Self-adaptive training method and system based on service data closed-loop feedback

Abstract

The invention discloses an adaptive training method and system based on service data closed-loop feedback. The method comprises the following steps: acquiring performance data of employees from a business operation system, and identifying a structured question label when a preset condition is met; automatically matching and generating a personalized micro-learning training task according to the question label; after the employee finishes training, an instruction is automatically sent to a business operation system, and business operation which is subsequently executed by the employee and is the same as the question label in type is required to be preferentially evaluated; and acquiring subsequent performance data generated by the priority evaluation, and quantifying the improvement effect of the training task by comparing changes of related indexes in the performance data before and after the training task is executed. According to the method, the verification condition is actively created, so that the timeliness and correlation of the evaluation data are ensured, the accurate quantification of the training return on investment is realized, and the training accuracy and efficiency are improved.

Description

Technical Field

[0001] This application relates to the field of computer application technology, and in particular to an adaptive training method based on closed-loop feedback of business data. Background Technology

[0002] In business operations, employee training is a crucial element in enhancing organizational capabilities. Current technologies typically rely on Learning Management Systems (LMS) for training. The main model involves companies pre-defining training plans, assigning standardized online courses to employees, and using course completion rates and exam scores as evaluation metrics. These training systems are usually independent of the company's daily business operating systems, such as Customer Relationship Management (CRM) and quality control, in terms of data and processes.

[0003] With technological advancements, several solutions have emerged that attempt to integrate business data with training. For example, some systems can acquire employee behavior data within business systems to determine training needs and then assess training effectiveness by comparing employee performance data before and after training. However, these solutions typically rely on passive collection and post-event analysis of business data when evaluating training effectiveness. This approach has significant drawbacks: First, data collection for evaluation may be delayed, failing to reflect training effectiveness in a timely manner; second, passively collected data samples may be insufficient or poorly correlated with the training content, leading to inaccurate and subjective evaluation results; finally, due to the lack of reliable quantitative feedback, it is difficult to prove a direct link between training input and business output, making it difficult to measure the return on investment (ROI) of training.

[0004] Therefore, existing technologies generally lack a mechanism that can proactively, reliably, and promptly collect follow-up business performance data that is strongly related to the training content, and use this as a basis to achieve accurate and efficient quantification of training effectiveness. Summary of the Invention

[0005] The purpose of this application is to provide an adaptive training method and system based on closed-loop feedback of business data, so as to solve the problems of inaccurate, untimely and unreliable training effect evaluation caused by passive data collection in the prior art.

[0006] To achieve the above objectives, this application provides an adaptive training method based on closed-loop feedback of business data. This method is executed by a computer system and includes the following steps: acquiring performance data related to the business operations of a specific employee from at least one business operating system, and analyzing the performance data to derive benchmark indicators characterizing the severity of problems the employee encounters in their business operations; when the benchmark indicators meet preset conditions, identifying at least one structured problem tag from the performance data, the problem tag representing a specific problem point in the employee's business operations; automatically matching and selecting at least one micro-learning module corresponding to the problem tag from a pre-built knowledge base that maps and associates problem tags with micro-learning modules, to generate at least one personalized training task; after confirming that the employee has completed the training task, automatically sending an instruction to the business operating system requesting priority evaluation of subsequent business operations of the same type as the problem tag performed by the employee; and acquiring subsequent performance data generated by the priority evaluation, quantifying the improvement effect of the training task on business performance by comparing the changes in indicators related to the problem tag in the employee's performance data before and after performing the training task.

[0007] This application provides an adaptive training method based on closed-loop feedback of business data, which further includes generating or updating a personal competency profile associated with the employee based on the quantified improvement effect, and evaluating the effectiveness of the micro-learning modules in the knowledge base.

[0008] Preferably, the step of automatically matching and selecting at least one micro-learning module corresponding to the problem tag further includes: determining the severity level of the problem tag, statistically analyzing the employee's historical problem occurrence frequency, and obtaining at least one of their learning preferences; and dynamically selecting or combining from multiple matching micro-learning modules based on at least one of the severity level, historical problem occurrence frequency, and learning preferences.

[0009] Furthermore, the issue label is in JSON format and includes an issue identifier and an issue severity level.

[0010] Preferably, the instructions automatically sent to the business operating system include: specifying the employee's identity identifier, at least one issue tag that needs to be focused on, and the number of subsequent business operations that need to be prioritized for evaluation.

[0011] Preferably, the step of quantifying the improvement effect of the training task on business performance specifically includes:

[0012] The improvement effect is determined by calculating the change in the incidence of the problem label before and after performing the training task, or the change in the average score of business indicators related to the problem label.

[0013] Furthermore, the calculation of the change in the incidence rate of the problem label before and after training is specifically performed using the following formula: Improvement rate = (Incidence rate before training - Incidence rate after training) / Incidence rate before training.

[0014] Preferably, the method further includes: performing data interaction and instruction sending between the computer system and the business operating system via an application programming interface (API); and pushing the personalized training task to the employee's user terminal via an instant messaging tool; and / or, the business operating system is a machine vision inspection system; and / or, the employee's user terminal is augmented reality (AR) smart glasses; and / or, the micro-learning module is AR teaching content presented in the AR smart glasses for real-time guidance of the employee in physical operations.

[0015] This application also provides an adaptive training system based on closed-loop feedback of business data, configured to execute any of the above methods. The system includes: a data fusion and problem identification module, used to acquire performance data related to the business operations of a specific employee from at least one business operating system, and to analyze the performance data to derive benchmark indicators characterizing the severity of problems; when the benchmark indicators meet preset conditions, at least one structured problem tag is identified from the performance data; an intelligent push and personalized task engine module, used to automatically match and select micro-learning modules from a pre-built knowledge base according to the problem tags to generate personalized training tasks; and a closed-loop verification and effect quantification module, used to automatically send an instruction to the business operating system after confirming that the employee has completed the training task, requesting priority evaluation of subsequent business operations performed by the employee that are of the same type as the problem tag; and to acquire pre-training and subsequent performance data generated by the priority evaluation, quantifying the improvement effect of the training task by comparing the changes in relevant indicators in the employee's performance data before and after performing the training task.

[0016] Furthermore, the system also includes a profile generation and evaluation module, used to generate or update a personal competency profile associated with the employee based on the quantified improvement effect, and to evaluate the effectiveness of the micro-learning modules in the knowledge base.

[0017] Compared with the prior art, this application has the following beneficial effects:

[0018] 1. Achieve precise quantification of training return on investment (ROI): Through the core mechanism of "automatically triggered priority evaluation," reliable data verification conditions are proactively created, ensuring that the collected post-training data is strongly correlated with and timely in relation to the training content. Based on this, by comparing changes in specific business indicators before and after training, the vague "learning completion rate" is transformed into a clear and measurable "business indicator improvement rate," providing effective data support for corporate training decisions.

[0019] 2. Improve the accuracy and effectiveness of training: Training tasks are triggered by real business problem data, ensuring that each training session directly addresses the exposed skill gaps, thereby improving the efficiency of employee skill improvement and the utilization rate of training resources.

[0020] 3. Shorten the "problem-learning-application" cycle: It realizes a rapid closed loop of "instant discovery, instant push, instant learning, and instant verification", which shortens the traditional training cycle that may have lasted for several weeks to the hour or even minute level, effectively preventing the solidification of employees' wrong behaviors.

[0021] 4. Automated management reduces management costs: It fully automates the cumbersome diagnosis, matching, push, tracking and verification processes in personalized training, and can achieve dynamic coaching for all employees with "personalized guidance" without human intervention, effectively improving management efficiency. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a schematic diagram of the structure of the adaptive training system provided in an embodiment of this application.

[0024] Figure 2 This is a flowchart illustrating the adaptive training method provided in an embodiment of this application.

[0025] Figure 3 This is a timing diagram of signaling interaction in the adaptive training method provided in the embodiments of this application.

[0026] Figure 4 This is a schematic diagram illustrating the generation of dynamic path tasks provided in an embodiment of this application.

[0027] Figure 5 This is a schematic diagram of the closed-loop verification and effect quantification process provided in the embodiments of this application.

[0028] The main reference numerals in the accompanying drawings are explained as follows:

[0029] 10: Data Fusion and Problem Identification Module; 20: Knowledge Graph and Micro-Learning Library Module; 30: Intelligent Push and Personalized Task Engine Module; 40: Closed-Loop Verification and Effect Quantification Module; 50: Business Operating System; 60: Learning Management System; 70: User Terminal; 100: Adaptive Training System Based on Closed-Loop Feedback of Business Data; 310: Input Parameters; 320: Dynamic Weighting Algorithm; 330: Output Task; 331: Image and Text Cards; 332: Teaching Videos; 333: Simulated Exercises. Detailed Implementation

[0030] To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings. The described embodiments should not be regarded as limitations on this application. All other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0031] Unless otherwise defined, all technical and scientific terms used in the embodiments of this application have the same meaning as commonly understood by one of ordinary skill in the art. The terminology used in the embodiments of this application is for the purpose of describing the embodiments of this application only and is not intended to limit this application.

[0032] Before providing a further detailed description of the embodiments of this application, the nouns and terms involved in the embodiments of this application will be explained, and the nouns and terms involved in the embodiments of this application shall be interpreted as follows.

[0033] (1) Business Operating System: This refers to the computer software system used by an enterprise to execute, manage, or monitor specific business processes in its daily production and operation activities. For example, in the customer service field, it may be a Customer Relationship Management (CRM) system, an intelligent quality inspection system, or a work order system; in the manufacturing field, it may be a Manufacturing Execution System (MES), an equipment monitoring system, or a machine vision inspection system. In the embodiments of this application, the business operating system 50 is the source of performance data and also the object that receives and executes priority evaluation instructions.

[0034] (2) Performance data: refers to structured or unstructured data that reflects an employee's performance, quality, efficiency, or compliance during business operations. This data comes directly from the business operating system, such as the quality inspection results of call recordings, defect records from code reviews, defective product records on the production line, or the completion time of specific business processes. This data is an objective basis for identifying employee skill gaps.

[0035] (3) Problem tags: These are structured identifiers extracted and abstracted from raw performance data to characterize a specific, categorizable skill deficiency or behavioral deviation in an employee's business operations. Examples include "non-compliant opening remarks," "omission of key selling points," and "incorrect operation sequence." Problem tags transform unstructured business problems into standardized data units that machines can understand and process, serving as a crucial bridge connecting business performance and training content.

[0036] (4) Micro-learning modules: These are atomized, short learning resources that focus on solving a single specific problem or imparting a single knowledge point. They can take the form of short videos, graphic cards, interactive simulation exercises, AR teaching content, etc. The design concept is to enable employees to use fragmented time to quickly and accurately complete learning and immediately apply it to their work practice to solve the skill gaps identified by specific problem tags.

[0037] (5) Benchmark Indicators: These are one or a set of quantitative values ​​used to measure the severity of an employee's problems in a specific business area. These indicators can be calculated based on performance data, such as the frequency of occurrence of a specific problem label, the average score of business indicators related to the problem, or the rate of customer complaints caused by the problem. Benchmark indicators are used to trigger adaptive training processes. When their values ​​exceed a preset threshold, it indicates that the employee's problem has reached a level that requires intervention.

[0038] (6) Priority Assessment: This refers to a proactive assessment or review process initiated by the adaptive training system based on closed-loop feedback of business data in this application embodiment, focusing on the same type of business operations performed by specific employees after completing training. Unlike regular, random, or periodic assessments, priority assessment has a clear objective (verifying whether specific problem points related to the training content have improved) and a higher execution priority. This process ensures that follow-up performance data for quantifying training effectiveness can be collected in a timely and reliable manner.

[0039] To achieve the above objectives, this application provides an adaptive training method based on closed-loop feedback of business data. This method is executed by a computer system and includes the following steps: acquiring performance data related to the business operations of a specific employee from at least one business operating system, and analyzing the performance data to derive benchmark indicators characterizing the severity of problems the employee encounters in their business operations. This step, by integrating data with the employee's daily work system, achieves objective monitoring of the employee's actual work performance. The benchmark indicators can serve as triggers to determine whether training intervention is necessary.

[0040] When the benchmark indicator meets preset conditions, such as a business indicator falling below a threshold or a certain error occurring frequently, at least one structured problem label is identified from the performance data. This structured problem label is a data structure used to accurately and standardizedly characterize specific problems encountered by employees in their business operations. It transforms vague performance data into machine-readable, specific problem diagnoses, providing a foundation for subsequent targeted training.

[0041] Based on the problem tags, at least one micro-learning module corresponding to the problem tag is automatically matched and selected from a pre-built knowledge base that maps and associates problem tags with micro-learning modules, to generate at least one personalized training task. This step, through automation, links the identified specific problem points with the most relevant solutions (i.e., micro-learning modules), ensuring a high degree of relevance in the training content and avoiding the inefficient "one-size-fits-all" approach of traditional training.

[0042] After confirming that the employee has completed the training task, an instruction is automatically sent to the business operating system, requesting priority evaluation of subsequent business operations performed by the employee that are of the same type as the problem tag. This is one of the core steps of this application, transforming passive waiting for data into proactively creating verification conditions. Through this instruction, the system can ensure that, after training is completed, it can collect subsequent business performance data that is strongly related to the training content in a timely and reliable manner, thereby solving the problems of data collection delays, irrelevant or insufficient samples in the prior art.

[0043] The subsequent performance data generated by the prioritization assessment is obtained. By comparing the changes in indicators related to the problem tags in the employee's performance data before and after performing the training task, the improvement effect of the training task on business performance is quantified. This step completes the closed loop of "data-driven - precise training - effect verification". Through direct comparison of before and after data, the effect of training is presented with specific and quantifiable indicators (such as the percentage reduction in error rate and the increase in average score), providing direct data basis for measuring the return on investment of training.

[0044] In one possible implementation, the method further includes: generating or updating a personal competency profile associated with the employee based on the quantified improvement effect, and evaluating the effectiveness of the micro-learning modules in the knowledge base. This forms a higher-order feedback loop. On the one hand, the quantified effect data can continuously enrich the employee's personal competency profile, providing data support for long-term talent development and job matching. On the other hand, by analyzing the average improvement effect brought by different micro-learning modules, the system can perform self-optimization of the knowledge base content, automatically increasing the recommendation weight of efficient modules and prompting the administrator to review or update inefficient modules, thereby ensuring the continuous improvement of the overall training system content quality.

[0045] In one possible implementation, the step of automatically matching and selecting at least one micro-learning module corresponding to the problem tag further includes: determining the severity level of the problem tag, statistically analyzing the employee's historical problem occurrence frequency, and obtaining at least one of their learning preferences; and dynamically selecting or combining from multiple matching micro-learning modules based on at least one of the severity level, historical problem occurrence frequency, and learning preferences. This approach makes the generation of training tasks more intelligent and humanized, enabling differentiated intervention strategies. For example, for a newly occurring problem with a low severity level, the system can push a lightweight graphic card; while for a recurring problem with a high severity level, it can push a combined task including video explanations and mandatory simulation exercises. This not only improves the utilization efficiency of training resources but also enhances the employee's learning experience.

[0046] Optionally, the issue label is in JSON format and includes an issue identifier and an issue severity level. For example, an issue label can be defined as {"tagId": "PROD_A_MISS", "severity": "high"}, where tagId is the unique identifier of the issue and severity is the severity level of the issue. Using a standardized JSON format facilitates parsing and interaction between different systems, improving system compatibility and scalability.

[0047] Optionally, the instruction sent to the business operating system includes: specifying the employee's identity identifier, one or more issue tags requiring focused attention, and the number of subsequent business operations requiring priority evaluation. This clarifies the specific content of the instruction, making it executable. For example, the instruction could be {"employeeId": "zhangsan01", "focusTags": ["PROD_A_MISS"], "evaluationCount": 3}, meaning that the business system is required to prioritize the evaluation of the employee "zhangsan01's" three subsequent business operations, with a focus on checking whether the "PROD_A_MISS" issue has occurred. This ensures the accuracy and efficiency of the subsequent evaluation.

[0048] In one possible implementation, the step of quantifying the improvement effect of the training task on business performance specifically includes: determining the improvement effect by calculating the change in the incidence rate of the problem label before and after executing the training task, or the change in the average score of business indicators related to the problem label. This provides two specific quantitative dimensions, making the evaluation of the improvement effect more comprehensive. For example, in a call center call quality inspection scenario, the improvement effect can be evaluated by calculating the change in the frequency of occurrence of the label "non-standard opening remarks"; in a sales performance scenario, it can be evaluated by calculating the change in the average score of related product sales before and after training.

[0049] Furthermore, the calculation of the change in the incidence rate of the problem label before and after training is specifically performed using the following formula: Improvement rate = (Incidence rate before training - Incidence rate after training) / Incidence rate before training. This formula provides a clear and reproducible calculation standard, ensuring consistency and comparability in the quantitative results of training effectiveness.

[0050] In one possible implementation, the method further includes: data interaction and instruction sending between the computer system and the business operating system via an application programming interface (API); and pushing the personalized training tasks to employees' user terminals via instant messaging tools. API integration enables automated and real-time data and instruction flow between systems, providing a technical guarantee for the efficient operation of the entire closed-loop process. Pushing tasks via instant messaging tools commonly used by employees (such as WeChat Work and DingTalk) ensures that training information reaches employees instantly, shortening the delay from problem discovery to learning commencement and improving the timeliness of training.

[0051] In a specific application scenario, the business operating system is a machine vision inspection system; the employee's user terminal is augmented reality (AR) smart glasses; and the micro-learning module is AR teaching content presented in the AR smart glasses for real-time guidance of the employee in physical operations. This scenario demonstrates the application of this application in physical operation fields such as manufacturing. The machine vision inspection system captures worker operational errors in real time and generates problem labels. The system immediately pushes the corresponding AR teaching content to the worker's AR smart glasses, providing real-time error correction and guidance in the worker's field of vision through highlighting, animation, and other methods. This "on-the-job, immediate, and immersive" training method improves the training efficiency and product qualification rate of operation-intensive positions. This application also provides an adaptive training system based on closed-loop feedback of business data, which is configured to execute any of the above methods. The system includes: a data fusion and problem identification module, used to acquire performance data related to the business operations of a specific employee from at least one business operating system, and to analyze the performance data to derive benchmark indicators characterizing the severity of problems. When the benchmark indicators meet preset conditions, at least one structured problem tag is identified from the performance data; an intelligent push and personalized task engine module, used to automatically match and select micro-learning modules from a pre-built knowledge base according to the problem tags to generate personalized training tasks; and a closed-loop verification and effect quantification module, used to automatically send instructions to the business operating system after confirming that the employee has completed the training task, requesting priority evaluation of the employee's subsequent business operations of the same type as the problem tag; and to acquire pre-training and subsequent performance data generated by the priority evaluation, and to quantify the improvement effect of the training task by comparing the changes in relevant indicators in the employee's performance data before and after performing the training task.

[0052] In one possible implementation, the system further includes: a profile generation and evaluation module, which generates or updates a personal competency profile associated with the employee based on the quantified improvement effect, and evaluates the effectiveness of the micro-learning modules in the knowledge base. (Refer to...) Figures 1 to 5 This application provides an adaptive training method (hereinafter referred to as the "method") based on closed-loop feedback of business data, and a system for implementing the method (hereinafter referred to as the "system"). The method is executed by a computer system and aims to solve the problems of inaccurate, untimely, and unreliable training effectiveness evaluation in the prior art. It creates reliable verification conditions by proactively intervening in business processes, thereby achieving precise quantification of training effectiveness.

[0053] like Figure 1As shown in the diagram, this application provides a structural schematic of an adaptive training system 100 based on closed-loop feedback of business data. This system can be deployed on a cloud server or an enterprise local server and interact with multiple other systems within the enterprise via a network. Specifically, the system includes a data fusion and problem identification module 10, a micro-learning module intelligent push and personalized task engine module 30, and a closed-loop verification and effect quantification module 40. These modules work together to execute the adaptive training method based on closed-loop feedback of business data proposed in this application.

[0054] The adaptive training system 100, based on closed-loop feedback of business data, interacts with external business operating systems 50, learning management systems 60, and user terminals 70 used by employees. The business operating system 50 serves as the platform for employees' daily work and is also the source of performance data. The user terminal 70 is the device used by employees to receive and complete training tasks, such as a personal computer, smartphone, or dedicated AR smart glasses. The learning management system 60 can be an optional integration point for recording employees' overall learning history and profiles. Figure 2 The diagram shown illustrates the overall process of the adaptive training method based on closed-loop feedback of business data provided in this application embodiment. The core idea of ​​this method is to construct an automated closed loop from "identifying problems" to "solving problems" and then to "verifying effects." The following will combine... Figure 1 System architecture and Figure 2 The flowchart illustrates in detail the basic implementation of this method.

[0055] In a basic implementation, the method first performs step S101, which involves acquiring performance data. Specifically, the data fusion and problem identification module 10 is configured to acquire performance data associated with the business operations of a specific employee from at least one business operating system 50. This acquisition can be real-time streaming data access or periodic batch data retrieval. For example, in a call center scenario, the business operating system 50 could be an intelligent voice quality inspection system that automatically analyzes customer service personnel's call recordings to generate quality inspection reports containing various service quality indicators; these reports constitute the performance data.

[0056] After acquiring the performance data, the data fusion and problem identification module 10 analyzes the data to calculate a benchmark indicator representing the severity of the employee's problems in business operations. For example, the system can count the number of times an employee "failed to proactively confirm customer issues" occurred in the past week. When this benchmark indicator meets preset conditions, such as occurring more than 3 times, the system considers that a training intervention needs to be initiated.

[0057] Once the benchmark indicators meet the preset conditions, the method proceeds to step S102, which involves identifying problem tags. The data fusion and problem identification module 10 further identifies or parses at least one structured problem tag from the performance data triggered by the conditions. This problem tag characterizes a specific problem point in the employee's business operations. This process transforms raw, complex performance data into one or more standardized "symptom" tags that can be used for subsequent matching. For example, from a quality inspection report, the system can extract specific problem tags such as "inaccurate opening remarks" and "incorrect product knowledge."

[0058] Next, the method executes step S103, which involves matching the learning task. The data fusion and question identification module 10 sends the identified question tags to the intelligent push and personalized task engine module 30. The intelligent push and personalized task engine module 30 queries and matches the received question tags in the pre-built knowledge graph and micro-learning library module 20. The knowledge graph and micro-learning library module 20 stores a large number of micro-learning modules and establishes a mapping relationship between question tags and these micro-learning modules.

[0059] For example, the knowledge graph and micro-learning library module 20 may define the association between "problem tag: non-standard opening remarks" and "micro-learning module: standard opening remarks teaching video" and "micro-learning module: opening remarks simulated dialogue practice". The intelligent push and personalized task engine module 30 will automatically match and select at least one micro-learning module corresponding to the problem tag to generate at least one personalized training task.

[0060] After generating the training task, the method executes step S104, which is to push the task. The intelligent push and personalized task engine module 30 pushes the generated personalized training task to the employee's user terminal 70. The purpose of the push is to ensure that employees can receive learning content closely related to their current work problems in a timely manner. For example, the system can send a message containing a learning link to the employee through the company's internal instant messaging tools.

[0061] Subsequently, in step S105, the employee receives and completes the training task on their user terminal 70. The system monitors the completion status of the task through technical means, such as tracking the video playback progress and the completion score of the exercises. When the system confirms that the employee has completed the training task, for example, when the employee clicks the "Learning Completed" button or reaches the preset pass standard, the user terminal 70 sends a completion status signal to the adaptive training system 100 based on closed-loop feedback of business data.

[0062] After confirming that the employee has completed the training, the method proceeds to one of the core steps of this application, namely step S106, which involves sending a priority evaluation instruction. Upon receiving the confirmation signal that the task has been completed, the closed-loop verification and effect quantification module 40 will immediately and automatically send an instruction to the business operating system 50 corresponding to the employee. The core content of this instruction is to require the business operating system 50 to prioritize the evaluation of any subsequent business operations performed by the employee that are of the same type as the previously identified problem tags.

[0063] This step is the key difference between this application and existing technologies. Instead of passively waiting for the next evaluation, it proactively and purposefully creates a data collection opportunity to verify the training effectiveness. For example, if employee "Zhang San" receives training on the issue of "omitted selling points of Product A," then the closed-loop verification and effectiveness quantification module 40 will send an instruction to the intelligent quality inspection system, requesting it to prioritize and focus on quality inspection of "Zhang San's" subsequent calls involving Product A.

[0064] Upon receiving the instruction, the business operating system 50 executes step S107, which involves performing a priority assessment. The business operating system 50 adjusts its internal work queues or assessment logic to ensure that specific types of operations performed by designated employees are prioritized. For example, the intelligent quality inspection system might prioritize "Zhang San's" subsequent calls in the quality inspection queue and activate a refined analysis model targeting "Product A's selling point."

[0065] After the priority assessment is completed, the method executes step S108, which involves acquiring subsequent data. The business operating system 50 sends the subsequent performance data generated by the priority assessment back to the closed-loop verification and effectiveness quantification module 40 of the adaptive training system 100, which is based on closed-loop feedback of business data. This performance data is the most direct and relevant evidence for verifying the training effectiveness.

[0066] Finally, the method executes step S109, which quantifies the improvement effect. The closed-loop verification and effect quantification module 40 acquires pre-training performance data (from step S101) and post-training performance data (from step S108), and quantifies the actual improvement effect of the training task on employee business performance by comparing the changes in indicators related to problem labels in these two sets of data. For example, the system can calculate the change in the incidence rate of a certain problem label before and after the training task, thus obtaining a specific percentage improvement.

[0067] like Figure 3As shown in the figure, this diagram more clearly illustrates the signaling interaction sequence among the adaptive training system 100 based on business data closed-loop feedback, the business operating system 50, and the user terminal 70 in the method of this application. First, the business operating system 50 sends initial performance data or problem data to the adaptive training system 100 based on business data closed-loop feedback. After analysis, the adaptive training system 100 pushes training tasks to the employee's user terminal 70. After the employee completes the training, the user terminal 70 returns a task completion status to the adaptive training system 100 based on business data closed-loop feedback. This is the key signal triggering the subsequent verification process. Subsequently, the adaptive training system 100 based on business data closed-loop feedback immediately sends the core priority evaluation instruction to the business operating system 50. Finally, the business operating system 50 performs the evaluation and returns the evaluation result data to the adaptive training system 100 based on business data closed-loop feedback, completing the entire closed loop.

[0068] Through the above process, this application constructs an automated closed loop of "data-driven problem discovery -> precise delivery of learning tasks -> proactive triggering of effect verification -> quantitative evaluation of improvement effects." Its fundamental technical advantage lies in solving the core problems of passive data collection, irrelevant samples, and untimely and unreliable evaluation in traditional training effectiveness evaluation by automatically sending priority evaluation instructions to the business operating system. It proactively creates high-quality verification data, making the return on investment (ROI) of training no longer a vague estimate, but something that can be accurately and objectively quantified, providing effective data-driven decision-making support for corporate talent development and training resource investment.

[0069] Furthermore, this method enables rapid response and immediate intervention to employee skill gaps, shortening the potentially weeks-long cycle of "problem identification-training arrangement-effect evaluation" to hours or even minutes. This effectively prevents the entrenching of employee errors and improves the efficiency and accuracy of training. The entire process is highly automated, requiring no manual intervention, thus reducing the management costs of providing personalized coaching to all employees.

[0070] In a preferred embodiment, to achieve stable and efficient data interaction between the system of this application and external systems, the method further includes data interaction and instruction sending between the adaptive training system 100 based on closed-loop feedback of business data and the business operating system 50 through an Application Programming Interface (API). Using standardized APIs (such as RESTful APIs, a web service interface based on the REST [Representational State Transfer] architectural style) for communication ensures decoupling and good scalability between systems, allowing the system of this application to be easily connected to various existing heterogeneous business systems within an enterprise.

[0071] Furthermore, in this embodiment, to ensure that training tasks can reach employees in a timely and effective manner, the method also includes pushing the personalized training tasks to employees' user terminals 70 via instant messaging tools. For example, this can be achieved through robots (bots) integrated with office software such as WeChat Work and DingTalk. Compared to traditional email or LMS in-site notifications, instant messaging pushes have a higher open rate and faster response speed, which is more in line with the "instant and fragmented" characteristics of micro-learning. The technical effect of the above-mentioned technical features is that, through standardized APIs and efficient instant messaging tools, a stable, reliable, and fast system integration and information delivery channel is built, providing technical support for the smooth operation of the entire closed-loop process and improving the robustness of the system and user experience.

[0072] In another preferred embodiment, the instructions automatically sent to the business operating system 50 can contain more specific and structured information for more precise control. Specifically, the instructions may include: specifying the employee's identification (e.g., employee ID), at least one issue tag requiring focused attention, and the number of subsequent business operations requiring priority evaluation. For example, an instruction could be {"employeeId": "zhangsan01", "focusTags": ["PROD_A_MISS"], "evaluationCount": 3}. This design allows the business operating system 50 to clearly know who to evaluate, what to check, and how many times to check. The technical effect of this feature is that it improves the accuracy and efficiency of priority evaluation, ensuring that the collected subsequent performance data is highly focused on the issues to be verified, thereby making the final quantitative evaluation results more reliable and convincing.

[0073] In a more refined implementation, the step of automatically matching and selecting micro-learning modules based on question tags can be further optimized to achieve deeper personalization. For example... Figure 4 As shown, this step may include: First, determining the severity level of the problem label, or statistically analyzing the employee's historical problem occurrence frequency, or obtaining their learning preferences, etc., as input parameters 310. Then, based on at least one of the severity level, historical problem occurrence frequency, or learning preferences, a dynamic weighting algorithm 320 is used to dynamically select or combine from multiple matching micro-learning modules, ultimately outputting task 330.

[0074] For example, for a problem tag like "inappropriate opening remarks," the knowledge base might associate it with three different types of micro-learning modules: a simple text and image card 331, a standard instructional video 332, and a challenging simulation exercise 333. If the system detects that an employee is exhibiting this problem for the first time and the severity level is low, the dynamic weighting algorithm 320 might determine that only a lightweight reminder is needed, and therefore only push the text and image card 331. However, if another employee has exhibited this problem multiple times in a short period, and one instance resulted in a customer complaint (high severity level), the algorithm will determine that intensive intervention is needed, and might choose to push a combined task of instructional video 332 and simulation exercise 333, requiring the simulation exercise to achieve a score of 95 or higher to be considered complete. The technical effect of the above features is that they enable differentiated and precise delivery of training resources, avoiding a "one-size-fits-all" training approach for all problems. It can provide the "just right" level of intervention based on the specific context of the problem and the individual circumstances of the employee, which saves training resources and improves the effectiveness of solving stubborn or high-risk problems, making personalized training more intelligent and efficient.

[0075] Furthermore, to enable the issue tags themselves to carry richer information to support the aforementioned dynamic selection, the structured issue tags can adopt JSON format and include an issue identifier and an issue severity level. For example, a tag can be defined as {"tagId": "T01", "tagName": "Initial opening is non-compliant", "severity": "high"}. This standardized data structure facilitates the transmission and parsing of issue tags between systems and provides directly usable input parameters for the dynamic weight algorithm 320 of the intelligent push and personalized task engine module 30. The technical effect of this feature is that, through standardized data format, it improves the standardization and efficiency of communication between modules within the system and between the system and external systems, providing a data foundation for realizing more complex dynamic decision-making logic.

[0076] In the process of quantitatively evaluating training effectiveness, there are also preferred calculation methods. For example... Figure 5 The diagram illustrates the closed-loop verification and effect quantification process. This process begins with storing pre-training indicators (S401), triggering the verification process (S402), calculating post-training indicators (S403), and finally calculating the improvement rate (S404) and outputting the quantification result (S405). Specifically, the step of quantifying the improvement effect may include: determining the improvement effect by calculating the change in the incidence rate of the problem label before and after training, or the change in the average score of business indicators related to the problem label.

[0077] For example, if the problem label appeared M times in N similar operations performed before training, the occurrence rate before training is M / N. After training and prioritization assessment, if the problem label appeared L times in K similar operations performed after training, the occurrence rate after training is L / K. By comparing these two occurrence rates, the improvement can be clearly seen. The technical effect of this feature is that it provides a specific and actionable quantitative assessment dimension, transforming the vague concept of "effect" into a clear mathematical indicator, making the assessment results objective and comparable.

[0078] More specifically, to standardize the expression of improvement effects, the calculation of the change in the incidence of the problem label before and after training can be specifically calculated using the following formula:

[0079]

[0080] The pre-training incidence rate represents the frequency of a specific problem label appearing in a certain number of business operation samples before receiving training intervention; the post-training incidence rate represents the frequency of the same problem label appearing in business operation samples collected through priority evaluation after completing the training task. The calculation result of this formula intuitively reflects the improvement in employee performance on specific problem points due to this training intervention. For example, if the pre-training incidence rate is 10% and drops to 2% after training, the improvement rate is (10%-2%) / 10% = 80%. The technical advantage of this feature is that it provides a unified and standardized quantitative formula, allowing the improvement effects of different training tasks, different employees, and different problem points to be measured and compared on the same dimension, providing a foundation for subsequent system optimization and data analysis.

[0081] The closed-loop feedback mechanism constructed in this application can not only be used to improve employee capabilities, but the resulting quantitative results can also feed back into the system itself, forming a higher-order optimization cycle. In a preferred embodiment, the method further includes: generating or updating a personal capability profile associated with the employee based on the quantified improvement effect, and evaluating the effectiveness of the micro-learning modules in the knowledge base.

[0082] Specifically, when the system calculates that the improvement rate of a training session is 80%, it updates this information in the employee's personal competency profile, marking an improvement in their ability to address the corresponding problem. Simultaneously, the system records this successful training case (i.e., "Problem Tag A" + "Micro-learning Module B" => "Improvement Rate 80%)). After accumulating sufficient data, the system performs statistical analysis. If it finds that the average improvement rate of micro-learning module B for solving problem tag A is significantly higher than that of module C, the system automatically determines that module B is a more effective learning resource. Based on this, the system can automatically increase the recommendation weight of module B in future task matching or issue a warning to the content administrator, suggesting a review or optimization of the inefficient module C. The technical effect of this feature is that it establishes a higher-dimensional self-optimizing closed loop. It not only improves the "people" but also continuously improves the "knowledge base" itself through data feedback, achieving the survival of the fittest and continuous iteration of training content and strategies, ensuring the long-term effectiveness and efficiency of the entire training system.

[0083] The application scenarios of this application are very broad and are not limited to office environments. In one innovative implementation, this application can be applied to physical operation scenarios such as industrial manufacturing. In this implementation, the business operating system 50 can be a machine vision inspection system deployed on a production line; the employee's user terminal 70 can be augmented reality (AR) smart glasses worn by the employee; and the micro-learning module is AR teaching content presented in the AR smart glasses for real-time guidance of the employee in performing physical operations.

[0084] For example, on an electronics assembly line, a machine vision inspection system (business operating system 50) monitors workers' operations in real time. When it detects that worker "Li Si" is assembling a component in the wrong order, the system immediately generates an "incorrect operation sequence" error label and processes it through the adaptive training system 100 based on closed-loop feedback of business data. The intelligent push and personalized task engine module 30 matches the corresponding AR teaching module and immediately pushes it to the AR smart glasses (user terminal 70) worn by "Li Si". At this time, virtual arrows, highlighted boxes, or animations will immediately appear in "Li Si's" field of vision, clearly indicating the correct operation steps, which is equivalent to an "on-the-job, real-time" micro-learning and error correction. After "Li Si" completes the correct operation according to the AR guidance, the system records this intervention and quantifies the improvement in his operation standardization based on the subsequent continuous monitoring data of his operation by the machine vision system. The technical effect of this feature is that it extends the closed-loop feedback concept of this application from online business scenarios to offline physical operation scenarios. By combining it with cutting-edge technologies such as AR, it enables real-time, immersive, on-the-job training and error correction for manual laborers' skills, improving training efficiency and reducing operational error rates. This is of great significance for improving the product qualification rate and production efficiency of the manufacturing industry.

[0085] The following will fully demonstrate the workflow and overall technical effect of this application through a minimum protection scope embodiment in a call center scenario that integrates the above-mentioned preferred technical features.

[0086] In this embodiment, the adaptive training system 100 based on closed-loop feedback of business data is deeply integrated with the intelligent quality inspection system (business operating system 50) of the call center and enterprise instant messaging tools (such as WeChat Work) through a RESTful API. The knowledge graph and micro-learning library module 20 is pre-loaded with rich question tags related to customer service work and corresponding micro-learning modules. For example, the question tag {"tagId": "T02", "tagName": "Product A selling point omission", "severity": "high"} is associated with two micro-learning modules: M1 (a 3-minute video explaining the selling points of Product A) and M2 (an AI-simulated sales dialogue exercise).

[0087] The workflow is as follows:

[0088] Step 1: Problem Capture and Analysis. Customer service employee "Wang Wu" failed to fully explain the key selling points of product A during a call with a customer. After the call, the intelligent quality inspection system automatically analyzed the recording, identified the problem, and generated the aforementioned JSON-formatted problem tag. Because omitting a selling point could lead to a failed sale, the system rated its severity as "high." Subsequently, the intelligent quality inspection system sent this performance data via API to the data fusion and problem identification module 10 of the adaptive training system 100, which is based on closed-loop feedback of business data.

[0089] Step Two: Dynamic Task Generation and Push. The intelligent push and personalized task engine module 30 receives the problem tag. It queries "Wang Wu's" history and finds that this is the second time he has made this type of mistake this month (the historical problem has a high frequency). Combining the "high" severity level of the problem tag, the dynamic weight algorithm 320 determines that a reinforcement intervention is needed. Therefore, the dynamic weight algorithm 320 decides to combine M1 (explanation video) and M2 (simulated practice) into a training task, and sets the passing score for M2 to be 95 points. After the task is generated, the system sends a message to "Wang Wu's" user terminal 70 (mobile phone) through the Enterprise WeChat robot: "Hello, the system detected that you may have missed the key selling points of product A in your recent call. Please click the link to complete an 8-minute reinforcement learning task to help you consolidate your knowledge."

[0090] Step 3: Employee Learning and Status Confirmation. "Wang Wu" clicked the link, watched a 3-minute video, and then entered a simulated dialogue practice. After two attempts, his practice score reached 96 points, meeting the passing requirements. The learning application on user terminal 70 then sent a task completion confirmation signal to the closed-loop verification and effect quantification module 40 of the adaptive training system 100, which is based on closed-loop feedback of business data, via API callback.

[0091] Step 4: Closed-Loop Verification Triggered. Upon receiving the confirmation signal, the closed-loop verification and effect quantification module 40 immediately constructs a priority evaluation instruction with the content {"action": "requestPriorityEvaluation", "employeeId": "wangwu02", "focusTags": ["T02"], "evaluationCount": 5}. This instruction is sent to the intelligent quality inspection system via API, requesting priority and focused quality inspection of the subsequent five calls involving product A related to "Wang Wu".

[0092] Step 5: Quantifying Results and System Self-Optimization. Over the next half-day, the intelligent quality inspection system completed a priority evaluation of "Wang Wu's" five relevant calls and returned the results. The returned data showed that the "product A selling point omission" issue did not reappear in any of these five calls. The closed-loop verification and results quantification module 40 calculated based on this data: Before training, "Wang Wu" encountered this issue twice in the most recent 10 relevant calls, with a pre-training incidence rate of 20%. After training, it occurred 0 times in the five calls, with a post-training incidence rate of 0%. Using the formula Improvement Rate = (20% - 0%) / 20%, the improvement rate was calculated to be 100%.

[0093] Step 6: Application of Results. This 100% improvement rate result was used in several ways: First, it was recorded in "Wang Wu's" personal competency profile, showing that his weakness in "Product A knowledge" had been addressed through this training. Second, this success story of "(M1+M2) combined task achieving 100% improvement rate for T02 problem" was used to evaluate the effectiveness of the two micro-learning modules, M1 and M2, increasing their recommendation weight in the knowledge base. Finally, a report containing the intervention cost (e.g., 8 minutes of employee time) and quantified benefits (100% problem improvement rate) was automatically generated and made available to his supervisor and training department.

[0094] This embodiment achieves comprehensive technical effects by combining all the aforementioned technical features. First, compared to implementing only a basic solution, dynamic task generation makes training intervention more targeted and avoids resource waste. Second, structured instructions and JSON tags improve the accuracy of communication between systems. Third, standardized quantitative formulas make the evaluation results of training effectiveness clear, reliable, and comparable. Most importantly, by using the quantitative results to update individual profiles and evaluate learning resources, the system forms a higher-order, self-iterable, and optimized intelligent closed loop, which not only continuously improves employee performance but also helps maintain the high effectiveness of the training system's content and strategies, thereby improving the return on investment in human capital.

[0095] This application has a wide range of applications and can meet the needs of different industries and positions. In the customer service industry, this application can be used to improve the service quality, sales skills, and compliance of customer service personnel. By analyzing performance data such as call recordings, online chat logs, and customer satisfaction questionnaires, the system can automatically identify problems such as "failure to effectively soothe customer emotions" and "failure to follow standard response procedures," and push corresponding micro-learning modules such as emotion management skills and communication techniques.

[0096] In industries with high compliance requirements, such as finance and insurance, this application can be used for employee compliance risk management. By connecting to the business review system and analyzing employee operation logs during business transactions, the system can identify compliance risk points such as "insufficient risk warnings" and "incomplete KYC (Know Your Customer) information collection," and immediately push relevant mandatory learning tasks such as interpretations of laws and policies and compliance operating procedures, thereby effectively reducing the company's compliance risks.

[0097] In the field of software development, this application can be used to improve developers' code quality and development efficiency. By integrating code repositories (such as the distributed version control system Git), continuous integration (CI) tools, and static code analysis tools, the system can identify problem tags such as "no design patterns used," "SQL injection risk," and "excessive cyclomatic complexity" from code review comments, compilation failure records, and code smell reports, and push relevant micro-learning content such as design pattern explanations, safe coding standards, and code refactoring techniques to developers.

[0098] In the healthcare field, this application can be used to ensure the standardization and safety of medical procedures. By analyzing data from Electronic Medical Record (EMR) systems and nursing record systems, behavioral deviations by doctors or nurses during the diagnosis and treatment process can be identified, such as "prescribing without following clinical pathways" or "performing nursing procedures without strictly adhering to aseptic principles." Relevant training content, including treatment guidelines and operational videos, can then be disseminated to improve the quality of medical services and ensure patient safety.

[0099] In the retail industry, this application can be used to improve the sales capabilities of store sales staff. By connecting to point-of-sale (POS) systems and customer flow analysis systems, it can analyze performance data such as sales conversion rate, average order value, and sales rate of related products for each salesperson, identify problems such as "weak ability to recommend new products" and "lack of cross-selling skills," and push relevant learning tasks such as product knowledge, sales scripts, and success story sharing.

[0100] In summary, the method and system provided in this application, through its unique "proactively triggered priority evaluation" mechanism, helps solve the core challenge of quantifying the effectiveness of traditional corporate training. It transforms employee training from an isolated, cost-center activity into a data-driven closed loop tightly integrated with business processes, capable of creating measurable value. Its high degree of automation, accuracy, and quantifiability allows it to adapt to various complex business scenarios, providing effective technical support for enterprises to build organizational capabilities for continuous learning and improvement. Those skilled in the art should understand that any modifications, equivalent substitutions, or improvements made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. An adaptive training method based on business data closed-loop feedback, the method being executed by a computer system, characterized in that, Includes the following steps: Retrieve performance data related to the business operations of a specific employee from at least one business operating system, and analyze the performance data to derive benchmark indicators characterizing the severity of the problems in the employee's business operations; When the benchmark indicator meets the preset conditions, at least one structured problem label is identified from the performance data. The problem label represents the specific problem points that the employee has in business operations. Based on the question tag, at least one micro-learning module corresponding to the question tag is automatically matched and selected from a pre-built knowledge base that maps and associates question tags with micro-learning modules, so as to generate at least one personalized training task. After confirming that the employee has completed the training task, an instruction is automatically sent to the business operating system to request that the subsequent business operations performed by the employee that are of the same type as the problem tag be given priority evaluation. as well as Obtain subsequent performance data generated by the prioritization assessment, and quantify the improvement effect of the training task on business performance by comparing changes in indicators related to the problem label in the employee's performance data before and after performing the training task.

2. The method of claim 1, wherein, The method further includes: Based on the quantified improvement effect, generate or update the personal competency profile associated with the employee, and evaluate the effectiveness of the micro-learning modules in the knowledge base.

3. The method according to claim 1 or 2, characterized in that, The step of automatically matching and selecting at least one micro-learning module corresponding to the question label further includes: Determine the severity level of the problem label, statistically analyze the historical frequency of the employee's problems, and obtain at least one of the following: Dynamically select or combine from multiple matching micro-learning modules based on at least one of the severity level, frequency of historical problem occurrence, and learning preferences.

4. The method according to claim 3, characterized in that, The issue labels are in JSON format and include an issue identifier and an issue severity level.

5. The method according to claim 1 or 2, characterized in that, The instructions automatically sent to the business operating system include: Specify the employee's identification, at least one issue label that requires special attention, and the number of follow-up business operations that need to be prioritized for evaluation.

6. The method according to claim 1 or 2, characterized in that, The steps for quantifying the improvement effect of the training task on business performance specifically include: The improvement effect is determined by calculating the change in the incidence of the problem label before and after performing the training task, or the change in the average score of business indicators related to the problem label.

7. The method according to claim 6, characterized in that, The change in the incidence rate of the problem label before and after training is specifically calculated using the following formula: Improvement rate = (Incidence rate before training - Incidence rate after training) / Incidence rate before training.

8. The method according to claim 1 or 2, characterized in that, The method further includes: Data interaction and command transmission are conducted between the computer system and the business operating system via application programming interface (API); and personalized training tasks are pushed to employees' user terminals via instant messaging tools; and / or, The operating system is a machine vision inspection system; and / or, The employee's user terminal is augmented reality (AR) smart glasses; and / or, The micro-learning module consists of AR teaching content presented in the AR smart glasses, used to guide the employees in performing physical operations in real time.

9. An adaptive training system based on closed-loop feedback of business data, used to implement the method according to any one of claims 1-8, characterized in that, include: The data fusion and problem identification module is used to obtain performance data related to the business operations of a specific employee from at least one business operating system, and to analyze the performance data to derive a benchmark indicator that characterizes the severity of the problem. When the benchmark indicator meets preset conditions, at least one structured problem label is identified from the performance data. The intelligent push and personalized task engine module is used to automatically match and select micro-learning modules from a pre-built knowledge base based on the question tags to generate personalized training tasks. The closed-loop verification and effect quantification module is used to automatically send an instruction to the business operating system after confirming that the employee has completed the training task, requesting that the business operations performed by the employee subsequently of the same type as the problem label be given priority evaluation; and to obtain the performance data before training and the subsequent performance data generated by the priority evaluation, and to quantify the improvement effect of the training task by comparing the changes of relevant indicators in the employee's performance data before and after performing the training task.

10. The system according to claim 9, characterized in that, Also includes: The profile generation and evaluation module is used to generate or update the personal competency profile associated with the employee based on the quantified improvement effect, and to evaluate the effectiveness of the micro-learning module in the knowledge base.

Images