An adaptive personalized learning path generation method and system

Abstract

The application provides a self-adaptive personalized learning path generation method and system, which comprises the following steps: extracting a target ability list from an existing training model or post competence requirement, and constructing an ability graph; generating a user ability state vector by using a weighted scoring mechanism based on user historical course test results and actual operation task completion conditions; performing priority scoring on the ability nodes, screening the ability nodes to form a learning path according to the priority scoring, allocating learning resources in combination with user preferences, and generating a task scheduling plan; issuing tasks according to the task scheduling plan, and collecting behavior data; calculating a behavior effectiveness score, identifying abnormal tasks with a behavior effectiveness score lower than a threshold, and adding a behavior deviation label; matching a correction strategy based on the behavior deviation label of the abnormal task, adjusting the path and the scheduling plan, and outputting the corrected learning path and scheduling table.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of online education, and particularly relates to a self-adaptive personalized learning path generation method and system. BACKGROUND

[0002] With the deepening of enterprise digital transformation and the increase in the number of knowledge-based positions, how to quickly improve the position competence and ability level of employees has become the core goal of enterprise training and education system. Traditional learning systems often focus on providing content recommendations or course lists. Although this approach can provide users with certain learning resources, it lacks path planning capabilities and cannot reflect the structural dependency relationship between learning content, making it difficult to ensure progressive and goal-oriented learning. In practice, users often face problems such as mismatch between recommended content and position goals, fragmented learning process, and uneven learning burden. For example, although enterprise training platforms can record course completion, subsequent learning recommended by the system is often based on static matching of a single tag, without considering the user's actual mastery of certain ability points and the logical order of knowledge points. This fragmented recommendation mode not only leads to low learning efficiency, but also easily causes deviation in the achievement of ability goals. In recent years, some systems have attempted to use knowledge graphs or simple machine learning models to improve path recommendations, but these methods still have many shortcomings. First, learning paths lack dynamism and are often difficult to update based on actual user performance after initial generation. Second, there is a lack of effective monitoring at the task execution level, which cannot determine whether the user's learning behavior is truly effective, but only uses course completion rate or click behavior as a standard. Third, path correction methods are too rough, often only allowing for overall rearrangement or regeneration, making it difficult to make targeted local adjustments while maintaining path stability. These shortcomings make it difficult for existing technologies to meet the needs for learning effectiveness, path rationality, and execution controllability in high-demand scenarios such as enterprise training. SUMMARY

[0003] The present application aims to design a self-adaptive personalized learning path generation method and system that can achieve a complete link from ability target modeling to path generation, from task execution monitoring to path correction, significantly improving the rationality, execution effectiveness, and dynamic adaptability of learning paths.

[0004] To achieve the above-mentioned purpose, in a first aspect of the present application, a self-adaptive personalized learning path generation method is provided, which comprises:

[0005] Extract a list of target capabilities from existing training models or job competency requirements, and construct a capability graph containing a set of capability nodes and a set of edges based on the sequential dependencies between capabilities; collect the descriptive text corresponding to each capability node, process it through word segmentation and word embedding, and then input it into a fully connected neural network to generate node semantic vectors; based on the user's historical course test results and practical task completion status, use a weighted scoring mechanism to generate a user capability status vector, and fill missing data with the average value of all users.

[0006] The system prioritizes capability nodes based on a structure-aware ranking function, which integrates the differences in users' mastery of nodes, the structural dependencies reflected by the in-degree and out-degree of nodes in the graph, and the complexity of the node's semantic vector. Based on the priority scores, capability nodes are selected to form learning paths. After topological validity checks, learning resources are allocated in combination with user preferences, and a task scheduling plan is generated based on the user's daily available learning time model.

[0007] Tasks are issued according to the task scheduling plan, and four types of behavioral data are collected: task dwell time, content integrity flag, evaluation score, and number of interruptions. A behavioral validity score is calculated using a behavioral validity function with an anomaly detection regularization term. Abnormal tasks with behavioral validity scores below the threshold are identified and behavioral deviation labels are added.

[0008] Based on the behavior deviation label matching correction strategy for abnormal tasks, a path perturbation cost function is introduced to control the correction scale. Under the constraint of total perturbation budget, the path and scheduling plan are adjusted, and the corrected learning path and scheduling table are output.

[0009] Furthermore, the sequential dependencies between capabilities are determined by fusing the hierarchical relationships annotated by domain experts with the sequence pattern mining results of historical user learning paths. The sequence pattern mining uses a sliding window algorithm to extract combinations of high-frequency continuous learning nodes.

[0010] Furthermore, the node semantic vectors are processed by a pre-trained 300-dimensional word embedding model to obtain a word vector matrix from the descriptive text, and then output through two layers of fully connected neural networks with ReLU activation functions, resulting in a vector dimension of 64 dimensions.

[0011] Furthermore, the specific steps of the weighted scoring mechanism include: dividing the user's ability evaluation dimensions into theoretical learning performance and practical task completion, and pre-setting the weight of each dimension according to the ability type; extracting the user's test scores and answer accuracy data in theoretical learning, and the completion quality scores and step compliance data in practical tasks, and mapping the two types of data to obtain the normalized scores of the theoretical dimension and the practical dimension; multiplying the normalized score of the theoretical dimension by the corresponding weight, and adding the result of multiplying the normalized score of the practical dimension by the corresponding weight, to obtain the comprehensive score of that ability node.

[0012] Further, the specific steps of the structure-aware ranking function include: calculating a mastery difference score of the user for the node, which is the difference between the score of the corresponding node in the current ability state vector and the target mastery threshold; calculating a structure dependence score, which is the weighted sum of the in-degree value and the out-degree value of the node in the ability graph, with the in-degree value being given a higher weight by a preset adjustment factor; calculating a complexity score of the node semantic vector, which is the information entropy value calculated based on the dispersion degree of the vector elements; and fusing the mastery difference score, the structure dependence score, and the complexity score by a preset proportion to generate a priority score for each node.

[0013] Further, the specific steps of allocating learning resources based on user preferences include: constructing a user preference vector based on the selection frequency and completion quality of different resource types in the user's historical learning records, with the vector elements corresponding to the preference weights of each type of resource; extracting the resource demand label of the current ability node and matching the resource type with the highest weight in the preference vector; if the number of the matched resource type in the resource library is less than a preset threshold, starting the supplementary process: first checking the content keyword matching degree between the graphic-text type resource and the current node, and selecting this type when the matching degree exceeds the threshold; if the graphic-text type resource does not meet the requirements, then checking the knowledge point coverage integrity of the PPT type resource and the scene adaptation degree of the mixed case type resource in turn, until a resource type that meets the requirements is selected; and recording the resource allocation result for updating the weights of the user preference vector.

[0014] Further, the abnormality detection regular term is based on the group historical average evaluation score of the current ability node, and the deviation of the user's score from the group average is regulated by a penalty coefficient.

[0015] Further, the path disturbance cost function calculates the disturbance cost by integrating the position offset before and after task correction, the time length variation ratio, and the depth of the node in the graph, and the structure penalty coefficient controls the weight of each disturbance term.

[0016] Further, the behavior deviation label is obtained based on the following steps:

[0017] When the evaluation score is lower than the preset evaluation score lower limit, the stay time is not less than the preset stay time lower limit, and the content is complete, the behavior deviation label is the ability deficiency type, and the correction strategy is intensive training;

[0018] When the stay time is lower than the preset stay time lower limit, the number of interruptions exceeds the preset number of interruptions upper limit, and the evaluation score is not lower than the preset evaluation score lower limit, the behavior deviation label is the attention dispersion type, and the correction strategy is rhythm adjustment;

[0019] If the content is not complete and the stay time is lower than the preset stay time lower limit, the behavior deviation label is the incomplete execution type, and the correction strategy is process simplification.

[0020] If the evaluation score is lower than the preset evaluation score lower limit and the number of interruptions exceeds the preset upper limit of the number of interruptions, the behavior deviation label is comprehensive disorder type, and the correction strategy is combined intervention.

[0021] In a second aspect of the application, a self-adaptive personalized learning path generation system is provided, the system comprising:

[0022] A state construction module is configured to extract a target ability list from an existing training model or job competency requirement, construct an ability graph containing an ability node set and an edge set according to the dependency relationship between abilities, collect the description text corresponding to each ability node, input the processed text into a fully connected neural network after word segmentation and word embedding, and generate a node semantic vector; based on the historical course test results and practical operation task completion of the user, a user ability state vector is generated using a weighted scoring mechanism, and missing data is filled with the average of all users;

[0023] A path generation module is configured to prioritize the ability nodes based on a structure-aware ranking function that takes into account the differences in the user's mastery of the nodes, the structural dependencies reflected by the in-degree and out-degree of the nodes in the graph, and the complexity of the node semantic vectors; based on the priority score, the ability nodes are filtered to form a learning path, and after a topological validity check, learning resources are allocated based on user preferences, and a task scheduling plan is generated based on the user's daily available learning time model;

[0024] A behavior evaluation module is configured to issue tasks based on the task scheduling plan, collect four types of behavior data: task stay time, content integrity flag, evaluation score, and number of interruptions; calculate the behavior effectiveness score using a behavior effectiveness function containing an anomaly detection regular term, and identify abnormal tasks with a behavior effectiveness score below a threshold and add a behavior deviation label;

[0025] A path deployment module is configured to match a correction strategy based on the behavior deviation label of the abnormal task, introduce a path disturbance cost function to control the correction size, adjust the path and scheduling plan under the constraint of the total disturbance budget, and output the corrected learning path and scheduling table.

[0026] The application has at least the following beneficial technical effects:

[0027] To solve the above problems, the application provides a kind of adaptive personalized learning path generation method and system, by constructing ability atlas and user ability state vector, learning goal is converted into directed graph structure, and the semantic representation of node is formed by combining the historical learning behavior and task performance of user, to ensure the goal-oriented nature and structural integrity of path generation.On this basis, the system introduces the ranking function of graph structure perception, considers the user ability difference, the structural position of node in atlas and its semantic complexity, and gives priority to the priority score of task node in path, generates the learning path sequence that meets the logical dependence and individual demand, and generates task scheduling plan in combination with user preference and learning time model.The system uses front-end burying point technology to collect the residence time, completion, test results and behavior stability of task in the execution phase, and through the design of behavior effectiveness function with abnormal detection regular term, it can ensure that false learning conditions such as "hang up completion", "abnormally high score" or "unstable learning" can be identified.In the path correction phase, the system selects replacement, insertion, splitting or postponement and other correction methods according to the abnormal label of task, and introduces path disturbance cost function to control the correction scale, to realize fine adjustment of local task while maintaining the overall stability of path.Through the above design, the application realizes the whole-process closed-loop management of learning path, that is, the complete link from ability target modeling to path generation, from task execution monitoring to path correction, which significantly improves the rationality, execution effectiveness and dynamic adaptive ability of learning path, and is especially suitable for enterprise training and high-demand capacity training scenarios. BRIEF DESCRIPTION OF DRAWINGS

[0028] The application is further described by means of the accompanying drawings, but the embodiments in the drawings do not constitute any limitation on the application, and other drawings can be obtained by those skilled in the art without creative labor on the basis of the following drawings.

[0029] Figure 1 A flowchart of the adaptive personalized learning path generation method of the application.

[0030] Figure 2 A system framework diagram of the adaptive personalized learning path generation system of the application. DETAILED DESCRIPTION

[0031] The embodiments of the application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference signs represent the same or similar elements or elements with the same or similar functions throughout. The embodiments described below by reference to the drawings are exemplary and are only used to explain the application, and cannot be understood as a limitation on the application.

[0032] In one or more embodiments, as Figure 1As shown, a self-adaptive personalized learning path generation method is disclosed, which comprises the following steps:

[0033] S1: Extract the target ability list from the existing training model or job competence requirement, construct the ability graph containing the ability node set and the edge set according to the dependence relationship between the abilities; collect the description text corresponding to each ability node, input the full connection neural network after word segmentation and word embedding processing, and generate the node semantic vector; based on the user's historical course test results and practical operation task completion, a weighted scoring mechanism is used to generate a user ability state vector, and the missing data is filled with the average value of all users;

[0034] Specifically, this step aims to provide structured input for path planning, including two core components: one is the ability graph structure representing the learning goal, and the other is the ability state vector reflecting the user's current ability mastery. This step is the basis of the entire path system, and its construction result will be directly input for subsequent path generation and execution scheduling, so it needs to have complete structural expression, accuracy and actual operability.

[0035] Firstly, the system extracts the target ability list from the existing training model or job competence requirement of the enterprise or platform. For example, the ability requirements set by a certain enterprise for "construction project manager" may include "construction progress control", "contract risk identification", "cost control", "team communication and cooperation", etc. The system administrator inputs these ability items and their mutual dependence relationship in a structured form in the background, such as "construction progress control" needs "engineering plan preparation" as a prerequisite, forming a directed dependence relationship between nodes. The system constructs an ability graph wherein is the set of ability nodes, is the edge set representing the dependence relationship. The graph structure is stored in the form of adjacency matrix or adjacency list for subsequent calling by the graph neural network module.

[0036] Each ability node needs to be mapped to a structured learnable representation, so we collect the descriptive text of its corresponding course content, task description, evaluation standard, etc. These texts can be automatically extracted from the enterprise learning platform (such as LMS), for example, the system extracts the description text of the course "construction drawing design audit" as "master the standard of building construction drawing reading, and can identify and propose modification suggestions for design errors in construction drawing." All descriptions will be processed by standard word segmentation and mapped to word vectors, using a pre-trained 300-dimensional word embedding model (such as GloVe or Word2Vec), to form a word vector matrix, which is input into a two-layer fully connected neural network (64 dimensions per layer, ReLU activation), and outputs a fixed-length ability node semantic vector The embedding vector will participate in the path node priority learning as the initial representation of the node in the subsequent graph neural network.

[0037] At the same time of building the capability graph, the system needs to obtain the current mastery degree of each user on each capability point, i.e. the capability state vector The generation of this vector is based on the user's historical learning behavior data, mainly including two parts: one is the course record and test results completed by the user on the learning platform, and the other is the completion of the practical operation tasks or projects participated by the user. These data are usually provided by the learning platform API interface, such as the "learning completion", "answering score" and other record fields under SCORM or xAPI protocol, or can be exported from the performance system or training system structured data table. Taking the "construction progress control" capability as an example, if the user has completed the related course and passed the class test (such as scoring 85 points), and has recorded in the actual engineering task management system that he has led a project plan preparation task, the system will record his theoretical score and practical completion status respectively.

[0038] To unify the indicators from different sources, the system adopts a weighted scoring mechanism to form a comprehensive score for each capability point :

[0039] ;

[0040] Among them: represents the user's mastery of the capability node , with a value range of ; represents the user's average test score in the course corresponding to the node, from the record in the LMS, normalized to ; represents whether the user has completed the practical operation task corresponding to the node, with 1 for completion and 0 for non-completion, from the project system; respectively represent the weight coefficients of theoretical learning and practical operation completion, which are preset by the system according to the type of capability, for example , ;

[0041] Since and are dimensionless (normalized), the weight is a real number, and the dimensions of the numerator and denominator are consistent, so the result of is dimensionless, meeting the requirement of dimensional consistency.

[0042] For missing data items, such as the user never participates in a certain course or has no relevant practical operation record, the system uses the mean value of the node among all users to fill in, avoiding the occurrence of untrainable items, and reducing the priority of the node in the path generation.

[0043] Finally, the semantic embedding representation of each node in the ability graph needs to be constructed, which is used for subsequent graph calculation. The calculation process is as follows:

[0044]

[0045] Wherein: is the ability node is the corresponding text embedding matrix, is the number of words in the text description of the node; , is the weight matrix of the two-layer fully connected network, using ReLU activation; is the output node semantic vector, representing the semantic features of the ability node;

[0046] Since the network input and output are both unified vector dimensions, the embedding process does not involve physical quantity units, and the dimensional consistency is guaranteed in the network design.

[0047] This structural semantic representation method combined with the graph structure provides a structural and semantic unified node representation form for subsequent graph calculation.

[0048] S2: Priority score of the ability node based on the structure-aware ranking function, which comprehensively considers the difference in the user's mastery of the node, the structural dependence reflected by the in-degree and out-degree of the node in the graph, and the complexity of the node semantic vector; According to the priority score, the ability nodes are filtered to form a learning path, which is checked for topological validity, combined with user preference to allocate learning resources, and a task scheduling plan is generated according to the user's daily available learning time model;

[0049] Specifically, this step is based on the ability graph structure and the user ability state vector constructed in the previous stage, combined with the ability node semantic representation vector , to generate a learning path sequence for the current user , and generate a task scheduling plan for the path. The generation of this path is not only the selection and sorting of unmastered content, but also based on the structural dependence in the task graph, combined with the user's current mastery state difference, ability node semantic expression, task learning load estimation and user preference, to perform overall sorting and scheduling. The goal is to generate an executable path that meets the structural logic, ability difference, user preference, and adaptive time rhythm, as the main control sequence of the learning system.

[0050] This step completely depends on the output variables of the previous stage: ability graph provides the learning structural dependence, user ability state vector ​It provides a score for the level of mastery of each node, while the node semantic vector... Structured semantic input features are provided. All inputs are no longer preprocessed; these values ​​are used only in the graph structure for path sorting and time matching.

[0051] The core task of path prioritization is to determine the "learning urgency" of each capability node, that is, how critical a particular capability is to the user's current goal achievement path. Traditional methods often only consider the degree of mastery. inverse value (e.g.) Using graph structure propagation as a ranking metric ignores the impact of propagation and semantic complexity on the learning path. In this embodiment, there is a clear "stage-module-subtask" structure between capabilities. For example, "construction stage management" contains three sub-modules: "progress," "quality," and "safety." These sub-modules are further subdivided into multiple parallel capability points. If the propagation of the graph structure is not considered, the path ranking will lack learning progression and task concentration.

[0052] Therefore, this step introduces a structure-aware sorting function that integrates the following three types of features:

[0053] Ability difference (by) (Reflection: the worse the understanding, the higher the weight).

[0054] Graph structure dependency (the higher the propagation strength of a node as a predecessor / successor node in a multi-path, the higher its weight).

[0055] Node semantic complexity (by The semantic vector complexity measure.

[0056] Based on the following priority scoring formula:

[0057] ;

[0058] in: Represents capability nodes The current user's learning urgency score is used as a path ranking metric. This indicates the user's level of understanding of the capability node calculated in the previous stage, with a value range of [value range missing]. ; For nodes The in-degree represents the number of other nodes that the node depends on, reflecting its learning depth; For nodes The out-degree indicates the scope of the node's influence on other tasks as a basic capability; This is the semantic vector of the capability node, generated by the previous network step. Indicate its Norm, as a metric for semantic complexity; , Adjustment factors for graph structure terms (e.g.) , (This indicates an emphasis on the ability to depend on predecessors). This is a semantic complexity adjustment term used to control the weight of "language expression complexity" in the ranking; each part of this expression is a dimensionless value or a vector norm, therefore the final... Maintaining dimensionlessness ensures dimensional consistency.

[0059] This scoring function incorporates both structural topology and semantic complexity terms as bias enhancement factors. For example, a user may have a poor grasp of "safety solution communication," but this capability node has an out-degree of 0 and low semantic complexity, indicating weak propagation of subsequent capabilities and simple representation, thus its ranking priority is appropriately reduced. Conversely, another user may have a poor grasp of "construction organization design," but this capability is the starting point for multiple modules and has high semantic complexity, resulting in a higher ranking score and priority learning.

[0060] After sorting, the system arranges the data from highest to lowest. Value selection Each task node serves as the learning path for this round. And perform a topology validity check on it to ensure that all preceding nodes have been included or have been mastered (i.e. The threshold is generally taken as This check ensures that the path is structurally executable.

[0061] After the path sorting is completed, the system allocates a specific set of content resources to each node in conjunction with the task resource library. And estimate the task execution time. Resource selection is prioritized based on user preference vectors. For example, if a user prefers video learning, the system will prioritize video content; if this content type is missing, a recommendation priority supplement strategy will be used (e.g., text > PPT > mixed case studies).

[0062] Next, the scheduling module allocates time periods for each task using a calendar. The system also includes a user learning time model. This represents the user's available daily study time (units are uniformly normalized to a proportional time period). The system will calculate the estimated time for the task. Tasks are sequentially loaded into consecutive dates. If the current date is already saturated, the task is postponed to the next day. This scheduling method does not require a dynamic programming algorithm but is implemented through a sliding window allocation, balancing efficiency and pace.

[0063] S3: Issue tasks according to the task scheduling plan, and collect four types of behavioral data: task dwell time, content integrity flag, evaluation score, and number of interruptions; calculate the behavioral effectiveness score using a behavioral effectiveness function with an anomaly detection regularization term, identify abnormal tasks with behavioral effectiveness scores below the threshold, and add behavioral deviation labels;

[0064] Specifically, this step is based on the learning path generated in the previous step. and scheduling table Based on this, learning tasks are actually distributed on the terminal, and key behavioral data is collected during the user's task execution. The core purpose is to determine whether the learning task has been effectively completed through automatic data collection and modeling, without relying on subjective user feedback, and to provide clear quantitative basis for subsequent path correction and capability status updates.

[0065] In actual deployment, the system uses the scheduling table. The tasks are then pushed to the user's terminal sequentially, and behavior monitoring begins. Behavioral data collection is achieved through automatic event tracking on the client side, requiring no active user intervention and offering non-intrusiveness and stability. However, due to the "task-driven, highly standardized, and time-restricted" characteristics of corporate training or pre-employment learning scenarios, the system cannot collect a large amount of multi-dimensional data like in internet scenarios. Therefore, the behavioral data collection in this step has been structurally simplified, retaining only four types of behavioral indicators that can directly map learning effectiveness:

[0066] :Task The actual dwell time on the device is automatically timed by the terminal front end;

[0067] : Content integrity completion indicator, such as whether a video has played more than 95% or a PDF has been fully viewed, with a value of 0 or 1;

[0068] If the task includes assessments, then collect the scores (e.g., multiple choice, true / false, quizzes, etc.), and normalize them. ;

[0069] The number of times a user interrupts or switches tasks reflects the learning stability; more than one interruption is considered abnormal.

[0070] Considering that users may engage in behaviors such as "idling while watching videos" or "skipping content quickly" in real-world applications, directly using the aforementioned metrics for weighted summation would lead to severe scoring distortion. Therefore, this step introduces a structure-enhanced behavioral effectiveness scoring function. Used to comprehensively evaluate each task The execution effect of the model is evaluated, and a structure with an explanatory weight regularization term is introduced to enhance its robustness and discriminability for enterprise training scenarios.

[0071] ;

[0072] wherein: is the behavior effectiveness score of the task , whose value range is not limited, and the higher the value, the more sufficient the task completion is; is the actual stay duration (collected by the system front end); is the recommended learning duration (calculated by the path generation module), which is used for normalization; is the content completion flag, which is judged by the system through the position of the scroll bar or the video playback progress, and the value is 0 or 1; is the task test score, which is normalized to , and if there is no test, it is taken as a blank; is the number of jumps or the number of interruption behaviors (such as switching windows, closing and re-entering); is the current ability node of the user, and the historical average score of other users in the same post group is used for behavior anomaly detection; to is the weight coefficient of the score item, which is recommended to be configured as , , , ; is the regularization factor of the anomaly detection item, which is recommended to be set to 0.2, indicating that when the user score deviates from the group mean, a reasonable penalty is given; all items are dimensionless data, which are normalized or designed to have consistent dimensions.

[0073] The formula introduces , which is used to identify the situation of "false completion" of learning. Taking enterprise training as an example, if a user's test score in the "construction plan progress control" task is much higher than the historical average, but the behavior trajectory shows that the stay time is very short and the interruption times are high, the system will mark such behavior as "possibly completed with external assistance" or "not truly mastered", and this regularization term will automatically lower its value, effectively enhancing the robustness and credibility of the behavior score. This discrimination mechanism has not appeared in existing path recommendation systems, and has obvious differences and practicality.

[0074] After calculating the of all task nodes, the system generates a score sequence . Then, according to the score lower limit set by the system (such as 0.6), the behavior abnormal tasks are identified to form an abnormal task set And further add behavior deviation labels to each abnormal task, such as "low duration", "frequent interruption", "abnormal score", etc., to facilitate classification processing in the next path correction stage.

[0075] S4: Based on the behavior deviation label matching of the abnormal task, the correction strategy is introduced, the path disturbance cost function is introduced to control the correction scale, the path and the scheduling plan are adjusted under the total disturbance budget constraint, and the corrected learning path and the scheduling table are output.

[0076] Specifically, this step aims to correct and reconstruct the original learning path and the scheduling table based on the behavior score sequence output by the previous stage . Unlike the traditional recommendation system strategy of rearrangement or full replacement, the correction mechanism designed in this scheme emphasizes the fine identification of task execution deviation and the protection of structural continuity. Under the premise of meeting the learning goal coverage, the path structure is locally fine-tuned, and finally the corrected path and scheduling table are output. This step not only completes the closed-loop update of the path, but also provides a key behavior adjustment channel for the adaptive learning system, which is the core module of realizing the intelligent evolution and continuous optimization of the path.

[0077] This step additionally inputs the original path : used to identify the context sequence and knowledge structure position of the affected tasks; and the original scheduling table : used to identify the time arrangement, learning rhythm and possible scheduling conflicts of each task.

[0078] The system first determines the type of each abnormal task . This determination does not rely on a deep model, but uses a rule model to construct an abnormal label set , each label representing a specific abnormal mode, such as "low duration and high score", "high jump rate and low score", "score deviates from the group mean", etc. The judgment is completely based on the behavior data collected in the third step, for example:

[0079] If and , it can be marked as ;

[0080] If and , it is marked as ;

[0081] If and , it indicates that the mastery degree of the ability point does not match the result, and is marked as ;

[0082] The above-mentioned anomaly type identification logic enables the system to distinguish between two root causes: "content resource mismatch" and "user not aware of the problem," and to perform targeted corrective operations.

[0083] For each type of abnormal task, the system designs different correction strategies, forming a task correction operation set. And execute it in conjunction with the path context. Taking "resource mismatch" as an example, the system uses the capability graph... Find nodes resource pool And based on user preference vectors (Defined in step one) Reselect content resources If marked as "Learning Failure", then in Insert preceding capability node (Depend on Supplementary tasks consisting of edges determined in the middle) If the learning curve of a task fluctuates drastically, the task can be broken down into subtasks, such as splitting "construction progress coordination" into two microtasks: "node control principle" and "schedule Gantt chart drawing".

[0084] To control the correction intensity and maintain the path structure, a path perturbation cost function was designed. This is used to measure the offset of each correction operation from the path structure.

[0085] ;

[0086] in: For the task The cost of correcting disturbances; and These represent the index positions (i.e., sequence numbers) in the path before and after the task correction, used to measure the magnitude of the sorting change; The change in total duration after task adjustment, i.e. ; The depth of the capability node in the graph (measured from the root node) indicates the range of the correction effect; The structural penalty coefficient controls the weights of each disturbance term; it is typically set to... , , ;

[0087] The system is based on Path correction is performed to meet constraints. The total disturbance budget is configurable and can be set according to the flexibility of the company's training plan, such as... The original path length cost. Correction tasks that satisfy the constraints will update the path one by one. , replace as a correction task , and synchronize the schedule , consider the recommended time of the new task and the scheduling conflict window, and insert using the postponement mechanism.

[0088] In one or more embodiments, as Figure 2 indicated, an adaptive personalized learning path generation system is disclosed, the system comprising:

[0089] A state construction module is configured to extract a target ability list from an existing training model or job competency requirement, construct an ability graph containing an ability node set and an edge set according to the dependency relationship between abilities, collect the description text corresponding to each ability node, input the processed text into a fully connected neural network after word segmentation and word embedding, and generate a node semantic vector; based on the historical course test results and practical task completion of the user, a weighted scoring mechanism is used to generate a user ability state vector, and missing data is filled with the average of all users;

[0090] A path generation module is configured to prioritize the ability nodes based on a structure-aware ranking function that integrates the differences in the user's mastery of the nodes, the structural dependencies reflected by the in-degree and out-degree of the nodes in the graph, and the complexity of the node semantic vectors; the learning path is formed by filtering the ability nodes according to the priority score, and after topological legality check, the learning resources are allocated in combination with the user preferences, and the task scheduling plan is generated according to the user's daily available learning time model;

[0091] A behavior evaluation module is configured to issue tasks according to the task scheduling plan, collect four types of behavior data: task stay time, content integrity flag, evaluation score, and interruption frequency; calculate the behavior effectiveness score using a behavior effectiveness function containing an anomaly detection regular term, identify abnormal tasks with a behavior effectiveness score below a threshold, and add a behavior deviation label;

[0092] A path deployment module is configured to match a correction strategy based on the behavior deviation label of the abnormal task, introduce a path disturbance cost function to control the correction size, adjust the path and the scheduling plan under the constraint of the total disturbance budget, and output the corrected learning path and schedule.

[0093] It is worth noting that the specific workflow of the adaptive personalized learning path generation system provided by the embodiments of the present application is the same as the workflow of the adaptive personalized learning path generation method described in the above embodiments, and will not be repeated here.

[0094] The embodiment of the present application further provides a self-adaptive personalized learning path generation device, comprising a processor, a memory and a computer program stored in the memory and configured to be executed by the processor, wherein the processor implements the steps in the above-mentioned self-adaptive personalized learning path generation method embodiment, for example, steps S1-S4 in the above-mentioned embodiment, when executing the computer program. Figure 1 The processor implements the functions of the modules in the above-mentioned system embodiments when executing the computer program.

[0095] For example, the computer program can be divided into one or more modules, which are stored in the memory and executed by the processor to complete the present application. The one or more modules can be a series of computer program instruction segments capable of completing a specific function, which are used to describe the execution process of the computer program in the self-adaptive personalized learning path generation device.

[0096] The self-adaptive personalized learning path generation device can be a desktop computer, a notebook computer, a palm computer, a cloud server and other computing devices. The self-adaptive personalized learning path generation device can include, but is not limited to, a processor and a memory. Those skilled in the art can understand that the self-adaptive personalized learning path generation device can also include an input / output device, a network access device, a bus and the like.

[0097] The processor can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASAC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc. The processor is the control center of the self-adaptive personalized learning path generation device, which connects all parts of the self-adaptive personalized learning path generation device through various interfaces and lines.

[0098] The memory can be used to store the computer program and / or modules, and the processor realizes various functions of the adaptive personalized learning path generation device by running or executing the computer program and / or modules stored in the memory, and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, at least one application required by a function, etc.; and the data storage area can store data created according to the running of the air conditioner controller, etc. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0099] The modules integrated in the adaptive personalized learning path generation device can be stored in a computer-readable storage medium if they are realized in the form of software function units and sold or used as independent products. Based on this understanding, all or part of the processes in the above-mentioned embodiment methods can also be completed by a computer program instructing related hardware, and the computer program can be stored in a computer-readable storage medium. The computer program can implement the steps of the above-mentioned various method embodiments when executed by a processor. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or some intermediate forms, etc. The computer-readable medium can include any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc.

[0100] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by a computer program instructing related hardware, and the program can be stored in a computer-readable storage medium, and the program can include the processes of the above-mentioned various method embodiments when executed. The storage medium can be a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM), etc.

[0101] The above is the preferred embodiment of the present application, it should be pointed out that, for those skilled in the art, without departing from the principles of the present application, can also make a number of improvements and refinements, these improvements and refinements are also considered to be within the scope of the present application.

Claims

1. A method for generating adaptive personalized learning paths, characterized in that, The method comprises: extracting a target ability list from an existing training model or job competency requirement, constructing an ability graph containing an ability node set and an edge set according to the sequential dependence between abilities; collecting the description text corresponding to each ability node, inputting the text after word segmentation and word embedding processing into a fully connected neural network to generate a node semantic vector; based on the user's historical course test results and practical task completion, a weighted scoring mechanism is used to generate a user ability state vector, and missing data is filled with the average of all users; the specific method steps of the weighted scoring mechanism include: dividing the user ability evaluation dimension into theoretical learning performance and practical task completion, and presetting the weight of each dimension according to the type of ability; extracting the test score and answer accuracy data of the user in theoretical learning, and the completion quality score and step compliance data in practical tasks, respectively mapping the two types of data to obtain the normalized score of the theoretical dimension and the normalized score of the practical dimension; multiplying the normalized score of the theoretical dimension by the corresponding weight, adding the result after multiplying the normalized score of the practical dimension by the corresponding weight, and obtaining the comprehensive score of the ability node; based on the priority score of the ability node, a structure-aware ranking function is used to score the priority of the ability node, the structure-aware ranking function comprehensively considers the difference in the user's mastery of the node, the structural dependence reflected by the in-degree and out-degree of the node in the graph, and the complexity of the node semantic vector; according to the priority score, the ability nodes are filtered to form a learning path, after topological legality check, learning resources are allocated combined with user preferences, and a task scheduling plan is generated according to the user's daily available learning time model; the specific steps of the structure-aware ranking function include: calculating the difference score of the user's mastery of the node, which is the difference between the score of the corresponding node in the current ability state vector and the target mastery threshold; calculating the structural dependence score, counting the in-degree value and out-degree value of the node in the ability graph, and then weighting and summing the in-degree value and out-degree value by giving the in-degree value a higher weight through a preset adjustment factor; calculating the complexity score of the node semantic vector, calculating the information entropy value based on the dispersion degree of the vector elements; the difference score, the structural dependence score and the complexity score are weighted and fused according to the preset proportion to generate the priority score of each node; according to the task scheduling plan, tasks are issued, and four types of behavior data are collected, including task stay time, content integrity flag, evaluation score and interruption times; an abnormal detection regular term is used to calculate the behavior effectiveness score, and abnormal tasks with a behavior effectiveness score lower than a threshold are identified and added with a behavior deviation label; based on the behavior deviation label of the abnormal task, a correction strategy is matched, a path disturbance cost function is introduced to control the correction size, the path and the scheduling plan are adjusted under the constraint of the total disturbance budget, and the corrected learning path and scheduling table are output; the path disturbance cost function calculates the disturbance cost by comprehensively considering the position offset before and after task correction, the length change ratio and the depth of the node in the graph, and the structure penalty coefficient controls the weight of each disturbance term; the behavior deviation label is obtained based on the following steps: When the evaluation score is lower than the preset evaluation score lower limit, the stay duration is not lower than the preset stay duration lower limit, and the content is complete, the behavior deviation label is the ability deficiency type, and the correction strategy is the reinforcement training; When the stay duration is lower than the preset stay duration lower limit, the number of interruptions exceeds the preset number of interruption upper limit, and the evaluation score is not lower than the preset evaluation score lower limit, the behavior deviation label is the attention dispersion type, and the correction strategy is the rhythm adjustment; If the content is incomplete and the stay duration is lower than the preset stay duration lower limit, the behavior deviation label is the incomplete execution type, and the correction strategy is the process simplification. If the evaluation score is lower than the preset evaluation score lower limit and the number of interruptions exceeds the preset number of interruption upper limit, the behavior deviation label is the comprehensive obstacle type, and the correction strategy is the combined intervention.

2. The method of claim 1, wherein, The inter-capability dependency relationship is determined by fusing a hierarchical relationship marked by a domain expert and a sequence pattern mining result of a historical user learning path, wherein the sequence pattern mining uses a sliding window algorithm to extract a high-frequency continuous learning node combination. 3.The adaptive personalized learning path generation method of claim 1, wherein, The node semantic vector is obtained by processing description text through a 300-dimensional word embedding model, and then output by a two-layer fully connected neural network with a ReLU activation function, with a vector dimension of 64.

4. The method of claim 1, wherein, The specific steps of allocating learning resources based on user preferences include: constructing a user preference vector based on the selection frequency and completion quality of different resource types in the user's historical learning records, and the vector elements correspond to the preference weights of each type of resource; extracting the resource demand label of the current ability node, and matching the resource type with the highest weight in the preference vector; if the number of matched resource types in the resource library is lower than a preset threshold, a supplementary process is started: first check the content keyword matching degree of the picture-text type resource and the current node, and select this type when the matching degree exceeds the threshold; if the picture-text type resource does not meet the requirement, check the knowledge point coverage integrity of the PPT type resource and the scene adaptation degree of the mixed case type resource in sequence until a resource type that meets the requirement is selected; record the resource allocation result for updating the weight of the user preference vector.

5. The method of claim 1, wherein, The abnormality detection regular term is based on the group historical average evaluation score of the current ability node, and adjusts the deviation degree of the user score from the group average value through a penalty coefficient.

6. An adaptive personalized learning path generation system, characterized by, The system comprises: The state construction module is configured to extract a target ability list from an existing training model or a post competency requirement, construct an ability graph containing a set of ability nodes and a set of edges according to a prior and dependent relationship between abilities, collect description texts corresponding to each ability node, input the description texts into a fully connected neural network after word segmentation and word embedding processing, and generate a node semantic vector; based on a user historical course test result and a practical operation task completion condition, a weighted scoring mechanism is used to generate a user ability state vector, and missing data is filled with an average value of all users; the specific method steps of the weighted scoring mechanism include: dividing a user ability evaluation dimension into a theoretical learning performance and a practical operation task completion condition, presetting a weight of each dimension according to an ability type; extracting test score and answer accuracy data of the user in theoretical learning, and completion quality score and step compliance data of the user in practical operation tasks, and respectively mapping the two types of data to obtain a theoretical dimension normalized score and a practical operation dimension normalized score; multiplying the theoretical dimension normalized score by the corresponding weight, adding the result after multiplying the practical operation dimension normalized score by the corresponding weight, and obtaining a comprehensive score of the ability node; The path generation module is configured to score the ability nodes in priority based on a structure perception ordering function, the structure perception ordering function comprehensively reflecting a difference in a user's mastery of the nodes, a structural dependence reflected by in-degree and out-degree of the nodes in the graph, and a complexity of the node semantic vector; filtering the ability nodes to form a learning path according to the priority score, checking the topological validity, allocating learning resources in combination with user preferences, and generating a task scheduling plan according to a user daily available learning time model; the specific steps of the structure perception ordering function include: calculating a difference in the user's mastery of the nodes, the difference in the user's mastery of the nodes being a difference between a score of a corresponding node in the current ability state vector and a target mastery threshold; calculating a structural dependence score, counting in-degree values and out-degree values of the nodes in the ability graph, and weighting and summing the in-degree values and the out-degree values by giving the in-degree values a higher weight through a preset adjustment factor; calculating a complexity score of the node semantic vector, calculating an information entropy value based on a discrete degree of a vector element; weighting and fusing the difference in the user's mastery of the nodes, the structural dependence score and the complexity score according to a preset proportion to generate a priority score of each node; The behavior evaluation module is configured to issue a task according to the task scheduling plan, collect four types of behavior data including a task stay time, a content integrity flag, an evaluation score and a number of interruptions, and calculate a behavior effectiveness score by using a behavior effectiveness function containing an abnormality detection regular term to identify an abnormal task with a behavior deviation label added when the behavior effectiveness score is lower than a threshold value; The path deployment module is configured to match a correction strategy based on the behavior deviation label of the abnormal task, introduce a path disturbance cost function to control a correction scale, adjust the path and the scheduling plan under a total disturbance budget constraint, and output a corrected learning path and a scheduling table; the path disturbance cost function comprehensively calculates a disturbance cost according to a position offset before and after task correction, a time length change proportion and a depth of the node in the graph, and a structure penalty coefficient controls weights of each disturbance term; the behavior deviation label is obtained based on the following steps: When the evaluation score is lower than the preset evaluation score lower limit, the stay duration is not lower than the preset stay duration lower limit, and the content is complete, the behavior deviation label is the ability deficiency type, and the correction strategy is the reinforcement training; When the stay duration is lower than the preset stay duration lower limit, the number of interruptions exceeds the preset number of interruption upper limit, and the evaluation score is not lower than the preset evaluation score lower limit, the behavior deviation label is the attention distraction type, and the correction strategy is the rhythm adjustment; If the content is incomplete and the stay duration is lower than the preset stay duration lower limit, the behavior deviation label is the incomplete execution type, and the correction strategy is the process simplification; If the evaluation score is lower than the preset evaluation score lower limit and the number of interruptions exceeds the preset number of interruption upper limit, the behavior deviation label is the comprehensive obstacle type, and the correction strategy is the combined intervention.

Images