Multi-terminal interaction intelligent training management method and device based on enterprise knowledge base

Abstract

This invention relates to the field of computer technology and provides a multi-terminal interactive intelligent training management method and device based on an enterprise knowledge base. The method includes: receiving training requests from the management end and parsing them to obtain matching training projects; verifying the login information of the trainee and loading the corresponding job competency model; retrieving response information from the enterprise knowledge base based on the trainee's question information and the job competency model; filtering and matching training projects and combining them with the response information before pushing them to the trainee; receiving learning progress query requests from team leaders and statistically sorting them for feedback to the management end; generating supplementary learning suggestions based on reminders and sending them to the corresponding trainees; and updating the job competency model based on the training feedback information from the trainees. This invention achieves closed-loop management of the entire process from training demand input, training project generation, learning process support, team progress supervision to continuous model optimization.

Description

Technical Field

[0001] This invention relates to the technical field of integrating enterprise digital learning systems with artificial intelligence, and in particular to a multi-terminal interactive intelligent training management method and device based on an enterprise knowledge base, applicable to scenarios requiring professional training such as retail sales staff training, insurance sales staff training, new energy vehicle sales staff training, and pharmaceutical sales staff training. Background Technology

[0002] As enterprises accelerate their digital transformation, the demand for employees to quickly adapt their skills to business changes is increasing. Corporate training is evolving from standardized and modular approaches to intelligent, personalized, and closed-loop systems. Against this backdrop, enterprise-level intelligent training systems that integrate Natural Language Interface (LUI) and Graphical User Interface (GUI) have become a research and application hotspot.

[0003] In existing technologies, mainstream SaaS training platforms generally adopt a modular architecture, requiring users to configure elements such as courses, exams, and learning paths through multi-step form filling. Some systems introduce NLP-based question-answering robots that can parse user questions and return knowledge point summaries or course links. Their core relies on a preset process engine to drive the creation and execution of training projects. The system integrates basic components such as organizational structure management, capability model library, and learning resource library, forming a relatively complete training management ecosystem.

[0004] However, the aforementioned technologies still have significant limitations: First, training program creation heavily relies on manual operation and experience-based judgment, resulting in cumbersome processes, delayed responses, and difficulty in supporting agile business iterations. Second, AI-assisted recommendations are limited to information retrieval, lacking modeling and integration of job competency characteristics, and thus unable to achieve project-level structured arrangement and dynamic generation. Third, the learning process lacks a precise intervention mechanism based on competency gaps; team leader supervision remains at the level of progress visibility and passive reminders, failing to form a closed loop with competency assessment, learning suggestions, and feedback optimization. Fourth, the system as a whole lacks a full-link intelligent management mechanism encompassing demand input, solution generation, learning support, supervisory intervention, and model evolution, leading to misallocation of training resources, insufficient personalization, and low business conversion rates. Therefore, existing intelligent learning methods for enterprise employees suffer from low learning efficiency. Summary of the Invention

[0005] This invention provides a multi-terminal interactive intelligent training management method and device based on an enterprise knowledge base, aiming to solve the problem of low learning efficiency in existing intelligent learning methods for enterprise employees.

[0006] In a first aspect, embodiments of the present invention provide a multi-terminal interactive intelligent training management method based on an enterprise knowledge base. The management method is applied to the server side of a management system, which includes the server side, a management terminal, a student terminal, and a team leader terminal. The management terminal, the student terminal, and the team leader terminal are respectively connected to the server side for communication. The method includes: Upon receiving the training request information input from the management terminal, the training request information is parsed to obtain training programs that match the training request information; If the login information from the trainee's end is verified, the job competency model corresponding to the login information is retrieved from the preset model library and loaded. If a question is received from the student's end, the pre-stored enterprise knowledge base is searched based on the question and the job competency model to obtain a matching answer. Training programs that match the response information are selected and combined with the response information before being pushed to the student's device; If a learning progress query request is received from the team leader, the learning progress of each student is counted, sorted, and the sorting results are fed back to the management terminal. Obtain the study reminder information fed back by the team leader based on the sorting result, generate supplementary learning suggestions based on the job competency model corresponding to the study reminder information, and send them to the student's end corresponding to the study reminder information; If training feedback information is received from the trainee's end, the job competency model in the model library is updated according to the preset update rules and the training feedback information.

[0007] Secondly, embodiments of the present invention also provide a multi-terminal interactive intelligent training management device based on an enterprise knowledge base. The management method is applied to the server side of a management system, which includes the server side, a management terminal, a student terminal, and a team leader terminal. The management terminal, the student terminal, and the team leader terminal are respectively connected to the server side for communication. The device is used to execute the multi-terminal interactive intelligent training management method based on an enterprise knowledge base as described in the first aspect. The device includes: The training project acquisition unit is used to receive training request information input by the management terminal, parse the training request information to obtain training projects that match the training request information; The job competency model loading unit is used to retrieve and load the job competency model corresponding to the login information from the preset model library if the login information verification from the trainee's terminal is successful. The response information acquisition unit is used to retrieve matching response information from the pre-stored enterprise knowledge base based on the question information and the job competency model if it receives the question information input by the trainee. The push unit is used to filter training programs that match the response information and combine them with the response information before pushing them to the student's end; The sorting result feedback unit is used to, upon receiving a learning progress query request input from the team leader, statistically analyze the learning progress of each student, sort them, and then feed the sorting result back to the management terminal. The supplementary learning suggestion sending unit is used to obtain the learning reminder information fed back by the team leader end according to the sorting result, generate supplementary learning suggestions according to the job competency model corresponding to the learning reminder information, and send them to the student end corresponding to the learning reminder information; The job competency model update unit is used to update the job competency model in the model library according to the preset update rules and the training feedback information if training feedback information is received from the trainee.

[0008] Thirdly, embodiments of the present invention also provide an electronic device, which includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the method described in the first aspect above.

[0009] Fourthly, embodiments of the present invention also provide a computer-readable storage medium storing a computer program, the computer program including program instructions that, when executed by a processor, can implement the method described in the first aspect.

[0010] This invention provides a multi-terminal interactive intelligent training management method based on an enterprise knowledge base. This method enables a natural language-driven training project creation mechanism, avoiding the cumbersome form-filling process of traditional systems and significantly improving the efficiency of training project creation; establishes a personalized learning foundation based on job competency models, allowing the system to accurately understand the current competency status of trainees; achieves intelligent knowledge matching integrating job competency models, overcoming the problem of inaccurate training project recommendations caused by keyword matching; constructs a dual-channel learning support mode of Q&A + recommendation, pushing targeted learning resources while answering questions; provides visualized team learning status monitoring capabilities, supporting team leaders to promptly grasp overall learning progress; achieves precise intervention based on competency gaps, breaking through the bottleneck of traditional learning reminders that only offer reminders without substantive learning guidance; and establishes a closed-loop feedback mechanism to continuously optimize the accuracy of job competency model assessment, thereby continuously improving the accuracy and business adaptability of subsequent training and learning. Attached Figure Description

[0011] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0012] Figure 1 A flowchart illustrating the multi-terminal interactive intelligent training management method based on an enterprise knowledge base provided in this embodiment of the invention; Figure 2 A schematic block diagram of a multi-terminal interactive intelligent training management device based on an enterprise knowledge base provided in an embodiment of the present invention; Figure 3 A schematic block diagram of an electronic device provided in an embodiment of the present invention; Figure 4 This is a schematic diagram illustrating an application scenario of the multi-terminal interactive intelligent training management method based on an enterprise knowledge base provided in an embodiment of the present invention. Detailed Implementation

[0013] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0014] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0015] It should also 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 invention. 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.

[0016] It should also be further understood that the term "and / or" as used in this specification and appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes such combinations. This invention provides a multi-terminal interactive intelligent training management method and apparatus based on an enterprise knowledge base. For specific application scenarios of this multi-terminal interactive intelligent training management method based on an enterprise knowledge base, please refer to... Figure 4 , Figure 4This is a schematic diagram illustrating an application scenario of the multi-terminal interactive intelligent training management method based on an enterprise knowledge base provided in an embodiment of the present invention. The multi-terminal interactive intelligent training management method based on an enterprise knowledge base is applied in, for example... Figure 4 In the application scenario, the multi-terminal interactive intelligent training management method based on enterprise knowledge base is applied to the server side 10 of the management system. The management system includes the server side 10, the management terminal 20, the student terminal 30, and the team leader terminal 40. The management terminal 20, the student terminal 30, and the team leader terminal 40 are respectively connected to the server side 10. The server side 10 can be implemented using a standalone server or a server cluster composed of multiple servers. The management terminal 20, the student terminal 30, and the team leader terminal 40 can be, but are not limited to, electronic devices such as servers, smartphones, tablets, and desktop computers. The invention will be described in detail below through specific embodiments.

[0017] Figure 1 This is a flowchart illustrating a multi-terminal interactive intelligent training management method based on an enterprise knowledge base, provided in an embodiment of the present invention. Figure 1 As shown, the method includes the following steps S110-S170.

[0018] S110. Upon receiving the training request information input by the management terminal, the training request information is parsed to obtain training programs that match the training request information.

[0019] Upon receiving the training request information input from the management terminal, the system parses the training request information to obtain training projects that match it. The training request information can be raw request data input by the administrator on the management terminal via natural language text, structured forms, or speech-to-text, containing information about training objectives, target audience, duration, and business scenarios; for example, it could be "Design a two-week anti-money laundering compliance training program for newly hired tellers." In this embodiment, this technical feature serves as the initial trigger condition for the entire training project generation process. Its parsing result will be used for subsequent operations such as calling the enterprise knowledge base for retrieval, resource matching, and generating structured solutions, constituting the starting point input of the method flow.

[0020] In a specific embodiment, step S110 includes the following sub-steps: performing intent parsing on the training request information to obtain the corresponding question form and feeding it back to the management terminal; obtaining the form data fed back by the management terminal based on the question form, and performing a search and matching on the pre-stored enterprise knowledge base to obtain the corresponding training project.

[0021] Specifically, a Questionnaire Form can refer to a structured data carrier used to dynamically render interactive form items in the management GUI interface. Its content is generated by the AI ​​Agent based on the semantic information of the current training request. For example, based on the information corresponding to "new hire" and "hiring time" in the Questionnaire Form, single-choice options such as "within 2 months", "2-4 months", and "4-6 months" are generated. Based on the information corresponding to "teller" and "position" in the Questionnaire Form, multiple-choice options such as "trainee teller", "junior teller", and "intermediate teller" are generated. An input box corresponding to "training difficulty" is also generated in the Questionnaire Form. The Questionnaire Form can be a JSON Schema format object containing field definitions, data type constraints, and validation rules. The information in the format supports single-choice, multiple-choice, and input box types. In this embodiment, the role of the Questionnaire Form is to actively identify and complete the key business parameters missing in the user's natural language input, transforming fuzzy semantics into structured input that can be parsed and executed by the system, thereby supporting the accurate generation of subsequent training programs.

[0022] Furthermore, form data can refer to a set of structured business parameters filled in and submitted by the management user based on the question form, with its fields strictly consistent with the definition of the question form; the Enterprise Knowledge Base can refer to a set of structured and unstructured knowledge stored on the server, including but not limited to organizational structure data, job competency tags, course resource metadata, exam question banks, intelligent tutoring scripts, enterprise policy documents, and learning map configurations; in this embodiment, form data serves as the primary key and constraint for retrieval, used to locate training projects in the Enterprise Knowledge Base that highly match business needs. Its role is to transform manually confirmed structured input into an executable training plan prototype, ensuring that the generated results are strongly coupled with the enterprise's real business scenarios.

[0023] For example, in this application, a management user can input training request information based on natural language description, such as "I want to conduct new employee onboarding training," in a GUI interface and submit the corresponding form data. After receiving the training request information, the server uses a lightweight intent classification model to determine that it belongs to the onboarding training project creation type. It then calls the named entity recognition module to extract the implicit job keyword "new employee" and business domain keyword "onboarding" from the training request information. This triggers the question form tool to generate fields to be completed (such as department, expected completion time, training difficulty, and onboarding time), ultimately forming structured request data with job dimensions and business constraints, which serves as the input basis for subsequent matching of training projects.

[0024] The system further searches and matches the pre-stored enterprise knowledge base based on the obtained form data to obtain training programs corresponding to the form data. Search criteria can be directly set based on the form data (e.g., generating job search criteria based on data corresponding to "job positions" in the form data) to search the enterprise knowledge base. If no training program in the enterprise knowledge base contains the corresponding search tags, the search criteria can be used to determine whether the search tags of the training program meet the corresponding conditions, and then the search can retrieve training programs that match the search criteria.

[0025] Alternatively, a feature vector corresponding to the form data can be extracted. For example, the extracted feature vector can be represented as {0.3, 0.8, 0.4, ...}. The similarity value between the feature vector and the features of each training project can be calculated, and the training project with the highest similarity value can be selected as the training project that matches the form data. The similarity value can be a cosine similarity value.

[0026] S120. If the login information from the trainee's end is verified, the job competency model corresponding to the login information is retrieved from the preset model library and loaded.

[0027] If the login information from the student's end is verified, the job competency model corresponding to the login information is retrieved from the pre-set model library and loaded. Login information can refer to the identity authentication credentials submitted by the student, including but not limited to account password, SMS verification code, WeChat / DingTalk login-free token, or biometric identifier. Each student has a corresponding job competency model in the model library. The job competency model can be a structured data object stored in the model library, containing multiple competency dimensions and their corresponding scores, such as compliance awareness, communication skills, job knowledge, market trends, and adaptability. The value range for each dimension is [0,1]. In this embodiment, this technical feature serves to establish an individual student profile, providing data support for subsequent personalized Q&A, resource matching, and learning suggestion generation based on competency differences.

[0028] Specifically, if the student logs in using WeChat Work by scanning a QR code, the system obtains the employee's ID and job code through the OAuth2.0 protocol. Based on this, it retrieves the corresponding job competency model from the model library and loads it into the runtime context of the AI-assisted learning service for subsequent steps.

[0029] S130. If a question is received from the student's terminal, the pre-stored enterprise knowledge base is searched according to the question and the job competency model to obtain a matching answer.

[0030] If a question is received from the trainee's terminal, the pre-stored enterprise knowledge base is searched based on the question and the job competency model to obtain a matching response. The question can be a technical, business, or procedural question entered by the trainee in the LUI interface in text or voice format. The enterprise knowledge base can be a unified knowledge platform deployed on a server, containing policy documents, operation manuals, FAQ collections, compliance guidelines, a typical case library, and approved learning resource metadata. In this embodiment, this technical feature enables targeted knowledge retrieval based on job competency profiles, ensuring that the response not only meets semantic relevance but also aligns with the trainee's current competency level and job responsibilities.

[0031] In a specific embodiment, step S130 includes the following sub-steps: obtaining the question feature vector corresponding to the question information; combining the model features of the job competency model with the question feature vector to obtain a corresponding combination vector; and retrieving knowledge from the enterprise knowledge base based on the combination vector to obtain knowledge that matches the combination vector as the corresponding response information.

[0032] Specifically, a question embedding vector can be a fixed-dimensional numerical vector obtained by processing the natural language question information input by the learner through a text encoding model, used to represent the position and meaning of the question in the semantic space. This question embedding vector can be obtained by embedding and encoding text-type question information through a lightweight text encoder (such as a fine-tuned version of Sentence-BERT); it can also be generated by performing word segmentation, word vector lookup, and pooling aggregation operations on the question text; or it can be a context-aware encoding method that implicitly captures the potential business intent while preserving the core semantics of the question.

[0033] Furthermore, the model feature of the position competency model can refer to the structured feature representation extracted from the position competency model corresponding to the student's login information, loaded from a pre-built model library. It is in the form of a fixed-dimensional numerical vector, with each dimension corresponding to a quantitative score (range [0,1]) for competency dimensions such as compliance awareness, communication skills, job knowledge, market trends, and adaptability. The model feature of this position competency model is combined with the question feature vector to obtain a combined vector. This combined vector constitutes the basis for a composite query. The combination method does not change the physical meaning of any original vector; it is only used to construct a joint semantic representation that integrates the user's competency profile and the real-time question intent. Specifically, the combined vector is obtained by sequentially concatenating the obtained question feature vector with the model feature. For example, if the question feature vector is 1×M dimensions and the model feature is 1×N dimensions, the concatenated vector will be 1×(M+N) dimensional.

[0034] Specifically, the process involves retrieving knowledge from the enterprise knowledge base based on the combined vector, and obtaining knowledge that matches the combined vector as the corresponding response information. This includes: calculating the similarity value between the combined vector and the knowledge vectors corresponding to each piece of knowledge in the enterprise knowledge base; and obtaining the knowledge with the highest similarity value as the knowledge that matches the combined vector.

[0035] Furthermore, the Enterprise Knowledge Base can refer to a collection of structured and unstructured knowledge pre-stored on the server side, including but not limited to corporate policy documents (such as the "Anti-Money Laundering Operation Guidelines"), business process descriptions, typical violation cases, standard script templates, compliance exam question bank metadata, intelligent tutoring scripts, etc. All knowledge items have been pre-vectorized and stored in a vector database. The combined vector serves as a unified query vector, and an Approximate Nearest Neighbor Search is performed in the vector database. The cosine similarity between the knowledge vector corresponding to each knowledge in the Enterprise Knowledge Base and the combined vector can be calculated, and the K knowledge items with the highest cosine similarity can be returned. The most matching knowledge item is selected as the response information. This process ensures that the search results not only meet the semantic matching of the question but also adapt to the current ability level of the learner—for example, for learners with lower compliance awareness scores, the system tends to return knowledge content with stronger explanations and more detailed steps.

[0036] For example, this application could involve a trainee inputting "What should I do if a customer doesn't have their ID card?" into the LUI. The system would then convert this into a question feature vector, combine it with the model features of the trainee's job competency model to obtain a combined vector, match and obtain the knowledge corresponding to the combined vector as the clauses in the "Customer Identification Management Measures" regarding the use of auxiliary documents, and generate explanatory response information. For example, the generated response information could include "According to the system requirements, temporary ID cards or social security cards and other auxiliary documents can be used for identity verification," while also indicating that the response comes from Article 3.2.1 of the enterprise knowledge base.

[0037] S140. Select training programs that match the response information and combine them with the response information before pushing them to the student's end.

[0038] Training programs matching the response information are selected and combined with the response information before being pushed to the learner. Training programs can refer to structured teaching units already published in a training program library. Each program includes stage divisions, job suitability tags, a list of associated resources, and learning path configurations. In this embodiment, this technical feature extends abstract knowledge answers into actionable learning tasks, achieving a closed-loop transformation from knowing what to do to how to do it, thus improving the efficiency of knowledge application.

[0039] In a specific embodiment, step S140 includes the following sub-steps: based on the business keywords in the response information, perform tag matching in the training project library to filter out projects containing the same keywords or synonymous tags; select the Top-N projects based on the semantic similarity ranking between the response information and the description text of each training project; and, in conjunction with the trainee's current job competency model, prioritize training projects that can make up for their weak dimensions.

[0040] Specifically, in response to the aforementioned reply "What should I do if the customer doesn't have their ID card?", the system identifies the keywords "auxiliary documents" and "identity verification", matches the special situation handling process stage under the "Teller Compliance Operation Improvement Training" project in the training project library, and encapsulates the training project card (including stage name, estimated time, and associated micro-course ID) together with the reply information into a JSON format message, which is then pushed to the student's homepage pop-up window.

[0041] S150. If a learning progress query request is received from the team leader, the learning progress of each student is counted, sorted, and the sorting results are fed back to the management terminal.

[0042] If a learning progress query request is received from the team leader, the learning progress of each student is statistically analyzed, sorted, and the sorting results are fed back to the management terminal. Learning progress can refer to the proportion of learning tasks completed by a student in a specified training program relative to the total number of tasks, or a weighted comprehensive completion indicator. The calculation basis includes course viewing time, exam scores, frequency of tutoring interactions, and resource click-through rate. In this embodiment, this technical feature provides team leaders with a quantifiable view of the team's learning status and highlights individuals with abnormal progress through a sorting mechanism, supporting data-driven management decisions.

[0043] In a specific embodiment, step S150 includes the following sub-steps: aggregating the learning behavior logs of each student by project dimension, counting the number of courses completed, the number of exams passed, and the number of practice sessions to achieve the target, and generating a weighted comprehensive progress score; assigning higher weight to recent learning behaviors based on the time decay function and dynamically updating the progress value; and making competency-oriented corrections to the progress value based on the improvement of each dimension in the job competency model.

[0044] Specifically, when a branch manager clicks to view the progress of the XX training project on the team leader's end, the system retrieves the learning logs of all tellers participating in the training from the database, calculates the number of courses completed / total number of courses for each person (weight 0.4), the average exam score (weight 0.3), and the practice pass rate (weight 0.3), generates a progress score and sorts it in descending order, generates a table containing name, position, progress percentage, and ranking, and pushes it synchronously to the data dashboard module on the management end.

[0045] S160. Obtain the study reminder information fed back by the team leader based on the sorting result, generate supplementary learning suggestions based on the job competency model corresponding to the study reminder information, and send them to the student's end corresponding to the study reminder information.

[0046] The system obtains the learning reminder information fed back by the team leader based on the sorting results, generates supplementary learning suggestions based on the job competency model corresponding to the learning reminder information, and sends them to the student's end corresponding to the learning reminder information. The learning reminder information can refer to an operation instruction initiated by the team leader targeting a specific student or student group, carrying a target student identifier and an optional intervention reason tag; for example, the team leader can select one or more students with lagging learning progress and set an intervention reason tag (such as lagging learning progress), generating the learning reminder information based on the target student identifier corresponding to the selected students. The supplementary learning suggestions can refer to structured suggestions generated based on the weak dimensions of the job competency model, with clear learning objectives and resource orientation; the role of this technical feature in this embodiment is to transform the team leader's subjective intervention intention into an executable learning intervention strategy, achieving precise coupling between supervisory actions and the student's competency growth path.

[0047] In a specific embodiment, step S160 includes the following sub-steps: obtaining the preferred strategy corresponding to the scores of each dimension in the job competency model according to the preset learning strategy; generating corresponding supplementary learning suggestions according to the preferred strategy and sending them to the student's terminal corresponding to the learning reminder information.

[0048] Specifically, the learning strategy is a set of rules pre-set on the server side, used to map the quantitative scores of each dimension in the job competency model to the corresponding learning intervention type; the scores of each dimension in the job competency model can refer to continuous values ​​in the range of [0,1] on dimensions such as compliance awareness, communication skills, job knowledge, market changes, and flexible response; the preferred strategy can refer to the behavioral guidance type with a clear learning orientation determined for the scores of multiple dimensions, such as obtaining one or more dimensions with the lowest scores as target dimensions, and matching a strategy in the learning strategy that matches the score of the target dimension as the preferred strategy.

[0049] Furthermore, supplementary learning suggestions are generated based on the optimized strategy. For example, if the optimized strategy is "compliance awareness consolidation learning," then matching training programs can be obtained based on this optimized strategy, and corresponding supplementary learning suggestions can be generated. If the generated supplementary learning suggestion is "This employee scored low in the 'compliance awareness' dimension, and it is recommended to supplement with the study of 'Analysis of Typical Violation Cases'," the generated supplementary learning suggestion will be sent to the learner's end corresponding to the reminder message.

[0050] S170. If training feedback information is received from the trainee's end, the job competency model in the model library is updated according to the preset update rules and the training feedback information.

[0051] If training feedback information is received from the trainee's end, the job competency model in the model library is updated according to the preset update rules and the training feedback information. Training feedback information can refer to unstructured data submitted by trainees during exams after completing training programs, including answers to corresponding test questions. Update rules can refer to a set of logical configurations preset on the server side, including components such as a tag configuration table, normalization function, and competency assessment model. This technical feature in this embodiment serves to construct a dynamic evolution mechanism for the job competency model, enabling the model to continuously optimize based on actual training effectiveness feedback, thus avoiding a disconnect between the static model and business development.

[0052] In a specific embodiment, step S170 includes the following sub-steps: marking the training feedback information according to the marking configuration table in the update rule to obtain the marking information corresponding to each dimension; normalizing the marking information of each dimension according to the normalization function in the update rule to obtain the corresponding normalized features; combining the normalized features with the model features of the job competency model to obtain the corresponding combined features; evaluating and analyzing the combined features according to the competency assessment model in the update rule to obtain the corresponding analysis score; and updating the job competency model according to the analysis score.

[0053] Specifically, the Mapping Configuration Table is used to map unstructured or semi-structured training feedback information to assign dimension tags to each question in the training feedback information. For example, if the associated dimensions of question 1 are communication skills and job knowledge, then dimension tags corresponding to communication skills and job knowledge will be added to question 1; if the associated dimensions of question 2 are job knowledge and adaptability, then dimension tags corresponding to job knowledge and adaptability will be added to question 2.

[0054] Furthermore, the labeled information for each dimension is normalized using a normalization function. Each question corresponds to a score; a correct answer earns the corresponding score. The scores for each dimension are then accumulated based on the labeled information and the score for each question, and the accumulated scores for each dimension are normalized. For example, if question 1 has a score of 4 points, and its labeled information is communication skills and job knowledge, a correct answer to question 1 earns 2 points (4 / 2) in the communication skills dimension and 2 points (4 / 2) in the job knowledge dimension. The scores for each question corresponding to each dimension are then accumulated and calculated. Afterward, normalization is performed, such as by using a normalization function like s... r =x r -x min / (x max -x min ), s r x is the normalized value corresponding to dimension r. r x is the cumulative score corresponding to dimension r. max The maximum score in dimension r (all answers are correct), x max The minimum score for dimension r (all answers are incorrect). The normalized values ​​for each dimension range from [0,1]. Combining the normalized values ​​for each dimension yields the corresponding normalized features.

[0055] The obtained normalized features are combined with the model features of the job competency model to obtain combined features. These combined features are then input into the competency assessment model for evaluation and analysis, resulting in corresponding analytical scores.

[0056] Furthermore, the competency evaluation model can be a lightweight machine learning model deployed on a server. Its input is a combination of features, and its output is the updated suggested scores for each competency dimension. The updated suggested scores for each competency dimension are combined to form the corresponding analytical score. The competency evaluation model can be one of the following: Multilayer Perceptron (MLP), Gradient Boosting Tree (GBDT), or rule engine.

[0057] The existing scores in the job competency model are updated based on the analysis scores. This update can refer to smoothly adjusting the current scores for the corresponding dimensions of the job competency model using the analysis scores as an adjustment parameter. The update can be performed using an Exponentially Weighted Moving Average (EWMA) method, with the following formula: ; in, For the first The updated rating values ​​for each dimension The output of the capability assessment model Update suggestion scores for each dimension, The first one in the job competency model before the update The rating values ​​corresponding to each dimension The learning rate parameter (e.g., 0.3) serves to achieve the gradual evolution of the job competency model: it fully absorbs the competency change trends reflected in the feedback from this training, and suppresses model oscillations caused by accidental or one-sided feedback through a smoothing mechanism, ensuring the long-term stability and business credibility of the model.

[0058] Furthermore, after receiving the question information input by the student terminal, the method further includes: determining whether the question information is related to the dialogue practice skills according to preset dialogue judgment rules; if the question information is not related to the dialogue practice skills, performing the step of searching the pre-stored enterprise knowledge base according to the question information and the job competency model; if the question information is related to the dialogue practice skills, generating a simulated dialogue corresponding to the question information and the job competency model according to the preset intelligent coaching model and the enterprise knowledge base and pushing it to the student terminal; if the interaction information fed back by the student terminal according to the simulated dialogue is received, obtaining an adjustment strategy matching the interaction information and the simulated dialogue; generating a simulated adjustment dialogue corresponding to the adjustment strategy, the question information and the job competency model according to the intelligent coaching model and pushing it to the student terminal.

[0059] The system uses dialogue judgment rules to determine whether a question is related to dialogue practice skills. Specifically, it obtains the question's feature vector and calculates the cosine similarity between this feature vector and the standard feature vector in the dialogue judgment rules. The standard feature vector corresponds to the training category of dialogue practice skills. The system then checks if the cosine similarity between the question's feature vector and the standard feature vector is greater than a similarity threshold set in the dialogue judgment rules. If it is greater, the question is considered related to dialogue practice skills; otherwise, it is considered not related.

[0060] If the question is not related to dialogue practice skills, the subsequent steps of searching the enterprise knowledge base continue. If it is related to dialogue practice skills, a simulated dialogue corresponding to the question information and the job competency model is generated based on the intelligent coaching model and the enterprise knowledge base. Specifically, the enterprise knowledge base can be searched based on the question information and the job competency model to obtain the response information with the highest matching degree. The simulated dialogue corresponding to the response information is generated based on the intelligent coaching model. The simulated dialogue includes simulated question content and simulated response content. The simulated dialogue uses the response information as the simulated response content, and intelligently generates matching simulated question content based on the simulated response content. For example, the intelligent coaching model obtains all question information corresponding to the simulated response content and performs semantic integration, such as obtaining the question feature vectors of all question information and averaging them to achieve semantic integration; the integrated and encoded language text content is obtained as the corresponding simulated question content. The generated simulated dialogue is pushed to the trainee's end, where the trainee displays the simulated question content in the simulated dialogue.

[0061] Students using the learning platform can interact with simulated dialogues, providing feedback information—the text information generated from their responses to simulated questions. The system then identifies adjustment strategies that match this interaction information and the simulated dialogue. Specifically, it calculates the similarity between the interaction information and the simulated responses in the dialogue, obtaining a similarity value. Based on this similarity value, a corresponding adjustment strategy is obtained. For example, each adjustment strategy corresponds to a similarity value range; the system can then identify an adjustment strategy that matches the calculated similarity value within that range, serving as the appropriate adjustment strategy for the interaction information and the simulated dialogue.

[0062] If the calculated similarity value is 0.56, which falls within the similarity value range of [0.40, 0.59], and the adjustment strategy that matches this similarity value range is "reduce difficulty", then the matching adjustment strategy is determined to be "reduce difficulty".

[0063] Based on the intelligent coaching model, a simulated adjustment dialogue is generated and adjusted according to the question information and the job competency model. Specifically, the enterprise knowledge base is searched based on the question information and job competency model to obtain multiple matching responses. These responses are then filtered based on the difficulty of the previous round of responses and the adjustment strategy. Each response also includes a corresponding difficulty tag, such as "Level 1 Difficulty," "Level 2 Difficulty,"... "Level 5 Difficulty," etc. If the difficulty tag of the previous response was "Level 4 Difficulty," then the current filtering condition is determined to be "Level 3 Difficulty." The response with the highest similarity to "Level 3 Difficulty" is selected from the multiple responses, and a simulated adjustment dialogue is generated accordingly. The generation process of the simulated adjustment dialogue is similar to that of the simulated dialogue. Upon receiving the interactive information from the trainee based on the simulated adjustment dialogue, the step of obtaining the matching adjustment strategy is repeated.

[0064] Furthermore, after updating the job competency models in the model library according to the preset update rules and the training feedback information, the method further includes: obtaining the job target competencies that match the job competency models of each trainee; comparing and analyzing the job competency models of each trainee with the corresponding job target competencies to obtain the corresponding weak competency ranking; obtaining reinforcement training suggestions corresponding to the weak competency ranking according to the preset reinforcement strategy and pushing them to the trainee end corresponding to the job competency model.

[0065] Each trainee is assigned a specific job competency model. Furthermore, based on the trainee's position and department within the organizational structure, the corresponding job target competencies are determined. These job target competencies represent the basic skill requirements the trainee must achieve. The dimensions of these job target competencies are the same as those of the job competency model.

[0066] A comparative analysis is performed between the job competency model and the obtained job target competencies. This involves calculating the score difference between the job competency model and the job target competencies across each dimension, and ranking the dimensions based on these score differences to obtain a ranking of weak competencies. Dimensions with the largest gap from the job target competencies and negative score differences are ranked higher. Based on pre-set reinforcement strategies, reinforcement training suggestions corresponding to the weak competency ranking are generated, such as for the three dimensions with the highest ranking of weak competencies. The reinforcement strategy includes multiple sets of training suggestions, each corresponding to a different dimension. The set of training suggestions that matches the three dimensions with the highest ranking of weak competencies can be obtained as the reinforcement training suggestion, such as "Recommend learning the training project of the 'Customer Objection Handling' micro-course". The generated reinforcement training suggestions are then pushed to the learner's end corresponding to the job competency model.

[0067] The multi-terminal interactive intelligent training management method based on an enterprise knowledge base disclosed in the above embodiments includes the following steps: receiving training request information input by the management terminal, parsing the training request information to obtain training projects matching the training request information; if the login information verification of the student terminal is successful, retrieving the job competency model corresponding to the login information from a preset model library and loading it; if a question is received from the student terminal, searching the pre-stored enterprise knowledge base according to the question and the job competency model to obtain matching answer information; filtering training projects matching the answer information and combining them with the answer information and pushing them to the student terminal; if a learning progress query request is received from the team leader terminal, statistically analyzing and sorting the learning progress of each student, and feeding back the sorting result to the management terminal; obtaining the learning reminder information fed back by the team leader terminal according to the sorting result, generating supplementary learning suggestions according to the job competency model corresponding to the learning reminder information and sending them to the student terminal corresponding to the learning reminder information; if training feedback information is received from the student terminal, updating the job competency model in the model library according to preset update rules and the training feedback information.

[0068] Figure 2 This is a schematic block diagram of a multi-terminal interactive intelligent training management device based on an enterprise knowledge base, provided as an embodiment of the present invention. Figure 2 As shown, corresponding to the above-mentioned multi-terminal interactive intelligent training management method based on enterprise knowledge base, the present invention also provides a multi-terminal interactive intelligent training management device based on enterprise knowledge base, the device being configured in, as shown in... Figure 4 In application scenarios. For details, please refer to [link / reference]. Figure 2 The multi-terminal interactive intelligent training management device 700 based on an enterprise knowledge base includes: The training project acquisition unit 701 is used to receive training request information input by the management terminal, parse the training request information to obtain training projects that match the training request information; The job competency model loading unit 702 is used to retrieve and load the job competency model corresponding to the login information from the preset model library if the login information verification from the trainee's terminal is successful. The response information acquisition unit 703 is used to retrieve matching response information from the pre-stored enterprise knowledge base based on the question information and the job competency model if it receives the question information input by the trainee terminal. The push unit 704 is used to filter training programs that match the response information and combine them with the response information before pushing them to the student's end; The sorting result feedback unit 705 is used to, upon receiving a learning progress query request input from the team leader, statistically analyze the learning progress of each student, sort them, and then feed the sorting result back to the management terminal. The supplementary learning suggestion sending unit 706 is used to obtain the learning reminder information fed back by the team leader end according to the sorting result, generate supplementary learning suggestions according to the job competency model corresponding to the learning reminder information, and send them to the student end corresponding to the learning reminder information. The job competency model update unit 707 is used to update the job competency model in the model library according to the preset update rules and the training feedback information if training feedback information is received from the trainee.

[0069] It should be noted that those skilled in the art can clearly understand that the specific implementation process of the above-mentioned multi-terminal interactive intelligent training management device based on enterprise knowledge base and each unit can be referred to the corresponding description in the foregoing method embodiments. For the sake of convenience and brevity, it will not be repeated here.

[0070] The aforementioned multi-terminal interactive intelligent training management device based on an enterprise knowledge base can be implemented as a computer program, which can be used in various ways, such as... Figure 3 It runs on the electronic device shown.

[0071] Please see Figure 3 , Figure 3 This is a schematic block diagram of an electronic device provided in an embodiment of the present invention. The electronic device 800 can be a terminal or a server. The terminal can be an electronic device with communication functions. The server can be a standalone server or a server cluster composed of multiple servers.

[0072] See Figure 3 The electronic device 800 includes a processor 802, a memory, and a network interface 805 connected via a system bus 801. The memory may include a non-volatile storage medium 803 and internal memory 804.

[0073] The non-volatile storage medium 803 can store an operating system 8031 ​​and a computer program 8032. The computer program 8032 includes program instructions that, when executed, cause the processor 802 to execute a multi-terminal interactive intelligent training management method based on an enterprise knowledge base.

[0074] The processor 802 provides computing and control capabilities to support the operation of the entire electronic device 800.

[0075] The internal memory 804 provides an environment for the execution of the computer program 8032 in the non-volatile storage medium 803. When the computer program 8032 is executed by the processor 802, the processor 802 can execute a multi-terminal interactive intelligent training management method based on an enterprise knowledge base.

[0076] This network interface 805 is used for network communication with other devices. Those skilled in the art will understand that... Figure 3 The structure shown is merely a block diagram of a portion of the structure related to the present invention and does not constitute a limitation on the electronic device 800 to which the present invention is applied. The specific electronic device 800 may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0077] The processor 802 is used to run the computer program 8032 stored in the memory to implement the steps included in the above-mentioned multi-terminal interactive intelligent training management method based on enterprise knowledge base.

[0078] It should be understood that, in this embodiment of the invention, the processor 802 may be a Central Processing Unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0079] It will be understood by those skilled in the art that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program includes program instructions and can be stored in a storage medium, which is a computer-readable storage medium. The program instructions are executed by at least one processor in the computer system to implement the process steps of the embodiments of the above methods.

[0080] Therefore, the present invention also provides a storage medium. This storage medium can be a computer-readable storage medium. The storage medium stores a computer program, wherein the computer program includes program instructions. When executed by a processor, the program instructions cause the processor to perform the steps included in the above-described multi-terminal interactive intelligent training management method based on an enterprise knowledge base.

[0081] The storage medium can be any computer-readable storage medium capable of storing program code, such as a USB flash drive, portable hard drive, read-only memory (ROM), magnetic disk, or optical disk.

[0082] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0083] In the several embodiments provided by this invention, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For example, the division of each unit is merely a logical functional division, and there may be other division methods in actual implementation. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed.

[0084] The steps in the method of this invention can be adjusted, merged, or reduced in order according to actual needs. The units in the device of this invention can be merged, divided, or reduced according to actual needs. Furthermore, the functional units in the various embodiments of this invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0085] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause an electronic device (which may be a personal computer, a terminal, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention.

[0086] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A multi-terminal interactive intelligent training management method based on an enterprise knowledge base, characterized in that, The management method is applied to the server side of the management system, which includes the server side, management side, student side and team leader side; The management terminal, the student terminal, and the team leader terminal are respectively connected to the server terminal for communication. The method includes: Upon receiving the training request information input from the management terminal, the training request information is parsed to obtain training programs that match the training request information; If the login information from the trainee's end is verified, the job competency model corresponding to the login information is retrieved from the preset model library and loaded. If a question is received from the student's end, the pre-stored enterprise knowledge base is searched based on the question and the job competency model to obtain a matching answer. Training programs that match the response information are selected and combined with the response information before being pushed to the student's device; If a learning progress query request is received from the team leader, the learning progress of each student is counted, sorted, and the sorting results are fed back to the management terminal. Obtain the study reminder information fed back by the team leader based on the sorting result, generate supplementary learning suggestions based on the job competency model corresponding to the study reminder information, and send them to the student's end corresponding to the study reminder information; If training feedback information is received from the trainee's end, the job competency model in the model library is updated according to the preset update rules and the training feedback information.

2. The multi-terminal interactive intelligent training management method based on an enterprise knowledge base according to claim 1, characterized in that, The step of parsing the training request information to obtain training programs that match the training request information includes: The training request information is analyzed to obtain the corresponding question form and then fed back to the management end. The management system retrieves the form data returned by the question form and searches and matches it against the pre-stored enterprise knowledge base to obtain the corresponding training program.

3. The multi-terminal interactive intelligent training management method based on an enterprise knowledge base according to claim 1 or 2, characterized in that, The step of searching the pre-stored enterprise knowledge base based on the question information and the job competency model to obtain matching response information includes: Obtain the question feature vector corresponding to the question information; The model features of the job competency model are combined with the question feature vector to form a corresponding combination vector; The knowledge in the enterprise knowledge base is retrieved based on the combined vector, and the knowledge that matches the combined vector is obtained as the corresponding response information.

4. The multi-terminal interactive intelligent training management method based on an enterprise knowledge base according to claim 3, characterized in that, The step of generating supplementary learning suggestions based on the job competency model corresponding to the learning reminder information and sending them to the student's end corresponding to the learning reminder information includes: The optimal strategy corresponding to the scores of each dimension in the job competency model is obtained based on the preset learning strategy. Based on the preferred strategy, corresponding supplementary learning suggestions are generated and sent to the student's end corresponding to the learning reminder message.

5. The multi-terminal interactive intelligent training management method based on an enterprise knowledge base according to claim 4, characterized in that, The step of updating the job competency model in the model library according to the preset update rules and the training feedback information includes: The training feedback information is dimension-labeled according to the label configuration table in the update rules to obtain the label information corresponding to each dimension. The label information of each dimension is normalized according to the normalization function in the update rule to obtain the corresponding normalized features; The normalized features are combined with the model features of the job competency model to obtain the corresponding combined features; The combined features are evaluated and analyzed according to the capability assessment model in the update rules to obtain the corresponding analysis score; The job competency model is updated based on the analysis scores.

6. The multi-terminal interactive intelligent training management method based on an enterprise knowledge base according to claim 5, characterized in that, After receiving the question information input by the student, the method further includes: Determine whether the question information is related to the dialogue practice skills based on the preset dialogue judgment rules; If the question information is not related to the dialogue practice skills, the step of retrieving the pre-stored enterprise knowledge base based on the question information and the job competency model is executed; If the question information is related to the dialogue practice skills, a simulated dialogue corresponding to the question information and the job competency model is generated based on the preset intelligent coaching model and enterprise knowledge base and pushed to the trainee's terminal; If the interaction information received from the student terminal based on the simulated dialogue is obtained, an adjustment strategy matching the interaction information and the simulated dialogue is obtained; The intelligent coaching model generates a simulated adjustment dialogue corresponding to the adjustment strategy, the question information, and the job competency model, and pushes it to the trainee's end.

7. The multi-terminal interactive intelligent training management method based on an enterprise knowledge base according to claim 6, characterized in that, After updating the job competency models in the model library according to preset update rules and training feedback information, the method further includes: Obtain the target competencies for each trainee that match their job competency model; A comparative analysis was conducted on the job competency models described by each trainee and the corresponding job target competencies to obtain a ranking of their weak competencies. Based on the preset reinforcement strategy, reinforcement training suggestions corresponding to the ranking of weak capabilities are obtained and pushed to the trainee's end corresponding to the job capability model.

8. A multi-terminal interactive intelligent training management device based on an enterprise knowledge base, characterized in that, The management method is applied to the server side of the management system, which includes the server side, the management side, the student side, and the team leader side; the management side, the student side, and the team leader side are respectively connected to the server side for communication; the device is used to execute the multi-terminal interactive intelligent training management method based on an enterprise knowledge base as described in any one of claims 1-7, and the device includes: The training project acquisition unit is used to receive training request information input by the management terminal, parse the training request information to obtain training projects that match the training request information; The job competency model loading unit is used to retrieve and load the job competency model corresponding to the login information from the preset model library if the login information verification from the trainee's terminal is successful. The response information acquisition unit is used to retrieve matching response information from the pre-stored enterprise knowledge base based on the question information and the job competency model if it receives the question information input by the trainee. The push unit is used to filter training programs that match the response information and combine them with the response information before pushing them to the student's end; The sorting result feedback unit is used to, upon receiving a learning progress query request input from the team leader, statistically analyze the learning progress of each student, sort them, and then feed the sorting result back to the management terminal. The supplementary learning suggestion sending unit is used to obtain the learning reminder information fed back by the team leader end according to the sorting result, generate supplementary learning suggestions according to the job competency model corresponding to the learning reminder information, and send them to the student end corresponding to the learning reminder information; The job competency model update unit is used to update the job competency model in the model library according to the preset update rules and the training feedback information if training feedback information is received from the trainee.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the multi-terminal interactive intelligent training management method based on an enterprise knowledge base as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, which includes program instructions. When executed by a processor, the program instructions cause the processor to perform the multi-terminal interactive intelligent training management method based on an enterprise knowledge base as described in any one of claims 1-7.

Images