Enterprise personnel task intelligent matching system based on inference model

Abstract

The application discloses an enterprise personnel task intelligent matching system based on an inference model, comprising: a data fusion module for acquiring and fusing historical task data and personnel execution data of an enterprise; a cause-effect chain construction module for constructing a historical task cause-effect chain; a fracture identification module for generating a historical cause-effect fracture point set; an ability portrait construction module for constructing a personnel fracture repair ability portrait; a task analysis module for analyzing task semantics and task constraints of to-be-assigned task data; a fracture inference module for generating a current task potential fracture point set through a cause-effect fracture inference model; a matching calculation module for generating a fracture matching result; and a task allocation module for determining a recommended execution object capable of blocking a potential fracture point propagation path and outputting a task allocation result. The application utilizes a cause-effect fracture inference method to realize task intelligent matching, and has the advantages of strong risk prediction capability, high matching precision and good execution stability.

Description

Technical Field

[0001] This invention relates to the field of task matching, and in particular to an intelligent task matching system for enterprise personnel based on a reasoning model. Background Technology

[0002] As enterprise tasks become more complex and collaborative, existing task allocation methods largely rely on human experience or simple rule-based matching, assigning tasks based on personnel skill tags, historical performance, or job attributes. Some systems introduce data analysis or recommendation algorithms to statically match task requirements with personnel capabilities, but overall, these methods still primarily rely on statistical results or superficial feature correlations, lacking systematic modeling of the task execution process, abnormal evolution paths, and multi-node collaborative relationships, making it difficult to reflect the dynamic risk transmission characteristics during task execution.

[0003] Existing technologies struggle to depict the causal transmission of anomalies from upstream nodes to downstream result nodes during task execution. They are unable to identify critical breakpoints and their impact on the final result, and lack structured modeling of the actual risk mitigation capabilities of enterprise personnel during anomaly handling. This results in the inability to predict potential break risks in advance and match execution objects with targeted handling capabilities during task allocation, easily leading to problems such as task delays, increased rework, or decreased quality. Overall matching accuracy and task execution stability need to be improved.

[0004] Therefore, how to provide an intelligent matching system for enterprise personnel tasks based on reasoning models is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0005] One objective of this invention is to propose an intelligent task matching system for enterprise personnel based on a reasoning model. This invention utilizes a causal break reasoning method to achieve intelligent task matching, and has the advantages of strong risk prediction capability, high matching accuracy, and good execution stability.

[0006] An intelligent task matching system for enterprise personnel based on a reasoning model, according to an embodiment of the present invention, includes: The data fusion module is used to acquire and merge historical task data and personnel execution data of the enterprise to generate a historical task fusion dataset. The causal chain building module is used to extract historical task execution nodes and the flow relationships between nodes based on the historical task fusion dataset, and to build historical task causal chains. The fracture identification module is used to perform abnormal cause reasoning on the abnormal influence path between abnormal nodes and result nodes based on the historical task causal chain, identify the causal fracture points that lead to abnormal results of historical tasks and generate a set of historical causal fracture points. The capability profile building module is used to trace back the abnormal handling behaviors and results of enterprise personnel in corresponding historical tasks based on the set of historical causal breakpoints, and to build a personnel breakpoint repair capability profile. The task parsing module is used to receive task data to be assigned, perform task semantic parsing and task constraint parsing on the task data to be assigned, and generate the current task feature set; The fracture reasoning module is used to input the causal chain of historical tasks and the feature set of the current task into the causal fracture reasoning model to perform potential fracture reasoning and generate a set of potential fracture points for the current task. The matching calculation module is used to calculate the risk blocking matching degree of candidate enterprise personnel based on the current set of potential breakpoints and the personnel break repair capability profile, and generate break matching results based on the risk blocking matching degree. The task allocation module is used to determine the recommended execution objects that can block the propagation path of potential breakpoints based on the breakpoint matching results, and output the task allocation results.

[0007] Optionally, the historical task data includes historical task basic data, historical task execution process data, and historical task exception data. The personnel execution data includes personnel identification, historical participation in tasks, task execution roles, exception handling records, collaboration records, communication records, delivery results, and task repair records. The fusion processing includes data cleaning, time alignment, entity normalization, and node relationship mapping.

[0008] Optionally, the causal chain construction module includes: Based on the historical task fusion dataset, historical task execution nodes are identified and divided into task nodes, personnel nodes, collaboration nodes, approval nodes, delivery nodes, exception nodes, and result nodes according to their source and function. Based on the occurrence time, execution order, and business flow relationship of historical task execution nodes in the historical task fusion dataset, extract the flow relationship between nodes; Based on the flow relationship between nodes, a directed connection relationship is established between historical task execution nodes. The historical task execution node located at the forefront of the task advancement direction is taken as the source node, and the target historical task execution node determined based on the occurrence time, execution order and business flow relationship is taken as the target node, forming a point flow structure. Based on the positions of abnormal nodes and result nodes in the node flow structure, the propagation path between abnormal nodes and result nodes is marked to obtain the abnormal result propagation path. Based on the node flow structure and abnormal result propagation path, the flow relationship between nodes is labeled with causal association, and a historical task causal chain is constructed.

[0009] Optionally, the fracture detection module includes: Based on the historical task causal chain, abnormal nodes and result nodes are located, and path tracing is performed along the causal relationship between abnormal nodes and result nodes to generate a set of abnormal impact paths. Based on the set of abnormal impact paths, the upstream execution nodes in each abnormal impact path are determined, and the time distance of abnormal occurrence, the strength of causal relationship, the number of abnormal triggers, and the degree of result correlation are extracted for the upstream execution nodes; The node contribution of upstream execution nodes is calculated based on the time distance of anomaly occurrence, the strength of causal relationship, the number of anomaly triggers, and the degree of correlation of results, and a node contribution sequence corresponding to the anomaly impact path is generated. The contribution of upstream execution nodes is determined based on the node contribution sequence. Upstream execution nodes whose node contribution is greater than the average contribution of all upstream execution nodes in the same abnormal impact path are identified as candidate causal breakpoints. Based on the node type, node flow relationship type, and abnormal result type corresponding to the candidate causal breakpoints, the breakpoints are classified into breakpoint types, and causal breakpoints with breakpoint type identifiers are generated. The causal breakpoints with breakpoint type identifiers are merged according to historical task identifiers, abnormal impact paths, and breakpoint type identifiers to generate a set of historical causal breakpoints.

[0010] Optionally, the capability profile building module includes: Based on the historical causal breakpoint set and the historical task fusion dataset, the abnormal handling behavior of enterprise personnel after the occurrence of the causal breakpoint and before the formation of the abnormal result is traced back, and the sequence of enterprise personnel breakpoint handling behavior is generated. Based on the sequence of enterprise personnel fracture handling behaviors and fracture type identifiers in the historical causal fracture point set, establish the correspondence between handling behaviors and fracture types, and generate enterprise personnel fracture repair behavior mapping data; Based on the enterprise personnel fracture repair behavior mapping data, extract the number of treatment behaviors covered by each enterprise personnel under each fracture type, the number of nodes affected by the treatment behaviors, the number of abnormal impact paths associated with the treatment behaviors, and the cumulative results of the contribution of the corresponding nodes of the treatment behaviors, and generate enterprise personnel fracture treatment intensity data. Based on the delivery results, task repair records and historical task anomaly data in the historical task fusion dataset, the delay reduction results, rework reduction results, failure avoidance results and quality recovery results corresponding to the enterprise personnel breakage handling behavior sequence are extracted to generate enterprise personnel breakage handling result data. Based on data on the intensity and outcome of fracture treatment for enterprise personnel, the risk mitigation capability of enterprise personnel under different fracture types is calculated, and a fracture repair capability matrix for enterprise personnel is generated. The personnel fracture repair capability matrix is ​​merged according to the personnel identification of the enterprise, and combined with the task execution role, collaboration record, communication record and task repair record to generate personnel fracture repair capability profile.

[0011] Optionally, the task parsing module includes: Receive task data to be assigned, identify fields such as task requirement text, task type, task complexity, participating departments, collaborating parties, time requirements, resource conditions and delivery goals in the task data to be assigned, and generate task field data to be assigned. Based on the data of the task to be assigned, the task requirement text is parsed to extract the semantics of the task target, task content, task object, task execution, and task delivery, and the semantics of the task requirement are generated. Based on the data in the task to be assigned field, the task constraint is parsed for participating departments, collaborating objects, time requirements, resource conditions, and delivery goals, generating the current task collaboration relationship, current task time constraint, current task resource constraint, and current task delivery requirements; Based on the current task identifier, the current task type, current task complexity, current task requirement semantics, current task participating departments, current task collaboration relationships, current task time constraints, current task resource constraints, and current task delivery requirements are associated and organized to generate a current task feature set.

[0012] Optionally, the fracture inference module includes: Training samples for a time series graph are constructed based on the historical task causal chain and the set of historical causal breakpoints. The training samples of the time series graph are input into the time series graph neural network for training to obtain the causal break inference model; Perform graph structure transformation on the current task feature set to construct the current task execution graph; Input the historical task causal chain and the current task execution graph into the causal break inference model, and perform potential break inference on the task execution flow relationship in the current task execution graph based on the abnormal transmission relationship of the historical task, and generate potential break inference results. Based on the results of potential breakage inference, candidate potential breakpoints in the current task are determined, and the risk level of the candidate potential breakpoints is judged to generate candidate potential breakpoint data. Based on the candidate potential breakpoint data and the current task execution graph, the propagation path of the candidate potential breakpoints to the current task result node is traced to generate the current task potential breakpoint set.

[0013] Optionally, the matching calculation module includes: Based on the profile of personnel fracture repair capabilities, the risk prevention capabilities of personnel in candidate enterprises under different fracture types are extracted, and data on the risk prevention capabilities of personnel in candidate enterprises are generated. Based on the potential breakpoint types in the current task potential breakpoint set, establish a type correspondence with the risk blocking capabilities in the candidate enterprise personnel risk blocking capability data, and generate candidate enterprise personnel breakpoint capability corresponding data. Based on the candidate enterprise personnel's breakpoint capability data, the potential breakpoint risk level and potential breakpoint propagation path in the current task's potential breakpoint set, calculate the risk blocking matching degree of the candidate enterprise personnel for the current task's potential breakpoint set. Candidate company personnel identifiers are sorted according to risk blocking matching degree, and the corresponding data of the personnel's breakpoint capabilities are combined to generate breakpoint matching results.

[0014] Optionally, the task allocation module includes: Based on the breakpoint matching type of the candidate enterprise personnel in the breakpoint matching results, and the potential breakpoint type in the potential breakpoint set of the current task, a coverage match is performed to filter the candidate enterprise personnel that can cover the potential breakpoint type in the potential breakpoint propagation path, and generate candidate data of recommended execution objects. The recommended implementation targets are determined based on the candidate data of the recommended implementation targets and the ranking sequence of the risk prevention capabilities of the personnel of the candidate enterprises. Based on the coverage of the potential breakpoint propagation path by the recommended execution objects, the ranking position of the recommended execution objects in the candidate enterprise personnel risk blocking capability ranking sequence, and the candidate enterprise personnel task adaptation results, the task allocation priority is determined. Based on the breakpoint matching type, risk blocking capability, and potential breakpoint propagation path of the recommended execution object, generate matching reasons and risk blocking prompts; The task allocation result is generated based on the current task identifier, the set of potential breakpoints of the current task, the recommended execution object, the task allocation priority, the matching reason, and the risk blocking prompt.

[0015] The beneficial effects of this invention are: This invention constructs a causal chain of historical tasks by structurally modeling the historical task execution process, including task execution nodes and their relationships. Based on this chain, it identifies the impact paths and causal breakpoints of anomalies, enabling a precise depiction of the transmission process of task anomalies from upstream nodes to downstream result nodes. Compared to existing methods that rely solely on result statistics or static feature matching, this invention reveals the causes and propagation paths of anomalies at the task execution mechanism level. This allows the system to identify potential risk sources earlier when facing complex multi-node collaborative tasks, thereby improving the interpretability and risk prediction capabilities of the task execution process.

[0016] In terms of personnel capability assessment, this invention traces the specific behaviors and corresponding results of enterprise personnel during anomaly handling based on a historical causal breakpoint set, constructing a personnel breakpoint repair capability profile. This profile structurally expresses personnel's repair capabilities, circumvention capabilities, suppression capabilities, and recovery capabilities under different breakpoint types. This capability profile not only reflects personnel's performance but also their actual risk mitigation capabilities in different anomaly scenarios. This allows the system to perform refined matching based on potential breakpoint types in specific tasks, avoiding the matching bias caused by traditional methods that rely solely on experience or single indicators, thereby significantly improving the matching accuracy between tasks and personnel.

[0017] Furthermore, this invention constructs a causal break inference model, jointly modeling the causal chain of historical tasks with the feature set of the current task. It then performs potential break inference on the current task execution process, generating a set of potential breakpoints. Based on the risk level and propagation path of these potential breakpoints, it calculates the risk blocking matching degree of candidate enterprise personnel. On this basis, it determines recommended execution targets and generates task allocation results, transforming the task allocation process from passive response to proactive prevention. This allows for risk identification and capability matching to be completed before task execution. Through this mechanism, the risks of task delays, rework, and failures can be effectively reduced, improving task execution stability and overall efficiency, and enhancing the enterprise's collaborative capabilities and management level in complex task environments. Attached Figure Description

[0018] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a schematic diagram of the structure of an intelligent matching system for enterprise personnel tasks based on a reasoning model proposed in this invention; Figure 2 This is a flowchart illustrating the process of constructing a profile of personnel fracture repair capabilities in an enterprise personnel task intelligent matching system based on a reasoning model, as proposed in this invention. Figure 3 This is a flowchart of causal break reasoning for an intelligent matching system for enterprise personnel tasks based on a reasoning model proposed in this invention. Detailed Implementation

[0019] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0020] refer to Figures 1-3 A task-matching system for enterprise personnel based on a reasoning model, comprising: The data fusion module is used to acquire and merge historical task data and personnel execution data of the enterprise to generate a historical task fusion dataset. The causal chain building module is used to extract historical task execution nodes and the flow relationships between nodes based on the historical task fusion dataset, and to build historical task causal chains. The fracture identification module is used to perform abnormal cause reasoning on the abnormal influence path between abnormal nodes and result nodes based on the historical task causal chain, identify the causal fracture points that lead to abnormal results of historical tasks and generate a set of historical causal fracture points. The capability profile building module is used to trace back the abnormal handling behaviors and results of enterprise personnel in corresponding historical tasks based on the set of historical causal breakpoints, and to build a personnel breakpoint repair capability profile. The task parsing module is used to receive task data to be assigned, perform task semantic parsing and task constraint parsing on the task data to be assigned, and generate the current task feature set; The fracture reasoning module is used to input the causal chain of historical tasks and the feature set of the current task into the causal fracture reasoning model to perform potential fracture reasoning and generate a set of potential fracture points for the current task. The matching calculation module is used to calculate the risk blocking matching degree of candidate enterprise personnel based on the current set of potential breakpoints and the personnel break repair capability profile, and generate break matching results based on the risk blocking matching degree. The task allocation module is used to determine the recommended execution objects that can block the propagation path of potential breakpoints based on the breakpoint matching results, and output the task allocation results.

[0021] In this embodiment, historical task data includes historical task basic data, historical task execution process data, and historical task anomaly data. Personnel execution data includes personnel identification, historical participation in tasks, task execution roles, anomaly handling records, collaboration records, communication records, delivery results, and task repair records. Fusion processing includes data cleaning, time alignment, entity normalization, and node relationship mapping.

[0022] In this embodiment, the causal chain construction module includes: Based on the historical task fusion dataset, historical task execution nodes are identified and divided into task nodes, personnel nodes, collaboration nodes, approval nodes, delivery nodes, exception nodes, and result nodes according to their source and function. Based on the occurrence time, execution order, and business flow relationship of historical task execution nodes in the historical task fusion dataset, the flow relationship between nodes is extracted. The flow relationship between nodes includes task advancement relationship, personnel participation relationship, collaboration and transmission relationship, approval flow relationship, delivery and acceptance relationship, exception triggering relationship and result association relationship. Based on the flow relationship between nodes, a directed connection relationship is established between historical task execution nodes. The historical task execution node located at the forefront of the task advancement direction is taken as the source node, and the target historical task execution node determined based on the occurrence time, execution order and business flow relationship is taken as the target node, forming a node flow structure to represent the historical task execution process. Based on the positions of abnormal nodes and result nodes in the node flow structure, the transmission path between abnormal nodes and result nodes is marked to obtain the abnormal result transmission path. The abnormal result transmission path consists of an abnormal node, an intermediate execution node located between the abnormal node and the result node, and a result node. It is used to represent the node path in which the results of task delays, task rework, task failures, or quality degradation are transmitted along the task execution process in historical tasks. Based on the node flow structure and abnormal result transmission path, the flow relationship between nodes is labeled with causal association, and a historical task causal chain is constructed. The causal association labeling is done by marking the flow relationship between adjacent historical task execution nodes in the abnormal result transmission path as a causal relationship, and marking the flow relationship between nodes not located in the abnormal result transmission path as a non-causal relationship, thus forming a historical task causal chain to represent the relationship between the task execution process and the abnormal result transmission.

[0023] In this embodiment, the fracture detection module includes: Based on the historical task causal chain, abnormal nodes and result nodes are located, and path tracing is performed along the causal relationship between abnormal nodes and result nodes to generate an abnormal impact path set; the abnormal impact path set includes abnormal nodes, result nodes, intermediate execution nodes located between abnormal nodes and result nodes, and corresponding node flow relationships; Based on the set of abnormal impact paths, the upstream execution nodes in each abnormal impact path are determined, and the time distance of abnormal occurrence, the strength of causal relationship, the number of abnormal triggers, and the degree of result correlation are extracted for the upstream execution nodes; The time interval between the occurrence of an anomaly is the time between the occurrence of the event corresponding to the upstream execution node and the occurrence of the event corresponding to the abnormal node; the number of times an anomaly is triggered is the number of times an anomaly record associated with the upstream execution node appears in the historical task anomaly data; The strength of causal relationship is obtained by weighted summation of flow connection frequency, co-occurrence frequency, and sequential consistency frequency. Among them, the flow connection frequency is determined by the ratio of the number of occurrences of node flow relationship between upstream execution node and abnormal node to the total number of occurrences of upstream execution node in the same type of historical task; the co-occurrence frequency is determined by the ratio of the number of times upstream execution node and abnormal node co-occur in the same historical task to the total number of the same type of historical task; and the sequential consistency frequency is determined by the ratio of the number of times upstream execution node occurs earlier than abnormal node occurrence time to the number of times upstream execution node and abnormal node co-occur. The correlation degree of the results is obtained by weighted summation of the frequency of co-occurrence of results, the frequency of consistent result order, and the normalized value of result severity. Among them, the frequency of co-occurrence of results is determined by the ratio of the number of times the upstream execution node and the abnormal result co-occur in the same historical task to the total number of times the upstream execution node occurs; the frequency of consistent result order is determined by the ratio of the number of times the upstream execution node occurs earlier than the time the abnormal result occurs to the number of times the upstream execution node and the abnormal result co-occur; the normalized value of result severity is determined by the ratio of the severity level of the abnormal result to the preset highest severity level. The node contribution of upstream execution nodes is calculated based on the time distance of anomaly occurrence, the strength of causal relationship, the number of anomaly triggers, and the degree of result correlation, generating a node contribution sequence corresponding to the anomaly impact path. The node contribution is obtained by multiplying the time impact value corresponding to the time distance of anomaly occurrence, the relationship impact value corresponding to the strength of causal relationship, the frequency impact value corresponding to the number of anomaly triggers, and the result impact value corresponding to the degree of result correlation by the time weight, relationship weight, frequency weight, and result weight, respectively, and then summing them. The time impact value is normalized based on the time distance of anomaly occurrence, the relationship impact value is the strength of causal relationship, the frequency impact value is normalized based on the number of anomaly triggers, and the result impact value is the degree of result correlation. The node contribution sequence is arranged according to the transmission order of upstream execution nodes in the anomaly impact path. The contribution of upstream execution nodes is determined based on the node contribution sequence. Upstream execution nodes whose node contribution is greater than the average contribution of all upstream execution nodes in the same abnormal impact path are identified as candidate causal breakpoints. Based on the node type, node flow relationship type, and abnormal result type corresponding to the candidate causal breakpoints, the breakpoints are classified into breakpoint types, and causal breakpoints with breakpoint type identifiers are generated. The categorization of breakpoint types includes classifying requirement deviations corresponding to task nodes as requirement comprehension breakpoints, information transmission anomalies corresponding to collaboration nodes as collaboration transmission breakpoints, process delays corresponding to approval nodes as approval flow breakpoints, resource insufficiency corresponding to task nodes and personnel nodes as resource allocation breakpoints, implementation anomalies corresponding to task nodes as technical implementation breakpoints, acceptance rework corresponding to delivery nodes as delivery and acceptance breakpoints, customer communication anomalies corresponding to collaboration nodes as customer communication breakpoints, and responsibility handover anomalies corresponding to personnel nodes as responsibility boundary breakpoints. The causal breakpoints with breakpoint type identifiers are merged according to historical task identifiers, abnormal impact paths, and breakpoint type identifiers to generate a historical causal breakpoint set. The historical causal breakpoint set includes historical task identifiers, the historical task execution nodes corresponding to the causal breakpoints, breakpoint type identifiers, node contribution, and abnormal impact paths.

[0024] In this embodiment, the capability profile construction module includes: Based on the historical task identifiers, historical task execution nodes and abnormal impact paths corresponding to the causal breakpoints in the historical causal breakpoint set, and the abnormal handling records, collaboration records, communication records and task repair records in the historical task fusion dataset, the abnormal handling behavior of enterprise personnel after the occurrence of the causal breakpoint and before the formation of the abnormal result is traced back, and the abnormal handling behavior sequence of enterprise personnel is generated. The retrospective analysis of abnormal handling behaviors includes: locating the corresponding historical task execution process based on historical task identifiers; determining the retrospective scope according to the abnormal impact path; determining the retrospective starting point based on the historical task execution node corresponding to the causal breakpoint; determining the retrospective ending point based on the result node corresponding to the abnormal result; tracing the execution nodes where the handling behavior occurred along the abnormal impact path node by node; extracting the personnel intervention behavior, behavior occurrence time, behavior target, and state changes before and after the behavior at each execution node; and connecting the handling behaviors in chronological order to form a sequence of enterprise personnel breakpoint handling behaviors covering the entire process from the causal breakpoint to the abnormal result. The sequence of enterprise personnel breakpoint handling behaviors is arranged according to the occurrence time of the handling behavior, the node where the handling behavior occurred, the type of breakpoint associated with the handling behavior, and the impact path of the handling behavior. The handling behaviors include abnormal repair behaviors, abnormal detour behaviors, abnormal suppression behaviors, and abnormal recovery behaviors. Based on the sequence of enterprise personnel fracture handling behaviors and fracture type identifiers in the historical causal fracture point set, establish the correspondence between handling behaviors and fracture types, and generate enterprise personnel fracture repair behavior mapping data; The enterprise personnel fracture repair behavior mapping data is used to represent the risk prevention methods of different enterprise personnel at different causal fracture points. Among them, the abnormal repair behavior corresponding to the demand understanding fracture point is mapped to the demand fracture repair capability; the abnormal repair behavior corresponding to the collaboration transmission fracture point is mapped to the collaboration fracture repair capability; the abnormal detour behavior corresponding to the approval flow fracture point is mapped to the approval fracture detour capability; the abnormal detour behavior and abnormal repair behavior corresponding to the resource allocation fracture point are mapped to the resource allocation capability; the abnormal suppression behavior corresponding to the technical implementation fracture point is mapped to the technical fracture suppression capability; the abnormal recovery behavior corresponding to the delivery and acceptance fracture point is mapped to the delivery fracture recovery capability; the abnormal repair behavior and abnormal recovery behavior corresponding to the customer communication fracture point are mapped to the customer communication fracture resolution capability; and the abnormal detour behavior and abnormal recovery behavior corresponding to the responsibility boundary fracture point are mapped to the multi-node fracture linkage processing capability. Based on the enterprise personnel fracture repair behavior mapping data, extract the number of treatment behaviors covered by each enterprise personnel under each fracture type, the number of nodes affected by the treatment behaviors, the number of abnormal impact paths associated with the treatment behaviors, and the cumulative results of the contribution of the corresponding nodes of the treatment behaviors, and generate enterprise personnel fracture treatment intensity data. The number of actions covered is the total number of effective actions taken by enterprise personnel under the same type of break. It is obtained by counting the actions that are within the backtracking range in the sequence of enterprise personnel's break actions, are related to the historical task execution nodes corresponding to the causal break points, and match the break type identifier. The number of nodes that affect the handling behavior is the number of historical task execution nodes that actually affect the handling behavior of the enterprise personnel. It is obtained by counting the number of nodes after deduplication of the historical task execution nodes corresponding to the handling behavior in the broken handling behavior sequence of the enterprise personnel. The number of abnormal impact paths associated with handling behaviors is the number of abnormal result transmission paths covered by the handling behaviors of enterprise personnel. It is obtained by statistically analyzing the abnormal impact path identifiers corresponding to the handling behaviors in the broken handling behavior sequence of enterprise personnel and then deduplicating them. The cumulative result of the node contribution of the disposal behavior is the sum of the node contribution of the causal breakpoints affected by the disposal behavior of the enterprise personnel. It is obtained by accumulating the node contribution of the causal breakpoints associated with each disposal behavior in the sequence of disposal behaviors of the enterprise personnel in the set of historical causal breakpoints. Based on the delivery results, task repair records and historical task anomaly data in the historical task fusion dataset, the delay reduction results, rework reduction results, failure avoidance results and quality recovery results corresponding to the enterprise personnel breakage handling behavior sequence are extracted to generate enterprise personnel breakage handling result data. Delay reduction results are obtained by comparing the estimated delay time before the intervention with the actual delay time after the intervention; rework reduction results are obtained by comparing the estimated number of reworks before the intervention with the actual number of reworks after the intervention; failure avoidance results are obtained by identifying situations where the task was at risk of failure before the intervention and the task was transformed into a completed or deliverable state after the intervention; quality recovery results are obtained by comparing the quality evaluation results before the intervention with the quality evaluation results after the intervention. Based on data on the intensity and outcome of personnel fracture treatment, the risk mitigation capability of personnel under different fracture types is calculated, and a matrix of personnel fracture repair capability is generated. The risk mitigation capability value is calculated by weighting the coverage of mitigation actions, the intensity of breakpoint handling, the effectiveness of abnormal path blocking, and the degree of improvement in mitigation results. The coverage of mitigation actions is determined by the ratio between the number of mitigation actions covered and the number of tasks historically participated in by enterprise personnel. The intensity of breakpoint handling is obtained by normalizing the cumulative contribution of the corresponding nodes of the mitigation actions. The effectiveness of abnormal path blocking is determined by the number of abnormal impact paths associated with the mitigation actions and the number of times the corresponding abnormal impact paths in the data of the mitigation results of the enterprise personnel are interrupted. The degree of improvement in mitigation results is obtained by weighting the normalized results of delay reduction, rework reduction, failure avoidance, and quality recovery. The Enterprise Personnel Fracture Repair Capability Matrix establishes a row-column correspondence relationship based on the enterprise personnel identifier and fracture type identifier. The risk blocking capability value of each enterprise personnel under each fracture type is filled into the corresponding matrix position to generate an Enterprise Personnel Fracture Repair Capability Matrix that characterizes the distribution of enterprise personnel’s ability to block different types of causal fracture points. The personnel breakdown capability matrix is ​​merged according to the personnel identification of the enterprise, and combined with the task execution role, collaboration record, communication record and task repair record to generate a personnel breakdown capability profile, including the ability to repair demand breakdown, the ability to repair collaboration breakdown, the ability to bypass approval breakdown, the ability to allocate resources, the ability to suppress technical breakdown, the ability to restore delivery breakdown, the ability to resolve customer communication breakdown, and the ability to handle multi-node breakdown linkage.

[0025] In this embodiment, the task parsing module includes: Receive task data to be assigned, identify fields such as task requirement text, task type, task complexity, participating departments, collaborating parties, time requirements, resource conditions and delivery goals in the task data to be assigned, and generate task field data to be assigned. Based on the data of the task to be assigned, the task requirement text is parsed to extract the semantics of the task target, task content, task object, task execution, and task delivery, and the semantics of the task requirement are generated. Based on the data in the task to be assigned field, the task constraint is parsed for participating departments, collaborating objects, time requirements, resource conditions, and delivery goals, generating the current task collaboration relationship, current task time constraint, current task resource constraint, and current task delivery requirements; The generation of current task collaboration relationships includes: identifying task transfer relationships, information confirmation relationships, approval cooperation relationships, and delivery connection relationships in the task field data of participating departments and collaborating objects based on the association information of participating departments and collaborating objects in the task field data; organizing the interaction order and dependency relationships between participating departments and collaborating objects; and forming current task collaboration relationships to characterize the multi-entity collaborative structure in the task execution process. The generation of current task time constraints includes: extracting task start time, task completion deadline and stage time nodes based on time requirements, determining the time urgency based on the difference between the task completion deadline and the preset standard execution cycle, integrating the time boundaries and stage time limits of the entire task execution process, and generating current task time constraints to characterize the task execution time limits; The generation of current task resource constraints includes: extracting human resource allocation, data resource availability, tool resource support capabilities, approval resource availability, and external support resource availability based on resource conditions; uniformly organizing the availability, acquisition conditions, and usage restrictions of various resources; and generating current task resource constraints to represent the resource condition restrictions for task execution. The generation of current task delivery requirements includes: extracting the type of delivery result, delivery quality requirements, delivery acceptance criteria and delivery risk requirements based on the delivery objectives, and uniformly mapping the completion form, quality evaluation criteria and acceptance judgment conditions of the delivery results to generate current task delivery requirements that characterize the final output requirements of the task. Based on the current task identifier, the current task type, current task complexity, current task requirement semantics, current task participating departments, current task collaboration relationships, current task time constraints, current task resource constraints, and current task delivery requirements are associated and organized to generate a current task feature set. Among them, the current task identifier is generated based on the task number, receipt time, and task source of the task to be assigned data; the current task type is determined based on the task type in the task to be assigned field data; the current task complexity is determined based on the task complexity in the task to be assigned field data; and the current task participating departments are determined based on the participating departments in the task to be assigned field data.

[0026] In this embodiment, the fracture reasoning module includes: The training samples of the time series graph are constructed based on the historical task causal chain and the set of historical causal breakpoints. The training samples of the time series graph include historical task execution nodes, causal relationships between nodes, abnormal influence paths, break type identifiers, node contribution, and abnormal result types. They are arranged according to the occurrence time of historical task execution nodes and node flow relationships, which are used to characterize the temporal relationship of abnormality in historical tasks being transmitted from upstream execution nodes to downstream result nodes through abnormal influence paths. The training samples of the time series graph are input into the time series graph neural network for training to obtain the causal break inference model; The causal break inference model is derived by: using historical task execution nodes from the time-series training samples as graph nodes, causal relationships between nodes as graph edges, node sequences corresponding to abnormal impact paths as path constraint information, break type identifiers and abnormal result types as supervision labels, and node contribution as node weight information input into the time-series graph neural network; performing temporal encoding on the graph nodes according to the occurrence time order of historical task execution nodes, and propagating information between adjacent graph nodes layer by layer based on node flow relationships to obtain the temporal feature representation of each graph node; performing path aggregation processing on the node sequences in the abnormal impact path based on the temporal feature representation of the graph nodes to obtain the abnormal transmission feature representation of the corresponding path; constructing the model output results based on the break type identifiers and abnormal result types, and comparing the model output results with the supervision labels. The degree of difference between nodes is used to construct a loss function, which consists of the prediction error of fracture type, the prediction error of abnormal results, and the weighted error of node contribution. Based on the loss function, the parameters in the temporal graph neural network are iteratively updated. The training parameters used in the iterative update process include the learning rate, the number of batch samples, the number of training epochs, and the parameter update step size. Specifically, the learning rate ranges from 0.001 to 0.01, the number of batch samples ranges from 16 to 128, the number of training epochs ranges from 50 to 300, and the parameter update step size ranges from 0.1 to 1. During each training epoch, the model parameters are adjusted according to the changes in the loss function. When the change in the loss function is lower than a preset threshold or reaches a preset maximum number of training epochs in multiple consecutive training epochs, the convergence condition is determined, and the trained network parameters are output to generate a causal fracture inference model. Perform graph structure transformation on the current task feature set to construct the current task execution graph; Graph structure transformation processing includes: mapping nodes to the current task feature set, including current task requirement semantics, participating departments, collaborative relationships, time constraints, resource constraints, and delivery requirements, to generate current task requirement nodes, participating nodes, collaborative nodes, time nodes, resource nodes, delivery nodes, and result nodes. The result nodes are generated based on the delivery type, acceptance criteria, and risk requirements in the delivery requirements. Task transfer and information exchange relationships between nodes are determined based on collaborative relationships, generating task flow edges. The execution order and stage connection relationships between nodes are determined based on time constraints, generating temporal constraint edges. Resource dependency relationships are determined based on resource constraints, generating resource dependence edges. Delivery dependency and result constraint relationships between nodes are determined based on delivery requirements, generating delivery constraint edges. Finally, nodes and edges are uniformly associated to form the current task execution graph. The historical task causal chain and the current task execution graph are input into the causal break inference model. Based on the historical task anomaly transmission relationship, potential break inference is performed on the task execution flow relationship in the current task execution graph to generate potential break inference results. The historical task anomaly transmission relationship includes anomaly triggering relationship, anomaly diffusion relationship, anomaly superposition relationship, and anomaly result association relationship. Among them, the anomaly triggering relationship is used to characterize the triggering effect of the upstream execution node on the downstream anomaly node, the anomaly diffusion relationship is used to characterize the propagation path of the anomaly among multiple execution nodes, the anomaly superposition relationship is used to characterize the joint effect of multiple anomalies on the same execution node, and the anomaly result association relationship is used to characterize the association relationship between the anomaly state and the final task result. The potential breakage inference includes: using a node feature encoding unit in the causal breakage inference model to encode the features of task execution nodes in the current task execution graph, generating a feature representation of the current task node; using a graph structure information propagation unit to propagate features between adjacent nodes in the task execution flow relationship in the current task execution graph based on the historical task anomaly propagation relationship in the historical task causal chain, obtaining an updated node association feature representation, where the task execution flow relationship includes task flow edges, temporal constraint edges, resource dependency edges, and delivery constraint edges; using a temporal dependency modeling unit to perform temporal modeling on the node association feature representation, combining the occurrence order of task execution nodes and node flow relationships, generating node temporal features reflecting the anomaly propagation trend; using a node state update unit to update the state of each task execution node according to the node temporal features, obtaining a node state representation containing breakage risk information; aggregating the temporal feature representations and anomaly propagation feature representations corresponding to historical task execution nodes under the same breakage type identifier in the historical causal breakage point set, obtaining node state features corresponding to each breakage type identifier; and using an output prediction unit to... Based on the degree of matching between the node state representation and the node state features corresponding to each fracture type identifier in the historical causal breakpoint set, the probability of fracture occurrence for each task execution node is calculated, and the fracture type matching result corresponding to each task execution node is determined according to the fracture type identifier with the highest probability of fracture occurrence. Based on the fracture type matching result corresponding to each task execution node, and along the node flow direction in the current task execution graph, the abnormal propagation direction corresponding to each task execution node is determined. Based on the abnormal propagation direction corresponding to each task execution node, the task flow relationship strength, temporal constraint strength, resource dependency strength, and delivery constraint strength between each task execution node and downstream task execution nodes are extracted, and the abnormal propagation strength corresponding to each task execution node is calculated. Based on the probability of fracture occurrence, abnormal propagation strength, current task time constraint, current task resource constraint, and current task delivery requirement corresponding to each task execution node, the fracture risk level corresponding to each task execution node is determined. Based on the probability of fracture occurrence, fracture type matching result, abnormal propagation direction, abnormal propagation strength, and fracture risk level corresponding to each task execution node, potential fracture inference results are generated. Based on the results of potential breakage inference, candidate potential breakpoints in the current task are determined, and the risk level of the candidate potential breakpoints is judged to generate candidate potential breakpoint data. The generation of candidate potential breakpoint data includes: filtering all task execution nodes based on the break occurrence probability corresponding to each task execution node in the potential break occurrence inference results, and selecting task execution nodes with a break occurrence probability greater than a preset probability threshold as candidate potential breakpoints; determining the type of candidate potential breakpoint based on the break type matching results corresponding to the candidate potential breakpoint; determining the task stage where the candidate potential breakpoint is located based on the node position of the candidate potential breakpoint in the current task execution graph and the corresponding node flow order; determining the associated execution node of the candidate potential breakpoint based on the task execution node identifier corresponding to the candidate potential breakpoint; and correcting the risk level of the candidate potential breakpoint based on the break risk level in the potential break occurrence inference results, combined with the break occurrence probability and abnormal propagation intensity of the candidate potential breakpoint, to generate candidate potential breakpoint data. Based on the candidate potential breakpoint data and the current task execution graph, the propagation path of the candidate potential breakpoints to the current task result node is traced to generate the current task potential breakpoint set. The current task potential breakpoint set includes the potential breakpoint type, the task stage where the potential breakpoint is located, the execution node associated with the potential breakpoint, the risk level of the potential breakpoint, and the potential breakpoint propagation path. The potential breakpoint propagation path is obtained by starting from the task execution node corresponding to the candidate potential breakpoint and tracing along the task flow edge, timing constraint edge, resource dependency edge, and delivery constraint edge in the current task execution graph to the current task result node.

[0027] In this embodiment, the matching calculation module includes: Based on the profile of personnel fracture repair capabilities, the risk prevention capabilities of personnel in candidate enterprises under different fracture types are extracted, and data on the risk prevention capabilities of personnel in candidate enterprises are generated. Based on the potential breakpoint types in the current task potential breakpoint set, establish a type correspondence with the risk blocking capabilities in the candidate enterprise personnel risk blocking capability data, and generate candidate enterprise personnel breakpoint capability corresponding data. Based on the candidate enterprise personnel's breakpoint capability data, the potential breakpoint risk level and potential breakpoint propagation path in the current task's potential breakpoint set, calculate the risk blocking matching degree of the candidate enterprise personnel for the current task's potential breakpoint set. The calculation of risk blocking matching degree includes: based on the candidate enterprise personnel's breakpoint capability data, finding the risk blocking capability dimension corresponding to the potential breakpoint type, and reading the capability value of the candidate enterprise personnel under this risk blocking capability dimension to obtain the capability type matching degree; converting the potential breakpoint risk level into the corresponding capability requirement value, comparing the capability value of the candidate enterprise personnel with the capability requirement value, and determining the risk level bearing degree as the highest value when the capability value of the candidate enterprise personnel is greater than or equal to the capability requirement value; when the capability value of the candidate enterprise personnel is less than the capability requirement value, determining the risk level bearing degree based on the difference between the capability value of the candidate enterprise personnel and the capability requirement value; counting the number of execution nodes associated with potential breakpoints in the potential breakpoint propagation path covered by the risk blocking capability of the candidate enterprise personnel, and calculating the ratio between this number and the total number of execution nodes in the potential breakpoint propagation path to obtain the propagation path blocking degree; and weighting and summing the capability type matching degree, risk level bearing degree, and propagation path blocking degree to obtain the risk blocking matching degree of the candidate enterprise personnel for the current task's potential breakpoint set. Candidate enterprise personnel identifiers are sorted according to risk blocking matching degree, and a breakpoint matching result is generated by combining the corresponding data of the breakpoint capabilities of the candidate enterprise personnel. The breakpoint matching result includes the candidate enterprise personnel identifier, the breakpoint matching type corresponding to the candidate enterprise personnel, the risk blocking capability ranking sequence, and the candidate enterprise personnel task adaptation result. The breakpoint matching type corresponding to the candidate enterprise personnel is determined according to the correspondence between the candidate enterprise personnel and the potential breakpoint types in the breakpoint capability corresponding data of the candidate enterprise personnel. The risk blocking capability ranking sequence is obtained by sorting from high to low risk blocking matching degree. The candidate enterprise personnel task adaptation result is determined according to whether the risk blocking matching degree reaches the preset matching threshold and whether the breakpoint matching type corresponding to the candidate enterprise personnel covers the potential breakpoint types in the potential breakpoint set of the current task.

[0028] In this embodiment, the task allocation module includes: Based on the breakpoint matching type of the candidate enterprise personnel in the breakpoint matching results, and the potential breakpoint type in the potential breakpoint set of the current task, a coverage match is performed to filter the candidate enterprise personnel that can cover the potential breakpoint type in the potential breakpoint propagation path, and generate candidate data of recommended execution objects. Recommended execution targets are determined based on candidate data and the ranking sequence of risk blocking capabilities of candidate enterprise personnel. The determination of recommended execution targets includes: when a single candidate enterprise personnel covers all potential breakpoint types in the current task's potential breakpoint set, and the task adaptation result of the candidate enterprise personnel reaches a preset adaptation threshold, the single candidate enterprise personnel is determined as the recommended personnel; when a single candidate enterprise personnel does not cover all potential breakpoint types, multiple candidate enterprise personnel are combined according to the potential breakpoint type coverage relationship in the potential breakpoint propagation path, and multiple candidate enterprise personnel are determined as the recommended team. The task allocation priority is determined based on the coverage of the potential breakpoint propagation path by the recommended execution objects, the ranking position of the recommended execution objects in the candidate enterprise personnel risk blocking capability ranking sequence, and the candidate enterprise personnel task adaptation results. Based on the breakpoint matching type, risk blocking capability, and potential breakpoint propagation path of the recommended execution object, generate matching reasons and risk blocking prompts; The generation of matching reasons includes: determining the types of potential breakpoints that the recommended execution object can handle based on the correspondence between the breakpoint matching type corresponding to the recommended execution object and the types of potential breakpoints in the current task's potential breakpoint set; extracting the capability value of the recommended execution object under the corresponding breakpoint matching type based on its risk blocking capability; determining the range of potential breakpoint propagation paths that the recommended execution object can block based on the range of potential breakpoint propagation paths covered by the recommended execution object; and combining the potential breakpoint type correspondence, risk blocking capability value, and propagation path coverage to generate matching reasons. The generation of risk blocking prompts includes: extracting the potential breakpoint types, risk levels, and propagation paths from the current task's potential breakpoint set; determining the risk blocking method for each potential breakpoint for the recommended execution object based on the breakpoint matching type; determining the risk blocking method for different potential breakpoint types based on the risk blocking capability of the recommended execution object; determining the priority potential breakpoints based on their risk levels; and combining the potential breakpoint types, risk levels, propagation paths, and corresponding risk blocking methods to generate risk blocking prompts. The task allocation result is generated based on the current task identifier, the set of potential breakpoints of the current task, the recommended execution object, the task allocation priority, the matching reason, and the risk blocking prompt. The task allocation result includes the target task identifier, the set of potential breakpoints of the current task, the recommended execution object identifier, the risk blocking capability corresponding to the recommended execution object, the task allocation priority, the matching reason, and the risk blocking prompt. The target task identifier is determined by the current task identifier, and the recommended execution object identifier is the recommended personnel identifier or the recommended team identifier.

[0029] Example 1: To verify the feasibility of this invention in practice, it was applied to a cross-departmental collaborative task allocation scenario in a large manufacturing enterprise. In its daily operations, this enterprise involves task execution processes with multiple departments. From requirement submission and solution design to approval processes and implementation, tasks involve multiple execution nodes, historically resulting in frequent delays, rework, and inconsistent quality. The enterprise's original task allocation method relied primarily on management experience, assigning tasks based on job positions and past performance. This lacked the ability to proactively identify potential risks during task execution and accurately match personnel with specific anomaly handling capabilities. Consequently, when anomalies occurred during task execution, multiple adjustments to personnel configuration were often necessary, leading to low efficiency.

[0030] In this scenario, the system first acquires historical task data and personnel execution data from the enterprise. Through fusion processing, a historical task fusion dataset is constructed, and further, historical task execution nodes and their flow relationships are extracted to form a historical task causal chain. Based on this, the abnormal impact paths between abnormal nodes and result nodes are analyzed to identify causal breakpoints leading to abnormal results, generating a set of historical causal breakpoints. Subsequently, the abnormal handling behaviors and corresponding results of enterprise personnel after the breakpoints occur are retrospectively analyzed to construct a personnel breakpoint repair capability profile. For new tasks to be assigned, the system parses the task requirement text, participating departments, time requirements, and resource conditions to generate a current task feature set. The task execution process is then analyzed using a causal breakpoint inference model to obtain a set of potential breakpoints for the current task. Based on this, combined with the personnel breakpoint repair capability profile, the risk blocking matching degree of candidate personnel is calculated, recommended execution objects that can cover the propagation paths of potential breakpoints are selected, and the task allocation results are output.

[0031] In practical application, the company deployed the system of this invention at its production base in East China during a certain quarter to replace the daily task allocation process. By comparing the operation before and after deployment, it can be observed that abnormal propagation paths during task execution were identified in advance, key breakpoints were addressed in a targeted manner, and the continuity of task execution was significantly improved. Historical data comparison shows that task delays were effectively mitigated, rework frequency was significantly reduced, personnel adjustments were decreased, and the overall execution process was more stable. Through statistical analysis of data from multiple task execution cycles, it can be verified that the invention has strong risk prediction and matching capabilities in complex collaborative environments, effectively solving the problems of traditional task allocation methods that cannot identify potential break risks and are not accurately matched with personnel capabilities.

[0032] Table 1. Performance Comparison of the Invention and Traditional Intelligent Personnel Task Matching Methods in Enterprises

[0033] As can be clearly seen from Table 1, the method of the present invention is superior to the traditional method in many indicators.

[0034] The average task allocation time of the method of this invention is reduced from 42.6 minutes in the traditional method to 31.8 minutes, a reduction of 10.8 minutes. The reason is that this invention does not rely solely on human experience for judgment, but rather performs matching calculations based on the set of potential fault points in the current task and the personnel's fault repair capability profile. This makes the candidate enterprise personnel screening process more direct and reduces the time required for repeated manual comparison of task types, personnel experience, and historical performance.

[0035] Regarding task delay rate, the traditional method has a delay rate of 18.4%, while the method of this invention has a delay rate of 13.7%, a reduction of 4.7 percentage points. This improvement stems from the causal break inference model's ability to identify the risk level and propagation path of potential breakpoints before task allocation. This allows the system to prioritize recommending execution objects with corresponding risk-blocking capabilities, thereby reducing the probability of demand comprehension breakdowns, collaboration transmission breakdowns, or resource allocation breakdowns continuing to propagate and resulting in delays.

[0036] The rework rate decreased from 11.2% to 8.6%, a drop of 2.6 percentage points. This is because the present invention traces back abnormal handling behaviors and results of enterprise personnel by using a set of historical causal breakpoints to construct a profile of personnel's ability to recover from setbacks. This allows task allocation to no longer only consider personnel positions or performance, but also to further consider personnel's actual recovery capabilities in scenarios such as delivery and acceptance breaks and technical implementation breaks. Therefore, it can reduce rework caused by mismatched capabilities.

[0037] The rate of secondary personnel adjustments decreased from 15.8% to 10.9%, a drop of 4.9 percentage points; the accuracy rate of initial task matching increased from 78.5% to 86.9%, an increase of 8.4 percentage points. This indicates that the fault matching results generated by this invention can more accurately reflect the correspondence between candidate enterprise personnel and potential fault points in the current task. The average anomaly response time decreased from 6.4 hours to 4.9 hours, and the on-time task delivery rate increased from 81.6% to 86.3%, demonstrating that this invention has a good overall effect in early risk identification, matching and blocking capabilities, and stable task execution.

[0038] The above are merely preferred embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A task-matching system for enterprise personnel based on a reasoning model, characterized in that, include: The data fusion module is used to acquire and merge historical task data and personnel execution data of the enterprise to generate a historical task fusion dataset. The causal chain building module is used to extract historical task execution nodes and the flow relationships between nodes based on the historical task fusion dataset, and to build historical task causal chains. The fracture identification module is used to perform abnormal cause reasoning on the abnormal influence path between abnormal nodes and result nodes based on the historical task causal chain, identify the causal fracture points that lead to abnormal results of historical tasks and generate a set of historical causal fracture points. The capability profile building module is used to trace back the abnormal handling behaviors and results of enterprise personnel in corresponding historical tasks based on the set of historical causal breakpoints, and to build a personnel breakpoint repair capability profile. The task parsing module is used to receive task data to be assigned, perform task semantic parsing and task constraint parsing on the task data to be assigned, and generate the current task feature set; The fracture reasoning module is used to input the causal chain of historical tasks and the feature set of the current task into the causal fracture reasoning model to perform potential fracture reasoning and generate a set of potential fracture points for the current task. The matching calculation module is used to calculate the risk blocking matching degree of candidate enterprise personnel based on the current set of potential breakpoints and the personnel break repair capability profile, and generate break matching results based on the risk blocking matching degree. The task allocation module is used to determine the recommended execution objects that can block the propagation path of potential breakpoints based on the breakpoint matching results, and output the task allocation results.

2. The enterprise personnel task intelligent matching system based on a reasoning model according to claim 1, characterized in that, The historical task data includes historical task basic data, historical task execution process data, and historical task anomaly data. The personnel execution data includes personnel identification, historical participation in tasks, task execution roles, anomaly handling records, collaboration records, communication records, delivery results, and task repair records. The fusion processing includes data cleaning, time alignment, entity normalization, and node relationship mapping.

3. The enterprise personnel task intelligent matching system based on a reasoning model according to claim 1, characterized in that, The causal chain construction module includes: Based on the historical task fusion dataset, historical task execution nodes are identified and divided into task nodes, personnel nodes, collaboration nodes, approval nodes, delivery nodes, exception nodes, and result nodes according to their source and function. Based on the occurrence time, execution order, and business flow relationship of historical task execution nodes in the historical task fusion dataset, extract the flow relationship between nodes; Based on the flow relationship between nodes, a directed connection relationship is established between historical task execution nodes. The historical task execution node located at the forefront of the task advancement direction is taken as the source node, and the target historical task execution node determined based on the occurrence time, execution order and business flow relationship is taken as the target node, forming a point flow structure. Based on the positions of abnormal nodes and result nodes in the node flow structure, the propagation path between abnormal nodes and result nodes is marked to obtain the abnormal result propagation path. Based on the node flow structure and abnormal result propagation path, the flow relationship between nodes is labeled with causal association, and a historical task causal chain is constructed.

4. The enterprise personnel task intelligent matching system based on a reasoning model according to claim 1, characterized in that, The fracture detection module includes: Based on the historical task causal chain, abnormal nodes and result nodes are located, and path tracing is performed along the causal relationship between abnormal nodes and result nodes to generate a set of abnormal impact paths. Based on the set of abnormal impact paths, the upstream execution nodes in each abnormal impact path are determined, and the time distance of abnormal occurrence, the strength of causal relationship, the number of abnormal triggers, and the degree of result correlation are extracted for the upstream execution nodes; The node contribution of upstream execution nodes is calculated based on the time distance of anomaly occurrence, the strength of causal relationship, the number of anomaly triggers, and the degree of correlation of results, and a node contribution sequence corresponding to the anomaly impact path is generated. The contribution of upstream execution nodes is determined based on the node contribution sequence. Upstream execution nodes whose node contribution is greater than the average contribution of all upstream execution nodes in the same abnormal impact path are identified as candidate causal breakpoints. Based on the node type, node flow relationship type, and abnormal result type corresponding to the candidate causal breakpoints, the breakpoints are classified into breakpoint types, and causal breakpoints with breakpoint type identifiers are generated. The causal breakpoints with breakpoint type identifiers are merged according to historical task identifiers, abnormal impact paths, and breakpoint type identifiers to generate a set of historical causal breakpoints.

5. The enterprise personnel task intelligent matching system based on a reasoning model according to claim 1, characterized in that, The capability profile construction module includes: Based on the historical causal breakpoint set and the historical task fusion dataset, the abnormal handling behavior of enterprise personnel after the occurrence of the causal breakpoint and before the formation of the abnormal result is traced back, and the sequence of enterprise personnel breakpoint handling behavior is generated. Based on the sequence of enterprise personnel fracture handling behaviors and fracture type identifiers in the historical causal fracture point set, establish the correspondence between handling behaviors and fracture types, and generate enterprise personnel fracture repair behavior mapping data; Based on the enterprise personnel fracture repair behavior mapping data, extract the number of treatment behaviors covered by each enterprise personnel under each fracture type, the number of nodes affected by the treatment behaviors, the number of abnormal impact paths associated with the treatment behaviors, and the cumulative results of the contribution of the corresponding nodes of the treatment behaviors, and generate enterprise personnel fracture treatment intensity data. Based on the delivery results, task repair records and historical task anomaly data in the historical task fusion dataset, the delay reduction results, rework reduction results, failure avoidance results and quality recovery results corresponding to the enterprise personnel breakage handling behavior sequence are extracted to generate enterprise personnel breakage handling result data. Based on data on the intensity and outcome of fracture treatment for enterprise personnel, the risk mitigation capability of enterprise personnel under different fracture types is calculated, and a fracture repair capability matrix for enterprise personnel is generated. The personnel fracture repair capability matrix is ​​merged according to the personnel identification of the enterprise, and combined with the task execution role, collaboration record, communication record and task repair record to generate personnel fracture repair capability profile.

6. The enterprise personnel task intelligent matching system based on a reasoning model according to claim 1, characterized in that, The task parsing module includes: Receive task data to be assigned, identify fields such as task requirement text, task type, task complexity, participating departments, collaborating parties, time requirements, resource conditions and delivery goals in the task data to be assigned, and generate task field data to be assigned. Based on the data of the task to be assigned, the task requirement text is parsed to extract the semantics of the task target, task content, task object, task execution, and task delivery, and the semantics of the task requirement are generated. Based on the data in the task to be assigned field, the task constraint is parsed for participating departments, collaborating objects, time requirements, resource conditions, and delivery goals, generating the current task collaboration relationship, current task time constraint, current task resource constraint, and current task delivery requirements; Based on the current task identifier, the current task type, current task complexity, current task requirement semantics, current task participating departments, current task collaboration relationships, current task time constraints, current task resource constraints, and current task delivery requirements are associated and organized to generate a current task feature set.

7. The enterprise personnel task intelligent matching system based on a reasoning model according to claim 1, characterized in that, The fracture reasoning module includes: Training samples for a time series graph are constructed based on the historical task causal chain and the set of historical causal breakpoints. The training samples of the time series graph are input into the time series graph neural network for training to obtain the causal break inference model; Perform graph structure transformation on the current task feature set to construct the current task execution graph; Input the historical task causal chain and the current task execution graph into the causal break inference model, and perform potential break inference on the task execution flow relationship in the current task execution graph based on the abnormal transmission relationship of the historical task, and generate potential break inference results. Based on the results of potential breakage inference, candidate potential breakpoints in the current task are determined, and the risk level of the candidate potential breakpoints is judged to generate candidate potential breakpoint data. Based on the candidate potential breakpoint data and the current task execution graph, the propagation path of the candidate potential breakpoints to the current task result node is traced to generate the current task potential breakpoint set.

8. The enterprise personnel task intelligent matching system based on a reasoning model according to claim 1, characterized in that, The matching calculation module includes: Based on the profile of personnel fracture repair capabilities, the risk prevention capabilities of personnel in candidate enterprises under different fracture types are extracted, and data on the risk prevention capabilities of personnel in candidate enterprises are generated. Based on the potential breakpoint types in the current task potential breakpoint set, establish a type correspondence with the risk blocking capabilities in the candidate enterprise personnel risk blocking capability data, and generate candidate enterprise personnel breakpoint capability corresponding data. Based on the candidate enterprise personnel's breakpoint capability data, the potential breakpoint risk level and potential breakpoint propagation path in the current task's potential breakpoint set, calculate the risk blocking matching degree of the candidate enterprise personnel for the current task's potential breakpoint set. Candidate company personnel identifiers are sorted according to risk blocking matching degree, and the corresponding data of the personnel's breakpoint capabilities are combined to generate breakpoint matching results.

9. The enterprise personnel task intelligent matching system based on a reasoning model according to claim 1, characterized in that, The task allocation module includes: Based on the breakpoint matching type of the candidate enterprise personnel in the breakpoint matching results, and the potential breakpoint type in the potential breakpoint set of the current task, a coverage match is performed to filter the candidate enterprise personnel that can cover the potential breakpoint type in the potential breakpoint propagation path, and generate candidate data of recommended execution objects. The recommended implementation targets are determined based on the candidate data of the recommended implementation targets and the ranking sequence of the risk prevention capabilities of the personnel of the candidate enterprises. Based on the coverage of the potential breakpoint propagation path by the recommended execution objects, the ranking position of the recommended execution objects in the candidate enterprise personnel risk blocking capability ranking sequence, and the candidate enterprise personnel task adaptation results, the task allocation priority is determined. Based on the breakpoint matching type, risk blocking capability, and potential breakpoint propagation path of the recommended execution object, generate matching reasons and risk blocking prompts; The task allocation result is generated based on the current task identifier, the set of potential breakpoints of the current task, the recommended execution object, the task allocation priority, the matching reason, and the risk blocking prompt.

Images