Intelligent training student data integrated management system

Abstract

A smart training student data integrated management system of the present invention belongs to the technical field of smart training and education data analysis, and comprises a learning track data acquisition module used for acquiring student interaction data of students in a training process; the dynamic cognition map construction module is used for constructing a dynamic cognition map representing the ability of the trainee by performing quantitative analysis on the dynamic association strength between skills in response to the trainee interaction data; the emergence capability identification and prediction module is used for calculating the emergence index of the skill cluster based on the dynamic cognitive map and identifying a capability emergence cluster; the self-adaptive evaluation engine module is used for dynamically updating the weight of the evaluation dimension in response to the emergence index, and robustness and universality of similarity calculation are guaranteed.

Description

Technical Field

[0001] The present invention relates to the technical field of smart training and education data analysis, and specifically to a comprehensive management system for smart training student data. Background Art

[0002] In the field of smart training, traditional student ability assessment methods mainly rely on fixed, reductionist skill assessment models. These models tend to break down students' overall abilities into a series of isolated skill units for independent assessment and measurement. Although this approach has a clear structure, there is a significant deviation between its assessment results and the actual comprehensive abilities demonstrated by students in complex tasks.

[0003] The root of the above problems lies in the limitations of the evaluation paradigm. Traditional models regard skills as static, isolated units, ignoring the nonlinear synergistic effects generated by the dynamic coupling of various skills in actual applications. This comprehensive ability, where the whole is greater than the sum of its parts, namely emergent ability, cannot be measured by simply measuring a single skill. Therefore, due to the nature of its methodology, existing technologies are fundamentally unable to capture and quantify this comprehensive ability generated by the dynamic coupling of multiple skills.

[0004] As a result, the assessment system cannot accurately reflect the students' true ability structure, resulting in a significant deviation between the assessment scores and the students' true abilities determined by domain experts. Furthermore, the personalized learning paths generated based on this inaccurate assessment are seriously mismatched with the students' nonlinear true ability development trajectory. This not only seriously affects the accuracy and effectiveness of learning path recommendations, but also hinders the realization of truly efficient and precise personalized intelligent training. Summary of the Invention

[0005] The purpose of the present invention is to provide a comprehensive management system for student data of smart training to solve the problems raised in the above background technology.

[0006] The technical solution of the present invention is, comprising: The learning trajectory data collection module is used to obtain the student interaction data during the training process; A dynamic cognitive map construction module is used to respond to student interaction data and quantitatively analyze the dynamic correlation strength between skills to construct a dynamic cognitive map that represents the student's ability; The emergent ability identification and prediction module is used to identify emergent ability clusters by calculating the emergence index of skill clusters based on the dynamic cognitive map; The adaptive evaluation engine module is used to dynamically update the weights of the evaluation dimensions in response to the emergence index.

[0007] Preferably, the process of the dynamic cognitive map construction module performing dynamic association strength quantitative analysis includes: Determine the co-occurrence of students' learning activities to represent learning behavior, the intrinsic semantic similarity to represent the intrinsic connection of skills, and the comprehensive task performance factor to represent comprehensive application performance; Combined with the preset weights, the co-occurrence of students’ learning activities, intrinsic semantic similarity and comprehensive task performance factors are comprehensively calculated to generate dynamic association strength.

[0008] Preferably, the intrinsic semantic similarity is obtained by processing the descriptive text of the skill through a preset language model; the comprehensive task performance factor is calculated based on the normalized score of the trainees completing the comprehensive task.

[0009] Preferably, the process of calculating the emergence index by the emergent capability identification and prediction module includes: Determine the number of skill nodes within the skill cluster; Determine the normalized eigenvector centrality of each skill node; The emergence index of the skill cluster is calculated by combining the number of skill nodes, normalized eigenvector centrality, and the dynamic correlation strength between nodes.

[0010] Preferably, the skill clusters are identified by executing a community discovery algorithm on a dynamic cognitive graph.

[0011] Preferably, the process of updating the evaluation dimension weights by the adaptive evaluation engine module includes: Determine the emergence index of the competency cluster most relevant to the assessment dimension; determine the average emergence index of all competency clusters; obtain the current weight of the assessment dimension; obtain the engagement gate that represents the student's learning willingness; based on the preset learning rate, based on the difference between the emergence index of the most relevant competency cluster and the average emergence index, and in response to the state of the engagement gate, adjust the current weight to generate the weight at the next moment.

[0012] Preferably, the process of determining the emergence index of the capability cluster most relevant to the evaluation dimension is as follows: Calculate the correlation between the evaluation dimension and each skill cluster, and determine the skill cluster with the largest correlation as the ability cluster most relevant to the evaluation dimension; set the emergence index of the most relevant ability cluster as the emergence index of the ability cluster most relevant to the evaluation dimension.

[0013] Preferably, the update logic of the evaluation dimension weight is as follows: When the emergence index of the competency cluster most relevant to the assessment dimension is higher than the average emergence index of all competency clusters, and the participation gate indicates that the learner has participated in the relevant learning activities, the weight of the assessment dimension is increased; Otherwise, the weight of the evaluation dimension will be weakened or kept unchanged.

[0014] The present invention provides a comprehensive management system for intelligent training student data, which has the following improvements and advantages compared with the prior art: 1. To accurately measure the synergistic relationship between skills, a dynamic association strength model is introduced. This model aims to accurately quantify the nonlinear coupling strength between skills by integrating the synergistic effect of learning behavior and the inhibitory effect of comprehensive task performance. Driven by collected student interaction data, this model dynamically quantifies the nonlinear coupling strength between any two skills, constructing a dynamic cognitive map that truly reflects the inherent structure of individual student abilities. The present invention maps the textual descriptions of skills into a high-dimensional vector space, bringing semantically similar skills closer together in this space. Its working principle is to map the textual descriptions of skills into a high-dimensional vector space, bringing semantically similar skills closer together in this space. This assumption ensures the robustness and universality of similarity calculations. 2. The system can objectively identify emerging clusters of capabilities from the data where the whole is greater than the sum of its parts, transforming implicit capabilities that are invisible in traditional assessments into measurable and comparable quantitative indicators. This is a fundamental improvement over existing technologies that rely on predefined skill sets by experts, achieving a leap from hypothesizing capability structures to discovering capability structures. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The present invention will be further explained below in conjunction with the accompanying drawings and Examples: Figure 1 This is a flow chart of a comprehensive management system for intelligent training student data of the present invention. DETAILED DESCRIPTION

[0016] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to specific embodiments.

[0017] Example 1: See also Figure 1 The present invention provides a comprehensive management system for intelligent training student data, comprising: The learning trajectory data collection module is used to obtain the student interaction data during the training process; A dynamic cognitive map construction module is used to respond to student interaction data and quantitatively analyze the dynamic correlation strength between skills to construct a dynamic cognitive map that represents the student's ability; The emergent ability identification and prediction module is used to identify emergent ability clusters by calculating the emergence index of skill clusters based on the dynamic cognitive map; An adaptive evaluation engine module for dynamically updating the weights of evaluation dimensions in response to the emergence index; The present invention provides a comprehensive management system for intelligent training student data. This system aims to overcome the discrepancy between assessment results and students' actual comprehensive abilities caused by the use of fixed, reductionist skill assessment models in existing technologies. By dynamically capturing the nonlinear coupling relationships between skills, the system identifies and quantifies emergent abilities, thereby enabling adaptive adjustment of the assessment system. This allows student assessment and learning path planning to more closely align with an individual's actual ability development trajectory. The system consists of four core modules working together: the learning trajectory data collection module, which aims to comprehensively record students' learning behavior. In this embodiment, it is configured to obtain multimodal data generated by students during their interaction with the learning system, including but not limited to student interaction data such as course clickstreams, exercise submissions, task completion status, and operation sequences in the simulation environment; The core purpose of the dynamic cognitive map construction module is to transform discrete learning behavior data into a topological network that represents the internal structure of student abilities. This module responds to student interaction data acquired by the learning trajectory data acquisition module and quantitatively analyzes the dynamic correlation strength between skills to construct a dynamic cognitive map with skills as nodes and the correlation strength between skills as weighted edges. This map forms the technical foundation for subsequent ability identification and assessment. The emergent competence identification and prediction module aims to discover and quantify, from the constructed cognitive map, comprehensive competences that arise from the synergy of multiple skills and cannot be measured by independently evaluating a single skill. Based on the topological structure of the dynamic cognitive map, this module calculates the emergence index of closely connected skill subsets within the map, i.e., skill clusters, to identify emergent competence clusters that represent specific comprehensive competences. The adaptive assessment engine module is designed to enable the assessment framework to dynamically respond to changes in student capabilities. This module dynamically updates the weights of each assessment dimension in the assessment system in response to the emergence index calculated by the emergent capability identification and prediction module. This update mechanism forms a closed-loop technology to ensure that the assessment focus remains on the key areas of competency that the student is currently developing or has emerged. These modules are interconnected and exchange data, forming a complete, self-consistent technical closed loop. The system begins by collecting raw learning data, constructing a capability map, identifying emergent capabilities, and providing feedback for dynamic adjustments to the evaluation framework, seamlessly implementing the entire process from data to insight to adaptive optimization. Through the synergistic effect of the above four modules, this system can break through the limitations of traditional linear assessment models and effectively identify and quantify emergent capabilities generated by the dynamic coupling of multiple skills; it solves the problem in existing technologies that the deviation between assessment results and students' actual abilities may be as high as 70% due to model paradigm limitations, and significantly reduces the mismatch rate between learning path recommendations and students' nonlinear learning processes, which may exceed 60%, thereby providing a system-level solution for achieving truly personalized and high-precision intelligent training.

[0018] The progress of this proposal lies in achieving a shift from a static, reductionist paradigm of student ability assessment to a cognitive paradigm of the emergence of a dynamic, holistic ability structure. Existing technologies deconstruct student abilities into isolated skill units for measurement. This method is essentially unable to capture the comprehensive abilities generated by the dynamic coupling of multiple skills in complex tasks, resulting in a significant deviation between the assessment results and the students' actual abilities, with a mismatch rate of up to 70 percent. This proposal effectively solves this core technical problem by introducing a set of interconnected dynamic analysis and adaptive adjustment mechanisms.

[0019] Example 2 The process of dynamic cognitive map construction module to conduct dynamic association strength quantitative analysis includes: Determine the co-occurrence of students' learning activities to represent learning behavior, the intrinsic semantic similarity to represent the intrinsic connection of skills, and the comprehensive task performance factor to represent comprehensive application performance; Combined with preset weights, the co-occurrence of students’ learning activities, intrinsic semantic similarity, and comprehensive task performance factors are comprehensively calculated to generate dynamic association strength; The intrinsic semantic similarity is obtained by processing the skill description text through a preset language model; the preset language model is preferably a pre-trained language model based on the Transformer architecture, such as the BERT model. The processing process is as follows: the description text of each skill is input into the pre-trained BERT model, and the corresponding [CLS] tag or word vector is extracted and averaged pooled to generate a high-dimensional, such as 768-dimensional, skill feature vector. and ; The intrinsic semantic similarity is obtained by calculating the cosine similarity of the two vectors ,Right now ;in, : the feature vector of skill i; : the feature vector of skill j; : represents skill i; : represents skill j; the comprehensive task performance factor is calculated based on the normalized score of the students completing the comprehensive task.

[0020] This example describes the implementation of a dynamic cognitive map construction module. The module's dynamic association strength quantitative analysis aims to establish a method that can accurately measure the strength of nonlinear, context-dependent connections between skills. This method overcomes the drawback of existing techniques that treat skills as isolated units and provides a solid data foundation for subsequent topological structure modeling. The core of this module lies in a dynamic association strength model. To achieve the above quantitative analysis, the process identifies three core factors; the first is the co-occurrence of students' learning activities, which refers to the normalized frequency of students learning or practicing two different skills at the same time within a specific time window. Its function is to characterize the correlation at the learning behavior level; the second is the intrinsic semantic similarity, which refers to the degree of intrinsic correlation between the two skills at the knowledge content level. Its function is to measure the fit of skills from the perspective of knowledge ontology; the third is the comprehensive task performance factor, which refers to the performance level of students in comprehensive tasks that require the simultaneous use of two skills to complete. Its function is to introduce a penalty item based on the actual application effect.

[0021] Based on the above factors, combined with a set of preset weights, the above three factors are comprehensively calculated to generate any two skill nodes and The dynamic correlation strength between them is calculated by the following correlation strength model:

[0022] in: :Skill and The dynamic association strength is a dimensionless value that is updated after each student learning activity to adjust the weight of the edge connecting the skill nodes in the cognitive map; :The co-occurrence degree of students’ learning activities, dimensionless, comes from the representation that within the time window, students are exposed to skills at the same time and Normalized frequency of learning materials; : Intrinsic semantic similarity is a dimensionless constant; in this embodiment, the source of intrinsic semantic similarity is a preset language model for skill and The default model is a deep learning model trained on a large-scale text corpus. The technology was chosen because it has been proven to be highly effective in capturing complex semantic relationships. After the skill description text is vectorized, the cosine similarity between the two is calculated to obtain ; : Comprehensive task performance factor, a range of In this embodiment, the comprehensive task performance factor is calculated based on the normalized score of the students completing the comprehensive task; when the students complete a task that requires the simultaneous use of skills and tasks and obtain a normalized score , the factor is calculated as:

[0023] The better the performance, The closer it is to 1, The closer it is to 0, the smaller the inhibitory effect on the association strength; : These are dimensionless, adjustable weights for co-occurrence, similarity, and performance factors, respectively. These preset weights are determined through regression analysis of training data from a small pilot user population, aiming to optimize the model's fit to the real-world ability associations. A feasible regression analysis method is to invite domain experts to manually mark the correlation between a series of skill pairs to obtain a reference correlation dataset; , intrinsic semantic similarity Penalty term for comprehensive task performance As the independent variable, the manually annotated reference correlation is used as the dependent variable to construct a multiple linear regression model; finally, the model is trained on the data collected from the pilot user group to solve the weight coefficient that can minimize the model prediction error. ; The constant 1 in the denominator of the formula is used to ensure the stability of the calculation and avoid the denominator being zero; By introducing an association strength model encompassing three dimensions: learning behavior, semantic content, and application performance, and specifying the calculation method for key factors, this system can more comprehensively and accurately quantify the dynamic synergy between skills. Compared to evaluation methods that rely solely on a single dimension, such as co-occurrence or similarity alone, the calculation results of this embodiment can more realistically reflect the student's ability structure in a specific context, significantly improving the accuracy and effectiveness of the dynamic cognitive map and providing a high-quality data foundation for precise decision-making in upper-level applications. To accurately measure the synergistic relationship between skills, a dynamic correlation strength model was introduced. This model is not a simple linear superposition, and its internal mechanism can be understood through a ratio of synergistic to inhibitory effects. Association strength Co-occurrence Similarity Performance factor Weight coefficient , in, : association strength; : co-occurrence degree; :similarity; :Performance factor; are the weight coefficients of co-occurrence, similarity, and performance factor respectively; the numerator of this formula , characterizes the synergistic factors that promote the association between two skills; among them, the co-occurrence of learning activities represents the association at the learning behavior level, while the intrinsic semantic similarity represents the inherent fit at the knowledge content level; the denominator represents the factors that inhibit skill association, among which the comprehensive task performance factor Quantified the barriers students face when integrating two skills; an inefficient performance can lead to The value increases, thereby suppressing the overall association strength; this model is driven by the collected student interaction data, dynamically quantifies the nonlinear coupling strength between any two skills, and constructs a dynamic cognitive map that can truly reflect the internal structure of the student's individual ability.

[0024] Among them, the preset weight coefficient They are not fixed. They are optimized and determined by regression analysis of training data from a small pilot user group. Their role is to balance the contributions of the three dimensions of learning behavior, content similarity, and actual application performance in the final association strength calculation; similarly, intrinsic semantic similarity The preset language model that the calculation relies on is pre-trained based on a massive text corpus. Its working principle is to map the text description of the skill into a high-dimensional vector space, so that semantically similar skills are closer in this space; this preset ensures the robustness and universality of the similarity calculation.

[0025] Example 3 The process of calculating the emergence index in the emergent capability identification and prediction module includes: Determine the number of skill nodes within the skill cluster; Determine the normalized eigenvector centrality of each skill node; The emergence index of the skill cluster is calculated by combining the number of skill nodes, normalized eigenvector centrality, and the dynamic correlation strength between nodes; Skill clusters are identified by executing a community discovery algorithm on a dynamic cognitive graph; In this embodiment, the process of calculating the emergence index in the emergent capability identification and prediction module is described. The purpose of this process is to conduct an in-depth analysis of the topological structure of the cognitive map to quantitatively identify those emergent capability clusters that represent comprehensive capabilities. The initial step of this process is to identify candidate skill clusters from the dynamic cognitive graph. A skill cluster is a subset of skill nodes in the graph that are densely connected internally but relatively sparsely connected to external nodes. In this embodiment, skill clusters are identified by executing a community discovery algorithm on the dynamic cognitive graph. Community discovery algorithms are graph algorithms that automatically detect community structures in complex networks. This embodiment uses mature community discovery algorithms such as the Louvain algorithm. The technical motivation is that such algorithms can efficiently and objectively extract natural clusters of skills from data-driven graph structures, avoiding the subjectivity and limitations of manually pre-set skill combinations. When using the Louvain algorithm, a key parameter, resolution, needs to be set, which affects the size of the skill clusters that are ultimately identified: higher resolutions tend to find more and smaller skill clusters, while lower resolutions tend to find fewer and larger skill clusters. In this solution, this parameter can be adjusted within a preset range, such as 0.5 to 1.5, and the average emergence index of the identified skill clusters can be selected. Achieve the maximum resolution value to obtain the most meaningful ability clustering results; Identifying skill clusters Finally, the calculation process of the emergence index includes the following steps: determining the number of skill nodes in the skill cluster ; Determine the normalized eigenvector centrality of each skill node Normalized eigenvector centrality is a measure of a node's influence in a network. It not only considers the number of direct connections to a node but also measures the importance of its adjacent nodes. The technical motivation for choosing this metric is that in a competency network, the importance of a skill depends not only on the skill itself but also on the importance of the other skills it connects to. The emergence index of a skill cluster is calculated by combining the number of skill nodes, the normalized eigenvector centrality, and the strength of the dynamic connections between nodes. This calculation is performed using the following emergence index model:

[0026] in: : No. Skill Clusters The emergence index is a dimensionless value. The higher the value, the stronger the comprehensive ability represented by the skill cluster. : Skill Cluster The number of skill nodes contained in is dimensionless and is obtained by counting the nodes in the cluster; :Skill node i; : kth skill cluster; : Index of skill cluster; :Skill Node The normalized eigenvector centrality of , dimensionless, is calculated by the standard graph computing library on the cognitive graph; :Skill and The dynamic correlation strength between them comes from the calculation output of the aforementioned dynamic correlation strength model and is the key input of this module; The technical benefits of this embodiment are as follows: By combining the community discovery algorithm and the emergence index model, this system establishes an objective and quantitative framework for identifying emergent capabilities. This framework not only automatically discovers potential capability combinations and skill clusters, but also accurately assesses the degree of emergence of these combinations, that is, the strength of their synergistic effects, through the emergence index model. This enables the system to accurately identify the high-value comprehensive capability directions that are crucial to student development from a complex landscape, providing clear and quantifiable goals for subsequent adaptive assessment and personalized recommendations. Based on the constructed dynamic cognitive map, this solution further defines an emergence index model to identify the ability combination with special value composed of multiple skill nodes, namely the ability emergence cluster. Emergence index Number of nodes in the cluster Node centrality Correlation strength:

[0027] in, ; :Skill cluster; : represents the skill node in the cluster; : represents a skill cluster; the derivation of this formula is not constructed out of thin air; its core is that the degree of emergence of a skill cluster depends not only on the simple accumulation of its internal correlation strength, but also on a nonlinear amplification effect; the logarithmic function The introduction of aptly expresses this effect: when the sum of the internal correlation strengths When it is low, the emergence index is approximately proportional to the sum; but as the internal synergy increases, its growth will tend to be flat, which is in line with the law of capability development; at the same time, multiplying the normalized eigenvector centrality of each node , which means that the core skills with greater influence in the network are given higher weights for their contribution to capability emergence; the whole formula is based on the size of the skill cluster. Normalization is performed to ensure the comparability of emergence indices between skill clusters of different sizes; Through this model, the system can objectively identify emerging clusters of capabilities from the data where the whole is greater than the sum of its parts, and transform implicit capabilities that are invisible in traditional assessments into measurable and comparable quantitative indicators. This is an essential improvement compared to the existing technology that relies on experts' predefined skill combinations, and it achieves a leap from hypothesizing capability structures to discovering capability structures.

[0028] Example 4 The process of updating the evaluation dimension weights in the adaptive evaluation engine module includes: Determine the emergence index of the competency cluster most relevant to the assessment dimension; determine the average emergence index of all competency clusters; obtain the current weight of the assessment dimension; obtain the engagement gate that represents the learner's willingness to learn; adjust the current weight based on the difference between the emergence index of the most relevant competency cluster and the average emergence index, and in response to the state of the engagement gate, according to a preset learning rate, to generate the weight at the next moment; The process of determining the emergence index of the most relevant competence cluster of the assessment dimension is as follows: Calculate the correlation between the evaluation dimension and each skill cluster, and determine the skill cluster with the greatest correlation as the ability cluster most relevant to the evaluation dimension; set the emergence index of the most relevant ability cluster as the emergence index of the ability cluster most relevant to the evaluation dimension; The update logic of evaluation dimension weights is as follows: When the emergence index of the competency cluster most relevant to the assessment dimension is higher than the average emergence index of all competency clusters, and the participation gate indicates that the learner has participated in the relevant learning activities, the weight of the assessment dimension is increased; On the contrary, the weight of the evaluation dimension will be weakened or kept unchanged; In this embodiment, the process of updating the evaluation dimension weights in the adaptive evaluation engine module is described. This process aims to establish a data-driven feedback mechanism to enable the evaluation framework to continuously and automatically focus on the most valuable competency development direction for the learner, thereby achieving personalized and dynamic evaluation. The core of updating the evaluation dimension weights lies in an evaluation dimension contribution model. The underlying logic is that when a competency cluster highly correlated with a particular evaluation dimension exhibits a strong emergent trend, and the learner demonstrates a willingness to learn in that competency direction, the system increases the weight of that evaluation dimension; otherwise, it decreases or remains unchanged. In order to establish the association between the evaluation dimension and the ability cluster, the correlation between the evaluation dimension and each skill cluster is calculated, and the skill cluster with the largest correlation is determined as the ability cluster most relevant to the evaluation dimension; Evaluate dimensions by calculation Semantic vectors and capability clusters It is determined by the cosine similarity between the average values ​​of all skill vectors within the skill vector. The method for generating the semantic vector of the evaluation dimension d is as follows: for each evaluation dimension d, for example, emergency handling capability, a set of core keywords is predefined, such as emergency, sudden, decision, and disposal; the same preset language model as that used to generate the skill vector is used, such as the aforementioned BERT model, to obtain the feature vector of each keyword, and these vectors are averaged. The result is the semantic vector of the evaluation dimension. ; Based on this, the emergence index of the most relevant capability cluster is Set as the emergence index of the capability cluster most relevant to the evaluation dimension, denoted as ; The weight updating process includes the following steps: First, determine the emergence index of the capability cluster most relevant to the evaluation dimension. Second, determine the average emergence index of all capability clusters , is used as a dynamic evaluation baseline; third, obtain the current weight of the evaluation dimension At the same time, obtain the participation gate that represents the students' learning willingness Engagement gate refers to a binary activation function that determines whether the learner has accepted the learning activities recommended by the system and related to the competency cluster and has shown high engagement. If so, the value is 1, otherwise it is 0. High participation can be judged by the following logic: when the system recommends the most relevant competence cluster to the trainee After the corresponding learning activity, within a preset time window, such as one week, if the student completes more than a predetermined proportion, such as 50% of the recommended learning tasks, or the cumulative learning time on the relevant learning materials exceeds a predetermined threshold, such as 30 minutes, then it is determined to be high engagement, and the engagement gate is then The value of is 1, otherwise it is 0; that is, the basis for judging high participation is the completion rate of recommended tasks or the cumulative learning time of students within the specified time;

[0029] in, : Engagement gate, which is a binary activation function with a value of 0 or 1; : represents a certain evaluation dimension. The setting of the gate ensures that the adjustment of the weight can reflect the active choice of the students; based on a preset learning rate , based on the difference between the emergence index of the most relevant ability cluster and the average emergence index, and in response to the state of the participation gate, the current weight is adjusted to generate the weight of the next moment; the preset learning rate is a dimensionless constant that controls the step size of weight adjustment. Its initial value is determined through A / B testing on a small pilot user group to select the value that can converge and stabilize the evaluation weight distribution in the shortest time. The weight update is performed by the following model:

[0030] in: : Evaluation dimension The weight at the next moment, dimensionless; : Evaluation dimension The current weight of is dimensionless and comes from the calculation result of the system at the previous moment; : learning rate, a dimensionless constant; : and evaluation dimensions The emergence index of the most relevant capability cluster is dimensionless and comes from the calculation results of the aforementioned emergence index model; : The average emergence index of all capability clusters, dimensionless, is obtained by The values ​​are averaged to obtain; : represents the evaluation dimension d; : represents the current moment; : represents the next moment; : Participation gate, the value is 0 or 1, and its value is determined by the students' interactive behavior data; The updating logic of the evaluation dimension weights is clearly defined: when the emergence index of the capability cluster most relevant to the evaluation dimension is higher than the average emergence index of all capability clusters ( ), and the participation gate indicates that the learner has participated in the relevant learning activities ( ), then the adjustment term of the update formula is positive, which increases the weight of the evaluation dimension; otherwise, or , if the adjustment item is negative or zero, the weight of the evaluation dimension will be weakened or kept unchanged.

[0031] Through the aforementioned weight update mechanism, the system establishes a complete, intelligent closed loop from identifying emerging abilities to adjusting the assessment framework. This mechanism ensures that assessment weights are allocated based on dual evidence—the student's objective ability development and emergence index—and their subjective learning willingness and engagement, rather than pre-set static rules. This enables a truly personalized and dynamic assessment framework, significantly improving the accuracy of identifying individual learning patterns and recommending learning paths, ultimately significantly enhancing the efficiency and effectiveness of the entire intelligent training system. To enable the assessment system to respond to changes in student capabilities, this solution designs an adaptive assessment engine module; its core assessment dimension weight update model can be regarded as an online learning and feedback control mechanism; Dimension weight learning rate most relevant cluster emergence index average emergence index participation gate:

[0032] : Evaluation dimension The current weight of is dimensionless and comes from the calculation result of the system at the previous moment; : learning rate, a dimensionless constant; : and evaluation dimensions The emergence index of the most relevant capability cluster is dimensionless and comes from the calculation results of the aforementioned emergence index model; : The average emergence index of all capability clusters, dimensionless, is obtained by The values ​​are averaged to obtain; : Participation gate, the value is 0 or 1, and its value is determined by the students' interactive behavior data; The inherent rationality of this update logic lies in that it binds the adjustment of the evaluation dimension to two key factors: the objective emergence of ability and the subjective willingness of the trainees; the difference item As a feedback signal, when a When the emergence level of the most relevant ability cluster is higher than the average level, the signal is positive, driving the system to increase attention to this dimension; otherwise, it will reduce attention; and the engagement gate The introduction of is crucial, as it ensures that the weight enhancement will only be triggered when the trainee shows actual participation in the relevant learning direction recommended by the system; this avoids the system forcing the trainee to develop competence directions that he or she is not interested in or not prepared for.

[0033] Preset learning rate Here, it plays the role of a dynamic balancer of the system. Its value is determined through A / B testing on pilot user groups to achieve the optimal balance between the rapid response and overall stability of the evaluation framework; Discovery of dynamic correlation: When students repeatedly practice instrument identification and crosswind correction in the simulator, the system calculates the high-pressure relationship between the two through the dynamic correlation strength model. value, the weight of the corresponding edge in the cognitive graph will increase; Identification of emergent capabilities: The community discovery algorithm identifies a tight cluster of skills on the graph that includes instrument recognition, crosswind correction, and tower communication. , and calculated by the emergence index model that it has a very high emergence index , the system marked it as an emerging ability cluster named complex situation decision-making; Adaptive assessment focus: The adaptive assessment engine module detected this high emergence index and found that it was highly correlated with the assessment dimension of emergency response capability. When trainees received and participated in more such comprehensive scenario training, After that, the system automatically and significantly increased the weight of the emergency response capability evaluation dimension. ; The assessment report provided by this solution is no longer a list of discrete skill scores. Instead, it presents a dynamically evolving capability radar chart centered on emergent capabilities such as complex situational decision-making. It not only identifies students' strengths and weaknesses but also reveals the synergistic structure among their various capabilities, increasing the accuracy of learning path recommendations to over 85%, enabling a precise portrayal and effective guidance of students' true comprehensive capabilities. To verify the technical effect of the present invention, a comparative experiment was conducted. 100 trainees were randomly divided into two groups: an experimental group of 50 people who used the intelligent training system of the present invention, and a control group of 50 people who used a traditional fixed assessment model system based on isolated skill point assessment. The experimental period was three months. Verification of the deviation of the evaluation results: After the experiment, all students participated in a comprehensive practical assessment designed by domain experts. The assessment results were regarded as the benchmark of the students' true comprehensive ability. The results showed that the average deviation rate between the evaluation scores of the control group system and the expert assessment scores was 68%, while the average deviation rate between the ability radar chart evaluation results output by the experimental group system and the expert assessment scores was only 15%. This proves that the present invention effectively solves the problem of deviation between the evaluation results and the students' true ability.

[0034] Verification of the accuracy of learning path recommendations: During the experiment, the system will recommend subsequent learning paths to students based on the evaluation results. We record the students' acceptance rate of the recommended content and the ability improvement rate after completion. The results show that the acceptance rate of the control group is about 40%, while the acceptance rate of the experimental group is as high as 92%. After completing the recommended learning, the emergence index of the experimental group students in the relevant skill clusters is The average improvement was 60% higher than that of the control group, confirming that the accuracy and effectiveness of the recommended path of the present invention reached a high level, such as above 85%.

[0035] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, and all of these should be included in the scope of protection of the present invention.

Claims

1. A comprehensive management system for student data of smart training, characterized by: include: The learning trajectory data collection module is used to obtain the student interaction data during the training process; A dynamic cognitive map construction module is used to respond to student interaction data and quantitatively analyze the dynamic correlation strength between skills to construct a dynamic cognitive map that represents the student's ability; The emergent ability identification and prediction module is used to identify emergent ability clusters by calculating the emergence index of skill clusters based on the dynamic cognitive map; The adaptive evaluation engine module is used to dynamically update the weights of the evaluation dimensions in response to the emergence index.

2. A smart training student data integrated management system according to claim 1, characterized in that: The process of the dynamic cognitive map construction module performing dynamic association strength quantitative analysis includes: Determine the co-occurrence of students' learning activities to represent learning behavior, the intrinsic semantic similarity to represent the intrinsic connection of skills, and the comprehensive task performance factor to represent comprehensive application performance; Combined with the preset weights, the co-occurrence of students’ learning activities, intrinsic semantic similarity and comprehensive task performance factors are comprehensively calculated to generate dynamic association strength.

3. The intelligent training student data integrated management system according to claim 2, characterized in that: The intrinsic semantic similarity is obtained by processing the skill description text using a preset language model; The comprehensive task performance factor is calculated based on the normalized score of the students completing the comprehensive task.

4. The intelligent training student data integrated management system according to claim 1, characterized in that: The process of calculating the emergence index by the emergent capability identification and prediction module includes: Determine the number of skill nodes within the skill cluster; Determine the normalized eigenvector centrality of each skill node; The emergence index of the skill cluster is calculated by combining the number of skill nodes, normalized eigenvector centrality, and the dynamic correlation strength between nodes.

5. The intelligent training student data integrated management system according to claim 4 is characterized in that: The skill clusters are identified by executing a community discovery algorithm on a dynamic cognitive graph.

6. The intelligent training student data integrated management system according to claim 1, characterized in that: The process of updating the evaluation dimension weights by the adaptive evaluation engine module includes: Determine the emergence index of the competency cluster most relevant to the assessment dimension; determine the average emergence index of all competency clusters; obtain the current weight of the assessment dimension; obtain the engagement gate that represents the student's learning willingness; based on the preset learning rate, based on the difference between the emergence index of the most relevant competency cluster and the average emergence index, and in response to the state of the engagement gate, adjust the current weight to generate the weight at the next moment.

7. The intelligent training student data integrated management system according to claim 6, characterized in that: The process of determining the emergence index of the most relevant capability cluster of the assessment dimension is as follows: Calculate the correlation between the evaluation dimension and each skill cluster, and determine the skill cluster with the largest correlation as the ability cluster most relevant to the evaluation dimension; set the emergence index of the most relevant ability cluster as the emergence index of the ability cluster most relevant to the evaluation dimension.

8. The intelligent training student data integrated management system according to claim 6, characterized in that: The update logic of the evaluation dimension weight is as follows: When the emergence index of the competency cluster most relevant to the assessment dimension is higher than the average emergence index of all competency clusters, and the participation gate indicates that the learner has participated in the relevant learning activities, the weight of the assessment dimension is increased; Otherwise, the weight of the evaluation dimension will be weakened or kept unchanged.

Images