path and the rationality of the hierarchical distribution, and makes global closed-loop correction based on the verification results, so as to achieve clear hierarchical division and effective structural separation of weak connection subgraphs and main process paths.

[0015] Preferably, step S1 includes:

[0016] Perform a global scan on the overall dependency graph to extract the connectivity and path span information of each node;

[0017] Based on connectivity information and path span information, adaptive structural entropy calculation is performed, and the weighting factors in the structural entropy calculation are dynamically adjusted.

[0018] Calculate the connection strength of each pair of nodes, and generate a connection strength index by fusing the connection degree, path span, and structural entropy weights.

[0019] The contribution factors of connectivity, path span, and structural entropy are dynamically adjusted through a preset adaptive algorithm, and iteratively optimized until the weights are stable.

[0020] Based on the connectivity strength index, a structural rank vector is constructed that includes connectivity degree, path span, structural entropy value, and connectivity strength.

[0021] Preferably, step S2 includes:

[0022] Based on the structural positional vector, the weight of each connection edge in the dependency graph is normalized.

[0023] A semantic confidence scoring mechanism is introduced based on the contextual dependencies corresponding to the connecting edges to generate a score value;

[0024] The normalized weights and semantic confidence scores are weighted according to a preset fusion algorithm to obtain the adjusted dependency weights.

[0025] Connection edges with fusion weights below a preset threshold are identified as weak dependencies, forming a set of low-coupling dependency weights.

[0026] Assign dependency labels and coupling level identifiers to the set of loosely coupled dependency weights for use in subsequent process layout and path optimization.

[0027] Preferably, step S3 includes:

[0028] Based on a set of low-coupling dependency weights, repulsion and attraction parameters are set for nodes in weakly connected subgraphs and main process paths;

[0029] Construct a multi-level iterative force field model, dynamically adjust the node positions based on repulsive and attractive forces, and iterate until the globally optimal layout is reached;

[0030] To assess visual interference, the force field parameters are adjusted based on minimum spatial distance, number of path intersections, path overlap rate, path curvature change rate, node cluster density, average length of connecting paths, and visual interference ratio.

[0031] Once the interference level meets the preset standard, the spatial location and connection path of the nodes are fixed, and the coupling level and spatial coordinates of each node are recorded.

[0032] Preferably, step S4 includes:

[0033] During the application of repulsive adjustment force, the physical length of the connection path and the direction of the curved path are dynamically adjusted.

[0034] Based on the path span and the positional differences between nodes, set the path curvature and the spatial location of the turning points;

[0035] After dynamic adjustment is completed, a global layout consistency check is performed on the path, and the path intersection, overlap and curvature continuity are detected based on geometric topology analysis and visual interference assessment.

[0036] By using a global optimization algorithm to fine-tune and correct paths with interference, a global layout with stable structure and clear visual appearance is formed.

[0037] Preferably, step S5 includes:

[0038] After the connection path is dynamically adjusted, the weak connection subgraph and its associated paths are extracted, and the node coordinates and path attributes are recorded.

[0039] The position of nodes within the subgraph is adjusted based on repulsive and attractive forces to dynamically optimize path direction, curvature, and inflection point positions.

[0040] Fine-grained corrections are applied to path length, rate of curvature change, number of inflection points, and path spacing to smooth subgraph boundaries;

[0041] After completion, verify the minimum spatial distance and path continuity between the subgraph and the main process path, and solidify the optimized layout data.

[0042] Preferably, step S6 includes:

[0043] After local incremental simulation optimization is completed, critical paths and auxiliary paths are identified based on the hierarchical distribution data model, and the number of levels and node depth are recorded.

[0044] Based on the salience of the critical path and the rationality of the hierarchical distribution, perform hierarchical consistency verification and detect spatial occlusion and hierarchical stacking;

[0045] For the defective parts, adjust the spatial position and hierarchical order of the critical path and weak connection subgraphs, and optimize the path direction and spacing;

[0046] After completing the global closed-loop correction, perform a full-map structure and hierarchy review to generate a hierarchy structure mapping diagram and layout data files.

[0047] The technical effects and advantages provided by the present invention in the above technical solution are as follows:

[0048] This invention effectively filters low-coupling dependencies through a dual recognition mechanism of structural positional vectors and semantic confidence scores. Combined with a multi-level iterative force field model and repulsive adjustment forces, it ensures that weakly connected subgraphs maintain a reasonable spatial distance from the core area of ​​the main process, avoiding visual interference. Simultaneously, dynamic path adjustment and local incremental simulation optimization guarantee the smoothness of path lines and the stability of the overall layout. Hierarchical consistency verification and global closed-loop correction further enhance the visual clarity and structural integrity of the flowchart. The overall solution not only improves the readability and interactive experience of complex flowcharts but also enhances the stability and adaptability of visualized processes in dynamic updates, industrial deployments, and large-scale application scenarios. Attached Figure Description

[0049] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0050] Figure 1 This is a flowchart of the automatic layout method for a visual process based on dependency graph optimization according to the present invention. Detailed Implementation

[0051] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the description of this disclosure will be more complete and fully convey the concept of the exemplary embodiments to those skilled in the art.

[0052] This invention provides, for example Figure 1 The automatic layout method for visual processes based on dependency graph optimization, as shown, includes the following steps:

[0053] S1 performs a global scan on the overall dependency graph and uses an adaptive structural entropy algorithm to extract the connectivity and path span information of each node in the dependency graph. Based on the connectivity and path span between nodes, a structural position vector reflecting the connection strength between nodes is constructed.

[0054] The method of performing a global scan of the overall dependency graph, extracting the connectivity and path span information of each node in the dependency graph using the adaptive structural entropy algorithm, and constructing a structural positional vector reflecting the connection strength between nodes based on the connectivity and path span of the nodes, specifically includes the following steps:

[0055] The overall dependency graph to be processed is used as input. Each node in the dependency graph represents a functional unit or logical operation in the process, and the directed edges between nodes represent the dependencies between them. In this stage, by traversing all nodes in the dependency graph, the number of connections between each node and other nodes is counted, i.e., the node's connectivity information. During this counting process, the path length of each connection path is recorded simultaneously. The path length refers to the span of the path from the current node to all directly and indirectly connected target nodes, and the path span is measured by the cumulative number of edges traversed on the path. Through this statistical process, the connectivity and path span of each node in the global dependency graph can be comprehensively collected, providing a data foundation for subsequent dependency strength modeling.

[0056] Based on the collected node connectivity and path span information, an adaptive structural entropy calculation is performed on the overall dependency graph. Structural entropy is a mathematical tool used to measure the complexity of a graph structure, essentially reflecting the degree of order and information uncertainty in the connections between nodes. To achieve adaptive structural entropy calculation, the weight factors in the structural entropy calculation are dynamically adjusted for nodes with different connectivity and path spans. Nodes with higher connectivity and smaller path spans have a higher influence weight in the entropy calculation, while nodes with lower connectivity or larger path spans are given relatively lower weights, thus reflecting the importance and centrality of nodes in the overall graph structure. Through the above adaptive weight adjustment of structural entropy calculation, not only can the global complexity of the dependency graph be accurately described, but the core role of nodes in the topology can also be highlighted, avoiding the one-size-fits-all treatment of node attributes in traditional structural entropy methods, and enhancing the ability to distinguish the importance of nodes in complex process structures.

[0057] After calculating the structural entropy, for each pair of nodes in the dependency graph, the connection strength between the nodes is calculated based on their connectivity information, path span information, and structural entropy weights. The connection strength calculation employs normalization, normalizing connectivity and path span to a unified scale, and then merging them with the corresponding structural entropy values ​​using a weighted function to generate a connection strength index for each pair of nodes. Specifically, in the connection strength calculation formula, the weight parameters of the connectivity contribution factor, path span contribution factor, and structural entropy contribution factor are dynamically adjusted using a pre-defined adaptive algorithm to ensure that the connection strength comprehensively reflects the actual degree of dependency between nodes and their relative position in the global graph. This step enables a quantitative evaluation of each pair of nodes in the dependency graph, providing a scientific basis for dependency weight determination and hierarchical division in the subsequent layout process.

[0058] A pre-defined adaptive algorithm dynamically adjusts the weight parameters of the connectivity contribution factor, path span contribution factor, and structural entropy contribution factor to ensure that the calculation of connectivity strength adaptively reflects the degree of dependency between different nodes and their global position in the graph based on the actual structural characteristics of the dependency graph. The pre-defined adaptive algorithm can employ methods such as weight optimization based on genetic algorithms, parameter optimization based on particle swarm optimization, dynamic weight adjustment based on gradient descent, or weight self-adjustment based on fuzzy control.

[0059] The specific application steps are as follows:

[0060] The initial values ​​of the weight parameters of the three components are initialized based on the overall topological features of the dependency graph, and an optimization objective function is set. The objective function is used to measure the consistency between the global dependency metric after the connection strength calculation and the layer-level distribution.

[0061] After each connection strength calculation, the fit between global dependency and local hierarchy is evaluated, and this is used as a feedback signal to input the adaptive algorithm.

[0062] Adaptive algorithms dynamically adjust the weight parameters of each contributing factor based on feedback signals. For example, genetic algorithms select the optimal weight combination through population evolution, particle swarm optimization continuously approximates the optimal value of individuals and the population, gradient descent methods adjust parameter weights in reverse based on fitting error, and fuzzy control sets adjustment strategies based on error magnitude and trend.

[0063] Iteratively perform weight adjustments and recalculate connection strength until the objective function converges or reaches a set threshold, obtaining a stable and optimal weight allocation ratio. This achieves a dynamic balance between connection degree, path span, and structural entropy in connection strength calculation, ensuring that dependency graphs with different structural characteristics can obtain accurate and effective dependency strength evaluations.

[0064] Based on the connection strength indices of all node pairs, a structural order vector reflecting the overall dependency graph structure is constructed. The structural order vector is a vectorized representation of the relative ranking and dependency strength distribution of each node in the global dependency graph. Its vector elements include the node's connectivity, path span, structural entropy value, and a summary index of connection strength with other nodes. Through the structural order vector, not only can the position and dependency weight distribution of each node in the overall graph structure be intuitively reflected, but it can also provide quantifiable input data for subsequent steps such as dependency weight normalization, semantic confidence scoring, and visual layout optimization.

[0065] This step lays a quantitative and structural foundation for the subsequent automatic layout of dependency weight determination and visualization processes. By performing a global scan of the entire dependency graph, the connectivity and path span information of each node in the graph can be fully grasped, thus accurately reflecting the connection density and level or depth of the node in the dependency network. An adaptive structural entropy algorithm is used to further model the entropy values ​​of connectivity and path span information, effectively measuring the overall topological complexity of the dependency graph and the organizational order among nodes, avoiding one-sided evaluations based solely on a single dimension such as the number of connections or path length. The structural order vector constructed based on this information can quantitatively express the degree of association between each node and the global graph and its positional order in the structural hierarchy. This not only reveals the potential dependency strength between nodes but also provides a basis for subsequent steps such as identifying weakly connected subgraphs, accurately adjusting dependency weights, and setting mechanical parameters in the force field model. This step transforms the originally complex and difficult-to-interpret dependency graph into a highly operable mathematical representation through a structured quantification method. It solves the problem of dependency graphs lacking a global perspective and fine-grained weight analysis in existing technologies, and is a key prerequisite for ensuring clear hierarchy, reasonable paths, and accurate identification of weak connections during automatic layout.

[0066] S2, based on the structural positional vector, normalizes the weight of each connection edge in the dependency graph, and introduces a semantic confidence scoring mechanism according to the contextual dependency relationship corresponding to the connection edge. Based on the normalized weight and the semantic confidence score, the dependency weight of the weak connection subgraph in the dependency graph is dynamically adjusted to form a set of low-coupling dependency weights for identifying weak connection subgraphs.

[0067] The specific steps for normalizing the weights of each connection edge in the dependency graph based on the structural positional vector, introducing a semantic confidence scoring mechanism based on the contextual dependencies corresponding to the connection edges, and dynamically adjusting the dependency weights of weakly connected subgraphs in the dependency graph based on the normalized weights and semantic confidence scores to form a set of low-coupling dependency weights for identifying weakly connected subgraphs are as follows:

[0068] Based on the structural order vector constructed in the preceding steps, the weights of each connection edge in the dependency graph are normalized. Specifically, for the nodes at both ends of a connection edge, their corresponding connectivity, path span, and structural entropy values ​​in the structural order vector are extracted. These parameters are then comprehensively calculated using a preset weight fusion function to generate preliminary connection strength values. Subsequently, based on the maximum and minimum connection strength values ​​of all connections across the entire graph, a linear normalization method is used to normalize the strength value of each connection edge to the range of 0 to 1, ensuring that connections at different scales are comparable globally. This normalization process avoids weight bias caused by individual nodes with extremely high connectivity or extremely small spans in a single graph, ensuring the stability and consistency of subsequent dependency weight adjustments.

[0069] For each normalized connection edge, a semantic confidence scoring mechanism is introduced based on the contextual dependencies of the nodes connecting its two ends in the original dependency graph. The semantic confidence score analyzes multi-dimensional information such as functional relevance, call frequency, interaction strength, and dependency tightness of the two nodes connected by the edge in actual application scenarios such as business processes, data flows, and control logic, forming an objective score value. The score value can be generated by combining a rule base built based on domain knowledge, dependency patterns mined from historical data, and prediction results from machine learning models, thereby achieving accurate quantification of the semantic strength of each connection edge. This step, by introducing semantic-level evaluation, overcomes the limitations of traditional methods that rely solely on structured data to judge connection strength, enhancing the sensitivity and accuracy of dependency weights to business semantics.

[0070] Semantic confidence scoring is an evaluation method that quantifies the reliability and tightness of dependencies at the semantic level based on the contextual dependencies between the two nodes connected by the edge in the actual application scenario when evaluating each connection edge in a dependency graph. In this step, the role of this mechanism is to supplement the business meaning and logical strength that cannot be reflected by simply relying on graph structure attributes (such as connectivity and path span). By analyzing the actual functional associations, call frequency, interaction strength, data dependencies, and historical interaction patterns of the nodes at both ends of the connection edge in specific business processes, data flows, or control logic, a score reflecting its true dependency value is formed. Connections with high scores represent not only tight structural connections but also deep business or functional dependencies; connections with low scores indicate that although there is a physical connection, the correlation is weak, calls are infrequent, or it belongs to a secondary path in the actual business process. Through this scoring mechanism, connections that are structurally similar but have different semantic strengths can be effectively distinguished, improving the accuracy of weak connection subgraph identification, avoiding the misclassification of non-core paths into the core area of ​​the main process due to structural misjudgments, and ensuring that reasonable layering and distribution are achieved simultaneously in terms of structure and business semantics during layout optimization.

[0071] Based on the combined result of normalized weights and semantic confidence scores, the dependency weights of weakly connected subgraphs in the dependency graph are dynamically adjusted. Specifically, the normalized weight of each connection edge and its corresponding semantic confidence score are weighted according to a predefined fusion algorithm to obtain the adjusted dependency weights. For connection edges whose fusion results are below a certain threshold, their corresponding connections are determined to be weak dependencies. Furthermore, subgraphs formed by these weakly dependent connections are identified in the graph, and all connection weights in these subgraphs are uniformly reduced in weight, forming a set of low-coupling dependency weights. In this process, the dynamic adjustment not only considers the attributes of individual connection edges but also comprehensively considers the overall dependency characteristics of all edges within the weakly connected subgraph, achieving refined control over the transformation of dependency weights from the single-connection level to the subgraph level and the identification of weak connections.

[0072] A fusion algorithm is an algorithm that combines the normalized weights of each connection edge in a dependency graph with its corresponding semantic confidence score according to a predetermined weighting strategy to generate adjusted dependency weights that reflect both structural dependency strength and semantic relevance. The purpose of this fusion algorithm is to establish a balance between the physical measurement of connection strength and semantic understanding, avoiding misjudgments caused by relying solely on structural information or local overfitting caused by relying solely on semantic scores. Existing fusion algorithms include weighted average, weighted geometric average, additive models, fuzzy weighting, analytic hierarchy process (AHP), and weight learning algorithms based on neural networks or support vector machines. The specific implementation steps are as follows:

[0073] Initial weight coefficients are set for normalized weights and semantic confidence scores, and the magnitudes of the weight coefficients are derived from domain experience or through training with historical dependency data.

[0074] For each connection edge, a weighted calculation is performed, specifically in the form of a weighted average method, which involves multiplying the normalized weight by its corresponding weight coefficient, multiplying the semantic confidence score by another weight coefficient, and adding the two together to obtain the fused dependency weight.

[0075] Based on the fusion results, the fusion weights of all connected edges are normalized to ensure that the weights are distributed between 0 and 1, so as to adapt to the numerical input requirements of subsequent layout algorithms.

[0076] Connections with fusion weights below a set threshold are labeled as weak dependencies and aggregated to form a set of low-coupling dependency weights. This step, through the organic combination of structural and semantic information, achieves a multi-dimensional comprehensive evaluation of dependency weights, improving the accuracy and robustness of weak connection identification.

[0077] After forming a set of low-coupling dependency weights, specific dependency labels and coupling level identifiers are assigned to the connecting edges and nodes within the set, serving as the basis for identification and processing during subsequent layout. This identifier allows for the application of differentiated repulsive adjustment forces to low-coupling dependencies within the force field model of automatic layout, ensuring that weakly connected subgraphs are physically separated from the main flow path and clearly hierarchically divided in spatial layout. Furthermore, this set and its labels provide structural data support for process visualization rendering, dynamic interactive prompts, and subsequent dependency updates and maintenance, avoiding identification errors and path confusion caused by opaque weight adjustments in traditional layout processes.

[0078] Specific dependency labels and coupling level identifiers refer to the categorical and hierarchical identification information assigned to each connection edge and its associated subgraph in the dependency graph, based on the fused dependency weights, to mark the nature of their dependencies and the strength of their coupling. Their function is to provide a clear description of dependency attributes for subsequent process layout, graphics rendering, interactive prompts, and process verification. This allows the layout algorithm to adopt differentiated force field parameters and layout strategies based on different dependency types and coupling strengths, thereby achieving effective visual and structural separation between the main process and weakly connected subgraphs. In this implementation, dependency tags are categorized into four types: "strong dependency," "medium dependency," "weak dependency," and "virtual dependency," corresponding to different dependency tightnesses from high to low fusion weights. Strong dependency refers to connections with a fusion weight higher than 0.75, representing tight connections on the core path; medium dependency corresponds to weights between 0.5 and 0.75, indicating connections with substantial business relevance but not on critical paths; weak dependency has a weight between 0.25 and 0.5, indicating peripheral or auxiliary dependencies; and virtual dependency is a connection with a weight lower than 0.25, typically used to express low-frequency calls or weakly related paths. Coupling level identifiers are divided into four levels: "high coupling," "medium coupling," "low coupling," and "loose coupling," directly mapping to the dependency tag classification. This guides the setting of attraction and repulsion parameters during the layout process, ensuring fine-grained management of the spatial tightness and path intersection control of connections at different levels. The design of these tags and identifiers not only enhances the hierarchy and logical clarity of the flowchart but also provides structured parameter basis for dynamic updates and incremental layout optimization, demonstrating clear engineering application value and technological innovation.

[0079] This step aims to accurately identify and classify weakly connected subgraphs by finely and dynamically adjusting the weights of each connection edge in the dependency graph. This provides a solid weight foundation for spatial distribution optimization, path intersection control, and hierarchical clarity in subsequent automatic flowchart layout. Normalizing the connection edge weights based on structural positional vectors effectively addresses numerical scale differences between nodes due to connectivity, path span, and other metrics. This ensures comparability and a unified measurement standard for dependency weights globally, avoiding interference from local node attributes in overall layout decisions. Introducing a semantic confidence scoring mechanism further incorporates the semantic aspect of dependencies into the evaluation system, overcoming the limitations of solely relying on topological structure for strong / weak dependency determination. Each connection edge, in addition to structural strength, also possesses credibility evaluations based on semantic dimensions such as business, function, and process, thus more accurately reflecting the actual business value and logical tightness of dependencies. Through the dynamic fusion calculation of normalized weights and semantic confidence scores, the adjustment of dependency weights goes beyond reflecting static attributes, enhancing the adaptability of weights in dynamic and complex dependency scenarios. Ultimately, based on this dynamically adjusted set of low-coupling dependency weights, a quantitative standard and screening basis are provided for the identification of weak connection subgraphs. This ensures that in the subsequent layout process, specific repulsive adjustment forces and path optimization strategies can be applied to low-coupling dependencies, effectively preventing weak connection subgraphs from being incorrectly clustered or confused with the main process path. This achieves the layout goal of clear process layer hierarchy, clear dependencies, and visual intuitiveness. This step, through the fusion of structural and semantic dual evaluation mechanisms, significantly improves the accuracy and adaptability of weak connection discrimination in complex dependency networks, possessing significant technological innovation value.

[0080] S3, based on a set of low-coupling dependent weights, constructs a multi-level iterative force field model. During the layout simulation of the visualization process, it applies a repulsive adjustment force to the identified weakly connected subgraphs, controlling the minimum visual interference distance between the weakly connected subgraphs and the core area of ​​the main process in the spatial layout.

[0081] The specific implementation steps for constructing a multi-level iterative force field model based on a set of loosely coupled dependent weights, and applying a repulsive adjustment force to the identified weakly connected subgraphs during the layout simulation of the visualization process to control the minimum visual interference distance between the weakly connected subgraphs and the core area of ​​the main process in the spatial layout are as follows:

[0082] Based on the set of low-coupling dependency weights obtained in the previous steps, global attribute labeling is performed on all nodes and edges in the dependency graph. All nodes and edges in the subgraphs marked as weakly connected are uniformly identified, and based on the distribution characteristics of the low-coupling weights, initial repulsion force parameters are preset between the nodes of the weakly connected subgraphs and the main process path. These repulsion force parameters are dynamically set according to the weight strength, number of nodes, and dependency hierarchy differences between the weakly connected subgraphs and the main process path, forming a basic repulsion force threshold. Simultaneously, the nodes and edges of the main process path are marked as attraction cores, and corresponding attraction weights are set to ensure the structural cohesion of the main process and the spatial concentration of the core area, laying the initial physical conditions for the subsequent multi-level construction of the force field model.

[0083] Based on the above parameter settings, a multi-level iterative force field model is constructed. This model treats all nodes in the dependency graph as entities with physical point mass properties, and defines the interaction forces between nodes according to a hybrid physical model of gravity and Coulomb repulsion. For nodes in the main flow path, an attractive force based on connection weights is applied between them and other nodes, driving the nodes to cluster towards the core region during the layout process. For weakly connected subgraphs in the low-coupling dependency weight set, a repulsive adjustment force based on a repulsive force threshold is applied between them and nodes in the main flow path. This repulsive force increases as the weight decreases, ensuring that the weakly connected subgraphs are gradually pushed away from the core region of the main flow. During iterative execution, the ratio coefficients of attractive and repulsive forces are dynamically adjusted based on the node positions and force balance states after each round of force field calculation, gradually converging to the globally optimal spatial layout state, forming a preliminary arrangement with hierarchy and structural stability.

[0084] The hybrid physical model of universal gravitation and Coulomb repulsion refers to defining the interaction forces between nodes in a dependency graph by simulating the interaction laws of universal gravitation and Coulomb repulsion in the physical world. This drives the dynamic arrangement of nodes during the visualization process layout, ensuring a clear structure and reasonable spatial distribution. Its function is to simulate the attraction and repulsion effects between nodes, causing nodes in the main flow path to attract each other due to strong dependencies, forming a compact, concentrated, and hierarchical core area; while nodes in weakly connected subgraphs, due to their low dependency weights on the main flow path, are effectively pushed away from the core area by repulsion, preventing aggregation and chaotic intersections. Specific steps include:

[0085] Based on the fused dependency weights, a gravitational force is applied between nodes in the main process path. The magnitude of the attraction is proportional to the weight of the connecting edge; the higher the weight, the greater the attraction, driving the core path nodes closer to each other.

[0086] Coulomb repulsion is applied between nodes in the identified weak connection subgraph and nodes in the main process path. The magnitude of the repulsion is inversely proportional to the low coupling dependency weight. The lower the weight, the stronger the repulsion, ensuring that the weak connection subgraph is far away from the core area of ​​the main process.

[0087] In each layout iteration, the total resultant force between all nodes is calculated in real time, and the position of the nodes is adjusted according to the resultant force. The iteration converges to the global minimum energy state, thereby achieving a stable arrangement of nodes.

[0088] By combining the dynamic balance of repulsion and attraction, we ensure that the weak connection subgraphs and the main process path are effectively separated in space and that visual interference is minimized. At the same time, the overall layout maintains structural stability and clear hierarchy, thereby improving the readability of the flowchart and the intuitive expression of business logic.

[0089] After the force field model converges iteratively, the visual interference level of the completed spatial layout is evaluated. If the evaluation results show that the visual interference level exceeds a preset threshold, the direction and magnitude of the repulsive adjustment force are readjusted based on the evaluation feedback. The repulsive effect is strengthened directionally for weakly connected subgraphs with severe interference, while the attractiveness to the main flow path is moderately enhanced, prompting the nodes to redistribute to the optimal spatial positions. This round of evaluation and feedback adjustment can be iterated multiple times until the visual interference level is reduced to within the range that meets the layout clarity standard.

[0090] After the force field model converges iteratively, several quantitative indicators are used to evaluate the visual interference of the completed spatial layout. These include: 1) Minimum spatial distance, which measures the physical distance between the weak connection subgraph and the nearest node in the main process path, ensuring effective spatial separation; 2) Number of path intersections, which counts the number of intersections between the weak connection subgraph and the main process path, avoiding confusion in process logic caused by path intersections; 3) Path overlap rate, which is the proportion of overlapping parts of connecting paths in the graph to the total path length, used to assess whether path lines are excessively overlapped; 4) Path curvature change rate, which detects the frequency and magnitude of curvature changes in connecting paths in space, preventing excessive path distortion from affecting visual tracking; 5) Node cluster density, which refers to the number of nodes per unit area in a local region, used to identify whether there is node crowding and visual clutter; 6) Average length of connecting paths, which reflects whether the connection between the weak connection subgraph and the main process path is too long or unbalanced, avoiding the connection span affecting visual uniformity; and 7) View interference ratio, which evaluates the degree of interference in different process areas in the view by analyzing the area overlap ratio of the weak connection subgraph and the main process path within the view window through projection analysis. The combined evaluation of these indicators can comprehensively and multidimensionally measure the visual clarity and logical layering effect of the spatial layout, providing an objective basis for subsequent repulsion enhancement and path optimization, and ensuring the intuitiveness of the flowchart presentation and the accuracy of the operation.

[0091] After completing the above balance adjustment of repulsion and attraction, the spatial positions and connection paths of the final nodes are laid out and solidified. The spatial distribution of the weak connection subgraph and the core area of ​​the main process are kept at a preset minimum interference distance in physical space. At the same time, the coupling level and spatial coordinates of each node are recorded in the layout data for subsequent path optimization and hierarchical consistency verification.

[0092] This step aims to address issues such as unreasonable spatial arrangement, high visual interference, and hierarchical confusion between weakly connected subgraphs and the main flow path during the automatic layout of the visualized process through simulation and multi-level iterative optimization of a physical force field model. It achieves effective isolation and a clear hierarchical display between the weakly connected subgraphs and the core area of ​​the main flow. Based on the previously identified set of low-coupling dependency weights, this step constructs a hybrid physical model combining gravity and Coulomb repulsion, treating each node in the flow diagram as a point mass with physical properties. The attraction and repulsion relationships between nodes are dynamically set according to the dependency weights. Furthermore, a repulsive adjustment force is applied between nodes in the weakly connected subgraphs and nodes in the main flow path. The magnitude of the repulsive force is inversely proportional to its low-coupling dependency weight, ensuring that nodes with weaker dependencies are pushed away from the core flow area during layout, avoiding erroneous clustering or path intersections caused by weight asymmetry or layout algorithm bias. Simultaneously, nodes within the main flow path move closer together through attraction based on dependency weights, maintaining the compactness and smoothness of the core flow. Through multiple iterations, the combined forces and positional adjustments between nodes are calculated in real time, gradually converging to a globally optimal stable layout state. Ultimately, this step ensures that the weakly connected subgraphs maintain a reasonable minimum spatial distance from the main flow path, reducing the possibility of path intersections and overlaps, improving the overall flowchart's hierarchical clarity, structural stability, and visual readability, and preventing users from misunderstanding or making operational errors due to structural confusion when understanding complex processes. This process provides a fundamental spatial and structural guarantee for subsequent path optimization, incremental simulation, and hierarchical consistency verification, possessing crucial value for layout guidance and refined arrangement control.

[0093] S4, during the process of applying repulsive adjustment force, dynamically adjust the physical length and curvature of the connection path in the dependency graph, and synchronously control the curvature and spatial position of the inflection point of the connection path according to the span of the connection path and the relative position difference between the nodes, so as to avoid visual intersection and overlap of connection paths between weak connection subgraphs and main flow paths.

[0094] During the application of repulsive adjustment force, the physical length and curvature of the connection paths in the dependency graph are dynamically adjusted. The curvature and spatial location of the turning points of the connection paths are controlled synchronously based on the span of the connection paths and the relative positional differences between nodes. This avoids visual intersections and overlaps between weak connection subgraphs and the main flow path. The specific steps are as follows:

[0095] While applying a repulsive adjustment force between the weakly connected subgraphs in the dependency graph and the main flow path, the physical length of the path and the parameters of the curved path are initialized for all connected paths, especially those involving weakly connected subgraphs. A dynamic adjustment factor for the path length is established by calculating the spatial straight-line distance between the nodes at both ends of the connected path, the path span, and the relative displacement of the nodes after the repulsive adjustment force is applied. This adjustment factor dynamically extends or compresses the physical length of the connected path based on changes in the path span and spatial distribution density, ensuring that the path maintains an appropriate spatial distance from the main flow path even after the weakly connected subgraph is repelled, preventing the path from forcibly intersecting the main flow area due to insufficient physical length and causing visual intersections.

[0096] Based on adjusting the path length, the curved path direction is determined according to the relative positional differences between nodes. Specifically, for each connecting path, the relative orientation and hierarchical differences between its starting and target nodes in the spatial coordinate system are analyzed. Combined with the distribution pattern of preceding repulsive and attractive forces, the optimal curved path direction is selected. The determination of the path direction follows the principle of minimum interference, that is, avoiding high-density node areas of the main process path and existing path clusters as much as possible. At the same time, spatially open or low-interference areas are prioritized as transition sections for path curvature, thereby reducing the probability of path intersections and the overlap rate of path lines.

[0097] For each connecting path, the curvature and spatial location of turning points are dynamically set based on the path span, distance between nodes, and the previously determined curved path orientation. The curvature setting varies depending on the path span; the larger the span, the gentler the curvature, to avoid excessive visual distortion caused by excessive curvature on long paths. For paths with smaller spans, the curvature is appropriately increased to ensure that the path can flexibly avoid the main flow path and other connecting lines. The spatial location of turning points is dynamically determined based on the node density and path density of the spatial area traversed by the path. In principle, turning points are placed in low-density, low-path-overlap areas to further reduce the risk of path intersections and overlaps, while ensuring the smoothness of the path lines and the continuity of visual guidance.

[0098] After dynamically adjusting path length, curved path direction, curvature, and turning points, a global layout consistency check is performed on the paths in the overall dependency graph. By simulating the projection effect of path lines on the visualization interface, issues such as intersections, overlaps, and excessive twists with the main flow path are detected. If interfering contacts or visual discontinuities are still found, iterative optimization is performed based on the dynamically adjusted path parameters until the global path network achieves a spatially minimized intersection, minimal overlap, and optimal smoothness.

[0099] A global layout consistency check is performed on the paths in the overall dependency graph, using a combination of geometric topology analysis and visual interference evaluation. The specific steps are as follows:

[0100] Geometric topology analysis is used to comprehensively examine the spatial distribution of all paths in the entire map, path intersections, overlapping areas, and minimum distance between paths and nodes. The contact points, intersection angles, and overlap lengths of each path with other paths or nodes are quantified to ensure that there are no unreasonable overlaps or intersections of connecting lines in physical space.

[0101] By combining visual interference assessment, the display effect of the flowchart under different scaling scales and viewing angles is simulated based on projection mapping technology. The path intersection density, node occlusion rate and path curvature continuity within the field of view are dynamically calculated to evaluate the visual coherence and readability of the layout.

[0102] By using rule-based layout consistency criteria, global consistency judgment is performed on indicators such as the curvature change rate, the number of turning points, and the balance of path length distribution, to ensure that the overall layout of the path meets the requirements of structural stability, clear hierarchy, and smooth path at both the physical and visual levels.

[0103] The global optimization algorithm automatically fine-tunes and corrects paths with interference or uneven layout until the overall dependency graph meets the preset consistency and optimization standards in terms of spatial hierarchy, visual guidance and topology, ensuring that the visualized flowchart has high readability and structural integrity.

[0104] The purpose of this step is to dynamically adjust the physical attributes and spatial orientation of connection paths during the dependency graph layout optimization process. This resolves visual interference and hierarchical confusion caused by path intersections, overlaps, and curvature imbalances in the visualization of weak connection subgraphs and the main flow path. When a repulsive adjustment force pushes weak connection subgraphs away from the core area of ​​the main flow, the relative positions of nodes change. If the length, curvature, and other attributes of the connection paths are not adjusted synchronously, paths can easily become too long, too short, or directly intersect the main flow area, creating visual path intersections or line overlaps that interfere with the user's intuitive understanding of the flow structure. Therefore, under the action of the repulsive force, this step first dynamically corrects the physical length of the connection paths, ensuring that the paths maintain a reasonable physical extension even after changes in node spacing, avoiding excessively short paths that cause intersecting or excessively long paths that cause visual fragmentation. Second, based on the relative positions of nodes and the path span, the curved path orientation is dynamically planned, prioritizing low-density areas and avoiding the core cluster area of ​​the main flow to reduce interference with other paths or nodes. Furthermore, the curvature of the path is dynamically adjusted based on its span and node distribution. Larger spans result in gentler curvature, while smaller spans result in moderately stronger curvature. Simultaneously, the positions of path turning points are strategically set to ensure smooth transitions and prevent abrupt turns or excessive curvature from disrupting readability. Ultimately, these dynamic adjustments ensure the visual smoothness, clarity, and isolation of paths in weakly connected subgraphs, avoiding confusion and interference in the overall layout. This achieves a high degree of structural, visual, and interactive coordination and readability in the flowchart, improving users' cognitive efficiency and operational accuracy in handling complex processes.

[0105] S5. After the connection paths in the dependency graph are dynamically adjusted, the local incremental simulation optimization mechanism is started to perform fine-grained optimization on the spatial position and path shape of the weak connection subgraphs and their corresponding associated paths that have been arranged, so as to improve the structural stability and boundary clarity of the overall layout of the visualization process.

[0106] After dynamically adjusting the connection paths in the dependency graph, a local incremental simulation optimization mechanism is initiated to perform fine-grained optimization on the spatial location and path shape of the already arranged weak connection subgraphs and their corresponding associated paths. The specific steps to improve the structural stability and boundary clarity of the overall layout of the visualization process are as follows:

[0107] After the connection paths in the dependency graph are dynamically adjusted, all arranged weakly connected subgraphs and all paths directly or indirectly connected to the main process path are extracted based on the pre-identified set of low-coupling dependency weights. An initial parameter set for local simulation optimization is constructed by recording the node spatial coordinates, path physical length, path curvature, and number of inflection points for each weakly connected subgraph. For each weakly connected subgraph, its boundary range in the global layout, minimum spatial distance from the main process path, and the spatial extension direction and length distribution of the connected paths are defined as the basic inputs for local incremental simulation optimization. This step ensures that the optimization process is performed only on the weakly connected subgraphs and their associated paths, without causing positional disturbances to the main process path and its closely connected nodes, thus guaranteeing the stability of the main process path and the overall layout.

[0108] Based on the initial parameters optimized by local simulation, fine-grained position adjustments are performed on the nodes within each weakly connected subgraph, using repulsive and attractive forces. Specifically, for all nodes within each weakly connected subgraph, a repulsive force is first applied between any two non-directly connected nodes to ensure these nodes maintain an appropriate spatial distance, avoiding excessive proximity that could lead to localized crowding or visual density. The strength of the repulsive force is dynamically adjusted based on the distance between nodes; the closer the nodes, the stronger the repulsive force, encouraging nodes to naturally distribute towards more open spaces. Simultaneously, for node pairs with direct dependencies, an attractive force is applied corresponding to their dependency weights; higher weights result in stronger attraction, ensuring that related nodes are relatively close in space, maintaining the integrity of the dependency logic and hierarchical structure within the weakly connected subgraph. While adjusting node positions, each path within the weakly connected subgraph is dynamically updated. The path's direction is replanned in real-time based on the new node positions, using smooth curves or natural interpolation to draw the paths, ensuring visually smooth lines and avoiding abrupt zigzags or excessive turns. When a path passes through densely populated areas, its detour strategy is automatically optimized, prioritizing passage through areas with lower node density to reduce interference with other nodes or paths. Similarly, for connections between weakly connected subgraphs and the main flow path, the path direction and curvature are dynamically optimized after node position changes. This ensures that these connecting lines spatially avoid the core area of ​​the main flow, preventing paths from intersecting, crossing, or overlapping the main flow path, thus avoiding visual interference with the main flow structure. Through this dynamic adjustment mechanism of repulsion and attraction, combined with synchronous path optimization and direction correction, the weakly connected subgraphs are ensured to have a balanced spatial distribution and smooth, natural paths, while maintaining clear visual isolation and logical independence from the main flow path. Simultaneously with node position adjustments, the path direction is adjusted in real-time for each subgraph's internal paths and the paths between subgraphs and the main flow path. Based on the latest node positions, the path curve shape is redesigned, controlling the path curvature within the range of 0.1 to 0.3 to prevent excessive bending. The turning points of the paths are set at positions no less than three times the average node spacing to ensure smooth paths and easy visual tracking.

[0109] For local areas that have undergone initial optimization of node positions and paths, further fine-grained modifications are made to the path morphology. Specifically, this includes balancing path lengths to ensure they remain within ±15% of the global average path length, preventing excessively long or short paths from affecting layout coordination. Simultaneously, the rate of curvature change for each path is controlled, limiting it to no more than 0.05 per unit path length to prevent sudden bends due to node fine-tuning. The number of turning points is controlled between 1 / 5 and 1 / 3 of the total path length, ensuring flexible path avoidance while preventing excessive turns that would complicate the path and make it difficult to read. The minimum spacing between paths and the main flow path is controlled to be more than twice the average node radius to prevent paths from being too close to or overlapping with the main flow in space. The overall boundary of weakly connected subgraphs is smoothed to give the subgraphs a streamlined or elliptical visual distribution, avoiding visual interference caused by uneven boundaries.

[0110] Finally, after completing the above local optimizations, a verification of the fusion of local and global layouts is performed. Verification includes checking whether the minimum spatial distance between the optimized weakly connected subgraphs and the main flow path reaches five times the average node diameter, whether the path intersects with other paths, whether the path curvature and turns are smooth and continuous, and the symmetry and hierarchical rationality of the overall layout. The optimization effect is confirmed by quantitative evaluation indicators such as a path overlap rate of less than 5%, a reduction in the number of path intersections of more than 30%, and a path average length fluctuation rate of less than 10%. If the evaluation does not meet the standards, the local positions and paths are automatically iterated and adjusted based on feedback parameters until the indicators meet the standards. Finally, the optimized node coordinates, path shapes, and subgraph boundary data are solidified into the global layout, ensuring that the entire flowchart is structurally stable, visually clear, and has distinct boundaries, possessing strong readability and smooth interaction. This fine-grained local incremental simulation optimization scheme ensures the spatial isolation effect and path standardization of the weakly connected subgraphs, improving the industrial-grade display quality and dynamic adaptability of the flowchart.

[0111] The purpose of this step is to further optimize the spatial position and path shape of the weakly connected subgraphs and their associated paths through a local incremental simulation optimization mechanism after the dynamic adjustment of the connection paths in the dependency graph. This addresses issues that may arise from the previous repulsive force field adjustment and path dynamic correction, such as uneven distribution of local nodes, discontinuous path curvature, excessive path distortion, and blurred subgraph boundaries. In the previous steps, the weakly connected subgraphs have already achieved initial spatial separation from the core area of ​​the main flow under the influence of the repulsive adjustment force. However, this layout method dominated by a global force field makes it difficult to ensure the structural stability and visual aesthetics of local areas. It can easily lead to problems such as uneven spatial distribution of local nodes causing them to cluster or scatter excessively, and path lines becoming messy, abruptly turning, and unbalanced in path spacing due to positional changes. These deficiencies affect the overall hierarchical clarity and readability of the flowchart. Therefore, this step uses local incremental simulation optimization to precisely target the weakly connected subgraphs and all their associated paths. Without disturbing the main flow path and the global layout, simulation is used to fine-tune and precisely correct the node positions, path lengths, curvatures, turning points, and path spacing in the local space. Through repeated simulation iterations, defects such as loose local structures, frequent path intersections, or unclear node boundaries caused by the initial layout are eliminated. This makes the weakly connected subgraphs more compact and orderly in structure, with natural and smooth path connections. At the same time, the visual boundaries between the subgraphs and the main flow path are clear and with minimal interference. This step not only improves the structural stability and space utilization of the overall layout of the visualized flowchart but also enhances the professionalism and aesthetics of complex processes during presentation. It ensures that users can quickly identify the hierarchical relationships and logical boundaries between the main flow and auxiliary subgraphs when understanding, analyzing, and operating the flowchart, significantly improving the application value and interactive efficiency of the flowchart.

[0112] S6. After the local incremental simulation optimization is completed, the hierarchical consistency verification process is executed. Based on the prominence of the critical path and the rationality of the overall process hierarchy distribution, the overall layout of the visualized process is verified for consistency. Based on the verification results, the overall layout is corrected globally in a closed loop, so as to achieve clear division of weak connection subgraphs and main process paths in terms of hierarchy and effective separation in terms of structure.

[0113] After the local incremental simulation optimization is completed, a hierarchical consistency verification process is executed. Based on the prominence of the critical path and the rationality of the overall process hierarchy distribution, the overall layout of the visualized process is verified for consistency. Based on the verification results, a global closed-loop correction is performed on the overall layout to achieve a clear hierarchical division and effective structural separation between the weak connection subgraph and the main process path. The specific implementation steps are as follows:

[0114] After completing local incremental simulation optimization, a global hierarchical extraction and path identification are performed on the entire dependency graph. By performing a depth-first traversal and analysis of all paths in the dependency graph, all possible paths from the process start point to the end point are extracted. Based on indicators such as path dependency weight, path span, and node connectivity, the critical path of the main process and all auxiliary paths are identified. For each path, the number of levels it traverses, the level depth of its nodes, and its intersection and parallelism with other paths are recorded, forming a complete hierarchical distribution data model. This step, by extracting critical path and hierarchical information, provides a data foundation for subsequent consistency verification, ensuring that the hierarchical relationship of the overall process structure can be accurately modeled and evaluated.

[0115] Based on the extracted hierarchical distribution data model, a hierarchical consistency verification process is initiated. This verification process evaluates based on two core dimensions: first, the salientity of the critical path, verifying whether the critical path is located in a visually central position within the overall layout, and whether it has the shortest path connectivity and the fewest path turns in spatial arrangement, ensuring high visual recognizability and path continuity of the main flow; second, the rationality of the hierarchical distribution of the overall flow, assessing whether the hierarchical depth of all paths progresses progressively from the starting point to the ending point, whether weakly connected subgraphs are located in the edge areas outside the main flow paths, and whether they maintain sufficient spatial hierarchy differences from the core hierarchy of the main flow. If spatial occlusion, unclear path cross-layers, or chaotic hierarchical stacking are found during verification, the corresponding layout defects and hierarchical inconsistencies are recorded.

[0116] To address the deficiencies identified during consistency verification, a global closed-loop correction process was implemented. Specifically, based on identified anomalies in hierarchical distribution and insufficient path salientity, the spatial positions and hierarchical order of critical paths and weakly connected subgraphs were readjusted. Nodes and their paths along the critical path were prioritized for placement within the core viewpoint of the flowchart, and their visual coherence and spatial centrality within the overall layout were strengthened by shortening node spacing and optimizing path routing. For weakly connected subgraphs, their spatial positions away from the main flow path were further adjusted based on their dependency weights and hierarchical depth, and their paths were replanned to ensure clear hierarchical boundaries and spatial isolation between them and the main flow path.

[0117] After the global closed-loop correction is completed, a full-map structure and hierarchy review is performed to confirm that the connectivity of all paths, the rationality of the hierarchy distribution, and the visual clarity of the spatial arrangement all meet the preset standards. A final hierarchy mapping map and layout data file are generated for the corrected overall layout, solidifying the optimization results and providing sustainable hierarchical basis and spatial parameters for subsequent dynamic dependency updates and layout adjustments.

[0118] In existing automated layout processes for visualized flowcharts based on dependency graph optimization, there are shortcomings in the layout identification and spatial arrangement of weakly connected subgraphs. Specifically, when a subgraph in a dependency graph is connected to the main flowchart only through a few virtual connections or long-distance edges spanning multiple levels, the layout algorithm, when performing dependency strength calculation and graph structure optimization, is prone to misclassifying such weakly dependent subgraphs as locally clustered blocks with strong coupling to the main flowchart based on the singularity of the connection and the physical existence of the connection path. Furthermore, during layout optimization, due to an imbalance in the distribution of attraction weights between nodes, this subgraph is untargetedly compressed to the spatial edges or corners of the flowchart, creating a visually apparent local high-density clustering phenomenon.

[0119] The purpose of this step is to review and correct the overall layout structure of the visualized process from a global perspective of hierarchy and path after the local incremental simulation optimization is completed. This ensures that the flowchart, after previous arrangement and local optimization, achieves a high degree of consistency and rationality in hierarchical structure, path continuity, and visual distinction between main and auxiliary processes. During the automatic layout of complex flowcharts, although weakly connected subgraphs undergo exclusionary adjustments and dynamic path optimization, achieving initial physical isolation from the main process path, they may still experience problems such as disordered hierarchical order, unclear critical paths, and blurred visual boundaries between subgraphs and the main process due to minor adjustments in local positions, path detours, or space constraints. These issues can affect the clear expression of the overall process and the user's cognitive efficiency. Therefore, this step verifies the prominence of the critical path through a hierarchical consistency verification process, ensuring that the main process path is always visually at the core of the layout, with clear node hierarchy progression, continuous and smooth path flow, and is not disturbed by path intersections or space compression of weakly connected subgraphs. Simultaneously, the evaluation assesses whether the hierarchical distribution of the overall process conforms to a progressive unfolding from the starting point to the end point, and whether weakly connected subgraphs are distributed around the periphery of the process and maintain spatial and logical isolation from the main process path. Based on these verification results, if situations are found where the critical path is not prominent, the hierarchical depth is disordered, or subgraphs are too close to the main process path, a global closed-loop correction will be triggered. Through global adjustments to node positions, path rearrangement, and hierarchical reordering, the clarity of the process hierarchy and the recognizability of the paths are restored and strengthened. Ultimately, this step ensures that the automatic layout not only achieves optimization effects in local details, but also meets users' intuitive needs for a clear process, well-defined paths, and a clear distinction between primary and secondary elements in terms of overall hierarchical structure and visual guidance. This achieves global closed-loop and consistency control of the layout effect, significantly improving the engineering practical value and interactive usability of the visualized flowchart.

[0120] The aforementioned automatic layout method for visualized flowcharts based on dependency graph optimization accurately identifies and effectively separates the structure, hierarchy, and spatial distribution between weakly connected subgraphs and the main flow path. This completely resolves issues in existing technologies, such as hierarchical ambiguity, path intersections, and layout instability caused by misjudgment of weakly connected subgraphs, unbalanced weight allocation, and chaotic path arrangement. This method effectively filters low-coupling dependencies through a dual identification mechanism of structural positional vectors and semantic confidence scoring. Combined with a multi-level iterative force field model and repulsive adjustment force, it ensures that weakly connected subgraphs maintain a reasonable spatial distance from the core area of ​​the main flow, avoiding visual interference. Simultaneously, dynamic path adjustment and local incremental simulation optimization ensure the smoothness of path lines and the stability of the overall layout. Hierarchical consistency verification and global closed-loop correction further enhance the visual clarity and structural integrity of the flowchart. The overall solution not only improves the readability and interactive experience of complex flowcharts but also enhances the stability and adaptability of visualized flowcharts in dynamic updates, industrial deployments, and large-scale application scenarios.

[0121] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

[0122] It should be noted that, in this document, the use of relational terms such as "first" and "second" is merely for distinguishing one entity or operation from another, and does not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the element.

[0123] It should be understood that in the various embodiments of this application, the order of the above-mentioned processes does not imply 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 this application.

[0124] Those skilled in the art will recognize that the units and algorithm steps of the various examples 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 implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art 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.

[0125] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0126] The units described as separate components may or may not be physically separate. The 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 the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0127] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0128] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0129] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A method for automatic layout of a visual process based on dependency graph optimization, characterized in that, Includes the following steps: S1, perform a global scan of the overall dependency graph, extract the connectivity and path span information of each node, and construct a structural positional vector of the connection strength between nodes; S2, based on the structural positional vector, normalizes the weights of the connecting edges and introduces semantic confidence scores in combination with contextual dependencies, dynamically adjusts the dependency weights of the weakly connected subgraphs, and forms a set of low-coupling dependency weights. S3, based on a set of low-coupling dependent weights, constructs a force field model to apply a repulsive adjustment force to the weakly connected subgraphs, controlling their visual interference distance from the core area of ​​the main process. S4, when applying a repulsive adjustment force, dynamically adjust the physical length of the connection path and the direction of the curved path, and simultaneously control the path curvature and turning point position according to the path span and node position to avoid visual intersection and path overlap. S5. After the path is dynamically adjusted, local incremental simulation optimization is initiated to optimize the spatial position and path shape of weakly connected subgraphs and associated paths in a fine-grained manner, thereby improving the stability and boundary clarity of the overall layout. S6, after local incremental simulation optimization, performs hierarchical consistency verification, verifies the overall layout based on the prominence of the critical path and the rationality of the hierarchical distribution, and makes global closed-loop correction based on the verification results, so as to achieve clear hierarchical division and effective structural separation of weak connection subgraphs and main process paths.

2. The automatic layout method for visual processes based on dependency graph optimization according to claim 1, characterized in that, Step S1 includes: Perform a global scan on the overall dependency graph to extract the connectivity and path span information of each node; Based on connectivity information and path span information, adaptive structural entropy calculation is performed, and the weighting factors in the structural entropy calculation are dynamically adjusted. Calculate the connection strength of each pair of nodes, and generate a connection strength index by fusing the connection degree, path span, and structural entropy weights. The contribution factors of connectivity, path span, and structural entropy are dynamically adjusted through a preset adaptive algorithm, and iteratively optimized until the weights are stable. Based on the connectivity strength index, a structural rank vector is constructed that includes connectivity degree, path span, structural entropy value, and connectivity strength.

3. The automatic layout method for visual processes based on dependency graph optimization according to claim 1, characterized in that, Step S2 includes: Based on the structural positional vector, the weight of each connection edge in the dependency graph is normalized. A semantic confidence scoring mechanism is introduced based on the contextual dependencies corresponding to the connecting edges to generate a score value; The normalized weights and semantic confidence scores are weighted according to a preset fusion algorithm to obtain the adjusted dependency weights. Connection edges with fusion weights below a preset threshold are identified as weak dependencies, forming a set of low-coupling dependency weights. Assign dependency labels and coupling level identifiers to the set of loosely coupled dependency weights for use in subsequent process layout and path optimization.

4. The automatic layout method for visual processes based on dependency graph optimization according to claim 1, characterized in that, Step S3 includes: Based on a set of low-coupling dependency weights, repulsion and attraction parameters are set for nodes in weakly connected subgraphs and main process paths; Construct a multi-level iterative force field model, dynamically adjust the node positions based on repulsive and attractive forces, and iterate until the globally optimal layout is reached; To assess visual interference, the force field parameters are adjusted based on minimum spatial distance, number of path intersections, path overlap rate, path curvature change rate, node cluster density, average length of connecting paths, and visual interference ratio. Once the interference level meets the preset standard, the spatial location and connection path of the nodes are fixed, and the coupling level and spatial coordinates of each node are recorded.

5. The automatic layout method for visual processes based on dependency graph optimization according to claim 1, characterized in that, Step S4 includes: During the application of repulsive adjustment force, the physical length of the connection path and the direction of the curved path are dynamically adjusted. Based on the path span and the positional differences between nodes, set the path curvature and the spatial location of the turning points; After dynamic adjustment is completed, a global layout consistency check is performed on the path, and the path intersection, overlap and curvature continuity are detected based on geometric topology analysis and visual interference assessment. By using a global optimization algorithm to fine-tune and correct paths with interference, a global layout with stable structure and clear visual appearance is formed.

6. The automatic layout method for visual processes based on dependency graph optimization according to claim 1, characterized in that, Step S5 includes: After the connection path is dynamically adjusted, the weak connection subgraph and its associated paths are extracted, and the node coordinates and path attributes are recorded. The position of nodes within the subgraph is adjusted based on repulsive and attractive forces to dynamically optimize path direction, curvature, and inflection point positions. Fine-grained corrections are applied to path length, rate of curvature change, number of inflection points, and path spacing to smooth subgraph boundaries; After completion, verify the minimum spatial distance and path continuity between the subgraph and the main process path, and solidify the optimized layout data.

7. The automatic layout method for visual processes based on dependency graph optimization according to claim 1, characterized in that, Step S6 includes: After local incremental simulation optimization is completed, critical paths and auxiliary paths are identified based on the hierarchical distribution data model, and the number of levels and node depth are recorded. Based on the salience of the critical path and the rationality of the hierarchical distribution, perform hierarchical consistency verification and detect spatial occlusion and hierarchical stacking; For the defective parts, adjust the spatial position and hierarchical order of the critical path and weak connection subgraphs, and optimize the path direction and spacing; After completing the global closed-loop correction, perform a full-map structure and hierarchy review to generate a hierarchy structure mapping diagram and layout data files.

Images