Development process management method based on knowledge graph

Abstract

The invention relates to the technical field of development process management, in particular to a knowledge graph-based development process management method, which comprises the following steps of: acquiring actual starting time and earliest starting time, identifying an abnormal path, updating a network relationship, sorting node priorities, adjusting a completion time limit, reconstructing a view and visually outputting the view. According to the method, a dependency path with negative total floating time is dynamically recognized in combination with a knowledge graph, the potential construction period risk is accurately exposed, node logic continuity is enhanced through dependency chain recursion rejection and network structure reconstruction, task priorities are clearly divided by utilizing topological sorting and a WBS hierarchy depth two-dimensional sorting mechanism, scheduling sequence disorder is avoided, and task scheduling efficiency is improved. A comparison mechanism of the latest completion time and the available construction period is introduced, potential delay nodes are identified in advance, the scheduling flexibility is improved, standardized visual expression is performed after a dependency state is uniformly updated, the perceptibility and response efficiency of a development state are enhanced, and the execution rigidity and plan controllability of project management are enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of development process management, and in particular to a development process management method based on knowledge graph. Background Art

[0002] The field of development process management encompasses various methods and approaches for planning, monitoring, coordinating, and optimizing all phases of the software development lifecycle. Its core content encompasses the orderly management of each stage of the development process, including requirements analysis, design, coding, testing, deployment, and maintenance. By standardizing processes, improving collaboration efficiency, and reducing development costs, it aims to ensure quality and controllable delivery of software projects. This often involves specific technologies such as configuration management, version control, process modeling, and resource scheduling.

[0003] Among them, the development process management method based on knowledge graph refers to constructing a knowledge graph containing knowledge entities related to the development process and their relationships, and using the graph to realize information integration, relationship modeling and decision support for development process management, covering knowledge acquisition, relationship extraction and semantic representation of requirements, design, implementation, testing, deployment and other links in the development process. Specifically, it is based on knowledge representation methods such as ontology modeling, semantic association reasoning and relationship extraction technology to establish a unified knowledge base to assist in the identification of development tasks, progress tracking and risk warning, and thus complete the structured management and knowledge support of the entire software development process.

[0004] Existing development process management approaches often rely on static path comparison or fixed plan analysis to identify schedule anomalies. These approaches lack adaptive dependency chain identification mechanisms, making it difficult to detect hidden schedule risk nodes within the path. This leads to a lag in the response between task execution status and plan deviations. Task priority is often determined simply by the planned order or task type, disregarding the task's actual position within the structural hierarchy and dependency network, resulting in scheduling that deviates from the project's logical structure. Task completion deadlines are not effectively linked to the project's actual calendar, and remaining time estimates for uncompleted tasks often misjudge schedule risks based on static plan baselines. Task status displays lack consistent presentation across views and lack granularity, leading project participants to frequently face multiple versions of task status and ambiguous priority information during execution. For example, in complex software projects, the same node may have varying priorities across multiple tasks. The lack of a consistent assessment mechanism leads to resource conflicts and response delays. This irrational task allocation ultimately creates bottleneck task backlogs, impacting the overall project delivery cycle. Summary of the Invention

[0005] The purpose of this invention is to solve the shortcomings of the existing technology and propose a development process management method based on knowledge graph.

[0006] In order to achieve the above objectives, the present invention adopts the following technical solution: a development process management method based on knowledge graph, comprising the following steps: S1: Obtain the actual start time of the WBS node in the knowledge graph, detect the earliest start time of the predecessor node, determine whether there is a dependency with a negative total float time, mark it as an abnormal path, and generate a list of abnormal dependencies; S2: Based on the abnormal dependency list, recursively remove all downstream development node dependencies of the abnormal path, release the abnormal connection of the marked development node, update the logical relationship of the project network diagram, and generate a baseline plan snapshot; S3: Based on the baseline plan snapshot, count the number of predecessors of the unfinished nodes, trace back to the starting development node, sort the development nodes in order according to the WBS level depth, and generate a management priority list; S4: Extract the latest completion time of each unfinished node based on the management priority list, determine whether there is a development node whose completion time exceeds the remaining available time, adjust the priority of the corresponding development node, and generate a collaborative task management priority sequence; S5: Based on the collaborative task management priority sequence, the dependency and priority status of all development nodes in the knowledge graph are uniformly updated, the development process management view is rebuilt, and visual output is performed to generate a development process management view model.

[0007] As a further solution of the present invention, the abnormal dependency list includes a negative floating dependency mark, a critical path conflict identifier, and an actual start time deviation node; the baseline plan snapshot includes an updated logical network diagram, a set of nodes with abnormal path removed, and corrected status information of the development node; the management priority list includes a development node sequence table generated according to the WBS hierarchy and topological relationship, a predecessor quantity indicator of each node, and a priority sorting index; the collaborative task management priority sequence includes the identification results of time-sensitive nodes, priority adjustment suggestions, and remaining duration matching status; the development process management view model includes a node status map, a task priority mapping diagram, and an information display structure that complies with the PMBOK6th standard.

[0008] As a further solution of the present invention, the specific steps of obtaining the abnormal dependency list are: S111: Obtain the actual start time of each WBS node in the knowledge graph and the configured predecessor node, detect the earliest start time of each predecessor node, and compare it with the actual start time of the corresponding WBS node, calculate the time difference between the two, and generate the predecessor time difference; S112: Based on the predecessor time difference, a difference judgment is performed on all WBS nodes with dependency relationships, with the planned duration as a reference, to analyze whether any time difference exceeds the planned duration, and all dependency paths that meet the conditions are screened. The corresponding dependency directions, node pairs, and offset magnitudes are recorded to obtain abnormal dependency marking results. S113: Based on the abnormal dependency marking result, all dependency paths whose deviation exceeds the planned duration are screened, corresponding WBS node combinations and dependency types are extracted, and an abnormal dependency list is established.

[0009] As a further solution of the present invention, the specific steps for obtaining the benchmark plan snapshot are: S211: Read the abnormal path in the abnormal dependency list, retrieve the corresponding abnormal path end node, locate all downstream connection relationships of the node, record all path numbers indirectly associated with the abnormal path, and generate an abnormal diffusion path number set; S212: Based on the abnormal diffusion path number set, all development node number information is counted, the logical connection relationship between each node is matched, and the node connection elimination strength value is calculated. If the node connection elimination strength value is lower than the judgment threshold, the node connection elimination is performed to obtain a logical connection update matrix; S213: Extract all node numbers and corresponding outgoing edge information according to the logical connection update matrix, reconstruct and update the structure, freeze the structure state according to the current time node, retain the adjusted global network connection information, mark the version number and connect the time tag, and establish a baseline plan snapshot.

[0010] As a further solution of the present invention, the specific steps for obtaining the management priority list are: S311: Based on the baseline plan snapshot, the completion flag fields of all nodes are extracted, the corresponding incoming edge connection path list of the unfinished nodes is read, the number of predecessor nodes of each node is counted, and a number-predecessor number mapping matrix is ​​constructed to obtain a predecessor relationship quantity table; S312: Based on the table of the number of immediate predecessors, all nodes with a number of immediate predecessors of zero are identified as topological starting points, all edge information in the network graph is sequentially scanned, a topological sorting process is performed, and a topological hierarchical order result is generated; S313: According to the topological hierarchy order result, extract the corresponding WBS coding information for each node, split the WBS number field, identify the WBS hierarchy depth value of each node, calculate the management priority value, arrange the node numbers from high to low according to the numerical value, and establish a management priority list.

[0011] As a further solution of the present invention, the specific steps for obtaining the collaborative task management priority sequence are: S411: Extract the latest planned completion time of each node from the unfinished nodes in the management priority list, synchronously obtain the remaining available time from the current time node to the project deadline, record the time difference, establish a mapping record between the node number and the overdue status, and generate a node overdue identification matrix; S412: Based on the numbers of all nodes marked as having time limit conflicts in the node overdue identification matrix, the original priority values ​​in the management priority list are updated, the original order of the nodes that have not expired is retained, and the updated priority information is paired with the node numbers to generate a priority adjustment value table; S413: Reorder all unfinished nodes according to the priority adjustment value table, use the adjusted priorities as the main sequence basis, and establish a collaborative task management priority sequence in combination with the original dependency structure and time limit conflict.

[0012] As a further solution of the present invention, the specific steps for obtaining the development process management view model are: S511: Based on the collaborative task management priority sequence, sequentially read the entity status of each development node in the current knowledge graph, extract the dependency structure and priority information of adjacent nodes, update the priority number and path frequency field of the dependent node, and generate priority dependency synchronization graph data; S512: Based on all node attributes in the priority-dependent synchronization graph data, a time series structure indexed by node priority is constructed, each node is divided into a corresponding management period, the distribution range and central tendency within the same period are determined, and a vertical arrangement rule is established in combination with the priority order to obtain a process layout coordinate matrix; S513: According to the node layout information recorded in the process layout coordinate matrix, the node graphic border style, the connection path line type and the field display method are selected, a visual task card is output for each development node, and a development process management view model is established.

[0013] Compared with the prior art, the advantages and positive effects of the present invention are: In the present invention, by combining the knowledge graph to dynamically identify dependency paths with negative total float time, potential construction period risks are accurately exposed, and the logical continuity of nodes is enhanced through recursive elimination of dependency chains and reconstruction of network structures. The dual-dimensional sorting mechanism of topological sorting and WBS hierarchical depth is used to clearly divide task priorities and avoid scheduling sequence confusion. A mechanism for comparing the latest completion time with the available construction period is introduced to identify potential delay nodes in advance and improve scheduling flexibility. After uniformly updating the dependency status, standardized visual expression is performed to enhance the perceptibility and response efficiency of the development status, and to enhance the execution rigidity and plan controllability of project management. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 It is a flow chart of the main steps of the present invention; Figure 2 Obtaining a flow chart for the abnormal dependency list of the present invention; Figure 3 Obtaining a flow chart for the benchmark plan snapshot of the present invention; Figure 4 Obtaining a flow chart for managing the priority list for the present invention; Figure 5 Obtaining a flow chart for the collaborative task management priority sequence of the present invention; Figure 6 A flow chart is obtained for the process management view model developed for the present invention. DETAILED DESCRIPTION

[0015] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0016] In the description of the present invention, it should be understood that the terms "length," "width," "up," "down," "front," "back," "left," "right," "vertical," "horizontal," "top," "bottom," "inside," "outside," and the like, indicating positions or relationships, are based on the positions or relationships shown in the accompanying drawings and are intended only to facilitate the description of the present invention and simplify the description. They do not indicate or imply that the devices or elements referred to must have a specific orientation, be constructed, or operate in a specific orientation. Therefore, they should not be construed as limiting the present invention. Furthermore, in the description of the present invention, "plurality" means two or more, unless otherwise expressly and specifically defined.

[0017] See also Figure 1 ,The development process management method based on knowledge graph includes the following steps: S1: Obtain the actual start time of the WBS node in the knowledge graph, detect the earliest start time of the predecessor node, determine whether there is a dependency with a negative total float time (the delay time of a task defined by PMBOK), mark it as an abnormal path, and generate a list of abnormal dependencies; S2: Based on the abnormal dependency list, recursively remove all downstream development node dependencies of the abnormal path, release the abnormal connection of the marked development node, update the logical relationship of the project network diagram, and generate a baseline plan snapshot; S3: Based on the baseline plan snapshot, count the number of predecessors of unfinished nodes, trace them back to the starting development node using a topological sorting algorithm, and sort each development node in order based on the WBS level depth (a core structural indicator in project management, referring to the number of decomposition levels from the project's overall goal (Level 1) to the work package (Work Package) in the work breakdown structure (WBS)) to generate a management priority list; S4: Based on the management priority list, extract the latest completion time of each unfinished node and compare it with the remaining available time in the project calendar to determine whether there are development nodes whose completion deadlines exceed the remaining available time. Adjust the priority of the corresponding development nodes and generate a collaborative task management priority sequence. S5: Based on the collaborative task management priority sequence, uniformly update the dependency and priority status of all development nodes in the knowledge graph, rebuild the development process management view according to the latest development node management status, and use the PMBOK6th version visualization specification (standardized requirements for information display in the project management knowledge system) for visualization output to generate a development process management view model.

[0018] The abnormal dependency list includes negative floating dependency markers, critical path conflict identifiers, and actual start time deviation nodes. The baseline plan snapshot includes the updated logical network diagram, the set of nodes with abnormal path removed, and the revised status information of the development nodes. The management priority list includes the development node sequence table generated according to the WBS hierarchy and topological relationship, the predecessor quantity indicator of each node, and the priority sorting index. The collaborative task management priority sequence includes the identification results of time-sensitive nodes, priority adjustment suggestions, and remaining duration matching status. The development process management view model includes the node status diagram, task priority mapping diagram, and information display structure that complies with the PMBOK6th standard.

[0019] See also Figure 2 , the specific steps of S1 are: S111: Obtain the actual start time of each WBS node in the knowledge graph and the configured predecessor node, detect the earliest start time of each predecessor node, and compare it with the actual start time of the corresponding WBS node, calculate the time difference between the two, and generate the predecessor time difference; Get the actual start time of the WBS node in the knowledge graph and the number information of its corresponding predecessor node. After extraction, establish an index mapping structure for the node number and node attribute, and uniformly convert the actual start time field into a daily value unit. For example, the actual start time of the node WBS-001 is the 30th day, and the earliest start time of its predecessor node PRE-001 is the 25th day. After unifying the data, perform a difference operation on the two time fields, that is, 30 minus 25 equals 5. The difference represents the time interval between the WBS node and its predecessor. Repeat this process for different node dependencies to calculate their time difference item by item, for example, WBS-002 and PRE-00 2, 50 minus 45 equals 5, WBS-003 and PRE-003 equals 70 minus 65 equals 5, and WBS-004 and its predecessor PRE-001 equals 90 minus 95 equals -5, which needs to be recorded as an abnormal dependency situation of early start. In actual deployment, batch import is usually completed through the progress execution data exported from the project management system. After structured table processing, batch time difference processing is directly performed to construct the time difference field. At the same time, ensure that each WBS node is mapped to its predecessor one by one to avoid missing dependency links. Through pair-by-pair extraction and calculation of time differences, a data set is formed, which can be further used as the core input data for subsequent dependency relationship judgment.

[0020] Table 1 Node dependency execution data table WBS node number Predecessor node number WBS actual start time (days) Earliest start time of the precursor (days) Planned construction period (days) Logical dependency quantization value WBS-001 PRE-001 30 25 6 2 WBS-002 PRE-002 50 45 4 1 WBS-003 PRE-003 70 65 5 3 WBS-004 PRE-001 90 95 8 2 As shown in Table 1, four groups of time-related parameters between WBS nodes and predecessor nodes are listed. On this basis, the complete predecessor time difference can be formed by the difference between the previous and next time fields.

[0021] S112: Based on the predecessor time difference, all WBS nodes with dependency relationships are judged with the planned duration as a reference. Analyze whether the time difference exceeds the planned duration. Filter all dependency paths that meet the conditions, record the corresponding dependency direction, node pair, and offset, and obtain the abnormal dependency marking result. Based on the numerical result of the predecessor time difference, an association operation is performed with the planned duration field to determine whether the node start offset is within the normal plan window. The absolute value of the time difference is subtracted from the corresponding planned duration and added to the logical dependency quantization value. If the result is less than zero, it is considered that there is an abnormal dependency path. For example, the absolute time difference between WBS-001 and PRE-001 is 5 days, the planned duration is 6 days, and the logical dependency value is 2. The calculated value is 5 minus 6 plus 2, which is 1, and the result is normal; while the time difference between WBS-004 and PRE-001 is -5 days, the absolute value is 5, which is 5 with a planned duration of 8 days. The difference is -3, and adding the logical dependency value 2 gives -1, which is a negative value and is marked as an abnormal dependency path. For example, in WBS-003, 5 minus 5 plus 3 equals 3, which is a positive value and a non-abnormal path. This type of calculation does not rely on external models, but instead uses the node's own time, duration, and dependency quantification values ​​to achieve result judgment, which facilitates embedded deployment in the actual project management process. By setting threshold judgment logic, path deviation anomalies can be quickly identified in batches, and abnormal identification fields can be uniformly added to the result fields to facilitate subsequent screening and processing. Structural data items indicating whether each dependency relationship is abnormal are obtained, and the abnormal dependency marking results are summarized and generated.

[0022] S113: Based on the abnormal dependency marking results, all dependency paths whose deviation exceeds the planned duration are screened, the corresponding WBS node combinations and dependency types are extracted, and a list of abnormal dependency relationships is established; According to the path information of all paths marked as abnormal in the abnormal dependency marking result field, the WBS node and predecessor node combination involved in the abnormal path is extracted and used as the unique identifier of the dependent path to build a list table. At the same time, the time difference, logical dependency type and planned duration fields of the path are summarized and sorted by node number, and repeated path combinations are removed. The same predecessor points to multiple successor nodes but has abnormal paths, and structured data records are constructed. The structured list format facilitates subsequent graphical highlighting, clustering statistics or archiving on the platform interface, provides data support for project progress management, and finally establishes an abnormal dependency list.

[0023] See also Figure 3 , the specific steps of S2 are: S211: Read the abnormal path in the abnormal dependency list, retrieve the corresponding abnormal path end node, locate all downstream connection relationships of the node, record all path numbers indirectly associated with the abnormal path, and generate an abnormal diffusion path number set; Based on each abnormal path listed in the abnormal dependency list, first extract the path number and match all the path end node fields in the network graph structure to form an initial abnormal end node set. For example, the abnormal path numbers are P02 and P04, and the corresponding end nodes are N02 and N04 respectively. After identifying such end nodes, retrieve the outgoing edge field of each node in the network graph structure, build an outgoing edge connection table, and then determine whether each outgoing edge connection is a downstream derivative path of the marked abnormal path. For example, if the N02 node connects N03 and N05, query its upstream path number through path tracing to see if it is included in the abnormal path number set. For example, if the upstream path of N02 is P02, its downstream path and N02 form a conduction link, requiring recursive judgment processing. Downstream path search and path source judgment are performed again for N03 and N05. The marking, path appending, and downstream identification operations are repeated. The current recursive tracking results are stored in a path number set structure to avoid infinite loops caused by repeated path entry. The maximum recursive depth is set to 10 to limit the number of path expansion layers. After the complete path recursion, the set of path numbers indirectly affected by the abnormal path conduction relationship is obtained. This set is used to complete the collection and identification of the downstream impact area and form the abnormal diffusion path number set.

[0024] S212: Based on the abnormal diffusion path number set, all development node number information is counted and the logical connection relationship between each node is matched using the formula: ; Calculate node connection culling strength value , if the node connection elimination strength value is lower than the judgment threshold, then the elimination is performed and the logical connection update matrix is ​​obtained, where, and Represents nodes respectively and nodes In the The logical position value in the path, Number the path The corresponding dependency ratio factor, is the total number of outgoing edges of the current node, For the The number of days of dependency delay of the connection edge, For the The logical connection type coefficient of the path; Based on all the path numbers extracted from the abnormal diffusion path number set, the corresponding connection nodes in the network structure are screened. For example, the abnormal numbers P02 and P04 correspond to nodes N02 and N04. Based on this, node pair structures such as (N02, N03) and (N04, N05) are constructed. The logical position value, dependency ratio factor, connection edge delay days, and logical connection type coefficient of the node pair in each path are read. According to the execution rules, the connection removal strength value is calculated for each group of node pairs and the removal operation is determined accordingly. The following is an example of the relevant node connection parameters: Table 2 Node connection relationship parameters Node number Path number Logical position value Dependency ratio factor Connection edge delay days Connection type coefficient N02 P02 3 0.6 3 2 N03 P02 2 0.6 3 2 Taking the node pair (N02, N03) as an example, the culling strength value is calculated using the following formula: ; Substitute each parameter into the position difference: , the dependency ratio factor is 0.6, the number of delay days is 3, and the connection type coefficient is 2. Substituting them into the equation, we get: ; The result is less than the judgment threshold of 3.5 (based on the empirical median of the average logical position difference of nodes in the logical structure multiplied by the dependency ratio factor, which is approximately 1.5 to 2.0. The average delay days on connection edges generally range from 2 to 4 days in project scheduling, and the connection type coefficient is mostly set between 1 and 3. Combined with these typical scenario parameter values, structural sensitivity analysis shows that when the calculated removal strength value is less than 3.5, most connection paths exhibit high delay ratios and weak dependency strength, making them logically feasible for removal. Removal will not cause a sudden change in the critical path structure of the entire graph. Therefore, the threshold of 3.5 is highly stable and has horizontal versatility. As the number of path delay days increases or the connection type coefficient increases, the threshold is more frequently reached, showing a trend directly linked to the risk of dependency lag). Therefore, this path meets the removal criteria, and the corresponding connection edge is removed from the network graph structure. At the same time, the upstream dependency field of the edge's endpoint node N03 is set to 0, and the node connection status is refreshed. After the update action is completed, all removed edge information is recorded in the structural update matrix, ultimately resulting in the logical connection update matrix.

[0025] The node connection removal strength value quantitatively assesses the rationality of retaining the logical connection between two nodes in the project network diagram. A larger value indicates a connection with stronger logical dependencies and timing constraints within the path structure, and a higher risk of removal. Conversely, a smaller value indicates a loose connection within the logical structure or significant impact from anomalous paths, making it a reasonable basis for removal. This value comprehensively considers the node's relative position within the path, the importance of path dependencies, the time delay along the connection, and the strength of the logical connection type. Through a series of weighting, root extraction, and normalization calculations, it transforms the structural connection relationship into a comparable quantitative metric. This serves as the core criterion for determining whether a connection is a low-impact path, can be replaced, or needs to be disconnected during removal. When this value falls below a preset threshold (e.g., 3.5), the system deems the connection eligible for removal and deletes it during the logical structure update.

[0026] The operation logic of this formula is to comprehensively evaluate the rationality of eliminating the connection relationship between two nodes. The addition, subtraction, multiplication, division and square root operations between parameters are set based on the actual engineering meaning of path structure and dependency characteristics. First, the left part It represents the difference in the logical positions of two nodes in the path multiplied by the proportion factor of the path in the overall dependency. The larger the difference in logical positions, the farther the distance between the nodes and the weaker the correlation. Multiplying by the proportion factor is used to reflect the importance of the path in all dependent paths. That is, if the proportion is high, it is not recommended to remove it even if the logical difference is small. The right part The overall dependency strength factor is the square of the number of days of delay for all connected edges multiplied by the weighted square sum of the logical dependency type coefficients, which are then averaged and squared. The square operation enhances sensitivity to large delays, and the square root operation compresses excessively high outliers, playing a balanced evaluation role. This item ultimately reflects the composite expression of the path delay and dependency strength carried by node connections in the network. Overall, it adopts a difference structure, that is, left minus right. This means that when the product of the logical difference of the node pair and the path proportion cannot offset the connection dependency strength, the formula calculation result is a negative value, indicating that the node connection is of weak significance and can be reasonably eliminated.

[0027] S213: Extract all node numbers and corresponding outgoing edge information based on the logical connection update matrix, reconstruct and update the structure, freeze the structure state at the current time node, retain the adjusted global network connection information, mark the version number and connect the time tag, and establish a baseline plan snapshot; According to the information of deleted and retained edge pairs summarized in the logical connection update matrix, the latest outgoing edge structure list of each node is first extracted, and the updated network structure diagram is constructed in combination with the current path number and node number pair. Then, the current operation time node of the system is used as the snapshot time label. For example, on May 28, 2024, a one-time data freeze processing is performed on the network graph structure, and all node structures and connection edge structures are generated according to the latest status. A unique version record is generated in the snapshot record, which includes the structure version number, adjusted edge set, and node set information. At the same time, a status identifier is set in the data structure to form a traceable structure version index record. This version structure serves as a reference map for subsequent scheduling, changes, and analysis, forming a structural static record, and finally establishing a baseline plan snapshot.

[0028] See also Figure 4 , the specific steps of S3 are: S311: Based on the baseline plan snapshot, extract the completion flag field of all nodes, set the identifier of all unfinished nodes to 1, and set the identifier of completed nodes to 0. Read the corresponding incoming edge connection path list for the unfinished nodes, count the number of predecessor nodes of each node, and construct a number-predecessor number mapping matrix to obtain the predecessor relationship quantity table; Obtain all node information recorded in the baseline plan snapshot, extract the completion status field of each node, filter out nodes marked as "unfinished", set such nodes as objects to be scheduled, and read and count all their incoming edge connections in the network structure. The statistical process reads the numbers of all paths whose end points are connected to the node, and de-duplicates the numbers of the source nodes of the paths, which are regarded as the number of predecessor nodes of the node. In this way, a mapping table of unique numbers and predecessor numbers is established for each unfinished node. In the actual example, if the WBS-01 node is connected to one predecessor node WBS-00, its predecessor number is 1. If WBS-02 is connected to three upstream nodes WBS-00, WBS-01, and WBS-03, its predecessor number is 3. After processing all unfinished nodes one by one, a structured statistical table is formed as the basic data for topological sorting, and a table of predecessor relationships is established.

[0029] S312: Based on the table of the number of immediate predecessors, all nodes with a immediate predecessor value of 0 are identified as topological starting points. All edge information in the network graph is scanned in sequence. After each node is processed, the immediate predecessor value of its subsequent node is decremented. If the immediate predecessor value of the subsequent node is reduced to 0, the node is pushed into the topological hierarchy linked list. This process is repeated until all nodes are sorted. The topological sorting process is executed to generate a topological hierarchy order result. According to the mapping value between the node number and the predecessor number in the predecessor relationship table, all nodes with a predecessor number of 0 are written into the sorted linked list as the initial level of the topology. The queue structure is used to maintain the scheduling process. The initial node is placed at the head of the queue and then the outgoing edge is scanned. Every time a downstream node is found, the predecessor number of the node is reduced by 1. If the predecessor number of a node is 0 after decrementing, it is pushed to the end of the queue and the status is updated. The process continues until the queue is empty, forming a topological hierarchical arrangement sequence. For example, the predecessor number of WBS-01 is 0, and it is connected to WBS-03 and WBS-04. After their predecessor values ​​are reduced by 1 and are 0, they are immediately included in the sorting. The following is the raw statistical information of some nodes: Table 3 Node topology relationship statistics Node number Number of immediate predecessor nodes Scheduling times 1 topological location Scheduling times 2 topological location WBS-01 1 1 2 WBS-02 3 2 1 WBS-03 2 3 3 WBS-04 4 4 5 WBS-05 2 5 4 As shown in Table 3, since WBS-01 has 1 immediate predecessor node, if its only predecessor node is a completed node, it can be used as the topological starting point and ranked in the first layer. It is then propagated through this node to nodes such as WBS-03 and WBS-04, completing the overall topological level calculation and establishing the topological level order result.

[0030] S313: Based on the topological hierarchy order, extract the corresponding WBS coding information for each node, split the WBS number field, and identify the WBS hierarchy depth value of each node. For example, split the number "1.2.3.1" into 4 layers, prioritize the topological hierarchy, and then sort the nodes in the same layer in ascending order of WBS depth using the formula: ; Calculating management priority values , arrange the node numbers from high to low according to the numerical value, and establish a management priority list, where, For the The node in The topological position number in the secondary schedule, For the The node in The square value of the WBS branch depth corresponding to the scheduling, is the WBS level depth, For the The number of immediate predecessors of a node, |F-2|, represents whether it is a critical node. If the number of immediate predecessors of a node is 2, the value is 0, reflecting its priority importance; Extract the WBS number field from each node and calculate its hierarchical depth according to the order obtained from the topological hierarchy order. For example, if the WBS-03 number is "1.2.3.1", its hierarchical depth is 4 layers. Then, call the topological position of the node in multiple scheduling processes and the square value of the WBS branch depth to perform comprehensive priority calculation. Taking the WBS-03 node as an example, its topological position in the two scheduling processes is 3 and 3, and the corresponding WBS branch depth square value is 16 and 16. The number of immediate predecessor nodes is 2, and the WBS hierarchical depth is 4. Substitute it into the formula: ; Continue to calculate WBS-04, the topological positions are 4 and 5, the depth squares are 25 and 36, the immediate predecessor is 4, and the level depth is 5. Substitute into the calculation: The summary is as follows: Table 4 Management priority calculation table Node number Topological location 1 Topological location 2 Branch depth squared 1 Branch depth squared 2 WBS level depth Number of immediate predecessor nodes Priority value WBS-01 1 2 4 9 2 1 3.75 WBS-02 2 1 9 4 3 3 2.833 WBS-03 3 3 16 16 4 2 3.5 WBS-04 4 5 25 36 5 4 2.857 WBS-05 5 4 36 25 3 2 3.917 As shown in Table 4, node WBS-05 is ranked first in the priority order due to its relatively low topological position but large WBS branch depth and high total value. Based on this ranking, a management priority list is established. This result indicates that nodes with higher priority values ​​are processed earlier in management, and their ranking order is determined by the node's topological position, WBS structure depth, and dependencies.

[0031] The management priority value is a comprehensive indicator that measures the degree to which each unfinished development node in the project network diagram needs to be given priority attention and processing in the current scheduling cycle. This value integrates key parameters of multiple dimensions such as the node's ranking position in the topological structure, the depth of the WBS structure hierarchy, the complexity of branches, and the number of immediate dependencies. It reflects the management sensitivity and scheduling urgency of the node in the entire project progress. The higher the value, the more likely the node is to be both front-end and have a complex structure or key dependencies, which can easily form a bottleneck for subsequent tasks. Therefore, it should be prioritized for scheduling and monitoring in resource allocation and schedule arrangement. This indicator can be used to quickly extract key nodes in a complex project environment with many nodes and intertwined dependencies, and assist in formulating a phased advancement strategy.

[0032] The operational logic of this formula is to integrate information from multiple dimensions to characterize the scheduling priority of the node. The numerator of the formula uses the topological position value in each scheduling. The corresponding WBS branch depth square value The square root of the nodes is added, and then the results of all scheduling batches are summed up to reflect the overall pre-position and structural complexity of the node under multiple scheduling perspectives. The topological position value reflects the front degree of the node in the dependency chain. The smaller the value, the closer it is to the starting point and the higher the timeliness. The square value of the branch depth reflects the complexity of the substructure under the node in the WBS structure. The square root of the square root is used to maintain the order of magnitude of the original value while weakening the impact of abnormal value fluctuations to avoid excessive dominance of structural deviations in sorting. The denominator is composed of the hierarchical depth of the node. and the number of immediately preceding nodes It is composed of the sum of the absolute values ​​of the deviations from the reference value of 2. The deeper the hierarchical depth, the higher the degree of detail of the task, and its priority should be appropriately lowered. The deviation value of the immediate predecessor compared with the conventional two predecessors can be used to identify the critical path concentration nodes. The greater the deviation, the lower the stability is considered and the priority attention should be paid. The overall logic increases the weight in the numerator through "predecessor topology + structural complexity" and controls the total value in the denominator through "structural hierarchy + dependency risk", realizing the comprehensive ranking of structure, sequence and dependency, and finally outputting a quantifiable management priority value.

[0033] See also Figure 5 , the specific steps of S4 are: S411: Based on the unfinished nodes in the management priority list, extract the latest planned completion time of each node, synchronously obtain the remaining available time from the current time node to the project deadline, record the time difference, establish a mapping record between the node number and the overdue status, and generate a node overdue identification matrix; According to the unfinished nodes that have been filtered out in the management priority list, the latest planned completion time set for each node in the current scheduling cycle is extracted, and the remaining available time between the current time and the project termination node in the project calendar is called as the comparison benchmark. In the operation, the units of both need to be unified as "days" for processing. Then, the latest planned completion time is subtracted from the remaining available time through node-by-node difference calculation. If the difference is greater than 0, it means that the remaining time of the current node is not enough to cover its expected completion time limit. The node needs to be marked as having a time limit conflict, otherwise it is marked as having no time limit conflict. Risk nodes, for example, the latest completion time of node A is 110 days, and the corresponding remaining available construction period is 100 days, with a difference of 10 days, which means it is overdue. The latest completion time of node B is 95 days, and the remaining duration is 100 days, with a difference of -5 days, which means it is not overdue. The comparison results of each node need to form a structured record, and a mapping structure between the node number and the overdue status field is constructed in the form of key-value pairs, and a duration difference field is attached for verification. After the operation is completed, the node number is summarized and a unified list is output for subsequent sorting priority adjustment operations. This process finally establishes a node overdue identification matrix.

[0034] S412: Based on the node numbers of all nodes marked as having time limit conflicts in the node expiration identification matrix, the original priority values ​​in the management priority list are updated, while retaining the original order of the nodes that have not expired. The updated priority information is paired with the node numbers to generate a priority adjustment value table; According to all the nodes marked as having time limit conflicts in the node overdue identification matrix, their corresponding priority values ​​in the original priority list are further read to determine the time difference on which the adjustment range is based, and the priority is numerically updated. In this implementation scenario, each day of overdue is regarded as the basis for the priority to increase by one level, and the difference field is directly added to the original priority value as a weighted item for cumulative update. For example, the original priority of node A is 4, and it exceeds 10 days, so it is adjusted to 4+10=14. The original priority of node C is 2, and it is still 2 if it has not exceeded the deadline. After completing the priority adjustment of all nodes, the updated priority values ​​are paired with the node numbers one by one to form a priority adjustment mapping set, and the values ​​of the original non-expired nodes are kept unchanged. The node unified list structure is updated to the mapping data format, and the output is used as the data source for subsequent sorting to finally generate a priority adjustment value table.

[0035] S413: Reorder all unfinished nodes according to the priority adjustment value table, use the adjusted priorities as the main sequence basis, and combine the original dependency structure and deadline conflicts to establish a collaborative task management priority sequence; The adjusted priority values ​​of all nodes in the priority adjustment value table are used as sorting reference items to rearrange the order of all nodes. During the sorting process, nodes with higher priority values ​​will be ranked in the front position. While performing the sorting operation, it is also necessary to review the link relationship of the immediate predecessor node, and evaluate the number of outgoing connection paths of the node in the topological network graph and the possible time delay risk value on the path. Then, multiple indicators are integrated to perform readjustment operations on the basis of priority sorting. By introducing evaluation items such as the compression amplitude of the immediate predecessor node and the task path delay rate, auxiliary fine-tuning processing is performed on the sorting results to determine whether certain nodes with concentrated immediate predecessor nodes and large total path delay meet the conditions for ranking increase. If so, their positions are adjusted; if not, the original adjustment value positions are maintained. The order reordering operation of all nodes is completed with the overall priority as the leading indicator and the path status as the adjustment factor, and the output is a unified structured sequence to obtain the collaborative task management priority sequence.

[0036] See also Figure 6 , the specific steps of S5 are: S511: Based on the collaborative task management priority sequence, read the entity status of each development node in the current knowledge graph in sequence, extract the dependency structure and priority information of adjacent nodes, update the priority number and path frequency field of the dependent node, and generate priority dependency synchronization graph data; Based on the collaborative task management priority sequence, for each development node in the knowledge graph, its status field and connection information are read. First, the node connected to the immediate predecessor is identified, and the priority value of the immediate predecessor node is read with the priority value of the current node, and a numerical comparison is performed. If the priority number of the current node is 4 and the priority of the immediate predecessor node is 2, it means that there is a priority inversion structure, and it is necessary to enter the next step of dependency strength judgment. The dependency strength parameter on the path is extracted. The dependency strength is 0.81. At the same time, the set scheduling benchmark value is 0.75. The path dependency strength is higher than the set benchmark, so the dependency adjustment operation is performed to reverse the original path direction. The operation is recorded in the path type field as "forced dependency" and marked as "forced dependency". The path frequency field records the number of visits as 3. This field will be used in subsequent path frequency normalization. The frequency value normalization is processed using the minimum-maximum method. If the minimum value of all path frequencies in the dataset is 1 and the maximum value is 6, then the normalized frequency of the current path is (3-1) / (6-1)=0.4. After the dependency adjustment is completed, the priority number fields of the two nodes are updated. The original immediate predecessor node number is updated from 2 to 3, and the current node number remains 4. The path direction field is also updated. If the number of path reversals is greater than 1, a "repeated reversal" flag is added to the path attribute. In addition, the statistical results of such processed paths are summarized to form the dependency state distribution data in the atlas. The relevant structure example is shown in the following table: Table 5 Dependency adjustment example table Node number Immediately preceding node number Current priority Immediate precedence priority Dependency strength Reversal Path frequency Path normalization value Type Identifier D-01 D-03 4 2 0.81 yes 3 0.4 Mandatory dependencies D-02 D-01 5 4 0.68 no 2 0.2 Normal Dependence As shown in Table 5, in the connection path established between node D-01 and its immediate predecessor node D-03, because its priority is inverted and the dependency strength reaches 0.81, exceeding the benchmark 0.75, the system reverses its dependency direction and synchronously updates the graph structure field, ultimately generating priority dependency synchronization graph data.

[0037] S512: Based on all node attributes in the priority-dependent synchronization graph data, a time series structure indexed by node priority is constructed, each node is divided into a corresponding management period, the distribution range and central trend within the same period are determined, and a vertical arrangement rule is established based on the priority order to obtain a process layout coordinate matrix; According to the priority dependency synchronization graph data, the priority field value of each node is extracted and inserted into the time series structure indexed by the priority number as the primary key. Assuming that the development task contains a total of 12 nodes, and their priority numbers range from 1 to 6, it is divided into 3 cycles with every 2 priority numbers as a cycle segment. Cycle one is numbered 6 to 5, cycle two is numbered 4 to 3, and cycle three is numbered 2 to 1. The five nodes are numbered A, B, C, D, and E and classified according to their priority numbers, such as A=6, B=5, C=3, D=2, and E=1, belonging to cycles one, two, and three respectively. After classifying the nodes in each cycle according to the cycle they belong to, the number of immediately connected paths for each node is read, such as node A is connected to 3 paths, B is connected to 2 paths, C is connected to 4 paths, D is connected to 1 path, and E is connected to 2 paths. At the same time, the activity information of these paths is extracted , with paths A1=12, A2=8, and A3=6, respectively. The total activity of node A is 26. The total activity of nodes A and B in period one is summed to 26+15=41, and the average number of connection paths is taken as (3+2) / 2=2.5. The deviation of the number of connection paths for each node in period one is calculated, which are |3−2.5|=0.5 and |2−2.5|=0.5, respectively, with an average deviation of 0.5. The vertical coordinate is established based on the node priority, where the vertical coordinate of node A is 1 and that of node B is 2. At the same time, the horizontal path density balance value of 2.5 is mapped to a unit spacing of 20 pixels. The horizontal spacing between nodes A and B is 20 pixels, and the vertical position difference is 1 unit. This is used to construct the spatial layout points (X=20, Y=20) and (X=40, Y=40), finally forming a two-dimensional coordinate matrix set and obtaining the process layout coordinate matrix.

[0038] S513: Based on the node layout information recorded in the process layout coordinate matrix, the node graphic border style, connection path line type, and field display method are selected, and a visual task card is output for each development node to establish a development process management view model; According to the node coordinate information in the process layout coordinate matrix, configure visual task cards for nodes A, B, and C respectively, extract field numbers such as A-01, B-02, and C-03, set the priority levels to 6, 5, and 3, set the dependent path number fields to 3, 2, and 4, and use a standard style for the border of each task card. Nodes with priority numbers greater than or equal to 5 are set to rounded rectangles, and nodes with numbers less than 5 are right-angled rectangles. The dependent path number is mapped to the length of the bar chart at the bottom of the card. Set the scale to 8 pixels per path unit, then the bar chart length of card A is 24 pixels, card B is 16 pixels, and card C is 32 pixels. Set the field number to Set the font size to bold 16, and draw the dependent path arrow line from coordinates A→B→C, where A to B is connected by a solid line, and B to C is connected by a dotted line. The corresponding annotation path type fields are "Normal" and "Delayed", and the path activity is mapped to the line transparency. The frequency values ​​are extracted from the dependent path attributes, which are A→B=5 and B→C=3 respectively. The transparency range is set to 0.3~0.9, 5 is mapped to 0.9, and 3 is mapped to 0.6. After completing the path line drawing and node position configuration, all cards are arranged according to the coordinate set to complete the layout output, and are uniformly encapsulated as SVG format development view files, and finally a development process management view model is established.

[0039] The above are merely preferred embodiments of the present invention and do not limit the present invention in any other form. Any technician familiar with the profession may use the technical content disclosed above to change or modify it into an equivalent embodiment with equivalent changes and apply it to other fields. However, any simple modification, equivalent change and modification made to the above embodiment based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall still fall within the scope of protection of the technical solution of the present invention.

Claims

1. The development process management method based on knowledge graph is characterized by: The following steps are involved: S1: Obtain the actual start time of the WBS node in the knowledge graph, detect the earliest start time of the predecessor node, determine whether there is a dependency with a negative total float time, mark it as an abnormal path, and generate a list of abnormal dependencies; S2: Based on the abnormal dependency list, recursively remove all downstream development node dependencies of the abnormal path, release the abnormal connection of the marked development node, update the logical relationship of the project network diagram, and generate a baseline plan snapshot; S3: Based on the baseline plan snapshot, count the number of predecessors of the unfinished nodes, trace back to the starting development node, sort the development nodes in order according to the WBS level depth, and generate a management priority list; S4: Extract the latest completion time of each unfinished node based on the management priority list, determine whether there is a development node whose completion time exceeds the remaining available time, adjust the priority of the corresponding development node, and generate a collaborative task management priority sequence; S5: Based on the collaborative task management priority sequence, the dependency and priority status of all development nodes in the knowledge graph are uniformly updated, the development process management view is rebuilt, and visual output is performed to generate a development process management view model.

2. The development process management method based on knowledge graph according to claim 1 is characterized in that: The abnormal dependency list includes negative floating dependency markers, critical path conflict identifiers, and actual start time deviation nodes. The baseline plan snapshot includes an updated logical network diagram, a set of nodes with abnormal path removed, and revised status information of development nodes. The management priority list includes a development node sequence table generated according to the WBS hierarchy and topological relationship, a predecessor quantity indicator for each node, and a priority sorting index. The collaborative task management priority sequence includes the identification results of time-sensitive nodes, priority adjustment suggestions, and remaining duration matching status. The development process management view model includes a node status diagram, a task priority mapping diagram, and an information display structure that complies with the PMBOK6th standard.

3. The development process management method based on knowledge graph according to claim 1 is characterized in that: The specific steps for obtaining the abnormal dependency list are: S111: Obtain the actual start time of each WBS node in the knowledge graph and the configured predecessor node, detect the earliest start time of each predecessor node, and compare it with the actual start time of the corresponding WBS node, calculate the time difference between the two, and generate the predecessor time difference; S112: Based on the predecessor time difference, a difference judgment is performed on all WBS nodes with dependency relationships, with the planned duration as a reference, to analyze whether any time difference exceeds the planned duration, and all dependency paths that meet the conditions are screened. The corresponding dependency directions, node pairs, and offset magnitudes are recorded to obtain abnormal dependency marking results. S113: Based on the abnormal dependency marking result, all dependency paths whose deviation exceeds the planned duration are screened, corresponding WBS node combinations and dependency types are extracted, and an abnormal dependency list is established.

4. The development process management method based on knowledge graph according to claim 1 is characterized in that: The specific steps for obtaining the benchmark plan snapshot are: S211: Read the abnormal path in the abnormal dependency list, retrieve the corresponding abnormal path end node, locate all downstream connection relationships of the node, record all path numbers indirectly associated with the abnormal path, and generate an abnormal diffusion path number set; S212: Based on the abnormal diffusion path number set, all development node number information is counted, the logical connection relationship between each node is matched, and the node connection elimination strength value is calculated. If the node connection elimination strength value is lower than the judgment threshold, the node connection elimination is performed to obtain a logical connection update matrix; S213: Extract all node numbers and corresponding outgoing edge information according to the logical connection update matrix, reconstruct and update the structure, freeze the structure state according to the current time node, retain the adjusted global network connection information, mark the version number and connect the time tag, and establish a baseline plan snapshot.

5. The development process management method based on knowledge graph according to claim 1 is characterized in that: The specific steps for obtaining the management priority list are: S311: Based on the baseline plan snapshot, the completion flag fields of all nodes are extracted, the corresponding incoming edge connection path list of the unfinished nodes is read, the number of predecessor nodes of each node is counted, and a number-predecessor number mapping matrix is ​​constructed to obtain a predecessor relationship quantity table; S312: Based on the table of the number of immediate predecessors, all nodes with a number of immediate predecessors of zero are identified as topological starting points, all edge information in the network graph is sequentially scanned, a topological sorting process is performed, and a topological hierarchical order result is generated; S313: According to the topological hierarchy order result, extract the corresponding WBS coding information for each node, split the WBS number field, identify the WBS hierarchy depth value of each node, calculate the management priority value, arrange the node numbers from high to low according to the numerical value, and establish a management priority list.

6. The development process management method based on knowledge graph according to claim 1 is characterized in that: The specific steps for obtaining the collaborative task management priority sequence are: S411: Extract the latest planned completion time of each node from the unfinished nodes in the management priority list, synchronously obtain the remaining available time from the current time node to the project deadline, record the time difference, establish a mapping record between the node number and the overdue status, and generate a node overdue identification matrix; S412: Based on the numbers of all nodes marked as having time limit conflicts in the node overdue identification matrix, the original priority values ​​in the management priority list are updated, the original order of the nodes that have not expired is retained, and the updated priority information is paired with the node numbers to generate a priority adjustment value table; S413: Reorder all unfinished nodes according to the priority adjustment value table, use the adjusted priorities as the main sequence basis, and establish a collaborative task management priority sequence in combination with the original dependency structure and time limit conflict.

7. The development process management method based on knowledge graph according to claim 1 is characterized in that: The specific steps for obtaining the development process management view model are: S511: Based on the collaborative task management priority sequence, sequentially read the entity status of each development node in the current knowledge graph, extract the dependency structure and priority information of adjacent nodes, update the priority number and path frequency field of the dependent node, and generate priority dependency synchronization graph data; S512: Based on all node attributes in the priority-dependent synchronization graph data, a time series structure indexed by node priority is constructed, each node is divided into a corresponding management period, the distribution range and central tendency within the same period are determined, and a vertical arrangement rule is established in combination with the priority order to obtain a process layout coordinate matrix; S513: According to the node layout information recorded in the process layout coordinate matrix, the node graphic border style, the connection path line type and the field display method are selected, a visual task card is output for each development node, and a development process management view model is established.

Images