Personalized vocational skill training system based on dynamic ability atlas

Abstract

The invention discloses a personalized vocational skill training system based on a dynamic ability map, and relates to the technical field of vocational skill training, and the system comprises an ability data collection module, a mapping rule definition module, a knowledge map construction module, an ability gap reasoning module and a personalized path customization module. The method comprises the following steps of: constructing association mapping rules of four dimensions of skills, knowledge, attainment and experience, constructing a cross-dimension knowledge graph, performing reverse tracing along an association path of the cross-dimension knowledge graph by taking a dominant capability short plate as a starting point based on a user capability data set, defining a substandard upstream node as a recessive root factor, and then calculating an association weight and an influence coefficient. A root problem list is generated according to the influence degree, the process depends on a four-dimensional associated cross-dimension knowledge graph structure, the internal relation of the capability of each dimension is established, dominant problems are cut into and traced to a hidden root layer by layer, accurate positioning of a user capability gap is realized, and a scientific basis is provided for formulating a training path.

Description

Technical Field

[0001] This invention relates to the field of vocational skills training technology, specifically to a personalized vocational skills training system based on dynamic competency mapping. Background Technology

[0002] With the upgrading of industrial structure and the development of refined division of labor, enterprises' demand for talent has gradually shifted from single skill orientation to composite ability orientation. Personalized vocational skills training has gradually become a key measure for enterprises to improve talent adaptability and enhance core competitiveness, and has also become an important way for professionals to achieve career advancement. Currently, the market demand for precise and differentiated vocational skills training continues to grow, and the traditional uniform training model is no longer able to match the diverse job requirements and individual ability status.

[0003] The application of existing technologies in vocational skills training has some shortcomings. Most training technologies focus only on the assessment of a single dimension of skills or knowledge, without constructing a four-dimensional correlation system of skills, knowledge, qualities, and experience. This makes it impossible to fully reflect the user's true ability level. Some technologies that introduce knowledge graphs have simple graph structures and lack dynamic expansion capabilities, making it difficult to uncover the hidden root causes behind explicit shortcomings through reverse reasoning. At the same time, the development of training paths often relies on matching fixed course libraries, without combining users' career positioning and learning abilities for differentiated design, and does not support dynamic adjustment of paths based on ability data updates. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of existing technologies and provide a personalized vocational skills training system based on a dynamic competency graph. This invention constructs a cross-dimensional knowledge graph by establishing association mapping rules across four dimensions: skills, knowledge, literacy, and experience. Based on a user competency dataset, it traces back along the association path of the cross-dimensional knowledge graph, starting with explicit competency weaknesses, defining upstream nodes that have not met the standards as implicit root causes. Then, it calculates association weights and influence coefficients, generating a list of root cause problems according to their degree of influence. This process relies on the four-dimensional association cross-dimensional knowledge graph structure to establish the intrinsic connections between competencies in each dimension. Starting from explicit problems, it traces back layer by layer to implicit root causes, achieving precise positioning of user competency gaps and providing a scientific basis for the formulation of training paths.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a personalized vocational skills training system based on dynamic competency graphs, the system comprising:

[0006] The competency data collection module collects user skill assessment scores, knowledge test results, professional quality assessment data, and practical experience records through assessment, behavioral surveys, and task records, and integrates them to form a complete user competency dataset.

[0007] Mapping rule definition module: For the four dimensions of skills, knowledge, literacy and experience, the module formulates the association mapping rules within the dimension and across dimensions. Within the dimension, the association is based on hierarchy, subordinate, classification and complexity. A mapping relationship of mastery-support, adaptation-influence and accumulation-improvement is established across dimensions.

[0008] Knowledge graph construction module: Utilize association mapping rules to build a four-dimensional association network, construct a cross-dimensional knowledge graph, and set weight allocation, evaluation criteria and thresholds to form a multi-dimensional capability evaluation index system;

[0009] Capability Gap Reasoning Module: Based on user capability datasets and cross-dimensional knowledge graphs, the module uses graph-based reverse reasoning algorithms to trace back along related paths starting from explicit capability gaps, locate hidden root causes, and generate a list of root cause issues in order of their impact.

[0010] Personalized Path Customization Module: Based on a list of root causes of problems, combined with the user's career positioning and learning ability, it matches job training content, sets the training pace and difficulty gradient, updates user ability data and cross-dimensional knowledge graph, and adjusts the training path.

[0011] Furthermore, in the capability data collection module, when collecting user capability data, the user's professional field and job type are first determined, and the corresponding skill, knowledge, competence, and experience data collection scope and indicators are clarified. For the skill and knowledge dimensions, assessment activities such as online written tests, offline practical operations, and case analysis are organized through assessment methods, and scores are recorded to form skill assessment scores and knowledge test results. For the competence dimension, questionnaires are distributed and evaluation opinions from superiors, teams, and clients are collected through behavioral surveys to form professional competence assessment data. For the experience dimension, user project participation information, task completion status, and work performance records are extracted from the project management system through task recording methods to form practical experience records.

[0012] Furthermore, in the mapping rule definition module, the basic constituent units of each dimension are first decomposed. Skills are divided into three levels: basic, advanced, and core. Knowledge is divided into three categories: general, professional, and cutting-edge. Qualities are divided into two categories: general qualities and professional-specific qualities. Experience is divided into two categories: basic task experience and complex project experience. The relationships between constituent units within the same dimension are then analyzed. Skills are associated by level, knowledge by subordination, qualities by category, and experience by complexity, forming a tree structure. The interaction relationships between constituent units in different dimensions are also analyzed, clarifying the correspondence between skill and knowledge mastery and support, skill and quality adaptation and influence, and skill and experience accumulation and improvement, thus forming standardized association mapping rules.

[0013] Furthermore, in the knowledge graph construction module, when constructing a cross-dimensional knowledge graph, a four-dimensional node system is defined. The skill dimension has three levels of nodes, the knowledge dimension has three types of nodes, the competency dimension has two types of nodes, and the experience dimension has two types of nodes. Each node is bound to the corresponding dimension's evaluation attributes. Node relationships are established based on association mapping rules. Within a dimension, nodes are linked according to hierarchy, subordination, classification, and complexity. Cross-dimensional nodes are linked according to mastery-support, adaptation-influence, and accumulation-improvement logic. Each relationship is then assigned a correlation degree value, and the four-dimensional nodes are integrated. The point system and node relationships form a four-dimensional network, thereby constructing a cross-dimensional knowledge graph containing node and relationship layers. Based on the user capability dataset, evaluation weights are assigned to nodes in each dimension according to the proportion of each dimension. The evaluation standards and thresholds for each dimension are set differently according to the job capability standards of different professional fields. The cross-dimensional knowledge graph supports the dynamic expansion of nodes and relationships. When new professional skill standards, cutting-edge industry knowledge content, new professional quality requirements, or complex project experience types are added, nodes in the corresponding dimensions are supplemented, and the relationship between new nodes and existing nodes is established according to the association mapping rules.

[0014] Furthermore, in the capability gap reasoning module, the user capability dataset is compared with the threshold of the multi-dimensional capability assessment index system. The cross-dimensional knowledge graph nodes corresponding to the unmet explicit shortcomings are selected as the starting point. All relevant nodes are traversed upstream along the association path of the cross-dimensional knowledge graph. The user capability data corresponding to the upstream nodes is extracted and compared with the threshold. The unmet nodes are marked, and the unmet upstream nodes are defined as the implicit root causes of the explicit shortcomings. All unmet upstream nodes are integrated, and the association weight between each unmet upstream node and the explicit shortcoming node is calculated through the graph back reasoning algorithm.

[0015] Furthermore, in the capability gap reasoning module, the calculation formula for the graph reverse reasoning algorithm is as follows: ,in, For the first The non-compliant upstream node and the first The association weight of each explicit weak node. For the first The non-compliant upstream node and the first The correlation degree of each explicit weak node is determined by cross-dimensional correlation mapping rules. For the first The evaluation weights of each non-compliant upstream node in the corresponding dimension are determined through a multi-dimensional capability evaluation indicator system. In order to be with the first The total number of upstream nodes that are not up to standard and are associated with each obvious bottleneck node.

[0016] Furthermore, in the capability gap reasoning module, the specific steps for generating a list of root causes of problems by sorting them according to their degree of impact are as follows: based on the association weights between each upstream node that fails to meet the standard and the obvious short-board node, the influence coefficient of each hidden root cause factor is calculated using the root cause influence coefficient formula, and the final sorting order of the list of root causes of problems is determined according to the magnitude of the influence coefficient.

[0017] Furthermore, in the capability gap reasoning module, the formula for the root cause influence coefficient is: ,in, For the first The influence coefficient of each latent root factor For the first The non-compliant upstream node and the first The association weight of each explicit weak node. For the first The proportion of each node corresponding to a hidden root cause factor in the core competency indicators of a job is determined by the authoritative standard document for job competency.

[0018] Furthermore, in the personalized path customization module, the root cause problem list is analyzed, and the implicit root cause factors with the highest matching degree with the user's career positioning are extracted as core gap items. Based on the core gap items, the training difficulty value corresponding to each core gap item is calculated using the training difficulty gradient calculation formula. Learning stages are divided according to the training difficulty value, and the training content and duration of each learning stage are set. The corresponding job training content is matched to form a training path. Then, based on the user's ability data after the stage training, the status of the corresponding nodes in the cross-dimensional knowledge graph is updated, and the priority and difficulty of subsequent training paths are adjusted. The adjustment cycle is set according to the user's learning stage: the adjustment cycle for the basic stage is 10-15 days, the adjustment cycle for the advanced stage is 20-30 days, and the adjustment cycle for the core stage is 50-60 days.

[0019] Furthermore, in the personalized path customization module, the formula for calculating the training difficulty gradient is: ,in, For the first The training difficulty value corresponding to each training stage. For the first The influence coefficient of each latent root factor The user's learning ability coefficient is determined by statistical results of the user's historical learning progress and knowledge absorption efficiency.

[0020] Compared with existing technologies, this personalized vocational skills training system based on dynamic competency maps has the following advantages:

[0021] I. This invention constructs a cross-dimensional knowledge graph by establishing association mapping rules for four dimensions: skills, knowledge, literacy, and experience. Based on a user capability dataset, it traces back along the association path of the cross-dimensional knowledge graph, starting with explicit capability shortcomings. The upstream nodes that fail to meet the standards are defined as implicit root causes. Then, the association weights and influence coefficients are calculated, and a list of root cause problems is generated according to the degree of influence. This process relies on the four-dimensional association cross-dimensional knowledge graph structure to establish the intrinsic connection between capabilities in each dimension. Starting from explicit problems, it traces back to implicit root causes layer by layer, achieving accurate positioning of user capability gaps and providing a scientific basis for the formulation of training paths.

[0022] Second, this invention analyzes a list of root causes of problems to extract the implicit root causes that best match the user's career positioning as core gap items. Based on these core gap items, it divides the learning into stages and sets the training content and duration to form an initial training path. Simultaneously, based on the user's ability data after each stage of training, it updates the node status in the cross-dimensional knowledge graph and adjusts the priority and difficulty of subsequent training paths. This process is based on the user's career development direction and personal learning characteristics to achieve personalized adaptation of training content and pace. Relying on the dynamic expansion capability of the cross-dimensional knowledge graph, the training path can be optimized in real time as the user's ability changes, ensuring that the training content always meets the user's ability improvement needs.

[0023] Other advantages, objectives and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination or study, or may be learned from the practice of the invention. Attached Figure Description

[0024] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0025] Figure 1 A flowchart of a personalized vocational skills training system based on dynamic competency graphs;

[0026] Figure 2 This is a framework diagram of a personalized vocational skills training system based on dynamic competency graphs.

[0027] Figure 3 is a flowchart of the competency gap reasoning module in a personalized vocational skills training system based on dynamic competency graphs. Detailed Implementation

[0028] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structure, features and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0029] Example 1: In a training scenario for improving the professional skills of R&D personnel in a technology company, 50 R&D personnel from the company were selected as training participants. The training objective was to accurately identify the skill gaps of the R&D personnel, customize targeted training paths, and help them adapt to the job requirements of the company's next-generation distributed system development. First, skill data collection was conducted to determine the scope and indicators of skill data collection for the R&D positions. The indicators covered four dimensions: skills, knowledge, qualities, and experience. For the skills and knowledge dimensions, three types of assessment activities were organized: online written tests, offline practical exercises, and case analysis. The online written tests covered programming language theories such as Python and Java, as well as professional knowledge such as data structures and algorithm principles. The offline practical exercises focused on module development, system debugging, and integration. Practical projects such as voice-to-speech and voice-to-speech (VPS) are used. Case studies of large-scale distributed architecture design cases are selected for R&D personnel to break down, analyze, and output optimization solutions. All assessment scores are recorded to form skills assessment scores and knowledge test results. Regarding the competency dimension, questionnaires are distributed to collect evaluation opinions from R&D personnel's direct supervisors and project team members. Evaluation indicators include innovation awareness, sense of responsibility, teamwork ability, and stress resistance, forming professional competency assessment data. Regarding the experience dimension, project participation information, task completion status, and work performance records of R&D personnel are extracted from the company's project management system, covering the number of projects participated in, project difficulty level, task delivery quality, milestone completion rate, etc., forming practical experience records. All data are integrated to form a complete user competency dataset, such as... Figure 1 As shown.

[0030] A mapping rule is established for the four dimensions of skills, knowledge, qualities, and experience. First, the basic components of each dimension are broken down. Skills are divided into three levels: basic skills, advanced skills, and core skills. Basic skills include coding and simple debugging; advanced skills include module development and interface design; and core skills include architecture design and system optimization. Knowledge is divided into three categories based on their hierarchy: general knowledge, professional knowledge, and cutting-edge knowledge. General knowledge includes computer fundamentals and network principles; professional knowledge includes data structures and algorithm principles; and cutting-edge knowledge includes artificial intelligence frameworks and microservice architecture. Qualities are divided into two categories based on their classification: general qualities and professional-specific qualities. General qualities include… It includes a sense of responsibility and teamwork skills; professional competence includes innovative thinking and resilience. Experience is categorized by complexity into basic task experience and complex project experience. Basic task experience includes single-module development, while complex project experience includes building large-scale distributed systems. The relationships between constituent units within the same dimension are analyzed: skills are linked hierarchically, knowledge is linked subordinately, competences are linked by category, and experience is linked by complexity, forming a tree structure. The interaction relationships between constituent units across different dimensions are then analyzed, clarifying the correspondence between skill and knowledge mastery and support, skill and competence adaptation and influence, and skill and experience accumulation and improvement, forming standardized association mapping rules, such as... Figure 2 As shown.

[0031] A four-dimensional association network is constructed using established association mapping rules to build a cross-dimensional knowledge graph. First, a four-dimensional node system is defined: the skill dimension has three levels of nodes: basic skills, advanced skills, and core skills; the knowledge dimension has three categories of nodes: general knowledge, professional knowledge, and cutting-edge knowledge; the competency dimension has two categories of nodes: general competency and professional competency; and the experience dimension has two categories of nodes: basic task experience and complex project experience. Each node is bound to the corresponding dimension's evaluation attributes. Based on the association mapping rules, node relationships are established. Within a dimension, nodes are linked by hierarchy, subordination, classification, and complexity; cross-dimensional nodes are linked by mastery-support, adaptation-influence, etc. The logic of accumulation and improvement establishes directed connections, assigning a correlation degree value to each connection, integrating the four-dimensional node system and node connections to form a four-dimensional connection network, thereby constructing a cross-dimensional knowledge graph containing node layers and connection layers; then, based on the user capability dataset, evaluation weights are assigned to nodes in each dimension according to the proportion of each dimension, and the evaluation standards and thresholds for each dimension are set differently according to the capability standards of technical R&D positions; this cross-dimensional knowledge graph supports the dynamic expansion of nodes and connections. When new professional skill standards related to the application of artificial intelligence large models are added, nodes in the corresponding dimensions are supplemented, and the connection between new nodes and existing nodes is established according to the association mapping rules.

[0032] This study conducts capability gap reasoning based on user capability datasets and cross-dimensional knowledge graphs. The user capability datasets are compared with thresholds in a multi-dimensional capability assessment indicator system to identify cross-dimensional knowledge graph nodes corresponding to explicit shortcomings. Taking a developer involved in distributed system development as an example, their explicit shortcoming is inadequate architecture design skills. Starting from this point, the study traverses all relevant nodes upstream along the cross-dimensional knowledge graph's association path, extracting user capability data corresponding to upstream nodes and comparing it with thresholds. Nodes that fail to meet the requirements are marked, and these upstream nodes are defined as the implicit root causes of the explicit shortcomings. The traversal reveals that the developer lacks sufficient knowledge of algorithm principles and experience in large, complex projects; these are identified as implicit root causes. All upstream nodes that fail to meet the requirements are integrated, and the association weight between each upstream node and the explicit shortcoming node is calculated using a graph back-inference algorithm. The calculation formula for the graph back-inference algorithm is as follows: ,in, For the first The non-compliant upstream node and the first The association weight of each explicit weak node. For the first The non-compliant upstream node and the first The correlation degree of each explicit weak node is determined by cross-dimensional correlation mapping rules. For the first The evaluation weights of each non-compliant upstream node in the corresponding dimension are determined through a multi-dimensional capability evaluation indicator system. In order to be with the first The total number of upstream nodes that are associated with each explicit bottleneck node; then, based on the association weight between each upstream node and the explicit bottleneck node, the influence coefficient of each implicit root cause factor is calculated using the root cause influence coefficient formula, which is: ,in, For the first The influence coefficient of each latent root factor For the first The non-compliant upstream node and the first The association weight of each explicit weak node. For the first The proportion of each hidden root cause factor's corresponding node in the core competency indicators of the position is determined by the authoritative standard document for the position's competency; the final sorting order of the root cause problem list is determined based on the magnitude of the influence coefficient, generating the root cause problem list, such as... Figure 3 As shown.

[0033] Based on the generated list of root causes of problems, a personalized path customization process was conducted. The list was analyzed to extract the implicit root causes that best matched the technical R&D personnel's career positioning as core gap items. These core gap items for the R&D personnel were insufficient mastery of algorithm principles and a lack of experience in large and complex projects. Based on these core gap items, the training difficulty value corresponding to each core gap item was calculated using the training difficulty gradient calculation formula: ,in, For the first The training difficulty value corresponding to each training stage. For the first The influence coefficient of each latent root factor The user's learning ability coefficient is determined by statistical results of the user's historical learning progress and knowledge absorption efficiency. Learning stages are divided based on the training difficulty value, and training content and duration are set for each stage. This is matched with corresponding job training content to form a training path. The basic stage includes advanced courses on algorithm principles, accompanied by after-class programming practice tasks. The advanced stage involves participation in the architecture design of the company's core distributed system project, with guidance from senior architects throughout. The core stage includes practical architecture design exercises, enabling independent design of small-scale distributed system architecture solutions. Based on the developer's ability data after each stage of training, the corresponding node status in the cross-dimensional knowledge graph is updated, adjusting the priority and difficulty of subsequent training paths. The adjustment cycle is set according to the learning stage: 10-15 days for the basic stage, 20-30 days for the advanced stage, and 50-60 days for the core stage.

[0034] In summary, this study targets the professional skills enhancement training scenarios for R&D personnel in technology companies. It collects data on personnel's skills, knowledge, qualities, and experience through multiple dimensions, formulates association mapping rules covering four dimensions, and builds a cross-dimensional knowledge graph. Starting with the explicit weakness of insufficient architectural design capabilities among R&D personnel, it traces the implicit root causes of weak algorithm principles and lack of experience in complex projects. Furthermore, it extracts core gaps based on job positioning, formulates phased training paths, and dynamically adjusts the plan according to the phased training results, achieving a precise match between R&D personnel's skill enhancement and job requirements.

[0035] Example 2: In a training scenario to enhance the professional skills of customer service personnel in an e-commerce company, 80 customer service personnel from the e-commerce company were selected as training participants. The training objective was to accurately identify the shortcomings in the customer service personnel's capabilities, customize personalized training programs, improve their customer service quality and problem-solving efficiency, and reduce customer complaint rates. First, capability data collection was conducted to determine the scope and indicators for capability collection corresponding to the e-commerce customer service positions, covering four dimensions: skills, knowledge, qualities, and experience. For the skills and knowledge dimensions, three types of assessment activities were organized: online written tests, offline practical exercises, and case analysis. The online written tests covered e-commerce platform rules, product knowledge, and after-sales service policies, while the offline practical exercises focused on customer inquiry responses and complaint handling. The training includes practical scenarios such as return and exchange procedures, simulating various customer communication situations. Case studies select complex complaint cases for customer service personnel to simulate handling and output solutions. All assessment scores are recorded to form skills assessment results and knowledge test results. For the competency dimension, questionnaires are distributed to collect evaluation opinions from customer service personnel's superiors, colleagues, and customers. Evaluation indicators include communication awareness, service attitude, patience, empathy, etc., forming professional competency assessment data. For the experience dimension, records such as the number of work orders processed, problem resolution rate, customer satisfaction rating, and repeat consultation rate of customer service personnel are extracted from the enterprise customer service management system to form practical experience records. All data are integrated to form a complete user competency dataset.

[0036] A mapping rule was established for the four dimensions of skills, knowledge, qualities, and experience. First, the basic components of each dimension were broken down. Skills were categorized into three levels: basic skills, advanced skills, and core skills. Basic skills include standardized communication scripts and work order entry; advanced skills include handling general inquiries and simple complaints; and core skills include handling complex complaints and customer retention. Knowledge was categorized into three types based on their hierarchy: general knowledge, professional knowledge, and cutting-edge knowledge. General knowledge includes communication skills and service etiquette; professional knowledge includes platform rules and product knowledge; and cutting-edge knowledge includes customer relationship management strategies and the application of intelligent customer service tools. Qualities were categorized into two types based on their classification: general qualities and professional-specific qualities. General competencies include patience, meticulousness, and a sense of responsibility, while specialized professional competencies include empathy and emergency response capabilities. Experience is categorized by complexity into two types: basic task experience and complex project experience. Basic task experience includes responding to general inquiries, while complex project experience includes handling a large number of customer complaints. The relationships between the constituent units within the same dimension are analyzed: skills are linked by level, knowledge by subordination, competencies by category, and experience by complexity, forming a tree structure. The interaction relationships between the constituent units in different dimensions are then analyzed to clarify the correspondence between skill and knowledge mastery and support, skill and competencies adaptation and influence, and skill and experience accumulation and improvement, thus forming standardized correlation mapping rules.

[0037] A four-dimensional association network is constructed using association mapping rules to build a cross-dimensional knowledge graph. First, a four-dimensional node system is defined: the skill dimension has three levels of nodes: basic skills, advanced skills, and core skills; the knowledge dimension has three categories of nodes: general knowledge, professional knowledge, and cutting-edge knowledge; the competency dimension has two categories of nodes: general competency and professional competency; and the experience dimension has two categories of nodes: basic task experience and complex project experience. Each node is bound to the corresponding dimension's evaluation attributes. Based on association mapping rules, node relationships are established. Within a dimension, nodes are linked by hierarchy, subordination, classification, and complexity; cross-dimensional nodes are linked by mastery-support, adaptation-influence, etc. The logic of accumulation and improvement establishes directed connections, assigning a correlation value to each connection, integrating the four-dimensional node system and node connections to form a four-dimensional connection network, thereby constructing a cross-dimensional knowledge graph containing node and connection layers. Based on the user capability dataset, evaluation weights are assigned to nodes in each dimension according to the proportion of each dimension, and the evaluation standards and thresholds for each dimension are set differently according to the capability standards of e-commerce customer service positions. This cross-dimensional knowledge graph supports the dynamic expansion of nodes and connections. When new skill standards related to cross-border e-commerce customer service are added, nodes in the corresponding dimensions are added, and the connection between the new node and the existing node is established according to the association mapping rules.

[0038] Based on user capability datasets and cross-dimensional knowledge graphs, capability gap reasoning was conducted. The user capability datasets were compared with thresholds in a multi-dimensional capability assessment indicator system to identify the cross-dimensional knowledge graph nodes corresponding to explicit shortcomings. Taking a customer service representative responsible for beauty products as an example, their explicit shortcoming was a lack of competence in handling complex complaints, a core skill. Starting from this point, the process traversed upstream along the cross-dimensional knowledge graph's association path, extracting user capability data corresponding to upstream nodes and comparing it with thresholds. Nodes that failed to meet the requirements were marked, and these upstream nodes were defined as the implicit root causes of the explicit shortcomings. The traversal revealed that the customer service representative lacked sufficient knowledge of beauty product ingredients and needed to improve empathy; these were identified as implicit root causes. All upstream nodes that failed to meet the requirements were integrated, and the association weight between each upstream node and the explicit shortcoming node was calculated using a graph back-inference algorithm. The calculation formula for the graph back-inference algorithm is as follows: Then, based on the correlation weights between each substandard upstream node and the explicit bottleneck node, the influence coefficient of each implicit root cause factor is calculated using the root cause influence coefficient formula, which is: The final sorting order of the root cause problem list is determined based on the magnitude of the impact coefficient, and the root cause problem list is generated.

[0039] Based on the generated list of root causes of problems, a personalized training path was developed. The list was analyzed to extract the implicit root causes that best matched the professional positioning of the e-commerce customer service representative as the core skill gaps. These core skill gaps for the representative were insufficient knowledge of beauty product ingredients and an need to improve empathy skills. Based on these core skill gaps, the training difficulty value corresponding to each core skill gap was calculated using a training difficulty gradient calculation formula. The training difficulty gradient calculation formula is as follows: The training is divided into stages based on difficulty level, with training content and duration set for each stage. This is matched with corresponding job-specific training content to form a training path. The basic stage includes in-depth courses on the ingredients of all beauty products, paired with product trial experience tasks. The advanced stage includes specialized training in empathic communication skills, simulating various customer complaint scenarios for script practice. The core stage includes complex complaint handling simulations with one-on-one guidance from senior customer service mentors. Based on the customer service personnel's ability data after each stage of training, the corresponding node status in the cross-dimensional knowledge graph is updated, adjusting the priority and difficulty of subsequent training paths. The adjustment cycle is set according to the learning stage: 10-15 days for the basic stage, 20-30 days for the advanced stage, and 50-60 days for the core stage.

[0040] In summary, this study focuses on providing vocational skills enhancement training for customer service personnel in e-commerce companies. By collecting relevant data on their capabilities through multiple channels, a cross-dimensional knowledge graph tailored to customer service roles is constructed. Taking the explicit weakness of insufficient complex complaint handling skills as a starting point, the study identifies the implicit root causes of weak product knowledge and insufficient empathy. Based on job requirements, core skill gaps are extracted and phased training plans are developed. The training path is dynamically optimized based on phased capability data to help customer service personnel fill skill gaps and better adapt to the service requirements of their positions.

[0041] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A personalized vocational skills training system based on dynamic competency graphs, characterized in that, The system includes: The competency data collection module collects user skill assessment scores, knowledge test results, professional quality assessment data, and practical experience records through assessment, behavioral surveys, and task records, and integrates them to form a complete user competency dataset. Mapping rule definition module: For the four dimensions of skills, knowledge, literacy and experience, the module formulates the association mapping rules within the dimension and across dimensions. Within the dimension, the association is based on hierarchy, subordinate, classification and complexity. A mapping relationship of mastery-support, adaptation-influence and accumulation-improvement is established across dimensions. Knowledge graph construction module: Utilize association mapping rules to build a four-dimensional association network, construct a cross-dimensional knowledge graph, and set weight allocation, evaluation criteria and thresholds to form a multi-dimensional capability evaluation index system; Capability Gap Reasoning Module: Based on user capability datasets and cross-dimensional knowledge graphs, the module uses graph-based reverse reasoning algorithms to trace back along related paths starting from explicit capability gaps, locate hidden root causes, and generate a list of root cause issues in order of their impact. Personalized Path Customization Module: Based on a list of root causes of problems, combined with the user's career positioning and learning ability, it matches job training content, sets the training pace and difficulty gradient, updates user ability data and cross-dimensional knowledge graph, and adjusts the training path.

2. The personalized vocational skills training system based on dynamic competency graphs according to claim 1, characterized in that, In the capability data collection module, when collecting user capability data, the user's professional field and job type are first determined, and the corresponding skill, knowledge, competence, and experience data collection scope and indicators are clarified. For the skill and knowledge dimensions, assessment activities such as online written tests, offline practical operations, and case analysis are organized through assessment methods, and scores are recorded to form skill assessment scores and knowledge test results. For the competence dimension, questionnaires are distributed and evaluation opinions from superiors, teams, and clients are collected through behavioral surveys to form professional competence assessment data. For the experience dimension, user project participation information, task completion status, and work performance records are extracted from the project management system through task recording methods to form practical experience records.

3. The personalized vocational skills training system based on dynamic competency graphs according to claim 1, characterized in that, In the mapping rule definition module, the basic constituent units of each dimension are first decomposed. Skills are divided into three levels: basic, advanced, and core. Knowledge is divided into three categories: general, professional, and cutting-edge. Qualities are divided into two categories: general qualities and professional-specific qualities. Experience is divided into two categories: basic task experience and complex project experience. The relationships between constituent units within the same dimension are then analyzed. Skills are associated by level, knowledge by subordination, qualities by category, and experience by complexity, forming a tree structure. The interaction relationships between constituent units in different dimensions are also analyzed, clarifying the correspondence between skill and knowledge mastery and support, skill and quality adaptation and influence, and skill and experience accumulation and improvement, thus forming standardized association mapping rules.

4. The personalized vocational skills training system based on dynamic competency graphs according to claim 1, characterized in that, In the knowledge graph construction module, when constructing a cross-dimensional knowledge graph, a four-dimensional node system is defined. The skill dimension has three levels of nodes, the knowledge dimension has three types of nodes, the competency dimension has two types of nodes, and the experience dimension has two types of nodes. Each node is bound to the corresponding dimension's evaluation attributes. Node relationships are established based on association mapping rules. Within a dimension, nodes are linked according to hierarchy, subordination, classification, and complexity. Cross-dimensional nodes are linked according to mastery-support, adaptation-influence, and accumulation-improvement. Each relationship is assigned a correlation value, and the four-dimensional node system and node relationships are integrated to form a four-dimensional association network. This constructs a cross-dimensional knowledge graph containing node and association layers. Based on the user competency dataset and the proportion of each dimension, evaluation weights are assigned to nodes in each dimension. The evaluation standards and thresholds for each dimension are set differently according to the job competency standards of different professional fields. Cross-dimensional knowledge graphs support the dynamic expansion of nodes and relationships. When adding new professional skill standards, cutting-edge industry knowledge content, new professional quality requirements, or complex project experience types, nodes of the corresponding dimensions are added, and the relationship between the new node and the existing node is established according to the association mapping rules.

5. A personalized vocational skills training system based on dynamic competency graphs according to claim 1, characterized in that, In the capability gap reasoning module, the user capability dataset is compared with the threshold of the multi-dimensional capability assessment index system. The cross-dimensional knowledge graph nodes corresponding to the unmet explicit shortcomings are selected as the starting point. All relevant nodes are traversed upstream along the association path of the cross-dimensional knowledge graph. The user capability data corresponding to the upstream nodes is extracted and compared with the threshold. The unmet nodes are marked, and the unmet upstream nodes are defined as the implicit root causes of the explicit shortcomings. All unmet upstream nodes are integrated, and the association weight between each unmet upstream node and the explicit shortcoming node is calculated through the graph back reasoning algorithm.

6. A personalized vocational skills training system based on dynamic competency graphs according to claim 5, characterized in that, In the capability gap reasoning module, the calculation formula for the graph reverse reasoning algorithm is as follows: ,in, For the first The non-compliant upstream node and the first The association weight of each explicit weak node. For the first The non-compliant upstream node and the first The degree of correlation of a node with obvious shortcomings. For the first The evaluation weight of each non-compliant upstream node in the corresponding dimension. In order to be with the first The total number of upstream nodes that are not up to standard and are associated with each obvious bottleneck node.

7. A personalized vocational skills training system based on dynamic competency graphs according to claim 5, characterized in that, In the capability gap reasoning module, the specific steps for generating a list of root causes of problems by sorting them according to their degree of impact are as follows: based on the correlation weight between each upstream node that fails to meet the standard and the obvious short-board node, the influence coefficient of each hidden root cause factor is calculated using the root cause influence coefficient formula, and the final sorting order of the list of root causes of problems is determined according to the magnitude of the influence coefficient.

8. A personalized vocational skills training system based on dynamic competency graphs according to claim 7, characterized in that, In the capability gap reasoning module, the formula for the root cause influence coefficient is: ,in, For the first The influence coefficient of each latent root factor For the first The non-compliant upstream node and the first The association weight of each explicit weak node. For the first The percentage of each hidden root cause factor in the core competency indicators of the job.

9. A personalized vocational skills training system based on dynamic competency graphs according to claim 1, characterized in that, In the personalized path customization module, the root cause problem list is analyzed, and the implicit root cause factors with the highest matching degree with the user's career positioning are extracted as core gap items. Based on the core gap items, the training difficulty value corresponding to each core gap item is calculated using the training difficulty gradient calculation formula. Learning stages are divided according to the training difficulty value, and the training content and duration of each learning stage are set. The corresponding job training content is matched to form a training path. Then, based on the user's ability data after each stage of training, the status of the corresponding nodes in the cross-dimensional knowledge graph is updated, and the priority and difficulty of subsequent training paths are adjusted. The adjustment cycle is set according to the user's learning stage: the adjustment cycle for the basic stage is 10-15 days, the adjustment cycle for the advanced stage is 20-30 days, and the adjustment cycle for the core stage is 50-60 days.

10. A personalized vocational skills training system based on dynamic competency graphs according to claim 9, characterized in that, In the personalized path customization module, the formula for calculating the training difficulty gradient is: ,in, For the first The training difficulty value corresponding to each training stage. For the first The influence coefficient of each latent root factor This represents the user's learning ability coefficient.

Images