Human resource organization and management method based on data driving

Abstract

The invention discloses a human resource organization management method based on data driving, which relates to the technical field of human resource data management, and comprises the following steps: constructing a weighted relation graph, converting the weighted relation graph into a sparse tensor, and integrating multiple target factors of post competency, cooperation effect, human cost and time stability to obtain a sparse tensor; an iterative normalization and rarefaction method is adopted to generate a man-post assignment matrix and a sensitivity scoring matrix; and generating a task performance signal through the man-post assignment matrix, carrying out dynamic feedback on the task performance signal, identifying an abnormal condition by adopting an online drift detection method, selecting a candidate set according to the sensitivity scoring matrix to execute local re-optimization, and updating the man-post assignment matrix. According to the invention, through improved multi-objective optimization assignment based on sparse tensor, the scientificity and comprehensiveness of man-post matching are improved.

Description

Technical Field

[0001] This invention relates to the field of human resource data management technology, and in particular to a data-driven human resource organization management method. Background Technology

[0002] In today's human resource management, as organizations expand in size and the complexity of tasks increases, data-driven human resource organization methods are gaining importance. Current conventional methods typically rely on rule-based systems or statistical analysis models. By integrating multi-source human resource data, static matching of personnel and positions is performed. Empirical rules and optimization algorithms are used to generate initial allocation schemes. In terms of standardized process processing, basic data integration, and static scenarios, a practical system is formed for the allocation of relevant resources.

[0003] Analysis of research and practice shows that conventional models commonly used in routine methods, when comprehensively considering multiple factors such as job competence, collaboration effect, human resource cost, and time stability, usually adopt fixed weight superposition or linear aggregation methods. These methods are difficult to effectively handle complex constraints under high-dimensional sparse data, have high computational complexity, and lack real-time feedback and online adjustment mechanisms, making it impossible to respond quickly to performance drift and abnormal situations during task execution. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a data-driven human resource organization and management method to address the problems of insufficient multi-objective collaborative optimization capabilities and lack of dynamic adjustment mechanisms in existing technologies.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] This invention provides a data-driven human resource organization management method, which includes:

[0008] Collect multi-source human resource data, perform event-based modeling on the multi-source human resource data, obtain a structured initial event table, preprocess the structured initial event table, and filter the qualified standardized event table and multidimensional table set according to weighted scores.

[0009] Based on the standardized event table and multidimensional table set of quality compliance, the edge weights between personnel and positions, personnel and skills, positions and tasks, and skills and tasks are calculated respectively, and a weighted relationship graph is constructed.

[0010] The weighted relationship graph is transformed into a sparse tensor. Taking into account multiple objective factors such as job competency, collaboration effect, human resource cost and time stability, the iterative normalization and sparsification methods are used to generate the job assignment matrix and sensitivity score matrix.

[0011] The task performance signal is generated by the job assignment matrix, the task performance signal is dynamically fed back, an online drift detection method is used to identify abnormal situations, a candidate set is selected based on the sensitivity scoring matrix to perform local re-optimization, and the job assignment matrix is ​​updated.

[0012] As a preferred embodiment of the data-driven human resource organization and management method described in this invention, the multi-source human resource data includes basic personnel information, attendance hours, performance evaluation, project tasks, external talent, and salary information.

[0013] As a preferred embodiment of the data-driven human resource organization and management method of the present invention, the steps of collecting multi-source human resource data, performing event-based modeling on the multi-source human resource data, and obtaining a structured initial event table are as follows.

[0014] Collect multi-source human resources data, generate unique identifiers for the multi-source human resources data and unify them with time, perform event-based modeling, and obtain an event-based multi-source human resources data set;

[0015] The event-based multi-source human resource data set is standardized and missing data is filled in to obtain a multi-source human resource data standard set. A quality score is calculated on the multi-source human resource data standard set to generate a structured initial event table.

[0016] As a preferred embodiment of the data-driven human resource organization and management method of the present invention, the multidimensional table set includes personnel dimension table, job dimension table, skill dimension table and task dimension table.

[0017] As a preferred embodiment of the data-driven human resource organization management method of the present invention, the steps of preprocessing the structured initial event table and selecting a set of standardized event tables and multidimensional tables of acceptable quality based on weighted scoring are as follows:

[0018] Perform field integrity checks, value validity checks, robust anomaly detection and replacement, quantile truncation standardization, missing data completion, consistency checks, and derived field generation on the structured event table to obtain an event table with auxiliary metrics;

[0019] The event tables with auxiliary metrics are filtered to obtain a set of standardized event tables and multidimensional tables that meet the required quality through weighted scoring.

[0020] As a preferred embodiment of the data-driven human resource organization management method described in this invention, the steps are as follows: Based on a standardized event table and multidimensional table set with acceptable quality, the edge weights between personnel and positions, personnel and skills, positions and tasks, and skills and tasks are calculated respectively to construct a weighted relationship graph.

[0021] Based on a set of standardized event tables and multidimensional tables that meet quality standards, a composite index is created to obtain a set of standardized event tables and multidimensional tables with the composite index.

[0022] Based on a standardized event table and a multidimensional table set with a composite index, calculate the edge weights of personnel and positions, and generate a set of personnel and position edges.

[0023] Combine the standardized event table with composite index and the multidimensional table set with the personnel and job edge set, calculate the personnel and skill edge weights, and obtain the personnel and skill edge set;

[0024] Combine the standardized event table with composite index and the multidimensional table set with the personnel and skill edge set to calculate the position and task edge weights and obtain the position and task edge set.

[0025] The standardized event table with composite index is combined with the multidimensional table set, along with the job and task edge set, to calculate the skill and task edge weights and obtain the skill and task edge set.

[0026] The sets of personnel and job positions, personnel and skills, job positions and tasks, and skills and tasks are normalized and trimmed, and then combined with multidimensional table sets to construct a weighted relationship graph.

[0027] As a preferred embodiment of the data-driven human resource organization and management method of the present invention, the step of converting the weighted relationship graph into a sparse tensor is as follows:

[0028] An index mapping is established based on a weighted relational graph, an index mapping set is obtained, and the index mapping set and the timestamp information in the weighted relational graph are divided into discrete time windows to generate a time window set.

[0029] By aggregating the time window set and the job and task edge set and skill and task edge set in the weighted relationship graph, a task requirement weight table is generated.

[0030] The task requirement weight table is integrated with the personnel and job edge set and personnel and skill edge set in the weighted relationship graph and filtered. After normalization, a sparse coordinate list is generated. The sparse coordinate list is then combined with the index mapping set and the time window set to generate a personnel-job-skill-time sparse tensor.

[0031] As a preferred embodiment of the data-driven human resource organization and management method described in this invention, the multi-objective factors, including comprehensive job competency, collaboration effect, labor cost, and time stability, are analyzed using iterative normalization and sparsification methods to generate a job assignment matrix and a sensitivity scoring matrix. The steps are as follows.

[0032] Expand the sparse tensor of personnel-position-skill-time and initialize it with a uniform probability distribution to obtain the initial personnel-position assignment matrix. Combine the initial personnel-position assignment matrix with the personnel and position edge set to calculate the position competency benefit and generate the position competency matrix.

[0033] By combining the job competency matrix with the personnel-skill edge set and the skill-task edge set, the collaborative effect benefit is calculated, and a collaborative effect matrix is ​​generated.

[0034] By combining the collaboration effect matrix with the personnel dimension table and the job dimension table, the human resource cost penalty is calculated, and a human resource cost matrix is ​​generated.

[0035] By combining the human resource cost matrix with the personnel and job assignment matrix of the previous time window in the historical time window, the time stability penalty is calculated and a time stability matrix is ​​generated.

[0036] The time stability matrix, job competency matrix, collaboration effect matrix and human resource cost matrix are weighted and combined, the resulting comprehensive benefit matrix is ​​iteratively normalized and sparsified to generate the job assignment matrix.

[0037] Based on the personnel-job assignment matrix, the degree of change of the comprehensive benefit matrix due to marginal disturbances is calculated, and the sensitivity score matrix is ​​obtained.

[0038] As a preferred embodiment of the data-driven human resource organization and management method of the present invention, the steps of generating task performance signals through a person-job assignment matrix and dynamically feeding back the task performance signals are as follows:

[0039] By combining the personnel-job assignment matrix, job dimension table, task dimension table, and standardized event table for quality compliance, a task performance signal table is generated.

[0040] The task performance signal table is compared with the historical baseline within a sliding time window to generate a feedback indicator table;

[0041] The task performance signal table and feedback indication table are merged to generate a task performance dynamic feedback record table.

[0042] As a preferred embodiment of the data-driven human resource organization and management method of the present invention, the steps of identifying abnormal situations using an online drift detection method, selecting a candidate set based on a sensitivity scoring matrix to perform local re-optimization, and updating the personnel-job assignment matrix are as follows:

[0043] Based on the task performance dynamic feedback record table, an abnormal task set is obtained through online drift detection method. The abnormal task set is then combined with a sensitivity scoring matrix for screening to obtain a candidate job matching set.

[0044] By combining the candidate job pair set and the comprehensive benefit matrix, a local objective function is constructed and solved. The local objective function table is iteratively normalized and sparsified to generate an updated job assignment matrix. The updated job assignment matrix is ​​compared with the job assignment matrix to generate a job assignment update record table.

[0045] The beneficial effects of this invention are as follows: by using an improved multi-objective optimization assignment based on sparse tensors, a unified representation of multi-dimensional data relationships and a cross-dimensional overall optimization expression are achieved, thereby improving the scientificity and comprehensiveness of person-job matching; and by using a person-job assignment matrix and a sensitivity scoring matrix, real-time response to task performance signals and multi-objective balanced optimization are achieved in a dynamic environment. Attached Figure Description

[0046] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0047] Figure 1 This is a flowchart of a data-driven human resource organization and management approach.

[0048] Figure 2 A flowchart for constructing a weighted relationship graph for a set of standardized event tables and multidimensional tables.

[0049] Figure 3 The flowchart for generating sparse tensors and person-job assignment matrices.

[0050] Figure 4 This is a flowchart of dynamic feedback and re-optimization based on the personnel-job assignment matrix. Detailed Implementation

[0051] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0052] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0053] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0054] Reference Figures 1-4 This is one embodiment of the present invention, which provides a data-driven human resource organization management method, including the following steps:

[0055] S1. Collect multi-source human resource data, perform event-based modeling on the multi-source human resource data, obtain a structured initial event table, preprocess the structured initial event table, and select a set of standardized event tables and multidimensional tables that meet the quality requirements based on weighted scores.

[0056] Multi-source human resources data includes basic personnel information, attendance hours, performance evaluations, project tasks, external talent, and salary information.

[0057] Collect multi-source human resource data, generate unique identifiers for the multi-source human resource data and unify them with time, perform event-based modeling, and obtain an event-based multi-source human resource data set.

[0058] Furthermore, records are obtained from sources such as human resources, attendance hours, performance evaluation, project tasks, external talent, and salary information. All records are collected in a structured manner according to a unified field directory, which includes personnel identifier, job identifier, skill identifier set, task identifier, timestamp, source type, and corresponding business attribute parameters. Multi-source human resources are generated. For each record in the multi-source human resources, an irreversible unique personnel identifier is generated based on the personnel name, employee number, date of birth, and organization code information. An event unique identifier is also generated based on the unique personnel identifier, event type code, record timestamp, and source type. Multi-source human resources data with unified identifiers is obtained. The local timestamps in the multi-source human resources data with unified identifiers are converted to Coordinated Universal Time (UTC). Differences in different time zones and daylight saving time are handled uniformly to generate multi-source human resources data with unified timestamps.

[0059] In multi-source human resources data with a unified timestamp, each record is transformed into an event record. An event is defined as a measurable behavior of an employee around a task at a unified point in time. Event fields include a unique employee identifier, a job identifier, a set of skill identifiers, a unified timestamp, and a business attribute vector. Records from different sources are mapped to a unified event quadruple format to generate a structured multi-source human resources data set. All event records in the structured multi-source human resources data set are then placed in a table according to the unique event identifier to establish a unified event-based multi-source human resources data set.

[0060] It should be noted that the event quadruple format refers to a set of unique personnel identifiers, job identifiers, skill identifiers, and a unified timestamp.

[0061] The event-based multi-source human resource data set is standardized and missing data is filled in to obtain a multi-source human resource data standard set. A quality score is calculated on the multi-source human resource data standard set to generate a structured initial event table.

[0062] Furthermore, based on the event-driven multi-source human resources data set, the data from all sources are unified in terms of fields, unified in encoding, numerical standardization, and timestamp formatting; the numerical fields in the business attribute vector are processed using an interval scaling method; the categorical fields are uniformly enumerated and mapped; the missing values ​​and incomplete records in the event-driven multi-source human resources data set are repaired using a data completion method; and the supplemented data from all sources are kept in a unified encoding and format to generate a multi-source human resources data standard set.

[0063] Based on a multi-source human resources data standard set, a data quality score is calculated for each event record using a weighted scoring function, and the multi-source human resources data standard set with quality scores is organized into a structured initial event table.

[0064] Perform field integrity checks, value validity checks, robust anomaly detection and replacement, quantile truncation standardization, missing data completion, consistency checks, and derived field generation on the structured event table to obtain an event table with auxiliary metrics.

[0065] Furthermore, in the structured event table, the required fields of each event record are checked to identify missing personnel unique identifiers, job identifiers, skill identifier sets, unified timestamps, and business attribute vectors. Missing event records are marked as to be completed. The validity of the value range of each field in the structured event table is verified, and data that fails the validity verification is removed.

[0066] Robust anomaly detection is performed on the structured event table after it has passed the validity check, and outliers in the robust anomaly detection are replaced by the average of adjacent time periods.

[0067] Perform quantile truncation standardization on numeric fields after replacing outliers, and then normalize the data within the truncated range.

[0068] For fields that still have missing values, a context-based completion method is used to complete the missing values.

[0069] After completing the missing information, the structured event table is then checked for consistency to remove any unreasonable event records introduced by the completion and abnormal replacement.

[0070] On the structured event table that has passed the consistency check, auxiliary metric fields are generated based on existing fields, including overtime rate, performance deviation value and task completion rate of personnel, to obtain the event table with auxiliary metrics.

[0071] The event tables with auxiliary metrics are filtered to obtain a set of standardized event tables and multidimensional tables that meet the required quality through weighted scoring.

[0072] Furthermore, by using a weighted scoring function, the quality score of each event record in the event table with auxiliary metrics is calculated, and a quality pass threshold is defined. When the quality score of the current event record is greater than the quality pass threshold, the current event record is retained and marked as a quality pass standardized event. All quality pass standardized events are counted to generate a quality pass standardized event table. Based on the quality pass standardized event table, information from different dimensions is extracted and a multidimensional table set is established.

[0073] It should be noted that the quality pass threshold is set based on the distribution of event record quality scores. The highest percentile of all event record quality scores is taken as the quality pass threshold value, and it is adjusted by combining percentile statistics with business rules.

[0074] The multidimensional table set includes personnel dimension tables, job dimension tables, skill dimension tables, and task dimension tables.

[0075] The personnel dimension table contains unique personnel identifiers and corresponding attribute information; the job dimension table contains unique job identifiers and corresponding business attributes; the skill dimension table contains skill identifiers and skill level information; and the task dimension table contains unique task identifiers and task-related information.

[0076] S2. Based on the standardized event table and multidimensional table set of quality compliance, calculate the edge weights between personnel and positions, personnel and skills, positions and tasks, and skills and tasks respectively, and construct a weighted relationship graph.

[0077] Based on a set of standardized event tables and multidimensional tables that meet quality standards, a composite index is created to obtain a set of standardized event tables and multidimensional tables with the composite index.

[0078] It should be noted that a composite index refers to a multi-field mapping structure established based on the key fields of each record in a high-quality standardized event table and a multidimensional table set, used for unique associations and fast retrieval between data records of different dimensions.

[0079] Furthermore, in the set of standardized event tables and multidimensional tables that meet quality standards, data fields that can be uniquely identified and efficiently linked are selected as the key fields of the composite index. The key fields of the composite index include unique event identifiers, unique personnel identifiers, unique job identifiers, unique skill identifiers, and unique task identifiers.

[0080] Use the event's unique identifier as the primary key.

[0081] Each unique identifier for a person, position, skill, and task is mapped one-to-one with the records in the corresponding personnel dimension table, position dimension table, skill dimension table, and task dimension table.

[0082] During the mapping process, a foreign key reference pointing to each dimension table is established for each event record, and a mapping relationship between the standardized event table and the multidimensional table set is established.

[0083] After completing the field mapping, based on the selected key fields, a hash index is used to construct a composite index structure with the event unique identifier as the primary key in the main index layer and the combined information of each dimension table recorded in the secondary index layer.

[0084] The composite index structure is combined with the standardized event table and stored in personnel dimension tables, job dimension tables, skill dimension tables, and task dimension tables to obtain a set of standardized event tables and multidimensional tables with composite indexes.

[0085] Based on a standardized event table with a composite index and a multidimensional table set, calculate the edge weights of personnel and positions, and generate a set of personnel and position edges.

[0086] Furthermore, based on a standardized event table and multidimensional table set with a composite index, within a unified time window, business events are grouped and summarized according to the co-occurrence relationship between the unique identifier of the personnel and the unique identifier of the position, generating a candidate list of personnel and positions.

[0087] The candidate lists of personnel and positions are combined with the standardized event table with joint index and the skill dimension table and position dimension table in the multidimensional table set. Based on the combination of each person's unique identifier and the unique identifier of the position, the intersection and union of the skill identifier set and the coverage ratio are calculated to generate a skill coverage metric table.

[0088] The system combines the timestamp information from the standardized event table with a joint index and the multidimensional table set with the candidate lists of personnel and positions. Based on the unique identifiers of personnel and positions, it combines the number of times the personnel have recently participated in position-related tasks with the time of the most recent occurrence. It then weights the earlier events according to the time decay weight to generate an experience timeliness measurement table.

[0089] By combining the candidate lists of personnel and positions with the standardized event table and multidimensional table set containing the joint index, the performance evaluation-related fields and task quality-related fields are combined. The standardized mean, stability, and timeliness indicators of task quality are calculated based on the combination of personnel unique identifier and position unique identifier, and a historical performance measurement table is generated.

[0090] It should be noted that the stability is calculated and expressed as: ;

[0091] in, Indicates stability. Indicates the number of tasks. Indicates the first The standardized quality value of each task. This represents the average quality of the tasks.

[0092] It should be noted that the standardized quality value of the task is obtained by extracting all original task quality records under the same combination of unique personnel identifier and unique job identifier, and using the range normalization method to uniformly map the task quality values ​​of different dimensions to the range of [0,1].

[0093] It should be noted that the calculation of the timeliness indicator specifically involves filtering task records from the standardized event table and multidimensional table set with composite indexes, based on the unique identifier of the personnel and the unique identifier of the position, and extracting the start timestamp of each task. and end timestamp Extract the planned completion time of the task from the task dimension table. The system calculates the delay based on available fields, compares the delay with zero to obtain the non-negative delay. When the delay is greater than zero, it is taken as the non-negative delay; when the delay is less than or equal to zero, it is taken as zero as the non-negative delay. Using exponential compression, the non-negative delay is mapped to a single task timeliness score. The task importance weight is obtained by counting the number of times the task is covered in the event records from the standardized event table. The timeliness index is obtained by weighted summarization based on the task importance weight.

[0094] Specifically, the delay is calculated based on available fields, meaning that when the task dimension table contains a planned completion time field... The delay is defined as: ;

[0095] When the task dimension table does not have a planned completion time field, but has an expected duration field. First, calculate the actual duration. Let's redefine the delay as: ;

[0096] When the task dimension table contains both a planned completion time field and an expected duration field, the planned completion time field should be used first.

[0097] The skill coverage metric, experience timeliness metric, and historical performance metric are normalized and weighted to generate a personnel and job weight table.

[0098] Sort the combinations in the personnel and job edge weight table from high to low and write them into the personnel and job structured result to generate the personnel and job edge set; the fields of the personnel and job structured result include personnel unique identifier, job unique identifier, and personnel and job edge weight.

[0099] By combining the standardized event table with composite index and the multidimensional table set with the personnel and job edge set, the personnel and skill edge weights are calculated, and the personnel and skill edge set is obtained.

[0100] Furthermore, the standardized event table with composite index is associated with the personnel and job edge set of the multidimensional table set. Within a unified time window, the job skill set is located based on the unique identifier of the personnel and the unique identifier of the job. The job skill set is merged with the unique identifier of the skills appearing in the event to generate a candidate list of personnel and skills. Based on the candidate list of personnel and skills, in the standardized event table with composite index and the multidimensional table set, the number of tasks completed, the number of task types covered, and the mean and variance of quality indicators related to the skill for each record in the candidate list of personnel and skills are counted according to the unique identifier of the personnel and skill. This generates a frequency quality metric table of personnel and skills.

[0101] Based on the personnel and skill frequency quality metric table, the system extracts the most recent occurrence time using a standardized event table with a composite index and a unified timestamp in a multidimensional table set, according to the unique identifier of the personnel and the unique identifier of the skill. It then calculates the activity density and time decay weighted frequency of the recent time window to generate the personnel and skill timeliness metric table.

[0102] It should be noted that the activity density for the recent time window is calculated as follows: ;

[0103] in, Indicates the activity density within a recent time window. This indicates the number of events recorded for personnel and skills within a time window. Indicates the length of the time window.

[0104] It should be noted that the number of event records is obtained by extracting all event records corresponding to the unique identifiers of personnel and skills, and then counting the total number of all event records.

[0105] The time-decay weighted frequency is calculated and expressed as: ;

[0106] in, Indicates the time-decay weighted frequency. This represents a set of historical event timestamps for personnel and skills. A specific timestamp representing a set of historical event timestamps for personnel and skills. Represents the current unified timestamp, This represents the time decay coefficient.

[0107] It should be noted that the historical event timestamp set for personnel and skills contains records of all time events that occurred with each person and skill. The current unified timestamp is used to calculate the time interval between historical events. Validation settings are configured based on historical event data. The value of satisfies the condition of maximizing prediction accuracy and minimizing error. It is a positive real number used to control the decay rate.

[0108] Based on the frequency and quality metrics and the timeliness metrics for personnel and skills, quantile truncation and interval normalization are performed on the three categories of indicators: frequency, quality, and timeliness. A personnel and skill proficiency metric is generated through weighted fusion. In the personnel and skill proficiency metric, each personnel and skill combination is mapped to a set of positions related to the personnel. A weighted average is then performed based on the edge weights in the personnel and position edge sets to generate a personnel and skill position relevance metric. A personnel and skill edge weight table is generated through fusion metric. The personnel and skill edge weight table is sorted from high to low according to the personnel and skill edge weights to generate a personnel and skill edge set.

[0109] It should be noted that the frequency index refers to the number of times the combination of personnel unique identifier and skill unique identifier appears in the standardized event table within a unified time window; the quality index refers to the statistical analysis of the performance of the combination of personnel unique identifier and skill unique identifier based on the average and stability of task quality; and the timeliness index refers to the measurement of the effectiveness of skill use of the combination of personnel unique identifier and skill unique identifier by combining the most recent occurrence time, activity density within the recent time window, and time decay weighted frequency.

[0110] It should be noted that frequency, quality, and timeliness are truncated separately. Specifically, the upper and lower quantiles of each of the three indicators are taken. When the value of the three indicators is less than the lower quantile of the current indicator, the indicator value of the current indicator is adjusted to the lower quantile. When the value of the three indicators is greater than the upper quantile of the current indicator, the indicator value of the current indicator is adjusted to the upper quantile. This process continues until the effective range of the current indicator set is limited to the corresponding upper and lower quantiles.

[0111] The upper and lower quantiles of the three types of indicators are set according to the actual data distribution characteristics and adjusted according to robust statistical principles.

[0112] It should be noted that the personnel and skill edge set includes a unique personnel identifier, a unique skill identifier, and personnel and skill edge weights.

[0113] By combining the standardized event table with composite index and the multidimensional table set with the personnel and skill edge set, the edge weights of positions and tasks are calculated, and the edge set of positions and tasks is obtained.

[0114] Furthermore, based on a standardized event table and a multidimensional table set with a composite index, all event records in the standardized event table are retrieved within a unified time window. The unique identifiers for positions and tasks are extracted from each event record. These unique identifiers are then grouped and statistically analyzed according to their combinations. If the unique identifier for a current position and the unique identifier for a current task appear at least once in the event record, a co-occurrence relationship is determined. Grouping is then performed based on the co-occurrence relationship between the unique identifiers for positions in the position dimension table and the unique identifiers for tasks in the task dimension table. Occasional combinations are then filtered out based on the number of event records, generating a candidate list of positions and tasks.

[0115] This approach combines a candidate list of job positions and tasks, a task dimension table, and a standardized event table with a composite index with a set of multidimensional tables. Tasks are grouped according to their unique identifiers in the task dimension table. Within each event record corresponding to a unique task identifier, a set of skill identifiers related to that unique task identifier is extracted and statistically analyzed. The frequency of each skill identifier appearing in the event record corresponding to the unique task identifier is counted to obtain the skill demand frequency. This frequency is then truncated and normalized. Combined with the task quality indicators in the standardized event table with a composite index, the average task quality and stability are calculated for the combination of skill identifiers and task identifiers. The average task quality and stability are mapped to skill importance. Finally, the skill demand frequency and skill importance are weighted and fused to generate a task skill demand metric table.

[0116] Specifically, the generation of a job skills supply metric table involves extracting and statistically analyzing the set of unique personnel identifiers related to the job identifiers in the job dimension table. This set of unique personnel identifiers is then associated with the set of personnel and skills edges. The unique skill identifiers and edge weights corresponding to each unique personnel identifier in the set of personnel and skills edges are obtained. Using the job identifier as the aggregation dimension, the skill edge weights corresponding to all unique personnel identifiers associated with the current job identifier are accumulated and weighted averaged to obtain the overall supply capacity of the job at the skill level. Combined with the distribution of skill edge weights, a job skills supply metric table is generated.

[0117] Align the job skill supply metric table and the task skill demand metric table in the skill identification dimension. Obtain the coverage ratio, important skill hit rate and intensity similarity through the unique job identifier and task identifier to obtain the job and task skill matching degree matrix. Extract the most recent occurrence time from the job and task candidate list, calculate the activity density and task urgency, and generate the job and task time decay factor and priority coefficient to obtain the job and task timeliness and priority modulation coefficient table.

[0118] Coverage ratio refers to the proportion of the skill set included in the job skill supply metric table that covers the skill set required in the task skill demand metric table. Obtaining the hit rate of important skills involves sorting skills by importance from high to low in the task skill demand metric table, defining a critical value for skill importance, selecting skills with importance greater than the critical value and statistically representing them as a set of skill identifiers, and using the proportion of this set of skill identifiers covered by the job skill supply metric table as the hit rate of important skills. The critical value for skill importance is determined using quantile statistics based on the distribution of skill importance values ​​corresponding to the unique identifiers of all skills in the task skill demand metric table. The setting of the critical value for skill importance ensures that the set of skills identified as key skills remains stable in quantity and reflects the actual demand intensity of the task at the key skill level. Intensity similarity refers to the degree of similarity between the supply intensity of each skill in the job skill supply metric table and the corresponding skill demand intensity in the task skill demand metric table, used to measure the rationality of the match between the strength and weakness of task skill requirements beyond the quantity of skills.

[0119] It should be noted that the active density is calculated as follows: ;

[0120] in, Indicates active density, This indicates the number of co-occurring event records of the unique identifier of the job and the unique identifier of the task within the time window.

[0121] It should be noted that the calculation of task urgency is as follows: the planned completion time of the unique identifier of the task is obtained from the task dimension table, the remaining time is calculated by combining it with the current unified timestamp, the remaining time is mapped to an urgency factor, the number of event records of the unique identifier of the task is counted within the same time window and normalized to obtain an activity factor, and the urgency factor and the activity factor are harmonic averaged to obtain the task urgency.

[0122] It should be noted that the generation of job and task time decay factors and priority coefficients is specifically as follows: within a unified time window, based on the combination of unique job identifiers and unique task identifiers in the job and task candidate lists, the most recent occurrence time is extracted. The time interval between job and task combinations is calculated by combining a unified timestamp. The time interval is mapped through an exponential decay function to obtain the job and task time decay factor. The job and task time decay factor is used to reflect the decay trend of the matching degree of job and task combinations over time. Based on the task urgency, the task urgency is fused with the job and task skill matching degree matrix. The task urgency of different tasks is compared, and the unique identifier of the task with higher task urgency is assigned a higher priority weight to obtain the priority coefficient.

[0123] In terms of unique job identifiers and unique task identifiers, the job and task skill matching matrix, job and task timeliness, and priority modulation coefficient table are aligned, and a functional job and task edge weight table is generated through fusion measurement.

[0124] The job and task edge weight table is sparsified and sorted from high to low according to the job and task edge weights, and then written into the job and task structured result. The job and task structured result includes the unique identifier of the job, the unique identifier of the task, and the job and task edge weights, generating a set of job and task edges.

[0125] The standardized event table with composite index is combined with the multidimensional table set, along with the job and task edge set, to calculate the skill and task edge weights and obtain the skill and task edge set.

[0126] Furthermore, the standardized event table with composite index is associated with the multidimensional table set and the job and task edge set. Using the unique task identifier as the primary key, within a unified time window, the skill identifier set in the task dimension table is extracted. The task occurrence records in the standardized event table with composite index are used for verification, occasional task combinations are eliminated, and a list of skill and task candidates is generated.

[0127] Based on the candidate lists of skills and tasks, in the standardized event table and multidimensional table set with composite index, groups will be established according to the combination of unique skill identifier and unique task identifier. Event records belonging to the same combination of unique skill identifier and unique task identifier will be assigned to the same group. Within each group, the number of times the skill appears in the task, the number of events covered, and the quality indicators related to the skill will be counted to generate a frequency and quality metric table for skills and tasks.

[0128] By weighting and aligning the skill and task frequency quality metric table with the job and task edge set, and using the job and task edge weights as the cumulative calculation of task-side strength, the task skill requirement metric table is updated. Based on a standardized event table and multidimensional table set with a joint index, within a unified time window, the global occurrence frequency and recent activity of each skill unique identifier in all tasks are statistically analyzed to construct skill scarcity and skill and task time decay factors, which are then merged into a skill scarcity and timeliness modulation coefficient table. In the dimensions of skill unique identifier and task unique identifier, the task skill requirement metric table is aligned with the skill scarcity and timeliness modulation coefficient table. A skill and task edge weight table is generated by fusing metrics. After sparsifying the skill and task edge weight table, it is sorted from high to low according to the skill and task edge weights to generate a skill and task edge set.

[0129] It should be noted that the construction of skill scarcity and skill-task time decay factors, and their merging into a skill scarcity and time-sensitivity modulation coefficient table, involves the following steps: Using the unique skill identifier as the grouping dimension, the number of event records for each unique skill identifier across all unique task identifiers is counted, and the global frequency of occurrence of the unique skill identifier is calculated. The reciprocal of the global frequency is taken as the skill scarcity, used to measure the scarcity of the current unique skill identifier across all unique task identifier sets. Within a unified time window, the timestamps of event records for each combination of unique skill identifier and unique task identifier in the standardized event table with a composite index are retrieved. The time of the most recent event record is extracted, and the time interval is calculated by combining it with the unified timestamp. The time intervals are then weighted and summed using an exponential decay function to obtain the skill-task time decay factor. The skill scarcity and the skill-task time decay factor are then merged to generate the skill scarcity and time-sensitivity modulation coefficient table.

[0130] It should be noted that the quality indicators related to skills include the average and stability of task quality. The skill-task edge set includes the unique identifier of the skill, the unique identifier of the task, and the edge weight of the skill and task. The global frequency of each unique identifier of the skill in all tasks is recorded as the global frequency of the skill in the task set.

[0131] The sets of personnel and job positions, personnel and skills, job positions and tasks, and skills and tasks are normalized and trimmed, and then combined with multidimensional table sets to construct a weighted relationship graph.

[0132] Furthermore, an edge weight pruning threshold is set. After interval normalization of the personnel-position edge set, personnel-skill edge set, position-task edge set, and skill-task edge set, pruning is performed according to the edge weight pruning threshold. Weakly related edge records below the edge weight pruning threshold are removed. The retained personnel unique identifier, position unique identifier, skill unique identifier, and task unique identifier are combined with the personnel dimension table, position dimension table, skill dimension table, and task dimension table in the multidimensional table set, respectively, to establish a cross-dimensional index relationship. Using the multidimensional table set as the node set and the normalized and pruned edge set as the edge set, a weighted relationship graph is constructed.

[0133] It should be noted that the edge weight pruning threshold is set using the quantile truncation method. The setting of the edge weight pruning threshold satisfies the sparsity and efficiency of the weighted relation graph. The weighted relation graph includes four types of nodes: personnel, positions, skills, and tasks. Weighted connections are established through personnel-position edges, personnel-skill edges, positions-task edges, and skills-task edges, respectively.

[0134] S3. Transform the weighted relationship graph into a sparse tensor. Considering multiple objective factors such as job competency, collaboration effect, human resource cost, and time stability, use iterative normalization and sparsification methods to generate a job assignment matrix and a sensitivity scoring matrix.

[0135] An index mapping is established based on a weighted relational graph, an index mapping set is obtained, and the index mapping set and the timestamp information in the weighted relational graph are divided into discrete time windows to generate a time window set.

[0136] Furthermore, sequential indexes are established for the unique identifiers of personnel, positions, skills, and tasks in the weighted relation graph. The generated personnel index table, position index table, skill index table, and task index table are combined into an index mapping set. Based on the unified timestamp of the event records in the weighted relation graph, the continuous time axis is divided into non-overlapping intervals with a fixed granularity. A corresponding discrete time window label is generated in each interval. The index mapping set is bound to the time window label to generate a time window set.

[0137] It should be noted that the time window set includes index information from the index mapping set and time slice information from the discrete time window labels.

[0138] By aggregating the time window set and the job and task edge set and skill and task edge set in the weighted relationship graph, a task requirement weight table is generated.

[0139] Furthermore, based on the time window set, each discrete time window is aggregated with the set of job and task edges and the set of skill and task edges in the weighted relationship graph. Within the same time window, the edge weights of the unique identifiers of job and task are accumulated to obtain the cumulative weight value of job and task. The edge weights of the unique identifiers of job and task are combined with the edge weights of the unique identifiers of skills and tasks under the corresponding tasks and then weighted and averaged. The cumulative weight values ​​of job and task are merged according to the unique identifier of task to obtain a multi-dimensional measure of the demand intensity, skill coverage and job relevance of the recorded task in different time windows, and a task demand weight table is generated.

[0140] It should be noted that the task requirement weight table uses the task's unique identifier as the primary key.

[0141] The task requirement weight table is integrated with the personnel and job edge set and personnel and skill edge set in the weighted relationship graph and filtered. After normalization, a sparse coordinate list is generated. The sparse coordinate list is then combined with the index mapping set and the time window set to generate a personnel-job-skill-time sparse tensor.

[0142] Furthermore, the task requirement weight table is fused with the personnel and job edge set and the personnel and skill edge set in the weighted relationship graph. The candidate association set is obtained by filtering according to the time window index in the task requirement weight table. The weight values ​​in the candidate association set are normalized to obtain a sparse coordinate list. The sparse coordinate list, index mapping set and time window set are combined and mapped, and uniformly encoded into a sparse tensor format to generate a personnel-job-skill-time sparse tensor.

[0143] It should be noted that during the fusion calculation process, only personnel-job-skill combinations that simultaneously meet the job requirement weight and skill requirement coverage are retained.

[0144] It should be noted that the filtering is based on the time window index in the task requirement weight table. Specifically, when merging the task requirement weight table with the personnel and job side set and the personnel and skill side set, only records whose time window index matches the current processing time window are retained, and the task requirement filtering range is controlled by the granularity of the time window.

[0145] Expand the sparse tensor of personnel-position-skill-time and initialize it with a uniform probability distribution to obtain the initial personnel-position assignment matrix. Combine the initial personnel-position assignment matrix with the personnel and position edge set to calculate the position competency benefit and generate the position competency matrix.

[0146] Furthermore, based on the personnel-position-skill-time sparse tensor, the candidate space of the personnel-position assignment matrix is ​​generated by expanding each time window in terms of personnel and position dimensions. The assignment probability matrix is ​​initialized to a uniform distribution to generate an initial personnel-position assignment matrix. The initial personnel-position assignment matrix is ​​then fused with the personnel and position edge set to calculate the position competency benefit. All position competency benefits are then summarized to generate a position competency matrix.

[0147] By combining the job competency matrix with the personnel-skill edge set and the skill-task edge set, the collaborative effect benefit is calculated, and a collaborative effect matrix is ​​generated.

[0148] Furthermore, by combining the job competency matrix with the personnel-skill edge set and the skill-task edge set, collaborative skill pairs are extracted. Through the mutual information approximation method, the collaborative effect benefits between personnel are calculated, and the collaborative effect benefits between all personnel are statistically analyzed to generate a collaborative effect matrix.

[0149] It should be noted that the benefit of collaboration among personnel is calculated using the mutual information approximation method, and is expressed as: ;

[0150] in, This indicates the benefits of collaboration among individuals. Indicates the first One skill, This represents the total number of skills. The weights of personnel and skills are represented and extracted from the set of personnel and skills edges. This indicates the intensity of skill requirements for a given position, obtained by generating a skill and task edge weight table through fusion metrics. This indicates the global frequency of a skill within a set of tasks.

[0151] By combining the collaboration effect matrix with the personnel dimension table and the job dimension table, the labor cost penalty is calculated, and a labor cost matrix is ​​generated.

[0152] Furthermore, the salary coefficient, working hour utilization rate, and experience level of each person are extracted from the personnel dimension table, and the budget constraint and job level weight of each job are extracted from the job dimension table. Based on the collaboration effect matrix, the salary coefficient of each person is compared with the budget constraint of the job to obtain the salary matching difference. The working hour utilization rate of each person is weighted with the job level weight to obtain the job load adjustment factor. By weighting and fusing the salary matching difference and the job load adjustment factor, the benefit value in the collaboration effect matrix is ​​penalized and corrected to generate the human resource cost matrix.

[0153] By combining the human resource cost matrix with the personnel and job assignment matrix of the previous time window in the historical time window, the time stability penalty is calculated, and a time stability matrix is ​​generated.

[0154] Furthermore, using the personnel-job assignment matrix of the previous time window in the historical time window as a reference, the personnel and job allocation results in the current human resource cost matrix are compared one by one. A stability weight parameter is defined. When the personnel-job assignment results of the same person in the previous time window and the current time window are consistent, the time stability penalty value is set to zero. When the personnel-job assignment results of the same person in the previous time window and the current time window are inconsistent, the time stability penalty value is calculated according to the stability weight parameter and the indicator function method. All time stability penalty values ​​are counted to generate a time stability matrix.

[0155] It should be noted that the stability weight parameter is calculated by using the correlation coefficient between job switching and performance decline in historical personnel and job allocation data.

[0156] The time stability matrix, job competency matrix, collaboration effect matrix, and human resource cost matrix are weighted and combined. The resulting comprehensive benefit matrix is ​​iteratively normalized and sparsified to generate the job assignment matrix.

[0157] Furthermore, the time stability matrix, job competency matrix, collaboration effect matrix, and human resource cost matrix are weighted and superimposed element by element under the same dimension to generate a comprehensive benefit matrix. The comprehensive benefit matrix is ​​then iteratively normalized by using the Sinkhorn-Knopp iterative method to normalize the rows and columns of the matrix respectively, and sparsification is applied to the comprehensive benefit matrix to generate a job assignment matrix.

[0158] Based on the personnel-job assignment matrix, the degree of change of the comprehensive benefit matrix due to marginal disturbances is calculated, and the sensitivity score matrix is ​​obtained.

[0159] Furthermore, based on the personnel-job assignment matrix, marginal perturbations are applied to the element positions of the personnel-job assignment matrix corresponding to personnel and jobs. The degree of change of the marginal perturbation on the comprehensive benefit matrix is ​​calculated by using the finite difference method to obtain the sensitivity score matrix.

[0160] It should be noted that the sensitivity score matrix is ​​obtained by calculating the degree of change in the comprehensive return matrix caused by marginal disturbances using the finite difference method, and is expressed as follows: ;

[0161] in, Represents the sensitivity rating matrix. Representing the comprehensive return matrix, This represents the personnel-job assignment matrix. Indicates marginal disturbance. This represents the unit perturbation matrix, which is only present at position. The value is 1. Indicates the first Individuals, Indicates the first One position, Indicates the first The person and the first The position of each job in the personnel-job assignment matrix.

[0162] S4. Generate task performance signals through the job assignment matrix, provide dynamic feedback on the task performance signals, identify abnormal situations using an online drift detection method, select a candidate set based on the sensitivity scoring matrix to perform local re-optimization, and update the job assignment matrix.

[0163] By combining the personnel-job assignment matrix, job dimension table, task dimension table, and standardized event table for quality compliance, a task performance signal table is generated.

[0164] Furthermore, the personnel-job assignment matrix is ​​aligned with the job dimension table and the task dimension table to obtain the correspondence between personnel and tasks. The correspondence between personnel and tasks is then associated with the standardized event table of qualified quality. The task completion time, task quality score, and personnel workload ratio of each task under the corresponding personnel assignment are extracted. The task performance signal value is calculated through a weighted function to generate a task performance signal table.

[0165] The task performance signal table is compared with the historical baseline within a sliding time window to generate a feedback indicator table.

[0166] Furthermore, based on the task performance signal table, it is divided into continuous sliding time windows according to timestamps. Within each sliding time window, the historical baseline of the task performance signal is calculated. The current task performance signal is compared with the historical baseline of the task performance signal, and the difference between the task performance signal within the sliding time window and the historical baseline of the task performance signal is calculated. Feedback indications are generated according to the positive or negative direction of the difference between the task performance signal within the sliding time window and the historical baseline of the task performance signal. When the difference is positive, the corresponding task of the current task performance signal is marked as performance improvement. When the difference is negative, the corresponding task of the current task performance signal is marked as performance decline. When the difference is zero, the corresponding task of the current task performance signal is marked as performance stability. At the task unique identifier dimension, the corresponding task markings of all task performance signals are summarized to generate a feedback indication table.

[0167] It should be noted that the historical baseline of the task performance signal includes the number-weighted average and standard deviation of the task performance signal.

[0168] The task performance signal table and feedback indication table are merged to generate a task performance dynamic feedback record table.

[0169] Furthermore, in terms of the unique identifier dimension of tasks, the task performance signal table and feedback indication table are aligned. Within the sliding time window, the task performance signal of each task is matched with the corresponding feedback indication, and a task record entry is established for each task. Under a unified time series, the record entries of all tasks are integrated to generate a dynamic feedback record table for task performance.

[0170] It should be noted that the task record entries include timestamps, task performance signal values, and corresponding feedback instructions, which are extracted from the feedback instruction table.

[0171] Based on the task performance dynamic feedback record table, an abnormal task set is obtained through online drift detection method. The abnormal task set is then combined with a sensitivity scoring matrix for screening to obtain a candidate job matching set.

[0172] Furthermore, the time series of task performance signals is extracted from the task performance dynamic feedback record table. The ADWIN method is used to perform online drift detection on the time series of task performance signals. When the distribution characteristics of the time series of task performance signals change abruptly, the corresponding task is marked as an abnormal task, and an abnormal task set is generated. A sensitivity threshold is defined. When the sensitivity score matrix is ​​greater than the sensitivity threshold, the corresponding person-job pair is retained. All person-job pairs are counted to generate a candidate person-job pair set.

[0173] Among them, the person-job pairing refers to the correspondence between personnel and positions, representing the position of each element in the person-job assignment matrix.

[0174] It should be noted that the sensitivity threshold is set by calculating the mean and standard deviation of the historical sensitivity rating matrix, and then summing the mean and standard deviation of the historical sensitivity rating matrix to obtain the sensitivity threshold value.

[0175] By combining the candidate job pair set and the comprehensive benefit matrix, a local objective function is constructed and solved. The local objective function table is iteratively normalized and sparsified to generate an updated job assignment matrix. The updated job assignment matrix is ​​compared with the job assignment matrix to generate a job assignment update record table.

[0176] Furthermore, based on the candidate-job pair set, corresponding elements in the comprehensive benefit matrix are extracted in the dimension of personnel-job correspondence, a local objective function is constructed, and a local objective function table is generated. Iterative normalization and sparsification operations are performed on the local objective function table to eliminate redundant allocation relationships and generate an updated personnel-job assignment matrix. The updated personnel-job assignment matrix is ​​compared with the original personnel-job assignment matrix to extract the changed personnel-job correspondences, which are recorded according to timestamps to generate a personnel-job assignment update record table.

[0177] This embodiment also provides a computer device applicable to the data-driven human resource organization management method, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the data-driven human resource organization management method proposed in the above embodiment.

[0178] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0179] This embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements the data-driven human resource organization and management method proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0180] In summary, this invention achieves a unified representation of multidimensional data relationships and a cross-dimensional overall optimization expression through an improved multi-objective optimization assignment based on sparse tensors, thereby enhancing the scientificity and comprehensiveness of person-job matching. Through the person-job assignment matrix and sensitivity scoring matrix, it realizes real-time response to task performance signals and multi-objective balanced optimization in dynamic environments.

[0181] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A data-driven based human resource organization management method, characterized in that: The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system.

2. The data-driven based human resource organization management method of claim 1, wherein: The application relates to a human resource management method and system.

3. The data-driven based human resource organization management method of claim 2, wherein: The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system.

4. The data-driven based human resource organization management method of claim 3, wherein: The application relates to a human resource management method and system.

5. The data-driven based human resource organization management method of claim 4, wherein: The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system.

6. The data-driven based human resource organization management method of claim 5, wherein: The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management method and system. The application relates to a human resource management The standardized event table with joint index is combined with the multi-dimensional table set, the post and task edge set, the skill and task edge set is calculated, and the skill and task edge set is obtained; The personnel and post edge set, the personnel and skill edge set, the post and task edge set, and the skill and task edge set are normalized and cut, and are combined with the multi-dimensional table set to construct a weighted relationship graph.

7. The data-driven based human resource organization management method of claim 6, wherein: The weighted relationship graph is converted into a sparse tensor, and the steps are as follows, Based on the weighted relationship graph, an index mapping is established, an index mapping set is obtained, and the index mapping set and the timestamp information in the weighted relationship graph are divided into discrete time windows to generate a time window set; By aggregating the time window set, the post and task edge set and the skill and task edge set in the weighted relationship graph, a task demand weight table is generated; The task demand weight table is fused with the personnel and post edge set and the personnel and skill edge set in the weighted relationship graph and is screened, a sparse coordinate list is generated after normalization, and the sparse coordinate list is combined with the index mapping set and the time window set to generate a personnel-post-skill-time sparse tensor.

8. The data-driven based human resource organization management method of claim 7, wherein: The multi-objective factors of post competence, cooperation effect, human cost and time stability are integrated, and an iterative normalization and sparsification method is used to generate a person-post assignment matrix and a sensitivity score matrix, and the steps are as follows, The personnel-post-skill-time sparse tensor is unfolded and initialized with a uniform probability distribution to obtain an initial person-post assignment matrix, and the initial person-post assignment matrix is combined with the personnel and post edge set to calculate the post competence benefit to generate a post competence matrix; The post competence matrix is combined with the personnel and skill edge set and the skill and task edge set to calculate the cooperation effect benefit to generate a cooperation effect matrix; The cooperation effect matrix is combined with the personnel dimension table and the post dimension table to calculate the human cost penalty to generate a human cost matrix; The human cost matrix is combined with the person-post assignment matrix of the previous time window in the historical time window to calculate the time stability penalty to generate a time stability matrix; The time stability matrix, the post competence matrix, the cooperation effect matrix and the human cost matrix are combined with weights, the generated comprehensive benefit matrix is iteratively normalized, and sparsification processing is performed to generate a person-post assignment matrix; Based on the person-post assignment matrix, the degree of change of the marginal disturbance to the comprehensive benefit matrix is calculated to obtain a sensitivity score matrix.

9. The data-driven based human resource organization management method of claim 8, wherein: The task performance signal is generated by the person-post assignment matrix, and the steps of dynamic feedback of the task performance signal are as follows, The task performance signal table is generated by combining the person-post assignment matrix, the post dimension table, the task dimension table and the standardization event table with qualified quality; The task performance signal table is compared with the historical baseline in a sliding time window to generate a feedback indication table; The task performance signal table and the feedback indication table are merged to generate a task performance dynamic feedback record table.

10. The data-driven based human resource organization management method of claim 9, wherein: The online drift detection method is used to identify abnormal situations, the candidate set is selected according to the sensitivity score matrix to perform local re-optimization, and the person-post assignment matrix is updated, and the steps are as follows, Based on the task performance dynamic feedback record table, the abnormal task set is obtained by the online drift detection method, and the abnormal task set is combined with the sensitivity score matrix to screen the candidate person-post pair set; The local objective function is constructed and solved in combination with the candidate post set and the comprehensive benefit matrix, the local objective function table is iteratively normalized and thinned, an updated post assignment matrix is generated, and a post assignment update record table is generated by comparing the updated post assignment matrix with the post assignment matrix.

Images