Learning task pushing method and device, equipment and storage medium

Abstract

The application relates to the technical field of logistics transportation, and particularly discloses a learning task pushing method and device, equipment and a storage medium, the method comprises the following steps: obtaining configuration information of at least one learning task, the configuration information comprising a trigger type, a trigger rule and a target object screening condition; obtaining business data by using a data processing mode corresponding to the trigger type, and determining a target object set according to the target object screening condition when it is determined that the business data meets the trigger rule; wherein the data processing mode comprises a real-time processing mode; generating a pushing message of the learning task, and sending the pushing message to a terminal of each target object in the target object set of the learning task. The scheme provided by the application can push learning tasks in real time, push corresponding learning tasks in time when employees have work problems or post scenarios are updated, efficiently process massive logistics business real-time data, and reduce data accumulation and processing lag problems.

Description

Technical Field

[0001] This application relates to the field of logistics management technology, and in particular to a learning task push method, apparatus, equipment and storage medium. Background Technology

[0002] With the deep integration of internet technology and online education, online learning platforms are widely used in corporate training, school education, and personal skills enhancement. The logistics industry is experiencing rapid technological iteration, with smart devices and new processes emerging constantly. Currently, it faces a prominent contradiction between the efficiency of talent development and the speed of business growth, necessitating the use of online learning platforms to push learning tasks to employees to meet the rapidly evolving business needs of modern logistics companies.

[0003] Currently, most existing methods for pushing learning tasks rely on offline data statistics. Specifically, the system typically pulls previous business data at set time points, performs offline cleaning and rule matching on the full or incremental data, and then generates corresponding learning tasks and pushes them to target employees. Alternatively, operations personnel may need to manually select target users and periodically import learning tasks for push notifications.

[0004] However, existing methods for pushing learning tasks suffer from poor real-time performance, failing to meet the dynamic empowerment needs of logistics positions. Specifically, the offline, periodic batch processing mode cannot capture dynamic changes in logistics business data in real time. Dynamic data such as employee errors, new job rotations, new business operation scenarios, and temporary performance failures cannot be collected and analyzed in a timely manner. This results in a lag between the pushed learning tasks and the current job learning needs and operational shortcomings of employees, leading to a decrease in the relevance of training and empowerment. Summary of the Invention

[0005] To address or partially address the problems existing in related technologies, this application provides a learning task push method, apparatus, device, and storage medium. It utilizes a real-time processing mode to generate business event streams based on business data. A first processing task consumes the real-time business event stream based on trigger rules to obtain a hit event stream. A second processing task consumes the hit event stream based on target object filtering conditions to obtain a set of target objects for each hit event corresponding to a learning task. The first and second processing tasks are executed in parallel, enabling real-time push of learning tasks. This allows for the immediate push of corresponding learning tasks when employees encounter operational problems or when job scenarios are updated. Furthermore, it can efficiently handle massive amounts of real-time logistics business data, reducing data accumulation and processing bottlenecks.

[0006] The first aspect of this application provides a method for pushing learning tasks, including: obtaining configuration information of at least one learning task, wherein the configuration information includes trigger type, trigger rules and target object filtering conditions;

[0007] The process involves acquiring business data using the data processing mode corresponding to the trigger type, and determining a target object set based on the target object filtering conditions when the business data satisfies the trigger rule. The data processing mode includes a real-time processing mode. The steps of acquiring business data using the data processing mode corresponding to the trigger type and determining the target object set based on the target object filtering conditions when the business data satisfies the trigger rule include: When the data processing mode corresponding to the trigger type is the real-time processing mode, the business data is acquired in real time; The business events in the business data are written to the first topic in real time to form the business event stream; A first global broadcast stream is generated based on the triggering rules; A second global broadcast stream is generated based on the target object filtering conditions; In the first processing task, the first global broadcast stream and the first topic are consumed based on the triggering rules, and the hit events that satisfy the triggering rules are written into the second topic; In the second processing task, the second global broadcast stream and the second topic are consumed based on the target object filtering conditions, and the target object set of the learning task corresponding to each hit event that meets the target object filtering conditions is written into the third topic; The first processing task and the second processing task are executed in parallel; Generate a push message for the learning task and send the push message to the terminal of each target object in the target object set of the learning task.

[0008] In conjunction with the first aspect, in one possible implementation of the first aspect, the triggering rule includes at least one of the following: single event threshold triggering rule, time-series window aggregation triggering rule, multi-dimensional composite condition triggering rule, baseline deviation triggering rule, and fixed triggering rule; In the single-event threshold triggering rule, the triggering condition is whether the indicator of a single business event exceeds a set threshold or is a preset value. In the time-series window aggregation triggering rule, the statistical data of continuous preset types of business events within a pre-set sliding time window are used as the triggering condition; In the multi-dimensional composite condition triggering rule, the logical relationships of AND, OR, and NOT are nested and combined to form the triggering condition, and the triggering condition includes multiple condition fields; In the baseline deviation triggering rule, the target object's personal historical operation error rate or the average operation error rate of the same position is used as the baseline. When it is determined from the target object's real-time operation error rate that it deviates from the baseline by a set amount based on the target object's real-time business data, the learning task is triggered. In the fixed triggering rule, a preset learning task is pushed to a preset target object according to a preset push cycle.

[0009] In conjunction with the first aspect, in one possible implementation of the first aspect, the data processing mode further includes an offline processing mode. The step of acquiring business data using the data processing mode corresponding to the trigger type, and determining a target object set based on the target object filtering conditions when it is determined that the business data satisfies the trigger rule, includes: When the data processing mode corresponding to the trigger type is the offline processing mode, the business data is acquired according to the preset acquisition cycle. Based on the triggering rules, a batch matching operation is performed on the business data to obtain the hit events that satisfy the triggering rules; The target object set for each hit event is determined based on the target object filtering criteria.

[0010] In conjunction with the first aspect, in one possible implementation of the first aspect, generating a push message for the learning task and sending the push message to the terminal of each target object in the target object set of the learning task includes: A push message for the learning task is generated, and the push message is sent to the terminal of each target object in the target object set of the learning task based on push control parameters. The push control parameters include at least one of push frequency limit, push time period and push exclusion list.

[0011] In conjunction with the first aspect, in one possible implementation of the first aspect, generating a push message for the learning task and sending the push message to the terminal of each target object in the target object set of the learning task includes: Acquire hardware operating status data, system and process operation data, and employee historical behavior data in the current work scenario; Based on the hardware operating status data, the system and process operating data, and the employee historical behavior data, calculate the attribution function value for each target object in the target object set; The cause of the anomaly of the target object is determined based on the attribution function value; When the cause of the anomaly is determined to be an employee capability issue, a push message for the learning task is generated and sent to the terminal of each target object in the target object set of the learning task.

[0012] In conjunction with the first aspect, in one possible implementation of the first aspect, the step of calculating the attribution function value of each target object in the target object set based on the hardware operating status data, the system process operating data, and the employee historical behavior data includes: A hardware status feature vector is determined based on the abnormal operating data in the hardware operating status data, wherein the abnormal operating data includes at least one of response delay data, fault code data, and sensor abnormal value data. The process operation status feature vector is determined based on the process complexity data in the system process operation data and the average error rate of all employees in similar businesses. Based on the historical operational error rate of each target object in the employee's historical behavior data, determine the personal skill feature vector of each target object; The attribution function value is calculated based on the hardware state feature vector, the process operation state feature vector, and the personal skill feature vector of each target object using a preset calculation method.

[0013] A second aspect of this application provides a learning task push device, comprising: an acquisition unit, configured to acquire configuration information of at least one learning task, the configuration information including trigger type, trigger rule and target object filtering conditions; A filtering unit is configured to acquire business data using a data processing mode corresponding to the trigger type, and, when determining that the business data satisfies the trigger rule, determine a set of target objects based on the target object filtering conditions; wherein, the data processing mode includes a real-time processing mode, and the step of acquiring business data using a data processing mode corresponding to the trigger type, and determining a set of target objects based on the target object filtering conditions when determining that the business data satisfies the trigger rule, includes: When the data processing mode corresponding to the trigger type is the real-time processing mode, the business data is acquired in real time; The business events in the business data are written to the first topic in real time to form the business event stream; A first global broadcast stream is generated based on the triggering rules; A second global broadcast stream is generated based on the target object filtering conditions; In the first processing task, the first global broadcast stream and the first topic are consumed based on the triggering rules, and the hit events that satisfy the triggering rules are written into the second topic; In the second processing task, the second global broadcast stream and the second topic are consumed based on the target object filtering conditions, and the target object set of the learning task corresponding to each hit event that meets the target object filtering conditions is written into the third topic; The first processing task and the second processing task are executed in parallel; The push unit is used to generate a push message for the learning task and send the push message to the terminal of each target object in the target object set of the learning task.

[0014] A third aspect of this application provides a computer device including a processor and a memory; wherein the memory stores a computer program adapted to be loaded by the processor and executed by the above-described learning task push method.

[0015] A fourth aspect of this application provides a computer-readable storage medium storing a plurality of instructions adapted for loading and execution by a processor of the above-described learning task push method.

[0016] The technical solution provided in this application may include the following beneficial effects: This application discloses a learning task push method, apparatus, device, and storage medium. The method includes: acquiring configuration information for at least one learning task, the configuration information including trigger type, trigger rule, and target object filtering conditions; acquiring business data using a data processing mode corresponding to the trigger type, and determining a target object set based on the target object filtering conditions when the business data satisfies the trigger rule; wherein the data processing mode includes a real-time processing mode, the real-time processing mode being used to generate a business event stream based on the business data, a first processing task consuming the real-time business event stream based on the trigger rule to obtain a hit event stream, a second processing task consuming the hit event stream based on the target object filtering conditions to obtain a target object set for each hit event, the first processing task and the second processing task being executed in parallel; generating a push message for the learning task, and sending the push message to the terminal of each target object in the target object set of the learning task. This application utilizes a real-time processing mode to generate business event streams based on business data. A first processing task consumes the real-time business event stream based on trigger rules to obtain a hit event stream. A second processing task consumes the hit event stream based on target object filtering conditions to obtain a set of target objects for each hit event corresponding to a learning task. The first and second processing tasks are executed in parallel, enabling real-time push of learning tasks. When employees encounter operational problems or job scenarios are updated, corresponding learning tasks are pushed immediately. It can also efficiently process massive amounts of real-time logistics business data, reducing data accumulation and processing bottlenecks.

[0017] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating the learning task push method shown in the embodiments of this application.

[0019] Figure 2 This is a schematic diagram of the learning task push device shown in the embodiments of this application.

[0020] Figure 3 This is a schematic diagram of the structure of a computer device shown in an embodiment of this application.

[0021] Figure 4 This is a schematic diagram of the structure of a computer-readable storage medium shown in an embodiment of this application. Detailed Implementation

[0022] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0023] The flowchart shown in the attached diagram is for illustrative purposes only and does not necessarily include all content and operations / steps, nor does it necessarily have to be performed in the order described. For example, some operations / steps can be broken down, combined, or partially merged, so the actual execution order may change depending on the actual situation.

[0024] It should be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the scope of the application. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0025] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0026] The following detailed description of some embodiments of this application is provided in conjunction with the accompanying drawings. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0027] It should be noted that all data involved in this application was obtained with the permission of the relevant users, complies with relevant policies and regulations, and will not infringe on user privacy.

[0028] This application provides a learning task push method, apparatus, device, and storage medium. It uses a real-time processing mode to generate business event streams based on business data. A first processing task consumes the real-time business event stream based on trigger rules to obtain a hit event stream. A second processing task consumes the hit event stream based on target object filtering conditions to obtain a set of target objects for each hit event. The first and second processing tasks are executed in parallel, enabling real-time push of learning tasks. This allows for the immediate push of corresponding learning tasks when employees encounter operational problems or when job scenarios are updated. It can also efficiently handle massive amounts of real-time logistics business data, reducing data accumulation and processing lag issues.

[0029] The method can be implemented using computer devices, which can be deployed on a single server or a cluster of servers. It can also be deployed on handheld terminals, laptops, wearable devices, or robots, etc.

[0030] It should be noted that the acquisition of any information involved in the provided methods is in compliance with relevant regulations and is carried out with the user's consent, and will not infringe on the user's privacy or violate relevant laws and regulations.

[0031] Please see Figure 1 , Figure 1 This is a flowchart illustrating the learning task push method shown in an embodiment of this application. Figure 1 As shown, the learning task push method includes steps S100 to S300.

[0032] Step S100: Obtain configuration information for at least one learning task.

[0033] The configuration information includes trigger type, trigger rules, and target object filtering conditions.

[0034] In some embodiments, the trigger type may include real-time event triggering, offline periodic triggering, and hybrid triggering. Real-time event triggering corresponds to real-time processing mode and is suitable for promptly pushing learning tasks after an employee's immediate operational error. Offline periodic triggering corresponds to offline processing mode and is suitable for periodic and systematic skills training scenarios. Hybrid triggering corresponds to both real-time and offline processing modes. Hybrid triggering can accommodate both immediate reinforcement learning for immediate operational errors and regular skills training scenarios, ensuring training effectiveness through high-frequency, multi-scenario learning tasks.

[0035] In some embodiments, the triggering rules may include at least one of the following: single event threshold triggering rules, time-series window aggregation triggering rules, multi-dimensional composite condition triggering rules, baseline deviation triggering rules, and fixed triggering rules. Each type of triggering rule can be used independently or configured in a nested combination.

[0036] In single-event threshold trigger rules, the trigger condition is whether the metric of a single business event exceeds a set threshold or a preset value. For example, configuring "the value of a single sorted item exceeding 500 yuan" as the trigger condition means that when the value of goods caused by a single sorting error by a target object (e.g., an employee) reaches the threshold, the corresponding learning task for the sorting specifications of valuable items is immediately triggered. Similarly, configuring "one violation of equipment operation safety rules" triggers a safety procedure training task. Single-event threshold trigger rules are simple to configure and have direct judgment, making them suitable for critical scenarios with high severity and requiring immediate intervention after a single occurrence. The calculation time for matching a single rule is small, and multiple single-event threshold trigger rules can run in parallel.

[0037] Furthermore, in the single-event threshold triggering rule, the metrics for a single business event may include the frequency of similar error events, the rate of metric achievement, the cause of damage, and the customer complaint rate.

[0038] In the time-series window aggregation trigger rule, the statistical data of consecutive preset types of business events within a pre-defined sliding time window are used as the trigger condition. For example, if the trigger condition is configured as "three consecutive failed scans within 15 minutes," a QR code learning task will be pushed out when the target object accumulates three failed scans within the 15-minute sliding time window. The time-series window aggregation trigger rule can identify the decline in ability reflected by consecutive employee errors, avoiding invalid pushes caused by occasional single errors.

[0039] In multi-dimensional composite condition triggering rules, nested combinations of AND, OR, and NOT logical relationships form the triggering conditions, which include multiple condition fields. These fields can include at least one of the following: event fields, device fields, and personnel attribute fields. Therefore, triggering conditions can include at least one of business indicator conditions, device conditions, and personnel attribute conditions. For example, configuring "(Number of errors ≥ 3 OR Error rate ≥ 5%) AND Position = Sorting worker AND Tenure < 6 months" as the triggering condition will only trigger the push of the learning task when both the business indicator condition and the personnel attribute condition are met. Multi-dimensional composite condition triggering rules can accurately target specific abnormal scenarios for specific groups of people, improving the accuracy of push notifications.

[0040] The baseline deviation trigger rule uses the target individual's historical operational error rate or the average operational error rate for the same position as the baseline. When the target individual's real-time operational error rate deviates from the baseline by a set margin based on real-time business data, a learning task is triggered. For example, a trigger condition could be configured as "an employee's daily operational error rate is 50% higher than their average error rate over the past 30 days," or "an employee's weekly error rate is 30% higher than the average level of all employees in the same position and with the same length of service." The baseline deviation trigger rule is based on relative deviation rather than absolute thresholds, allowing it to adapt to different employee skill baselines. It differentiates between occasional declines in high-skill employees and consistent performance of low-skill employees, avoiding a "one-size-fits-all" approach and improving the accuracy of push notifications.

[0041] In the fixed triggering rule, a preset learning task is pushed to a preset target object according to a preset push cycle.

[0042] In some embodiments, the target object filtering conditions may include basic personnel attribute filtering conditions, business performance dimension filtering conditions, learning behavior dimension filtering conditions, and tag-based hierarchical filtering conditions. Various filtering conditions can be logically combined and configured.

[0043] The basic personnel attribute filtering criteria are based on employees' static organizational attributes. Configurable AND, OR, and NOT logical combinations of attributes such as job type (e.g., sorter, warehouse clerk, driver, and customer service), department and team, length of service, job level, and employment type are allowed. For example, configuring "Job = Sorter AND Length of Service < 6 months AND Department = East China Sorting Center" can filter newly hired sorters as the target group. The basic personnel attribute filtering criteria offer stable data and low computational overhead, making them suitable for scenarios involving batch selection of training personnel based on organizational structure.

[0044] In the business performance dimension filtering criteria, filtering is based on employees' dynamic business performance data. Configurable criteria include periodic error rate, operational accuracy rate, performance score ranking, work order processing timeliness, and number of customer complaints. Multiple filtering methods are employed, such as threshold comparison, ranking percentage, and month-on-month changes. For example, filtering criteria can be set as "bottom 20% of error rate ranking in the last 7 days" or "monthly performance score below 80 points." This approach allows for the precise delivery of learning tasks to employees with poor business performance who urgently need skills enhancement.

[0045] In the learning behavior dimension filtering criteria, selection is based on employees' historical training data, and can be configured with conditions such as course completion status, historical exam scores, cumulative learning time, and skill certificate validity period. For example, filtering criteria such as "not completed the 'Sorting Safety Standards' course" or "not participated in refresher training in the past 90 days" can be configured to identify employees with lagging training progress and push corresponding learning tasks. In this way, the coverage and completion rate of training can be guaranteed, avoiding the repeated push of content that has already been mastered and reducing unnecessary interruptions.

[0046] The tiered, tag-based filtering system uses an employee skills tag database. Each employee corresponds to at least one skills tag in this database, such as "weak in barcode scanner operation," "unfamiliar with large item sorting," "unfamiliar with new processes," or "safety awareness needs improvement." Filtering can be based directly on tag combinations; for example, a filter could be configured with "tags include weak in barcode scanner operation AND do not include tags indicating training completed," accurately identifying employees with corresponding skill gaps who have not yet received training. This tiered, tag-based filtering system enables precise matching of employee competency profiles with learning tasks.

[0047] Understandably, an employee's business event must simultaneously meet both the triggering rules and the target object filtering conditions in order to trigger the push of the corresponding learning task to that employee.

[0048] Step S200: Obtain business data using the data processing mode corresponding to the trigger type, and determine the target object set based on the target object filtering conditions when it is determined that the business data meets the trigger rules.

[0049] The data processing mode includes a real-time processing mode, which is used to generate business event streams based on business data. The first processing task consumes the real-time business event streams based on triggering rules to obtain the hit event streams. The second processing task consumes the hit event streams based on target object filtering conditions to obtain the target object set of the learning task corresponding to each hit event. The first processing task and the second processing task are executed in parallel.

[0050] Existing technologies execute core processes such as data statistics, rule triggering, and personnel screening sequentially, resulting in lengthy processing chains and time-consuming computations. This makes it impossible to promptly push relevant learning content when employees encounter operational problems or when job scenarios change. Furthermore, the sequential processing mode is inefficient and prone to data backlog and processing bottlenecks.

[0051] In the above approach, the first processing task consumes the real-time business event stream based on trigger rules to obtain the hit event stream, and the second processing task consumes the hit event stream based on target object filtering conditions to obtain the target object set for each hit event's corresponding learning task. The first and second processing tasks are executed in parallel, achieving decoupled parallel processing between them. Event matching and personnel filtering are executed in layers, isolated by the event stream, enabling high-throughput, low-latency real-time processing. Furthermore, the trigger rules and target object filtering conditions can be modified independently without stopping the processing tasks.

[0052] In some embodiments, a stream processing engine can be used to implement the real-time processing mode. The stream processing engine can be Apache Flink, Apache Kafka Streams, Apache Samza, etc. The following example uses Apache Flink as the stream processing engine.

[0053] Specifically, when the data processing mode corresponding to the trigger type is real-time processing mode, step S200 includes steps S210 to S260.

[0054] Step S210: When the data processing mode corresponding to the trigger type is real-time processing mode, obtain business data in real time.

[0055] Step S220: Write the business events in the business data into the first topic in real time to form a business event stream.

[0056] For example, the first topic is also known as the first subject. When new or modified operations occur in the business data, the changes in the business data are captured in real time and converted into business events and written to the first topic. For example, operations such as order status changes, abnormal item registration, and performance point deductions are automatically converted into corresponding business events and written to the first topic.

[0057] Step S230: Generate the first global broadcast stream based on the triggering rules.

[0058] Specifically, step S230 includes steps S231 to S232.

[0059] Step S231: Write the triggering rule into the first MySQL data table.

[0060] Initially, the first MySQL table is empty. After the triggering rule is written to the first MySQL table, the first MySQL table is modified.

[0061] Step S232: In the first Flink CDC task, generate a first global broadcast stream including the triggering rule expression based on the first MySQL data table.

[0062] Specifically, a first Flink CDC task is started to monitor the first MySQL data table. When the trigger rule is modified, the trigger conditions in the first MySQL data table are converted into standardized Aviator expression data, generating a first global broadcast stream. In existing technologies, rule configuration is packaged in the task code. Modifying the rule requires stopping the task, repackaging, and deploying it online. The entire process is time-consuming and interrupts the real-time push service. This application achieves hot updates of trigger rules by using a first Flink CDC task to monitor changes in the first MySQL data table and generating a first global broadcast stream that is dynamically distributed to all compute nodes, without restarting the processing task.

[0063] Step S240: Generate a second global broadcast stream based on the target object filtering conditions.

[0064] Specifically, step S240 includes steps S241 to S242.

[0065] Step S241: Write the target object filtering criteria into the second MySQL data table.

[0066] Initially, the second MySQL table is empty. After the target object filtering criteria are written into the second MySQL table, the table is modified.

[0067] Step S242: In the second Flink CDC task, a second global broadcast stream including the target object filtering condition expression is generated based on the second MySQL data table.

[0068] Specifically, a second Flink CDC task is started to monitor the second MySQL data table. When the target object filtering conditions are modified, the triggering conditions in the second MySQL data table are converted into standardized Aviator expression data, generating a second global broadcast stream. This application enables hot updates of target object filtering conditions without restarting the processing task by monitoring changes to the second MySQL data table through the second Flink CDC task and generating a second global broadcast stream that is dynamically distributed to all computing nodes.

[0069] Step S250: In the first processing task, consume the first global broadcast stream and the first topic based on the triggering rules, and write the hit events that meet the triggering rules into the second topic.

[0070] For example, the second topic is also known as the second topic. The first processing task is a first-level Flink task. The first processing task is used to perform expression calculations on each business event based on the Aviator expression data of the triggering conditions, and write the hit events that meet the triggering conditions into the second topic.

[0071] Step S260: In the second processing task, based on the target object filtering conditions, consume the second global broadcast stream and the second topic, and write the target object set of the learning task corresponding to each hit event that meets the target object filtering conditions into the third topic.

[0072] For example, the second processing task is a second-level Flink task. This second processing task performs expression calculations on each hit event based on the Aviator expression data of the target object filtering criteria. Specifically, it determines the set of target objects for each hit event that meets the target object filtering criteria from the target object database, and writes the set of target objects for the learning task into the third topic. One learning task corresponds to one set of target objects.

[0073] Besides real-time emergency scenarios, logistics companies also have a large number of periodic training needs, such as monthly performance review training, quarterly safety training, and batch training for new employees. These scenarios involve large amounts of data and do not have high real-time requirements, making offline processing a more cost-effective and resource-efficient approach. Both offline and real-time processing modes share the same set of rule expressions, ensuring consistent judgment criteria for both scenarios.

[0074] In some embodiments, the data processing mode further includes an offline processing mode, and step S200 includes steps S270 to S290.

[0075] Step S270: When the data processing mode corresponding to the trigger type is offline processing mode, acquire business data according to the preset acquisition cycle.

[0076] For example, the preset acquisition period can be 1 hour, 12 hours, 24 hours or 7 days, etc.

[0077] Step S280: Perform batch matching operations on the business data based on the triggering rules to obtain the hit events that meet the triggering rules.

[0078] Step S290: Determine the set of target objects for each hit event corresponding to the learning task based on the target object filtering criteria.

[0079] Specifically, a batch matching process is performed using an SQL engine. First, the triggering rules are translated into corresponding first SQL query statements. Then, the target object filtering conditions are translated into corresponding second SQL query statements. The first and second SQL query statements are then combined to generate target SQL query statements. These target SQL query statements are used to perform query filtering on the business data to obtain the hit events that satisfy the triggering rules and the set of target objects for each hit event's corresponding learning task.

[0080] Step S300: Generate a push message for the learning task and send the push message to the terminal of each target object in the target object set of the learning task.

[0081] For real-time processing mode, push messages for learning tasks are generated based on the learning task events in the third topic and the set of target objects corresponding to the learning tasks.

[0082] In some embodiments, the learning task may include at least one of videos, documents, and hands-on tasks.

[0083] Unrestricted push of learning tasks can easily lead to employees receiving too many messages, causing annoyance, and even affecting normal work. It is necessary to use multi-dimensional push control parameters, such as push frequency, push time, and push exclusion lists, to achieve reasonable and appropriate precise pushes, balancing training effectiveness and employee experience.

[0084] In some embodiments, step S300 may include: generating a push message for a learning task, and sending the push message to the terminal of each target object in the target object set of the learning task based on push control parameters, wherein the push control parameters include at least one of push frequency limit, push time period and push exclusion list.

[0085] For example, the push frequency limit can be configured to "each person receives only one push for a single learning task". In this case, before generating the push message in step S300, it is first checked whether the employee has already received the learning task. If it has already been pushed, it is skipped directly to avoid duplicate pushes. This method is suitable for one-time mandatory courses, ensuring that each person receives only one notification and is not repeatedly disturbed.

[0086] For example, the push period can be configured as "weekdays 9:00-11:00, 14:00-16:00". In this case, step S300 checks the current time when generating the push message. If it is not within the push period, the push message is stored in a delay queue and sent when the next push period window opens. This method is suitable for non-urgent training content and can avoid disrupting normal work by pushing messages during peak employee work hours or rest periods.

[0087] For example, the push exclusion list can be configured as "employees on leave, resigned employees, and employees who have participated in training". In this case, before generating the push message, step S300 removes employees who do not meet the push criteria to obtain the final target set and then pushes the message. This method can avoid invalid pushes.

[0088] Traditional push notification methods send learning tasks to employees whenever a business error occurs. However, in actual logistics operations, the causes of errors are complex and varied: they may be due to hardware issues such as barcode scanner malfunctions or conveyor belt jams; they may be due to process problems such as unreasonable standard operating procedures or flawed system design; or they may be due to employee lack of proficiency or insufficient skills. Indiscriminately sending learning tasks to all employees not only renders training ineffective but also allows employees to receive notifications for errors not caused by their own actions, while masking the real equipment malfunctions or process defects that need to be addressed. This application quantifies the impact of hardware, process, and skill factors, and only sends learning tasks when the issue is determined to be an employee skill problem, achieving truly precise empowerment.

[0089] In some embodiments, step S300 may include steps S310 to S340.

[0090] Step S310: Obtain hardware operating status data, system process operating data, and employee historical behavior data in the current work scenario.

[0091] Specifically, the system acquires real-time hardware operational status data through an IoT platform. This data may include at least one of the following: response latency data for sorting barcode scanners, fault code data, sensor anomaly data, conveyor belt motor speed deviation, number of jams, zero-point drift of weighing equipment, and package measurement deviation. The system also acquires operational data for procedures and processes through a business process management system. This data may include directed graphs of processes, the number of steps in an ongoing process, the percentage of abnormal branches (and other process complexity data), and the average error rate of all employees executing the process. Finally, the system obtains historical employee behavior data from the human resources and training system. This data may include employees' historical skills assessment pass rates, operational accuracy rates over the past three months, and training course completion progress.

[0092] Step S320: Calculate the attribution function value of each target object in the target object set based on hardware operation status data, system process operation data, and employee historical behavior data.

[0093] Specifically, step S320 includes steps S321 to S324.

[0094] Step S321: Determine the hardware state feature vector based on the abnormal working data in the hardware operating status data.

[0095] Abnormal operating data includes at least one of response delay data, fault code data, and sensor abnormal value data.

[0096] In some embodiments, step S321 includes steps S3211 to S3213.

[0097] Step S3211: Calculate multiple statistics of device response delay time within a preset backtracking time window based on response delay data, normalize all statistics, and then determine the time-domain feature components in the hardware state feature vector based on the normalized multiple statistics.

[0098] Among them, several statistics include the mean, the maximum value, and the coefficient of variation. The mean is used to reflect the average performance of the equipment, the maximum value is used to reflect the extreme performance of the equipment, and the coefficient of variation is used to reflect the performance stability of the equipment.

[0099] Step S3212: Determine the weight of each equipment fault based on the equipment fault codes appearing in the backtracking time window in the fault code data and the preset correspondence between the equipment fault codes and the fault impact level. Calculate the fault frequency score based on the weights and occurrences of all faults. Finally, determine the fault feature components in the hardware state feature vector based on the fault frequency score.

[0100] For example, the weight of a fatal fault level is 1.0, the weight of a severe fault level is 0.7, the weight of a general fault level is 0.4, and the weight of a minor alarm level is 0.1.

[0101] The formula for calculating the fault frequency score is: ,in, The score represents the frequency of failures. The base of the natural logarithm. Indicates the first Weights of different types of faults Indicates the first The number of times this type of fault occurs. It converges to 1.

[0102] When the number of faults is small, the fault frequency score increases at a steep rate and is highly sensitive; when the number of faults is large, the score growth rate slows down, which can prevent numerical saturation caused by a single equipment fault infinitely increasing the score.

[0103] Minor faults usually do not affect the quality of work, while only serious faults will directly lead to operational errors by employees. The weighted fault frequency score can accurately reflect the causal contribution of different fault levels to errors and improve the accuracy of attribution.

[0104] In other embodiments, the occurrence count of all faults is determined based on the device fault codes that appear within the backtracking time window in the fault code data, and the occurrence count of all faults is normalized to obtain a fault frequency score.

[0105] Step S3213: Calculate the relative deviation of each sensor outlier value in the sensor outlier data from the corresponding standard reference value, then calculate the degree of deviation of the sensor outlier value based on the relative deviation, and finally determine the sensor outlier feature component in the hardware state feature vector based on the degree of deviation.

[0106] Specifically, for each sensor anomaly value, the relative deviation of that anomaly value is obtained by dividing the absolute value of the difference between the anomaly value and its corresponding standard reference value by the standard reference value. Then, a preset deviation calculation formula is used to calculate the deviation of all sensor anomalies for a single device. The deviation calculation formula is as follows: ,in, Indicates the degree of deviation. Indicates the first The proportion of the relative deviation of each sensor to the total relative deviation.

[0107] When a single sensor's value deviates significantly, the deviation is concentrated in one dimension, approaching zero, and can be identified as a single-point fluctuation rather than a systemic equipment failure. However, when multiple sensors simultaneously show deviations, the deviation distribution becomes diffuse, and the degree of deviation increases significantly, reflecting overall equipment performance degradation. The above method is robust to single-point numerical interference.

[0108] In some embodiments, step S321 further includes step S3214.

[0109] Step S3214: Calculate the average unit time error rate of the device based on the business data of all employees using the same device within the first preset time period, and calculate the contribution of the device's own factors based on the average unit time error rate and the preset baseline of the average error rate per person in the whole station. Finally, determine the contribution feature component in the hardware status feature vector based on the contribution.

[0110] The causal relationship between abnormal equipment parameters and employee operational errors is often difficult to determine. It could be that equipment malfunction leads to employee error, or that frequent accidental activation due to employee unfamiliarity with the equipment causes abnormal alarm parameters. Simply analyzing equipment parameters cannot distinguish the causal relationship, and attribution is always subject to systematic bias. By using the method described above, we can isolate human error by analyzing the operational performance of other employees on the same equipment, thus determining the net contribution of the equipment itself to the error.

[0111] Specifically, the formula for calculating contribution is: ,in, This indicates the contribution of factors inherent to the equipment itself. This represents the average error rate per unit of time of the equipment. This represents the baseline error rate per person across the entire station. A contribution score greater than 1 indicates a high overall error rate on this equipment, pointing to a problem with the equipment itself; a score equal to 1 indicates that the equipment is functioning normally.

[0112] Step S322: Determine the process operation status feature vector based on the process complexity data in the system process operation data and the average error rate of all employees in similar businesses.

[0113] Specifically, step S322 includes steps S3221 to S3222.

[0114] Step S3221: Calculate the structural complexity based on the directed graph of the process in the process complexity data, and determine the complexity component in the feature vector of the process running state based on the structural complexity.

[0115] Specifically, the formula for calculating structural complexity is: ,in, Indicates structural complexity. This represents the number of process nodes in a directed graph. Indicates the number of branches. This represents the number of nodes on the normal path of the main branch of a directed graph. The information entropy represents the branch paths of a directed graph of a process. Higher structural complexity indicates a greater susceptibility to errors in process design.

[0116] In some embodiments, historical process data is acquired, the execution probability of each branch path is determined based on the historical process data, and then the information entropy of the branch path is determined based on the execution probability of the branch path.

[0117] in, The more nodes and branches a process has, the higher the cognitive load on employees' memory and execution, and the more likely they are to make mistakes. Used to measure the degree of dispersion of the number of branches across nodes. When the workload is high, the branches are scattered across a large number of nodes, and employees need to continuously judge the path throughout the process, resulting in a high cognitive load and a high probability of error.

[0118] Step S3222: Calculate the relative deviation between the average error rate of all employees performing the same type of business in the current process and the preset error rate of the same position across the entire station, and determine the deviation component of the same type of business in the process operation status feature vector based on the relative deviation.

[0119] Specifically, the formula for calculating the relative deviation is: ,in, Indicates the relative deviation. This represents the average error rate of all employees performing similar tasks within the current process. This represents the error rate for the same position across the entire station. A positive relative deviation, and the larger the value, the higher the systematically high error rate across all personnel in that process, indicating a problem with the process itself.

[0120] In some embodiments, step S322 further includes step S3223.

[0121] Step S3223: Obtain historical business data of multiple different independent workstations executing the same process, calculate the coefficient of variation of the error rate of each workstation based on the historical business data, and then determine the workstation verification component in the process running status feature vector based on the coefficient of variation.

[0122] If the coefficient of variation is less than the preset threshold for workstation verification, it indicates that the error rate of different workstations executing the same process is highly consistent, ruling out individual factors and confirming it as a process issue. Otherwise, the errors are more likely due to differences in the capabilities of the personnel at each workstation.

[0123] Step S323: Determine the personal skill feature vector of each target object based on the historical operational error rate of each target object in the employee's historical behavior data.

[0124] Using a fixed threshold to judge employee ability is unfair to new employees, as their error rate is already higher than that of experienced employees. A fixed threshold would lead to an over-promotion of new employees. Therefore, step S323 includes steps S3231 to S3232.

[0125] Step S3231: Select historical behavior data of employees with the same job position as the target object and whose length of service does not differ from the target object by more than the second preset duration from the historical behavior data of employees as control data, and then calculate the average error rate and standard deviation of the control data as dynamic parameters for the ability assessment of the target object.

[0126] Step S3232: Calculate the target object's standard score based on the dynamic parameters of the ability assessment and the target object's historical operation error rate within the most recent second preset time period, and determine the target object's personal skill feature vector based on the standard score.

[0127] The formula for calculating the standard score of the target object is: ,in, Indicates the standard score. This represents the historical operational error rate of the target object. This represents the average error rate of the control data. This represents the standard deviation of the comparison data. A standard score greater than 0 indicates that the employee's error rate is higher than the average level for employees with similar experience; the higher the score, the weaker the ability. In this method, different dynamic parameters for ability assessment are used for fair evaluation of both new and experienced employees, which can reduce the interference of seniority factors.

[0128] Step S324: Calculate the attribution function value based on the hardware status feature vector, the process operation status feature vector, and the personal skill feature vector of each target object using a preset calculation method.

[0129] Step S324 includes steps S3241 to S3242.

[0130] Step S3241: Based on the preset weighted calculation method of each feature vector, calculate the hardware status score, process operation status score and personal skill score respectively based on the hardware status feature vector, process operation status feature vector and personal skill feature vector of each target object.

[0131] In the pre-weighted calculation method, the value in each vector is multiplied by a pre-defined calculation weight. All calculation weights can be configured manually or determined after training with an artificial intelligence model.

[0132] All parameters used in the calculations in this application have been normalized. For example, the Min-Max normalization algorithm is used to normalize all parameters used in the calculations.

[0133] In other embodiments, the hardware state feature vector, the process operation state feature vector, and the personal skill feature vector of each target object are input into a pre-trained artificial intelligence model to obtain the hardware state score, the process operation state score, and the personal skill score.

[0134] Step S3242: Calculate the attribution function value based on the hardware status score, process operation status score, and personal skill score.

[0135] Among these, hardware problems and process problems are given priority in attribution, and hardware problems or process problems will suppress the weight of capability problems in the attribution.

[0136] Specifically, the formula for calculating the attribution function value is as follows: ,in, This represents the attribution function value, indicating the confidence level that the error was caused by the employee's incompetence. Indicates individual skill score, This represents the preset hardware problem attenuation coefficient. Indicates hardware status. This represents the preset process problem attenuation coefficient. This indicates the process running status.

[0137] Step S330: Determine the cause of the anomaly of the target object based on the attribution function value.

[0138] Step S330 includes steps S331 to S334.

[0139] Step S331: Determine the judgment threshold based on the mean of the attribution function values ​​under preset historical normal operation.

[0140] In some embodiments, the mean of the attribution function values ​​under preset historical normal operation is determined as the judgment threshold.

[0141] In logistics operations, workload, equipment status, and environmental noise vary significantly across different time periods (e.g., peak sales periods versus regular periods, night shifts versus day shifts). Fixed judgment thresholds are often insufficiently accurate. In other embodiments, the current system concurrent order volume and equipment alarm frequency are obtained, and then a real-time noise figure is calculated based on these figures. Finally, the judgment threshold is determined based on the average of the real-time noise figure and the attribution function values ​​under historical normal operation.

[0142] Specifically, the formula for calculating the initial real-time noise figure is as follows: ,in, Represents the real-time noise figure. This indicates the current number of concurrent orders in the system. This represents the preset frequency weighting coefficient. This represents the preset historical average order volume. This represents the preset historical average device alarm frequency. Then, the initial real-time noise figure is normalized to obtain the final real-time noise figure. Finally, the mean of the attribution function values ​​under historical normal operation is multiplied by 1 and added to the sum of the final real-time noise figures to obtain the judgment threshold.

[0143] Step S332: When the attribution function value is greater than the judgment threshold, the cause of the anomaly is determined to be an employee's ability problem.

[0144] Step S333: When the attribution function value is not greater than the judgment threshold and the hardware status score is the highest, the cause of the anomaly is determined to be a hardware problem.

[0145] Step S334: When the attribution function value is not greater than the judgment threshold and the process running status score is the highest, the cause of the anomaly is determined to be a process problem.

[0146] Step S340: When the cause of the anomaly is determined to be an employee's ability problem, generate a push message for the learning task and send the push message to the terminal of each target object in the target object set of the learning task.

[0147] In some embodiments, when it is determined that the cause of the anomaly is not an employee capability issue, corresponding equipment repair feedback information or process optimization suggestions are generated based on the cause of the anomaly, but no push message for learning tasks is generated.

[0148] The above methods enable differentiated and precise handling of the root causes of errors, rather than indiscriminately pushing learning tasks to all employees. On the one hand, for errors that are indeed caused by insufficient employee capabilities, corresponding learning tasks are precisely pushed to ensure that training resources are invested in the personnel who truly need improvement. On the other hand, for errors caused by non-human factors such as equipment failures and process defects, equipment repair feedback information or process optimization suggestions are automatically generated and circulated to the corresponding functional departments to solve the problem at its root, while avoiding employee resistance due to receiving push messages for reasons not attributable to them.

[0149] This application discloses a learning task push method, comprising: obtaining configuration information for at least one learning task, the configuration information including trigger type, trigger rule, and target object filtering conditions; acquiring business data using a data processing mode corresponding to the trigger type, and determining a target object set based on the target object filtering conditions when the business data meets the trigger rule; wherein the data processing mode includes a real-time processing mode, the real-time processing mode being used to generate a business event stream based on the business data, a first processing task consuming the real-time business event stream based on the trigger rule to obtain a hit event stream, and a second processing task consuming the hit event stream based on the target object filtering conditions to obtain a target object set for each hit event, the first processing task and the second processing task being executed in parallel; generating a push message for the learning task, and sending the push message to the terminal of each target object in the target object set of the learning task. This application utilizes a real-time processing mode to generate business event streams based on business data. A first processing task consumes the real-time business event stream based on trigger rules to obtain a hit event stream. A second processing task consumes the hit event stream based on target object filtering conditions to obtain a set of target objects for each hit event corresponding to a learning task. The first and second processing tasks are executed in parallel, enabling real-time push of learning tasks. When employees encounter operational problems or job scenarios are updated, corresponding learning tasks are pushed immediately. It can also efficiently process massive amounts of real-time logistics business data, reducing data accumulation and processing bottlenecks.

[0150] Corresponding to the aforementioned application function implementation method embodiments, this application also provides a learning task push device, an electronic device, and corresponding embodiments.

[0151] Please see Figure 2 , Figure 2 This is a schematic diagram illustrating the structure of a learning task push device according to an embodiment of this application. This application provides a learning task push device 300, comprising: The acquisition unit 310 is used to acquire configuration information for at least one learning task, including trigger type, trigger rule and target object filtering conditions.

[0152] The filtering unit 320 is used to acquire business data using the data processing mode corresponding to the trigger type, and determine the target object set according to the target object filtering conditions when it is determined that the business data meets the trigger rules. The data processing mode includes a real-time processing mode and an offline processing mode. The real-time processing mode is used to generate a business event stream based on the business data. The first processing task consumes the real-time business event stream based on the trigger rules to obtain the hit event stream. The second processing task consumes the hit event stream based on the target object filtering conditions to obtain the target object set of the learning task corresponding to each hit event. The first processing task and the second processing task are carried out in parallel.

[0153] The push unit 330 is used to generate push messages for learning tasks and send the push messages to the terminals of each target object in the target object set of the learning task.

[0154] For other details regarding the implementation of the above technical solutions by each module in the learning task push device, please refer to the description of the learning task push method provided in the above embodiments of the invention, which will not be repeated here.

[0155] Based on the above-described learning task push method, this embodiment of the invention also provides a computer device. Figure 3 This is a schematic diagram of the structure of a computer device shown in an embodiment of this application, such as... Figure 3 As shown, the computer device 400 includes a processor 401 and a memory 402 coupled to the processor 401. The memory 402 stores a computer program, which, when executed by the processor 401, causes the processor 401 to perform the steps of the learning task push method in the above embodiments.

[0156] For other details regarding the implementation of the above technical solution by the processor 401 in the above computer device, please refer to the description in the learning task push method provided in the above embodiments of the invention, which will not be repeated here.

[0157] The processor 401 can also be called a CPU (Central Processing Unit). The processor 401 may be an integrated circuit chip with signal processing capabilities. The processor 401 may also be a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Data Array), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. The general-purpose processor may be a microprocessor, or the processor 401 may be any conventional processor.

[0158] This invention also provides a computer-readable storage medium. Figure 4 This is a schematic diagram illustrating the structure of a computer-readable storage medium according to an embodiment of this application. Figure 4 As shown, a computer-readable storage medium 500 stores a readable computer program 501. This computer program 501 can be stored in the storage medium as a software product, including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the methods of various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, magnetic disks or optical disks, ROM (Read-Only Memory), RAM (Random Access Memory), or terminal devices such as computers, servers, mobile phones, and tablets.

[0159] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces, indirect coupling or communication connection between apparatuses or modules, and may be electrical, mechanical, or other forms.

[0160] The modules described as separate components may or may not be physically separate. Similarly, the components shown as modules may or may not be physical modules; they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0161] Furthermore, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can be stored in a computer-readable storage medium.

[0162] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product.

[0163] A computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the flow or function according to the embodiments of this application is generated. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions may be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that a computer can store or a data storage device such as a server or data center that integrates one or more available media. The available medium may be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state disk (SSD)).

[0164] The technical solutions provided in this application have been described in detail above. Specific examples have been used in this application to illustrate the principles and implementation methods of this application. The description of the above embodiments is only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

[0165] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0166] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to this application. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0167] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0168] These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable apparatus for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0169] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for pushing learning tasks, characterized in that, include: Obtain configuration information for at least one learning task, the configuration information including trigger type, trigger rule, and target object filtering conditions; The process involves acquiring business data using the data processing mode corresponding to the trigger type, and determining a target object set based on the target object filtering conditions when the business data satisfies the trigger rule. The data processing mode includes a real-time processing mode. The steps of acquiring business data using the data processing mode corresponding to the trigger type and determining the target object set based on the target object filtering conditions when the business data satisfies the trigger rule include: When the data processing mode corresponding to the trigger type is the real-time processing mode, the business data is acquired in real time; The business events in the business data are written to the first topic in real time to form a business event stream; A first global broadcast stream is generated based on the triggering rules; A second global broadcast stream is generated based on the target object filtering conditions; In the first processing task, the first global broadcast stream and the first topic are consumed based on the triggering rules, and the hit events that satisfy the triggering rules are written into the second topic; In the second processing task, the second global broadcast stream and the second topic are consumed based on the target object filtering conditions, and the target object set of the learning task corresponding to each hit event that meets the target object filtering conditions is written into the third topic; The first processing task and the second processing task are executed in parallel; Generate a push message for the learning task and send the push message to the terminal of each target object in the target object set of the learning task.

2. The learning task push method according to claim 1, characterized in that, The triggering rules include at least one of the following: single-event threshold triggering rules, time-series window aggregation triggering rules, multi-dimensional composite condition triggering rules, baseline deviation triggering rules, and fixed triggering rules; In the single-event threshold triggering rule, the triggering condition is whether the indicator of a single business event exceeds a set threshold or is a preset value. In the time-series window aggregation triggering rule, the statistical data of continuous preset types of business events within a pre-set sliding time window are used as the triggering condition; In the multi-dimensional composite condition triggering rule, the logical relationships of AND, OR, and NOT are nested and combined to form the triggering condition, and the triggering condition includes multiple condition fields; In the baseline deviation triggering rule, the target object's personal historical operation error rate or the average operation error rate of the same position is used as the baseline. When it is determined from the target object's real-time operation error rate that it deviates from the baseline by a set amount based on the target object's real-time business data, the learning task is triggered. In the fixed triggering rule, a preset learning task is pushed to a preset target object according to a preset push cycle.

3. The learning task push method according to claim 1, characterized in that, The data processing mode also includes an offline processing mode. The step of acquiring business data using the data processing mode corresponding to the trigger type, and determining a target object set based on the target object filtering conditions when the business data meets the trigger rules, includes: When the data processing mode corresponding to the trigger type is the offline processing mode, the business data is acquired according to the preset acquisition cycle. Based on the triggering rules, a batch matching operation is performed on the business data to obtain the hit events that satisfy the triggering rules; The target object set for each hit event is determined based on the target object filtering criteria.

4. The learning task push method according to claim 1, characterized in that, The step of generating a push message for the learning task and sending the push message to the terminal of each target object in the target object set of the learning task includes: A push message for the learning task is generated, and the push message is sent to the terminal of each target object in the target object set of the learning task based on push control parameters. The push control parameters include at least one of push frequency limit, push time period and push exclusion list.

5. The learning task push method according to claim 1, characterized in that, The step of generating a push message for the learning task and sending the push message to the terminal of each target object in the target object set of the learning task includes: Acquire hardware operating status data, system and process operation data, and employee historical behavior data in the current work scenario; Based on the hardware operating status data, the system and process operating data, and the employee historical behavior data, calculate the attribution function value for each target object in the target object set; The cause of the anomaly of the target object is determined based on the attribution function value; When the cause of the anomaly is determined to be an employee capability issue, a push message for the learning task is generated and sent to the terminal of each target object in the target object set of the learning task.

6. The learning task push method according to claim 5, characterized in that, The calculation of the attribution function value for each target object in the target object set based on the hardware operating status data, the system process operating data, and the employee historical behavior data includes: A hardware status feature vector is determined based on the abnormal operating data in the hardware operating status data, wherein the abnormal operating data includes at least one of response delay data, fault code data, and sensor abnormal value data. The process operation status feature vector is determined based on the process complexity data in the system process operation data and the average error rate of all employees in similar businesses. Based on the historical operational error rate of each target object in the employee's historical behavior data, determine the personal skill feature vector of each target object; The attribution function value is calculated based on the hardware state feature vector, the process operation state feature vector, and the personal skill feature vector of each target object using a preset calculation method.

7. A learning task push device, characterized in that, include: The acquisition unit is used to acquire configuration information for at least one learning task, the configuration information including trigger type, trigger rule and target object filtering conditions; A filtering unit is configured to acquire business data using a data processing mode corresponding to the trigger type, and, when determining that the business data satisfies the trigger rule, determine a set of target objects based on the target object filtering conditions; wherein, the data processing mode includes a real-time processing mode, and the step of acquiring business data using a data processing mode corresponding to the trigger type, and determining a set of target objects based on the target object filtering conditions when determining that the business data satisfies the trigger rule, includes: When the data processing mode corresponding to the trigger type is the real-time processing mode, the business data is acquired in real time; The business events in the business data are written to the first topic in real time to form a business event stream; A first global broadcast stream is generated based on the triggering rules; A second global broadcast stream is generated based on the target object filtering conditions; In the first processing task, the first global broadcast stream and the first topic are consumed based on the triggering rules, and the hit events that satisfy the triggering rules are written into the second topic; In the second processing task, the second global broadcast stream and the second topic are consumed based on the target object filtering conditions, and the target object set of the learning task corresponding to each hit event that meets the target object filtering conditions is written into the third topic; The first processing task and the second processing task are executed in parallel; The push unit is used to generate a push message for the learning task and send the push message to the terminal of each target object in the target object set of the learning task.

8. A computer device, characterized in that, It includes a processor and a memory; wherein the memory stores a computer program adapted to be loaded by the processor and to execute the steps of the method as described in any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a plurality of instructions adapted for loading by a processor and executing the steps of the method as described in any one of claims 1-6.

Images