Adaptive cognitive learning network simulation system based on large language model

Abstract

The invention relates to the technical field of cognitive learning, in particular to an adaptive cognitive learning network simulation system based on a large language model. The system comprises a user cognitive learning feature vector conversion module, a hierarchical cognitive reasoning and learning network construction module, a cognitive learning interaction bottleneck simulation analysis module and a cognitive learning network feedback optimization module. The method comprises the following steps: acquiring user historical cognitive learning data, user cognitive interaction record data and user cognitive learning feedback information data, and performing cognitive feature vector fusion conversion to obtain a user cognitive learning feature vector; and performing hierarchical cognitive inference analysis and adaptive cognitive learning network construction based on the large language model to generate a user adaptive cognitive learning network initial architecture, and performing cognitive learning interaction bottleneck simulation analysis and cognitive learning network feedback optimization adjustment to generate a user adaptive cognitive learning optimization adjustment network. According to the invention, the learning efficiency and effect of the user cognitive learning process can be improved.

Description

Technical Field

[0001] The present invention relates to the field of cognitive learning technology, and in particular to an adaptive cognitive learning network simulation system based on a large language model. Background Art

[0002] With the rapid development of artificial intelligence technology, especially the application of large language models, the research on brain cognitive learning has gradually attracted attention. The simulation of brain cognitive learning network aims to simulate the cognitive learning process of the human brain to improve learning efficiency and knowledge acquisition ability. By taking the large language model as the core component and using its powerful natural language understanding and generation capabilities, the model provides rich semantic and knowledge support for the learning network. By analyzing the user input, the model can understand the learner's needs and knowledge background. It also introduces adaptive learning strategies to dynamically adjust the learning path and content according to the learner's cognitive characteristics, learning progress and feedback information. It also automatically optimizes the learning materials and difficulty by analyzing user performance in real time to ensure the personalization of the learning process. At the same time, by establishing a real-time feedback mechanism to provide instant feedback based on the learner's answers and performance, it helps them identify knowledge blind spots and improve their understanding ability. This mechanism can promote active learning and self-reflection. However, traditional cognitive learning network simulation systems are difficult to make personalized adjustments based on the cognitive learning characteristics and needs of individual learners, and cannot effectively handle diverse knowledge structures and learning tasks, resulting in a lack of flexibility and effectiveness in the cognitive learning process. Summary of the invention

[0003] Based on this, it is necessary for the present invention to provide an adaptive cognitive learning network simulation system based on a large language model to solve at least one of the above technical problems.

[0004] To achieve the above purpose, an adaptive cognitive learning network simulation system based on a large language model includes the following modules:

[0005] A user cognitive learning feature vector conversion module is used to obtain user historical cognitive learning data, user cognitive interaction record data, and user cognitive learning feedback information data, and perform cognitive feature vector fusion conversion based on the user historical cognitive learning data, user cognitive interaction record data, and user cognitive learning feedback information data to obtain a user cognitive learning feature vector;

[0006] The hierarchical cognitive reasoning and learning network construction module is used to perform hierarchical cognitive reasoning analysis on the user cognitive learning feature vector based on the large language model to obtain the user cognitive knowledge low-level learning tasks and the user cognitive knowledge high-level learning reasoning relationship; according to the user cognitive knowledge low-level learning tasks and the user cognitive knowledge high-level learning reasoning relationship, an adaptive cognitive learning network is constructed to generate the initial architecture of the user adaptive cognitive learning network;

[0007] The cognitive learning interaction bottleneck simulation analysis module is used to obtain the user's real-time cognitive learning behavior data, and based on the user's real-time cognitive learning behavior data, perform cognitive learning interaction bottleneck simulation analysis on the user's adaptive cognitive learning network initial architecture to obtain the bottleneck point of the user's cognitive learning interaction simulation process;

[0008] The cognitive learning network feedback optimization module is used to perform cognitive learning network feedback optimization adjustment on the initial architecture of the user adaptive cognitive learning network based on the bottleneck point of the user cognitive learning interaction simulation process to generate a user adaptive cognitive learning optimization adjustment network.

[0009] Furthermore, the user cognitive learning feature vector analysis module includes the following functions:

[0010] Obtaining historical cognitive learning data records from a user cognitive learning platform database to obtain user historical cognitive learning data;

[0011] Conduct cognitive learning trajectory analysis on the user's historical cognitive learning data to obtain the user's historical cognitive learning activity trajectory;

[0012] Based on the user's historical cognitive learning activity trajectory, cognitive interaction analysis is performed on the user's historical cognitive learning data and recorded to obtain user cognitive interaction record data;

[0013] A user cognitive interaction quality evaluation system is constructed based on user cognitive interaction record data, and cognitive learning feedback analysis is performed on user historical cognitive learning data based on the user cognitive interaction quality evaluation system to obtain user cognitive learning feedback information data;

[0014] The cognitive feature vector is fused and transformed according to the user's historical cognitive learning data, the user's cognitive interaction record data and the user's cognitive learning feedback information data to obtain the user's cognitive learning feature vector.

[0015] Furthermore, the cognitive learning trajectory analysis of the user's historical cognitive learning data includes:

[0016] Perform cognitive learning time and content extraction processing on the user's historical cognitive learning data to obtain the user's historical cognitive learning timestamp and the user's historical cognitive learning content;

[0017] Conduct cognitive learning theme mining and analysis on the user's historical cognitive learning content to obtain the theme of each user's historical cognitive learning activity;

[0018] Conduct correlation analysis on the themes of each user's historical cognitive learning activities to obtain the correlation relationship between the learning activity behaviors of each user's cognitive learning activity themes;

[0019] Based on the learning activity behavior association relationship between each user's cognitive learning activity theme, the cognitive learning behavior association map of the user's historical cognitive learning content is analyzed to generate a user's historical cognitive learning behavior association map;

[0020] Based on the user's historical cognitive learning timestamp, the cognitive learning trajectory analysis is performed on the user's historical cognitive learning behavior association map to obtain the user's historical cognitive learning activity trajectory.

[0021] Furthermore, the cognitive feature vector fusion conversion based on the user's historical cognitive learning data, the user's cognitive interaction record data, and the user's cognitive learning feedback information data includes:

[0022] Perform cognitive learning content feature analysis on the user's historical cognitive learning data to obtain the user's cognitive learning content features, wherein the user's cognitive learning content features include the user's cognitive learning content theme, the user's cognitive learning difficulty level, and the user's cognitive learning time;

[0023] Performing user interaction frequency feature analysis on user cognitive interaction record data to obtain user cognitive learning interaction frequency features, wherein the user cognitive learning interaction frequency features include user cognitive learning click-through rate, user cognitive learning viewing time, and user cognitive learning discussion participation;

[0024] Perform cognitive learning feedback feature analysis on user cognitive learning feedback information data to obtain user cognitive learning feedback sentiment features, wherein the user cognitive learning feedback sentiment features include a user cognitive learning positive feedback sentiment index, a user cognitive learning neutral feedback sentiment index, and a user cognitive learning negative feedback sentiment index;

[0025] The user cognitive learning content features, user cognitive learning interaction frequency features and user cognitive learning feedback sentiment features are transformed into cognitive feature vectors to obtain the user cognitive learning feature vector.

[0026] Furthermore, the hierarchical cognitive reasoning and learning network building module includes the following functions:

[0027] Based on the low-level cognitive modules in the large language model, low-level task recognition and analysis are performed on the user's cognitive learning feature vector to obtain the user's cognitive knowledge low-level learning tasks;

[0028] Perform task content classification processing on the user cognitive knowledge low-level learning tasks to obtain the user cognitive low-level learning task content classification matrix;

[0029] A high-level cognitive reasoning framework that can handle low-level learning tasks is constructed through high-level cognitive modules in the large language model. Based on the high-level cognitive reasoning framework, a high-level cognitive reasoning path analysis is performed on the corresponding low-level learning tasks in the user's cognitive low-level learning task content classification matrix to obtain the high-level cognitive reasoning path between each low-level learning task.

[0030] Based on the high-level cognitive reasoning path between each low-level learning task, the corresponding low-level learning task of user cognitive knowledge is adjusted and analyzed at a high level to obtain the high-level learning reasoning relationship of user cognitive knowledge;

[0031] An adaptive cognitive learning network is constructed according to low-level learning tasks of user cognitive knowledge and high-level learning reasoning relationships of user cognitive knowledge to generate an initial architecture of the user adaptive cognitive learning network.

[0032] Further, the high-level learning reasoning adjustment analysis of the corresponding user cognitive knowledge low-level learning tasks based on the high-level cognitive reasoning path between the low-level learning tasks includes:

[0033] Based on the high-level cognitive reasoning paths between the low-level learning tasks, the corresponding user cognitive knowledge low-level learning tasks are connected with the cognitive reasoning paths to generate a high-level cognitive reasoning path relationship map between the low-level learning tasks;

[0034] The learning task reasoning relationship strength calculation formula is used to evaluate and calculate the reasoning relationship strength of the high-level cognitive reasoning path relationship map between each low-level learning task, and the cognitive reasoning relationship strength between each low-level learning task is obtained;

[0035] Based on the strength of cognitive reasoning relationships between low-level learning tasks, a high-level learning reasoning adjustment analysis is performed on the high-level cognitive reasoning path relationship map between low-level learning tasks to obtain the high-level learning reasoning relationship of user cognitive knowledge.

[0036] Furthermore, the calculation formula of the learning task reasoning relationship strength is specifically:

[0037]

[0038] In the formula, R ij is the cognitive reasoning relationship strength between the i-th low-level learning task and the j-th low-level learning task, i and j are both item measurement parameters of low-level learning tasks, T is the upper limit of the time range for calculating the reasoning relationship strength, t is the time variable parameter, and w ij is the weight of the inference relationship between the i-th low-level learning task and the j-th low-level learning task, s ij(t) is the number of related reasoning paths between the i-th low-level learning task and the j-th low-level learning task at time t, α is the attenuation coefficient of the number of reasoning paths, c ij (t) is the task content similarity between the i-th low-level learning task and the j-th low-level learning task at time t, β is the attenuation coefficient of the task content similarity, p ij (t) is the context correlation parameter between the ith low-level learning task and the jth low-level learning task at time t, γ is the attenuation coefficient of the context correlation effect, and η is the correction coefficient of the cognitive reasoning relationship strength.

[0039] Furthermore, the adaptive cognitive learning network construction based on the low-level learning tasks of the user cognitive knowledge and the high-level learning reasoning relationship of the user cognitive knowledge includes:

[0040] Performing base node configuration on the user cognitive knowledge low-level learning task to obtain the user cognitive low-level learning task base node;

[0041] Performing adaptive learning node transformation on the user cognitive low-level learning task base node to obtain the user cognitive low-level adaptive learning node;

[0042] Based on the high-level learning and reasoning relationship of user cognitive knowledge, an adaptive cognitive learning network is constructed for the corresponding user cognitive low-level adaptive learning nodes to generate the initial architecture of the user adaptive cognitive learning network.

[0043] Furthermore, the cognitive learning interaction bottleneck simulation analysis module includes the following functions:

[0044] Obtain users’ real-time cognitive learning behavior data;

[0045] Perform learning behavior feature extraction and processing on the user's real-time cognitive learning behavior data to obtain the user's cognitive learning behavior features; perform cognitive learning interaction pattern recognition and analysis on the user's real-time cognitive learning behavior data based on the user's cognitive learning behavior features to obtain the user's cognitive learning interaction behavior pattern;

[0046] Based on the user cognitive learning interactive behavior mode, the user cognitive learning behavior features are divided and integrated to obtain the learning feature fusion vector corresponding to each user cognitive learning interactive behavior;

[0047] By inputting the learning feature fusion vector corresponding to each user's cognitive learning interaction behavior into the initial architecture of the user adaptive cognitive learning network to perform cognitive learning interaction simulation analysis, the cognitive learning interaction simulation process corresponding to each user's cognitive learning interaction behavior is obtained;

[0048] The interactive simulation process bottleneck identification analysis is performed on the cognitive learning interactive simulation process corresponding to each user's cognitive learning interactive behavior to obtain the bottleneck point of the user's cognitive learning interactive simulation process.

[0049] Furthermore, the cognitive learning network feedback optimization module includes the following functions:

[0050] Conduct cognitive learning interaction bottleneck feature analysis on the bottleneck points of the user cognitive learning interaction simulation process to obtain bottleneck feature description data of the user cognitive learning interaction process;

[0051] Based on the bottleneck feature description data of the user cognitive learning interaction process, the initial architecture of the user adaptive cognitive learning network is analyzed for learning network optimization objectives, and a list of user cognitive learning network optimization objectives is obtained;

[0052] Conduct learning network optimization strategy analysis on the user cognitive learning network optimization target list to obtain the user cognitive learning network optimization adjustment strategy plan;

[0053] According to the user cognitive learning network optimization adjustment strategy scheme, the initial architecture of the user adaptive cognitive learning network is optimized and adjusted by cognitive learning network feedback to generate a user adaptive cognitive learning optimization adjustment network.

[0054] Beneficial effects of the present invention:

[0055] The adaptive cognitive learning network simulation system based on the large language model proposed in the present invention is composed of a user cognitive learning feature vector conversion module, a hierarchical cognitive reasoning and learning network construction module, a cognitive learning interaction bottleneck simulation analysis module, and a cognitive learning network feedback optimization module. Compared with the prior art, the beneficial effect of the present application is that by obtaining historical cognitive learning data records from the user cognitive learning platform database, it can provide basic data support for subsequent analysis, because historical data is the key to understanding user cognitive learning behavior, preferences and effects. By collecting users' cognitive learning records in different time periods, including cognitive learning content, duration, frequency and grades, the user's cognitive learning process and state can be fully described, which not only helps the design of personalized learning paths, but also provides quantitative data basis for subsequent cognitive interaction processes. By performing cognitive interaction analysis and recording of user historical cognitive learning data, the user's interaction mode in the cognitive learning process can be deeply excavated. This analysis not only focuses on the relationship between users and learning content, but also emphasizes the interaction between users and platforms and other learners. By recording the user's interactive behavior in the learning process, detailed cognitive interaction record data can be constructed. These data provide an important perspective for understanding users' learning habits and social behaviors. For example, users repeatedly review, ask questions, or participate in discussions on specific content, indicating their interest or confusion in the content. Through cognitive interaction record data, a corresponding quality evaluation system is constructed to conduct feedback analysis on the user's historical cognitive learning data. The establishment of this system aims to evaluate the quality of user interaction during the learning process, including the frequency, depth, and diversity of interaction. Through these indicators, we can better understand how users interact with learning content and communities during the learning process, and evaluate the impact of these interactions on their learning outcomes, thereby providing basic data support for subsequent processing. At the same time, by fusion transformation of cognitive feature vectors based on user historical cognitive learning data, user cognitive interaction record data, and user cognitive learning feedback information data, a comprehensive understanding of user cognitive characteristics can be achieved. This fusion process integrates the user's learning history, interaction behavior, and feedback information to form a multi-dimensional cognitive feature vector. These feature vectors can not only reflect the user's knowledge level, learning preference, and interaction quality, but also provide strong data support for the subsequent construction of adaptive cognitive learning network simulation. Secondly, by using the low-level cognitive module within the large language model to identify and analyze the user's cognitive learning feature vector, the user's low-level learning tasks can be identified. Through this analysis, the large language model can deeply understand the user's knowledge structure and learning habits, thereby identifying the specific learning tasks that exist in the cognitive learning process.By conducting high-level learning reasoning analysis on low-level learning tasks of user cognitive knowledge, the reasoning paths between learning tasks can be analyzed to identify which learning tasks have a greater impact on the user's overall learning process. This analysis enables users to better understand the dependencies between various learning tasks and provide a basis for the corresponding cognitive learning strategy adjustments. More importantly, users can optimize their own cognitive learning plans based on high-level reasoning relationships to ensure that tasks that are critical to subsequent learning are completed first during the learning process. An adaptive cognitive learning network is constructed based on low-level learning tasks of user cognitive knowledge and high-level learning reasoning relationships of user cognitive knowledge to construct a corresponding adaptive cognitive learning network. This network is generated through personalized adjustment and optimization based on the specific cognitive learning characteristics and needs of users. It can dynamically adapt to the learning progress and changes of users. Through the establishment of an adaptive learning network, learning content and strategies can be adjusted in real time, thereby providing a personalized learning experience. Through this network, users can obtain learning resources and suggestions suitable for themselves at different learning stages, and can also effectively handle diverse knowledge structures and learning tasks, further improving the flexibility and effectiveness of the cognitive learning process. This networked learning structure also encourages users to conduct more in-depth exploration and practice in the cognitive learning process, so that they can be comprehensively improved in cognitive development, thereby providing sustainable data support for users' long-term cognitive development. Then, by obtaining real-time cognitive learning behavior data of users, this step collects users' learning behaviors in real time through various channels (such as learning platforms, mobile applications, sensors, etc.), including learning time, participation, question answering, and interaction frequency. These data provide real and dynamic information for subsequent analysis, which helps to understand users' learning habits and preferences. In addition, by simulating and analyzing the cognitive learning interaction bottleneck of the initial architecture of the user adaptive cognitive learning network based on the real-time cognitive learning behavior data of users, the system can be helped to discover bottlenecks and challenges in the user's learning process. The results of this step can not only reveal which links lead to poor cognitive learning effects, but also provide data support for educators to formulate effective intervention strategies. By identifying and resolving these bottlenecks, the user's cognitive learning process can be more effectively supported, ensuring timely help when encountering difficulties. The ultimate goal of this analysis is to optimize the learning experience, improve learning efficiency, and enable users to achieve better results in the cognitive learning process. Finally, by optimizing the initial architecture of the user's adaptive cognitive learning network based on the bottleneck points of the user's cognitive learning interaction simulation process, the system can have a deeper understanding of the bottleneck obstacles in the user's cognitive learning process, and formulate specific optimization strategies to effectively solve the previously identified learning bottlenecks.By dynamically optimizing the learning network architecture and making feedback adjustments based on previous analysis results, a more effective cognitive learning environment is formed. By implementing these optimizations, the cognitive learning network can respond to users' learning needs in real time and provide personalized learning resources and support. This feedback optimization not only enhances the user's learning experience, but also promotes the improvement of learning results, making the learning process more efficient and effective, thus providing users with a smoother and more effective learning experience, thereby achieving a higher level of cognitive learning achievement. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Other features, objects and advantages of the present invention will become more apparent from the detailed description of non-limiting embodiments thereof made with reference to the following drawings:

[0057] Figure 1 It is a module schematic diagram of the adaptive cognitive learning network simulation system based on a large language model of the present invention;

[0058] Figure 2 for Figure 1 Schematic diagram of the functional flow of the feature vector conversion module for user cognitive learning;

[0059] Figure 3 for Figure 1 Schematic diagram of the functional flow of the building blocks of the mid-level cognitive reasoning and learning network. DETAILED DESCRIPTION

[0060] The technical system of the present invention is described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.

[0061] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor systems and / or microcontroller systems.

[0062] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0063] To achieve this, please refer to Figures 1 to 3 The present invention provides an adaptive cognitive learning network simulation system based on a large language model, the system comprising the following modules:

[0064] A user cognitive learning feature vector conversion module is used to obtain user historical cognitive learning data, user cognitive interaction record data, and user cognitive learning feedback information data, and perform cognitive feature vector fusion conversion based on the user historical cognitive learning data, user cognitive interaction record data, and user cognitive learning feedback information data to obtain a user cognitive learning feature vector;

[0065] The hierarchical cognitive reasoning and learning network construction module is used to perform hierarchical cognitive reasoning analysis on the user cognitive learning feature vector based on the large language model to obtain the user cognitive knowledge low-level learning tasks and the user cognitive knowledge high-level learning reasoning relationship; according to the user cognitive knowledge low-level learning tasks and the user cognitive knowledge high-level learning reasoning relationship, an adaptive cognitive learning network is constructed to generate the initial architecture of the user adaptive cognitive learning network;

[0066] The cognitive learning interaction bottleneck simulation analysis module is used to obtain the user's real-time cognitive learning behavior data, and based on the user's real-time cognitive learning behavior data, perform cognitive learning interaction bottleneck simulation analysis on the user's adaptive cognitive learning network initial architecture to obtain the bottleneck point of the user's cognitive learning interaction simulation process;

[0067] The cognitive learning network feedback optimization module is used to perform cognitive learning network feedback optimization adjustment on the initial architecture of the user adaptive cognitive learning network based on the bottleneck point of the user cognitive learning interaction simulation process to generate a user adaptive cognitive learning optimization adjustment network.

[0068] In the embodiment of the present invention, please refer to Figure 1 FIG. 1 is a schematic diagram of an adaptive cognitive learning network simulation system based on a large language model of the present invention. In this example, the adaptive cognitive learning network simulation system based on a large language model includes the following modules:

[0069] S1: User cognitive learning feature vector conversion module, used to obtain user historical cognitive learning data, user cognitive interaction record data and user cognitive learning feedback information data, and perform cognitive feature vector fusion conversion based on user historical cognitive learning data, user cognitive interaction record data and user cognitive learning feedback information data to obtain user cognitive learning feature vector;

[0070] In an embodiment of the present invention, by connecting to the database of the user cognitive learning platform, an SQL query statement is used to extract historical cognitive learning record information related to the user from the learning record table, including fields such as user ID, learning topic, learning time, learning duration and completion status, thereby obtaining the user's historical cognitive learning data. By identifying and analyzing the learning trajectory of the acquired user historical cognitive learning data, the learning activities are sorted by timestamp using the time series analysis method, and a learning trajectory diagram is created to record the learning topics and corresponding learning time of each user in a specific time period. The user learning trajectory is displayed in the form of a line graph by combining the previously obtained cognitive learning activity trajectory, and the corresponding user historical cognitive learning data is subjected to cognitive learning interaction analysis, so as to model the interaction between users and content in learning activities by using graph analysis tools, and record the interactive behaviors in each learning activity, such as answering questions, taking notes, participating in discussions, etc., so as to form cognitive interaction record data. At the same time, a corresponding cognitive interaction quality evaluation system is established based on the user cognitive interaction records obtained by the previous analysis to define corresponding evaluation indicators, such as interaction frequency, interaction depth, and feedback effectiveness, etc., and based on the previously established user cognitive interaction quality evaluation system, a weight distribution method is used to set corresponding weights for each evaluation indicator, and a comprehensive evaluation is performed on the corresponding learning content in the user historical cognitive learning data, and the interaction quality score is calculated, so as to generate the user's cognitive learning feedback evaluation data, thereby obtaining the user's cognitive learning feedback information data. Then, by combining the user's historical cognitive learning data, user cognitive interaction record data and user cognitive learning feedback information data obtained in the previous analysis, the cognitive feature vector is fused and transformed, so that various types of data can be converted into feature data by adopting feature engineering methods. For example, features including learning content theme, learning difficulty level and learning time are obtained from the user's historical cognitive learning data, features including click-through rate, viewing time and discussion participation are obtained from the user's cognitive interaction record data, and features including positive, neutral and negative feedback sentiment index are obtained from the user's cognitive learning feedback information data. The features are fused to generate the final feature vector by using feature superposition, such as [A, B, C], where A is the user cognitive learning content feature set, B is the user cognitive learning interaction frequency feature set, and C is the user cognitive learning feedback sentiment feature set, and finally the user cognitive learning feature vector is obtained.

[0071] S2: Hierarchical cognitive reasoning and learning network construction module, used to perform hierarchical cognitive reasoning analysis on user cognitive learning feature vectors based on a large language model to obtain low-level learning tasks of user cognitive knowledge and high-level learning reasoning relationships of user cognitive knowledge; construct an adaptive cognitive learning network based on the low-level learning tasks of user cognitive knowledge and high-level learning reasoning relationships of user cognitive knowledge to generate the initial architecture of the user adaptive cognitive learning network;

[0072] In an embodiment of the present invention, a low-level cognitive module in a large language model is used to perform recognition and analysis of learning tasks on the user cognitive learning feature vector obtained by the previous fusion transformation, wherein the input feature vector includes information such as learning content features, learning interaction frequency features, and learning feedback sentiment features, so that these input feature vectors are analyzed in the low-level cognitive module through a specific machine learning algorithm (such as a decision tree or a support vector machine), and corresponding learning tasks are identified therefrom, and these learning tasks include information memory, concept understanding, etc., thereby obtaining low-level learning tasks of user cognitive knowledge. By classifying the task content of the previously identified user cognitive knowledge low-level learning tasks, the task content classification matrix is ​​constructed to classify the low-level learning tasks according to themes, difficulty and relevance, and the learning tasks are grouped by using clustering algorithms (such as K-means or hierarchical clustering), which can be divided into memory tasks, understanding tasks and application tasks, etc., and by using the high-level cognitive modules in the large language model, a high-level cognitive reasoning framework that can handle low-level learning tasks is constructed, so that the basic structure of high-level cognitive reasoning, including reasoning rules, logical relations and reasoning paths, is defined, and the high-level reasoning path analysis is performed on the user's low-level learning task content by using a reasoning engine (such as Prolog or a custom rule engine). The model derives the relationship between each low-level learning task through the established reasoning rules, thereby generating a high-level cognitive reasoning path. At the same time, the corresponding user cognitive knowledge low-level learning task is adjusted by combining the high-level cognitive reasoning path obtained by the previous reasoning analysis, so that the low-relationship strength path in the high-level reasoning path is deleted or reconstructed by using the adjustment method in graph theory, so as to optimize the learning reasoning relationship between learning tasks, thereby generating an optimized user cognitive knowledge high-level learning reasoning relationship. Then, an adaptive cognitive learning network is constructed based on the low-level learning tasks and high-level learning reasoning relationships of the user cognitive knowledge obtained from the previous analysis. The construction process includes defining the nodes and edges of the network, where nodes represent learning tasks and edges represent reasoning relationships between tasks. The network is optimized by using a graph neural network (GNN) or a reinforcement learning algorithm, so that the network can be adaptively adjusted according to user feedback, and finally the initial architecture of the user adaptive cognitive learning network is generated.

[0073] S3: Cognitive learning interaction bottleneck simulation analysis module, used to obtain real-time cognitive learning behavior data of users, and perform cognitive learning interaction bottleneck simulation analysis on the initial architecture of the user adaptive cognitive learning network based on the real-time cognitive learning behavior data of users, to obtain the bottleneck point of the user cognitive learning interaction simulation process;

[0074] In an embodiment of the present invention, the user's cognitive learning activities are monitored in real time by using sensors, a learning management system (LMS) and data collection tools. These tools include a learning progress tracker, an online assessment platform and interactive learning software. Through these systems, the data collected include the user's learning time, answering speed, number of wrong answers, frequency of participating in discussions and use of learning resources, etc., so as to integrate and obtain the user's real-time cognitive learning behavior data. The real-time cognitive learning behavior data of the user previously collected in real time is cleaned and preprocessed to remove noise and outliers, and the user's learning behavior is deeply analyzed by using feature extraction algorithms, such as principal component analysis (PCA) and clustering algorithms, and key features are extracted. These features include the user's learning efficiency (such as the ratio of the time required to complete the task to the accuracy rate), learning mode (such as the ratio of active learning to passive learning) and emotional state (such as the anxiety level during learning), etc. At the same time, the corresponding user cognitive learning behavior process in the user's real-time cognitive learning behavior data is cognitively learned by applying a machine learning model (such as a random forest or a support vector machine). Recognition and analysis of interaction patterns, through training models, to identify users' learning behavior interaction patterns, such as self-regulated learning, social learning or cooperative learning patterns, and by combining the user cognitive learning interaction behavior patterns obtained in the previous identification analysis, the corresponding learning behavior features are divided into the same interaction behavior pattern, and cluster analysis is used to group behavior features with similar interaction behavior patterns, and each group of behavior features is fused to generate corresponding feature fusion vectors. These vectors will contain the common characteristics of each group of user cognitive learning behaviors, such as typical performance and interaction methods in specific learning tasks, so as to obtain the learning feature fusion vectors corresponding to each user's cognitive learning interaction behavior. At the same time, by inputting the learning feature fusion vectors corresponding to each user's cognitive learning interaction behavior generated by the previous fusion into the previously constructed user adaptive cognitive learning network initial architecture, a simulation analysis of the cognitive learning interaction process is performed to predict each user's behavioral performance and learning effect in a specific learning environment. The network training uses the back propagation algorithm, and the network weights are optimized through multiple iterations to simulate an accurate cognitive learning interaction process. The simulation results show the user's expected path and existing learning outcomes in the learning process. Then, after completing the cognitive learning interaction simulation process corresponding to each user's cognitive learning interaction behavior, a bottleneck identification analysis of the interaction simulation process is performed to identify the bottleneck points in the cognitive learning interaction simulation process by applying data mining techniques, such as outlier detection and time series analysis. These bottleneck points include obstacles to understanding specific knowledge points, frequent erroneous responses, or obvious lags in learning progress, and finally the corresponding bottleneck points of the user cognitive learning interaction simulation process are obtained.

[0075] S4: Cognitive learning network feedback optimization module, used to perform cognitive learning network feedback optimization adjustment on the initial architecture of the user adaptive cognitive learning network based on the bottleneck point of the user cognitive learning interaction simulation process, so as to generate a user adaptive cognitive learning optimization adjustment network.

[0076] In an embodiment of the present invention, by statistically analyzing the characteristics of cognitive learning bottlenecks of the user cognitive learning interaction simulation process obtained by previous analysis (such as obstacles to understanding specific knowledge points, frequent erroneous responses, or obvious lags in learning progress), the bottleneck performance of users in specific cognitive learning situations is statistically analyzed to reveal the characteristic description factors that affect cognitive learning efficiency, such as user distraction, weak grasp of knowledge points, or low interactive participation, and to help the system to more deeply understand the obstacles that users encounter in the cognitive learning process, and by combining the description of the bottleneck characteristics of the user cognitive learning interaction process obtained by previous analysis, the initial architecture of the user adaptive cognitive learning network used in the cognitive learning interaction process is identified and analyzed for the optimization goals of the learning network, so that the bottleneck feature data is mapped to the structure of the learning network by applying the theory of system thinking, and the optimization goals are clarified, including reducing the user's cognitive load, improving the depth of knowledge mastery, and improving the initiative of learning, wherein each optimization goal corresponds to a specific quantitative indicator, such as a 20% reduction in error rate, a 15% increase in learning initiative, and by constructing an optimization goal list, at the same time, by optimizing the learning network after determining the optimization goal list of the user cognitive learning network Strategy analysis is to evaluate the performance of the current learning network by adopting model prediction technology and compare it with the optimization goal. Then, a variety of optimization strategies are designed using reinforcement learning algorithms. For example, for users with excessive cognitive load, an optimization strategy based on hierarchical teaching method is implemented to relieve learning pressure by providing step-by-step guidance and timely feedback. Simulation tools are used to evaluate strategies to ensure the effectiveness and operability of the strategies, thereby forming a comprehensive optimization adjustment strategy plan. Then, by combining the learning network optimization adjustment strategy plan obtained from the previous analysis, the corresponding user adaptive cognitive learning network initial architecture is optimized and adjusted to analyze the new data of the user in the learning process in real time by using machine learning algorithms, and respond to the user's learning needs by adjusting network parameters and learning paths. For example, if a user shows extremely high frustration in a specific link, the learning content will be dynamically adjusted, the corresponding basic knowledge review materials will be provided, and interactive exercises will be added to enhance cognitive learning confidence, thereby optimizing and adjusting the corresponding adaptive cognitive learning optimization network, which will effectively improve the user's learning efficiency and effect, achieve the goal of personalized cognitive learning, and finally optimize the generation of the user's adaptive cognitive learning optimization adjustment network.

[0077] Furthermore, the user cognitive learning feature vector analysis module includes the following functions:

[0078] Obtaining historical cognitive learning data records from a user cognitive learning platform database to obtain user historical cognitive learning data;

[0079] Conduct cognitive learning trajectory analysis on the user's historical cognitive learning data to obtain the user's historical cognitive learning activity trajectory;

[0080] Based on the user's historical cognitive learning activity trajectory, cognitive interaction analysis is performed on the user's historical cognitive learning data and recorded to obtain user cognitive interaction record data;

[0081] A user cognitive interaction quality evaluation system is constructed based on user cognitive interaction record data, and cognitive learning feedback analysis is performed on user historical cognitive learning data based on the user cognitive interaction quality evaluation system to obtain user cognitive learning feedback information data;

[0082] The cognitive feature vector is fused and transformed according to the user's historical cognitive learning data, the user's cognitive interaction record data and the user's cognitive learning feedback information data to obtain the user's cognitive learning feature vector.

[0083] As an embodiment of the present invention, refer to Figure 2 As shown, Figure 1 Schematic diagram of the functional flow of the user cognitive learning feature vector analysis module in the embodiment. In this embodiment, the functions of the user cognitive learning feature vector analysis module include:

[0084] S11: Obtaining historical cognitive learning data records from a user cognitive learning platform database to obtain user historical cognitive learning data;

[0085] In an embodiment of the present invention, by connecting to the database of the user cognitive learning platform, an SQL query statement is used to extract historical cognitive learning record information related to the user from the learning record table, including fields such as user ID, learning topic, learning time, learning duration and completion status. After the data is extracted, it is formatted to ensure that the time format of each record is consistent and data redundancy is avoided, and finally the user's historical cognitive learning data is obtained.

[0086] S12: Analyze the cognitive learning trajectory of the user's historical cognitive learning data to obtain the user's historical cognitive learning activity trajectory;

[0087] In an embodiment of the present invention, the learning trajectory of the acquired user historical cognitive learning data is identified and analyzed, and the learning activities are sorted by timestamp using the time series analysis method. The Pandas library in Python is used to create a learning trajectory graph to record the learning topic and corresponding learning time of each user in a specific time period. Combined with the graphic visualization tool Matplotlib, the user's learning trajectory is displayed in the form of a line graph, making the learning behavior pattern more intuitive, and finally obtaining the user's historical cognitive learning activity trajectory.

[0088] S13: performing cognitive interaction analysis and recording the user's historical cognitive learning data based on the user's historical cognitive learning activity trajectory to obtain user cognitive interaction record data;

[0089] In an embodiment of the present invention, cognitive learning interaction analysis is performed on the corresponding user historical cognitive learning data by combining the user historical cognitive learning activity trajectory obtained by previous analysis, so as to model the interaction between users and content in learning activities by using graph analysis tools, and record the interactive behaviors in each learning activity, such as answering questions, taking notes, participating in discussions, etc., and quantify these behaviors to form cognitive interaction record data, which are stored in a database in a structured form to finally obtain user cognitive interaction record data.

[0090] S14: constructing a user cognitive interaction quality evaluation system based on the user cognitive interaction record data, and performing cognitive learning feedback analysis on the user's historical cognitive learning data based on the user cognitive interaction quality evaluation system to obtain user cognitive learning feedback information data;

[0091] In an embodiment of the present invention, a corresponding cognitive interaction quality evaluation system is established based on the user cognitive interaction record data obtained by previous analysis to define corresponding evaluation indicators, such as interaction frequency, interaction depth, and feedback effectiveness, etc., and a weight distribution method is used based on the previously established user cognitive interaction quality evaluation system to set corresponding weights for each evaluation indicator, conduct a comprehensive evaluation of the corresponding learning content in the user's historical cognitive learning data, and calculate the interaction quality score, thereby generating the user's cognitive learning feedback evaluation score, and finally obtaining the user's cognitive learning feedback information data.

[0092] S15: Perform cognitive feature vector fusion conversion based on the user's historical cognitive learning data, the user's cognitive interaction record data, and the user's cognitive learning feedback information data to obtain the user's cognitive learning feature vector.

[0093] In an embodiment of the present invention, a fusion conversion of cognitive feature vectors is performed by combining the user's historical cognitive learning data, user cognitive interaction record data, and user cognitive learning feedback information data obtained by previous analysis, so as to convert various types of data into feature data by adopting a feature engineering method. For example, features including learning content theme, learning difficulty level, and learning time are obtained from the user's historical cognitive learning data, features including click-through rate, viewing time, and discussion participation are obtained from the user's cognitive interaction record data, and features including positive, neutral, and negative feedback sentiment indexes are obtained from the user's cognitive learning feedback information data. The features are then fused to generate a final feature vector by using a feature superposition method, for example, A, B, C], where A is a user cognitive learning content feature set, B is a user cognitive learning interaction frequency feature set, and C is a user cognitive learning feedback sentiment feature set, and finally a user cognitive learning feature vector is obtained.

[0094] Furthermore, the cognitive learning trajectory analysis of the user's historical cognitive learning data includes:

[0095] Perform cognitive learning time and content extraction processing on the user's historical cognitive learning data to obtain the user's historical cognitive learning timestamp and the user's historical cognitive learning content;

[0096] In an embodiment of the present invention, user historical cognitive learning data is extracted from the user's historical cognitive learning database, and the user historical cognitive learning data is screened by using SQL queries to obtain the timestamp and content of each learning activity. The timestamp will be converted into a standard format for time series analysis. The extracted learning content will be stored in a structured data table, which contains a unique identifier, learning topic, timestamp and content description for each learning activity. Then, the extracted data is cleaned and formatted by using Python's Pandas library to ensure that the correspondence between the timestamp and the content is clear, and finally the user's historical cognitive learning timestamp and the user's historical cognitive learning content are obtained.

[0097] Preferably, the cognitive learning theme mining and analysis is performed on the user's historical cognitive learning content to obtain the theme of each user's historical cognitive learning activity;

[0098] In an embodiment of the present invention, cognitive learning topics are mined and analyzed on the extracted historical cognitive learning content of the user, by using natural language processing (NLP) technology and applying the LDA (latent Dirichlet allocation) model for topic modeling, and word segmentation tools (such as jieba) are used to segment the learning content, and stop words are removed to form a bag-of-words model. The bag-of-words model is then input into the LDA model for training to identify the user's main learning topics, thereby generating a list of learning activity topics for each user, and finally obtaining each user's historical cognitive learning activity topic.

[0099] Preferably, a correlation analysis is performed on each user's historical cognitive learning activity theme to obtain a correlation relationship between learning activity behaviors of each user's cognitive learning activity theme;

[0100] In an embodiment of the present invention, by performing correlation identification and analysis on the themes of each user's historical cognitive learning activity, a topic co-occurrence matrix is ​​constructed to record the frequency of co-occurrence between different cognitive learning activity themes, and the Apriori algorithm is applied to extract frequent item sets by setting a minimum support threshold, and further identify the correlation between the user's learning activity themes. For each topic pair, its confidence is calculated to form behavioral association rules for user learning activities. This process will reveal the mutual influence correlation between themes in the user's learning process, and ultimately obtain the learning activity behavior correlation between the themes of each user's cognitive learning activity.

[0101] Preferably, based on the learning activity behavior association relationship between each user's cognitive learning activity theme, a cognitive learning behavior association map analysis is performed on the user's historical cognitive learning content to generate a user's historical cognitive learning behavior association map;

[0102] In an embodiment of the present invention, a learning behavior association graph is constructed by connecting the corresponding learning content themes in the user's historical cognitive learning content through the learning activity behavior association relationship between each user's cognitive learning activity themes obtained by previous association analysis. The graph structure in graph theory is used to take each corresponding learning theme in the user's historical cognitive learning content as a node, and the association relationship between the nodes as an edge. The graph is constructed using network analysis tools (such as NetworkX). By analyzing indicators such as the node degree and clustering coefficient of the graph, the user's learning preferences and behavior patterns are understood, thereby generating a corresponding cognitive learning behavior association graph to intuitively display the connection between different learning themes, and finally generating a user's historical cognitive learning behavior association graph.

[0103] Preferably, a cognitive learning trajectory analysis is performed on the user's historical cognitive learning behavior association map based on the user's historical cognitive learning timestamp to obtain the user's historical cognitive learning activity trajectory.

[0104] In an embodiment of the present invention, the learning process trajectory of the learning content subject nodes of the corresponding cognitive learning behavior association path in the user's historical cognitive learning behavior association map is identified and analyzed by combining the user's historical cognitive learning timestamps extracted previously, so that the extracted timestamps are arranged in chronological order and combined with the association map to identify the user's cognitive learning activity trajectory in the corresponding time sequence, thereby identifying and analyzing the user's specific learning path, and by calculating the user's time distribution on each learning topic, constructing the user's learning trajectory map, visually displaying the user's learning activities and their changes on different topics, and then revealing learning habits and potential problems, and finally analyzing the user's historical cognitive learning activity trajectory.

[0105] Furthermore, the cognitive feature vector fusion conversion based on the user's historical cognitive learning data, the user's cognitive interaction record data, and the user's cognitive learning feedback information data includes:

[0106] Perform cognitive learning content feature analysis on the user's historical cognitive learning data to obtain the user's cognitive learning content features, wherein the user's cognitive learning content features include the user's cognitive learning content theme, the user's cognitive learning difficulty level, and the user's cognitive learning time;

[0107] In an embodiment of the present invention, by collecting the user's historical cognitive learning data, including learning records, subject content and completion time, and organizing and cleaning the user's historical cognitive learning data by using data analysis tools (such as Python's Pandas library), natural language processing technology is used to extract learning content topics, and the TF-IDF algorithm is used to calculate the weight of the topic to determine the main content that the user is concerned about. At the same time, statistical analysis methods (such as mean and standard deviation) are used to evaluate the difficulty level of the learning content, and the duration of each learning is calculated, so as to analyze and generate the user's cognitive learning content characteristics, covering the learning content topic, learning difficulty level and learning duration, and finally obtaining the user's cognitive learning content characteristics.

[0108] Preferably, the user interaction frequency characteristics are analyzed on the user cognitive interaction record data to obtain the user cognitive learning interaction frequency characteristics, wherein the user cognitive learning interaction frequency characteristics include the user cognitive learning click-through rate, the user cognitive learning viewing time and the user cognitive learning discussion participation;

[0109] In an embodiment of the present invention, by collecting users' cognitive interaction record data, including click logs, viewing time and discussion records, and using data mining technology (such as time series analysis), the user's click rate is calculated to obtain the user's participation in different learning contents. The viewing time is identified by comparing the user's actual viewing time of each learning module with the preset time, and the discussion participation is evaluated by analyzing the user's speaking frequency and reply situation in the discussion area, thereby analyzing and obtaining the corresponding learning interaction frequency characteristics, including click rate, viewing time and discussion participation, and finally obtaining the user's cognitive learning interaction frequency characteristics.

[0110] Preferably, cognitive learning feedback feature analysis is performed on the user cognitive learning feedback information data to obtain the user cognitive learning feedback emotional feature, wherein the user cognitive learning feedback emotional feature includes the user cognitive learning positive feedback emotional index, the user cognitive learning neutral feedback emotional index and the user cognitive learning negative feedback emotional index;

[0111] In an embodiment of the present invention, a statistical analysis of learning feedback features is performed on the user cognitive learning feedback information data obtained by the previous feedback analysis, with emphasis on analyzing the emotional mathematical feature parameters, so as to perform emotional scoring on the user feedback information by utilizing the emotional analysis technology (such as VADER or TextBlob), classify the feedback content into positive, neutral or negative, and calculate the emotional index of each feedback type to obtain the user's emotional tendency in the learning process. The positive feedback emotional index is based on the number of positive evaluations divided by the total number of evaluations, the neutral feedback emotional index is based on the number of comments with no obvious tendency divided by the total number of evaluations, and the negative feedback emotional index is based on the number of negative evaluations divided by the total number of evaluations, so as to generate the corresponding learning feedback emotional features through statistical analysis, and finally obtain the user cognitive learning feedback emotional features.

[0112] Preferably, cognitive feature vector fusion conversion is performed on the user cognitive learning content features, the user cognitive learning interaction frequency features and the user cognitive learning feedback emotion features to obtain the user cognitive learning feature vector.

[0113] In an embodiment of the present invention, the user cognitive learning content features, user cognitive learning interaction frequency features, and user cognitive learning feedback emotion features obtained by the previous analysis are integrated, and the different features are normalized by adopting a feature standardization method to ensure that each feature is at the same level, and each feature is integrated in a feature superposition manner to generate a final feature vector, for example [A, B, C], where A is a set of user cognitive learning content features, B is a set of user cognitive learning interaction frequency features, and C is a set of user cognitive learning feedback emotion features. This vector can be used for subsequent model training and learning effect prediction, fully reflecting the user's cognitive state and learning habits, and finally obtaining a user cognitive learning feature vector.

[0114] Furthermore, the hierarchical cognitive reasoning and learning network building module includes the following functions:

[0115] Based on the low-level cognitive modules in the large language model, low-level task recognition and analysis are performed on the user's cognitive learning feature vector to obtain the user's cognitive knowledge low-level learning tasks;

[0116] Perform task content classification processing on the user cognitive knowledge low-level learning tasks to obtain the user cognitive low-level learning task content classification matrix;

[0117] A high-level cognitive reasoning framework that can handle low-level learning tasks is constructed through high-level cognitive modules in the large language model. Based on the high-level cognitive reasoning framework, a high-level cognitive reasoning path analysis is performed on the corresponding low-level learning tasks in the user's cognitive low-level learning task content classification matrix to obtain the high-level cognitive reasoning path between each low-level learning task.

[0118] Based on the high-level cognitive reasoning path between each low-level learning task, the corresponding low-level learning task of user cognitive knowledge is adjusted and analyzed at a high level to obtain the high-level learning reasoning relationship of user cognitive knowledge;

[0119] An adaptive cognitive learning network is constructed according to low-level learning tasks of user cognitive knowledge and high-level learning reasoning relationships of user cognitive knowledge to generate an initial architecture of the user adaptive cognitive learning network.

[0120] As an embodiment of the present invention, refer to Figure 3 As shown, Figure 1 A schematic diagram of the functional flow of the middle-level cognitive reasoning and learning network building module. In this embodiment, the functions of the middle-level cognitive reasoning and learning network building module include:

[0121] S21: Perform low-level task recognition analysis on the user's cognitive learning feature vector based on the low-level cognitive module in the large language model to obtain the user's cognitive knowledge low-level learning task;

[0122] In an embodiment of the present invention, a low-level cognitive module in a large language model is used to perform recognition and analysis of learning tasks on the user cognitive learning feature vector obtained by the previous fusion transformation, wherein the input feature vector includes information such as learning content features, learning interaction frequency features, and learning feedback sentiment features, so that these input feature vectors are analyzed in the low-level cognitive module through a specific machine learning algorithm (such as a decision tree or a support vector machine), and corresponding learning tasks are identified therefrom, and these learning tasks include information memory, concept understanding, etc., thereby forming a low-level learning task set of user cognitive knowledge, and finally obtaining the low-level learning tasks of user cognitive knowledge.

[0123] S22: performing task content classification processing on the user cognitive knowledge low-level learning task to obtain a user cognitive low-level learning task content classification matrix;

[0124] In an embodiment of the present invention, the task contents of previously identified user cognitive knowledge low-level learning tasks are classified, and a task content classification matrix is ​​constructed to classify the low-level learning tasks according to subject, difficulty and relevance. The learning tasks are grouped by using a clustering algorithm (such as K-means or hierarchical clustering) to generate identifications of different categories. For example, learning tasks can be divided into memory tasks, comprehension tasks, and application tasks, etc., thereby forming a clear classification matrix, identifying the characteristics of each task, and finally obtaining a user cognitive low-level learning task content classification matrix.

[0125] S23: Construct a high-level cognitive reasoning framework that can handle low-level learning tasks through high-level cognitive modules in the large language model, and perform high-level cognitive reasoning path analysis on the corresponding low-level learning tasks in the user's cognitive low-level learning task content classification matrix based on the high-level cognitive reasoning framework to obtain the high-level cognitive reasoning paths between the low-level learning tasks;

[0126] In an embodiment of the present invention, a high-level cognitive reasoning framework capable of handling low-level learning tasks is constructed by using high-level cognitive modules within a large language model, so as to define the basic structure of high-level cognitive reasoning, including reasoning rules, logical relationships and reasoning paths, and perform high-level reasoning path analysis on the user's low-level learning task content classification matrix by using a reasoning engine (such as Prolog or a custom rule engine). The model derives the relationship between each low-level learning task through the established reasoning rules, thereby generating a high-level cognitive reasoning path, so as to better understand the user's learning process and needs, and finally obtain the high-level cognitive reasoning path between each low-level learning task.

[0127] S24: performing high-level learning reasoning adjustment analysis on the corresponding user cognitive knowledge low-level learning tasks based on the high-level cognitive reasoning paths between the low-level learning tasks, so as to obtain the high-level learning reasoning relationship of the user cognitive knowledge;

[0128] In an embodiment of the present invention, the learning reasoning relationship of the corresponding low-level learning tasks of user cognitive knowledge is adjusted by combining the high-level cognitive reasoning paths between the low-level learning tasks obtained by previous reasoning analysis, and the low relationship strength paths in the high-level reasoning paths are deleted or reconstructed by using the adjustment method in graph theory to optimize the learning reasoning relationship between learning tasks. The reasoning paths can also be analyzed by using matrix operations to ensure that the logical relationship between each task is clear and efficient, thereby generating an optimized high-level learning reasoning relationship, and finally obtaining a high-level learning reasoning relationship of user cognitive knowledge.

[0129] S25: Construct an adaptive cognitive learning network based on the low-level learning tasks of the user cognitive knowledge and the high-level learning reasoning relationship of the user cognitive knowledge to generate an initial architecture of the user adaptive cognitive learning network.

[0130] In an embodiment of the present invention, an adaptive cognitive learning network is constructed by constructing low-level learning tasks and high-level learning reasoning relationships of user cognitive knowledge obtained based on previous analysis. The construction process includes defining nodes and edges of the network, where nodes represent learning tasks and edges represent reasoning relationships between tasks. The network is optimized by using a graph neural network (GNN) or a reinforcement learning algorithm so that the network can be adaptively adjusted based on user feedback, thereby connecting to form the initial architecture of the user adaptive cognitive learning network. The architecture can dynamically respond to user learning conditions, improve learning efficiency, and ultimately generate the initial architecture of the user adaptive cognitive learning network.

[0131] Further, the high-level learning reasoning adjustment analysis of the corresponding user cognitive knowledge low-level learning tasks based on the high-level cognitive reasoning path between the low-level learning tasks includes:

[0132] Based on the high-level cognitive reasoning paths between the low-level learning tasks, the corresponding user cognitive knowledge low-level learning tasks are connected with the cognitive reasoning paths to generate a high-level cognitive reasoning path relationship map between the low-level learning tasks;

[0133] In an embodiment of the present invention, the corresponding user cognitive knowledge low-level learning tasks are connected by combining the high-level cognitive reasoning paths between the low-level learning tasks obtained by previous analysis, and the node and edge construction method in graph theory is used to regard each low-level learning task as a node, and the reasoning path relationship between tasks as an edge. Each node is traversed by an algorithm (such as depth-first search or breadth-first search) to identify the reasoning relationship between them, and a high-level cognitive reasoning path relationship map is generated using a data visualization tool (such as Gephi), and the identified relationship is graphed to ensure that the reasoning relationship of all low-level learning tasks is presented in a structured form, and finally a high-level cognitive reasoning path relationship map between each low-level learning task is generated.

[0134] Preferably, the learning task reasoning relationship strength calculation formula is used to evaluate and calculate the reasoning relationship strength of the high-level cognitive reasoning path relationship map between each low-level learning task, so as to obtain the cognitive reasoning relationship strength between each low-level learning task;

[0135] In an embodiment of the present invention, a suitable learning task reasoning relationship strength calculation formula is formed by combining the time variable parameter, the reasoning relationship influence weight between each low-level learning task, the relevant number of reasoning paths, the influence attenuation coefficient of the number of reasoning paths, the task content similarity, the task content similarity influence attenuation coefficient, the context relevance parameter, the context relevance influence attenuation coefficient and related parameters to evaluate the reasoning relationship strength of the high-level cognitive reasoning path relationship map between each low-level learning task, so as to quantify the reasoning relationship strength between each task and finally obtain the cognitive reasoning relationship strength between each low-level learning task.

[0136] Preferably, based on the cognitive reasoning relationship strength between each low-level learning task, a high-level learning reasoning adjustment analysis is performed on the high-level cognitive reasoning path relationship map between each low-level learning task to obtain the high-level learning reasoning relationship of the user's cognitive knowledge.

[0137] In an embodiment of the present invention, the corresponding high-level cognitive reasoning path relationship map is analyzed and adjusted by combining the cognitive reasoning relationship strength between each low-level learning task obtained by previous quantitative calculation, and the strength-based weighted graph adjustment algorithm is applied to delete or adjust the edges with low cognitive reasoning relationship strength, and optimize the connection relationship between each low-level learning task to ensure that the map reflects a more accurate learning path. This process uses the clustering algorithm in machine learning to cluster tasks of similar strength in order to conduct an overall evaluation of the high-level reasoning relationship, thereby generating a new high-level learning reasoning relationship situation, thereby effectively displaying the best learning reasoning path taken by the user in the learning process, and finally obtaining the user's cognitive knowledge high-level learning reasoning relationship.

[0138] Furthermore, the calculation formula of the learning task reasoning relationship strength is specifically:

[0139]

[0140] In the formula, R ij is the cognitive reasoning relationship strength between the i-th low-level learning task and the j-th low-level learning task, i and j are both item measurement parameters of low-level learning tasks, T is the upper limit of the time range for calculating the reasoning relationship strength, t is the time variable parameter, and w ij is the weight of the inference relationship between the i-th low-level learning task and the j-th low-level learning task, s ij (t) is the number of related reasoning paths between the i-th low-level learning task and the j-th low-level learning task at time t, α is the attenuation coefficient of the number of reasoning paths, c ij (t) is the task content similarity between the i-th low-level learning task and the j-th low-level learning task at time t, β is the attenuation coefficient of the task content similarity, pij (t) is the context correlation parameter between the ith low-level learning task and the jth low-level learning task at time t, γ is the attenuation coefficient of the context correlation effect, and η is the correction coefficient of the cognitive reasoning relationship strength.

[0141] The present invention obtains a learning task reasoning relationship strength calculation formula by using a specific mathematical model and after verification, which is used to evaluate and calculate the reasoning relationship strength of the high-level cognitive reasoning path relationship map between each low-level learning task. The learning task reasoning relationship strength calculation formula combines the number of reasoning paths, task content similarity and context relevance, and provides a comprehensive framework to evaluate the cognitive reasoning relationship between low-level learning tasks, this multi-dimensional evaluation method. Through integral calculation (from 0 to T), the formula can take into account the changes in time and reflect the reasoning relationship strength of the learning task in different time periods. This dynamic characteristic enables the analysis to adapt to changes in the learning process. In addition, by introducing the reasoning relationship influence weight, the calculation formula can adjust the relative importance between different tasks according to actual conditions, thereby improving the flexibility and accuracy of the evaluation. By using the influence attenuation coefficient, the change of task relevance over time can be effectively controlled, so that the influence of the corresponding reasoning relationship on the current analysis is reduced, thereby ensuring that the results are more timely and relevant. By combining task content similarity with context relevance, the formula can more comprehensively evaluate the relationship between tasks, which is particularly important for complex learning tasks because they are often affected by multiple factors. In addition, by introducing a correction coefficient, the calculation results can be adjusted according to specific needs to adapt to specific application scenarios or compensate for potential deviations of the model, thereby improving the credibility of the results. In summary, this formula fully considers the strength of the cognitive reasoning relationship R between the i-th low-level learning task and the j-th low-level learning task. ij , the item measurement parameters i and j of the low-level learning task, the upper limit of the time range T for calculating the strength of the reasoning relationship, the time variable parameter t, the influence weight w of the reasoning relationship between the i-th low-level learning task and the j-th low-level learning task ij , the number of related reasoning paths s between the i-th low-level learning task and the j-th low-level learning task at time t ij (t), the attenuation coefficient α of the number of reasoning paths, the task content similarity c between the i-th low-level learning task and the j-th low-level learning task at time t ij (t), the attenuation coefficient of task content similarity β, the context correlation parameter p between the i-th low-level learning task and the j-th low-level learning task at time t ij (t), the attenuation coefficient of context relevance γ, the correction coefficient of cognitive reasoning relationship strength η, according to the cognitive reasoning relationship strength R between the i-th low-level learning task and the j-th low-level learning taskij The correlation between the above parameters forms a functional relationship:

[0142]

[0143] This formula can realize the calculation process of evaluating the reasoning relationship strength of the high-level cognitive reasoning path relationship map between each low-level learning task. At the same time, by introducing the correction coefficient η of the cognitive reasoning relationship strength, it can be adjusted according to the errors occurring in the calculation process, thereby improving the accuracy and applicability of the calculation formula for the reasoning relationship strength of learning tasks.

[0144] Furthermore, the adaptive cognitive learning network construction based on the low-level learning tasks of the user cognitive knowledge and the high-level learning reasoning relationship of the user cognitive knowledge includes:

[0145] Performing base node configuration on the user cognitive knowledge low-level learning task to obtain the user cognitive low-level learning task base node;

[0146] In an embodiment of the present invention, the base nodes of the user cognitive knowledge low-level learning tasks obtained by the previous identification and analysis are configured, and the relationship between the knowledge points that the user has mastered and the content of the task to be learned in the user cognitive knowledge low-level learning tasks is marked by using the knowledge graph method. For example, if the user is learning mathematics, the base nodes include "basic arithmetic", "geometric concepts", etc. During the configuration process, the user's learning content and knowledge points are mapped by using a classification algorithm (such as a decision tree or a random forest) to ensure that the definition and scope of each knowledge point are clear, thereby forming a systematic low-level learning task base node, and finally obtaining the user cognitive low-level learning task base node.

[0147] Preferably, the user cognitive low-level learning task base node is converted into an adaptive learning node to obtain a user cognitive low-level adaptive learning node;

[0148] In an embodiment of the present invention, an adaptive learning algorithm is used to analyze the learning style and preference of each user in the base node of the user's cognitive low-level learning task, and similar learning behaviors are classified through cluster analysis to identify variables that affect the learning effect (such as learning speed, memory, etc.). Subsequently, these variables are associated with the base nodes and transformed to generate adaptive learning nodes, and a genetic algorithm is used to optimize the relationship between the nodes to ensure that the learning nodes not only reflect the user's existing knowledge, but also include personalized learning paths and strategies. The formed adaptive learning nodes should have the ability to dynamically update to adapt to new needs generated by users during the learning process, and finally obtain user cognitive low-level adaptive learning nodes.

[0149] Preferably, an adaptive cognitive learning network is constructed for the corresponding user cognitive low-level adaptive learning nodes based on the user cognitive knowledge high-level learning reasoning relationship to generate an initial architecture of the user adaptive cognitive learning network.

[0150] In an embodiment of the present invention, the corresponding user cognitive low-level adaptive learning nodes are connected and integrated in the cognitive learning network by combining the high-level learning and reasoning relationships of the user cognitive knowledge obtained by previous analysis, so as to utilize the graph neural network technology, regard the low-level adaptive learning nodes as nodes in the graph, and use the learning and reasoning relationships as edges to design the network structure. In the construction process, a deep learning model is used to train the relationship between nodes, and the implicit features generated in the user learning process are extracted, so as to optimize the connection mode of the learning network. After multiple iterations and verifications, it is ensured that the network structure can effectively support the user's reasoning ability in complex cognitive learning tasks, thereby forming a dynamic and efficient adaptive cognitive learning network architecture, and finally connecting to generate the initial architecture of the user adaptive cognitive learning network.

[0151] Furthermore, the cognitive learning interaction bottleneck simulation analysis module includes the following functions:

[0152] Obtain users’ real-time cognitive learning behavior data;

[0153] In an embodiment of the present invention, the user's cognitive learning activities are monitored in real time by utilizing sensors, a learning management system (LMS) and data collection tools. These tools include a learning progress tracker, an online assessment platform and interactive learning software. Through these systems, the data collected include the user's learning time, answering speed, number of incorrect answers, frequency of participation in discussions and usage of learning resources, etc. All data are aggregated to a central database through an API interface to ensure the accuracy and timeliness of the data. This process provides a comprehensive view of the user's dynamic changes in the learning environment, and ultimately integrates and obtains the user's real-time cognitive learning behavior data.

[0154] Preferably, the user's real-time cognitive learning behavior data is subjected to learning behavior feature extraction processing to obtain the user's cognitive learning behavior features; based on the user's cognitive learning behavior features, cognitive learning interaction pattern recognition analysis is performed on the user's real-time cognitive learning behavior data to obtain the user's cognitive learning interaction behavior pattern;

[0155] In an embodiment of the present invention, the real-time cognitive learning behavior data of the user previously collected in real time is cleaned and preprocessed to remove noise and outliers, and the user learning behavior is deeply analyzed by using feature extraction algorithms, such as principal component analysis (PCA) and clustering algorithms, to extract key features, including the user's learning efficiency (such as the ratio of the time required to complete the task to the accuracy rate), learning mode (such as the ratio of active learning to passive learning) and emotional state (such as the anxiety level during learning), so as to obtain the user's cognitive learning behavior characteristics. At the same time, by combining the previously extracted user cognitive learning behavior characteristics, a machine learning model (such as a random forest or a support vector machine) is applied to the corresponding user cognitive learning behavior process in the user's real-time cognitive learning behavior data to perform recognition and analysis of cognitive learning interaction patterns, so as to identify the user's learning behavior interaction patterns, such as self-regulated learning, social learning or cooperative learning patterns, through the training model, and finally obtain the user's cognitive learning interaction behavior patterns.

[0156] Preferably, the user cognitive learning behavior features are divided and fused into interactive behavior features based on the user cognitive learning interactive behavior mode to obtain a learning feature fusion vector corresponding to each user cognitive learning interactive behavior;

[0157] In an embodiment of the present invention, the corresponding user cognitive learning behavior features are divided into the same interactive behavior pattern by combining the user cognitive learning interactive behavior pattern obtained by previous identification and analysis, and cluster analysis is used to group the behavioral features with similar interactive behavior patterns, and a corresponding feature fusion vector is generated for each group of behavioral features. These vectors will contain the common characteristics of each group of user cognitive learning behaviors, such as typical performance and interaction methods in specific learning tasks. The fusion vector is generated using a weighted average method to ensure that the degree of influence on each behavioral feature reflects its importance in the actual learning process, and finally the learning feature fusion vector corresponding to each user cognitive learning interactive behavior is obtained.

[0158] Preferably, the cognitive learning interaction simulation process corresponding to each user's cognitive learning interaction behavior is obtained by inputting the learning feature fusion vector corresponding to each user's cognitive learning interaction behavior into the user adaptive cognitive learning network initial architecture for cognitive learning interaction simulation analysis;

[0159] In an embodiment of the present invention, the learning feature fusion vectors corresponding to each user's cognitive learning interaction behavior previously generated by fusion are input into the previously constructed user adaptive cognitive learning network initial architecture. At this time, the user adaptive cognitive learning network framework is used to perform simulation analysis of the cognitive learning interaction process to predict each user's behavioral performance and learning effect in a specific learning environment. The network is trained using a back propagation algorithm, and the network weights are optimized through multiple iterations to simulate an accurate cognitive learning interaction process. The simulation results show the user's expected path and existing learning outcomes in the learning process, and finally the cognitive learning interaction simulation process corresponding to each user's cognitive learning interaction behavior is obtained.

[0160] Preferably, the cognitive learning interaction simulation process corresponding to each user's cognitive learning interaction behavior is subjected to interaction simulation process bottleneck identification analysis to obtain the bottleneck point of the user's cognitive learning interaction simulation process.

[0161] In an embodiment of the present invention, by performing bottleneck identification analysis of the interaction simulation process after completing the cognitive learning interaction simulation process corresponding to each user's cognitive learning interaction behavior, the bottleneck points in the cognitive learning interaction simulation process are identified by applying data mining techniques, such as outlier detection and time series analysis. These bottleneck points include obstacles to understanding specific knowledge points, frequent erroneous responses, or obvious lags in learning progress. During the analysis process, the bottleneck data is displayed in combination with visualization tools to make the identification results more intuitive, and finally the corresponding bottleneck points of the user's cognitive learning interaction simulation process are obtained.

[0162] Furthermore, the cognitive learning network feedback optimization module includes the following functions:

[0163] Conduct cognitive learning interaction bottleneck feature analysis on the bottleneck points of the user cognitive learning interaction simulation process to obtain bottleneck feature description data of the user cognitive learning interaction process;

[0164] In an embodiment of the present invention, by performing statistical analysis on the cognitive learning bottleneck characteristics of the bottleneck points of the user cognitive learning interaction simulation process obtained through previous analysis (for example, the user's stagnation time at a specific knowledge learning point is too long, etc.), the bottleneck performance of the user in a specific cognitive learning situation is statistically analyzed to reveal the characteristic description factors that affect the efficiency of cognitive learning, such as the user's distracted attention, weak grasp of knowledge points, or low interactive participation, etc., and to help the system to have a deeper understanding of the obstacles encountered by the user in the cognitive learning process. For example, when a user is learning mathematical concepts, he frequently makes mistakes due to lack of basic knowledge, resulting in low learning efficiency, and finally the bottleneck characteristic description data of the user's cognitive learning interaction process is obtained.

[0165] Preferably, based on the bottleneck feature description data of the user cognitive learning interaction process, the initial architecture of the user adaptive cognitive learning network is analyzed for learning network optimization objectives to obtain a user cognitive learning network optimization objective list;

[0166] In an embodiment of the present invention, the initial architecture of the user adaptive cognitive learning network used in the cognitive learning interaction process is identified and analyzed by combining the bottleneck feature description data of the user cognitive learning interaction process obtained in the previous analysis to identify and analyze the optimization goals of the learning network, so as to map the bottleneck feature data to the structure of the learning network by applying the system thinking theory, and clarify the optimization goals, including reducing the user's cognitive load, improving the depth of knowledge mastery, and improving the initiative of learning, wherein each optimization goal corresponds to a specific quantitative indicator, such as a 20% reduction in error rate, a 15% increase in learning initiative, etc., and by constructing an optimization goal list, the correlation and feasibility between the optimization goals are ensured, and finally a user cognitive learning network optimization goal list is obtained.

[0167] Preferably, a learning network optimization strategy analysis is performed on the user cognitive learning network optimization target list to obtain a user cognitive learning network optimization adjustment strategy solution;

[0168] In an embodiment of the present invention, after determining the optimization target list of the user cognitive learning network, a learning network optimization strategy analysis is performed to evaluate the performance of the current learning network by adopting model prediction technology and compare it with the optimization target. Then, a variety of optimization strategies are designed using a reinforcement learning algorithm. For example, for users with excessive cognitive load, an optimization strategy based on a tiered teaching method is implemented to alleviate learning pressure by providing step-by-step guidance and timely feedback. At the same time, a simulation tool is used to evaluate the strategy to ensure the effectiveness and operability of the strategy, thereby forming a comprehensive optimization adjustment strategy scheme to cover different cognitive learning links and individual differences of users. Finally, the user cognitive learning network optimization adjustment strategy scheme is obtained through analysis.

[0169] Preferably, the initial architecture of the user adaptive cognitive learning network is subjected to cognitive learning network feedback optimization adjustment according to the user cognitive learning network optimization adjustment strategy scheme to generate a user adaptive cognitive learning optimization adjustment network.

[0170] In an embodiment of the present invention, the cognitive learning network is optimized and adjusted on the corresponding user adaptive cognitive learning network initial architecture by combining the user cognitive learning network optimization and adjustment strategy scheme obtained by previous analysis, so as to perform real-time analysis on the new data of the user in the learning process by using a machine learning algorithm, and respond to the user's learning needs by adjusting the network parameters and learning paths. For example, if a user shows extremely high frustration in a specific link, the learning content will be dynamically adjusted, the corresponding basic knowledge review materials will be provided, and interactive exercises will be added to enhance cognitive learning confidence, so as to optimize and adjust the corresponding adaptive cognitive learning optimization network, which will effectively improve the user's learning efficiency and effect, achieve the goal of personalized cognitive learning, and finally optimize and generate the user's adaptive cognitive learning optimization adjustment network.

[0171] Therefore, no matter from which point of view, the embodiments should be regarded as exemplary and non-restrictive, and the scope of the invention is limited by the appended claims rather than the above description, so it is intended to include all changes that fall within the meaning and scope of the equivalent elements of the application documents. The above is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. An adaptive cognitive learning network simulation system based on a large language model, characterized in that: Includes the following modules: A user cognitive learning feature vector conversion module is used to obtain user historical cognitive learning data, user cognitive interaction record data, and user cognitive learning feedback information data, and perform cognitive feature vector fusion conversion based on the user historical cognitive learning data, user cognitive interaction record data, and user cognitive learning feedback information data to obtain a user cognitive learning feature vector; The hierarchical cognitive reasoning and learning network construction module is used to perform hierarchical cognitive reasoning analysis on the user cognitive learning feature vector based on the large language model to obtain the user cognitive knowledge low-level learning tasks and the user cognitive knowledge high-level learning reasoning relationship; according to the user cognitive knowledge low-level learning tasks and the user cognitive knowledge high-level learning reasoning relationship, an adaptive cognitive learning network is constructed to generate the initial architecture of the user adaptive cognitive learning network; The cognitive learning interaction bottleneck simulation analysis module is used to obtain the user's real-time cognitive learning behavior data, and based on the user's real-time cognitive learning behavior data, perform cognitive learning interaction bottleneck simulation analysis on the user's adaptive cognitive learning network initial architecture to obtain the bottleneck point of the user's cognitive learning interaction simulation process; The cognitive learning network feedback optimization module is used to perform cognitive learning network feedback optimization adjustment on the initial architecture of the user adaptive cognitive learning network based on the bottleneck point of the user cognitive learning interaction simulation process to generate a user adaptive cognitive learning optimization adjustment network.

2. The adaptive cognitive learning network simulation system based on a large language model according to claim 1 is characterized in that: The user cognitive learning feature vector analysis module includes the following functions: By acquiring historical cognitive learning data records from the user cognitive learning platform database, the user's historical cognitive learning data is obtained; Conduct cognitive learning trajectory analysis on the user's historical cognitive learning data to obtain the user's historical cognitive learning activity trajectory; Based on the user's historical cognitive learning activity trajectory, cognitive interaction analysis is performed on the user's historical cognitive learning data and recorded to obtain user cognitive interaction record data; A user cognitive interaction quality evaluation system is constructed based on user cognitive interaction record data, and cognitive learning feedback analysis is performed on user historical cognitive learning data based on the user cognitive interaction quality evaluation system to obtain user cognitive learning feedback information data; The cognitive feature vector is fused and transformed according to the user's historical cognitive learning data, the user's cognitive interaction record data and the user's cognitive learning feedback information data to obtain the user's cognitive learning feature vector.

3. The adaptive cognitive learning network simulation system based on a large language model according to claim 2 is characterized in that: The cognitive learning trajectory analysis of the user's historical cognitive learning data includes: Perform cognitive learning time and content extraction processing on the user's historical cognitive learning data to obtain the user's historical cognitive learning timestamp and the user's historical cognitive learning content; Conduct cognitive learning theme mining and analysis on the user's historical cognitive learning content to obtain the theme of each user's historical cognitive learning activity; Conduct correlation analysis on the themes of each user's historical cognitive learning activities to obtain the correlation relationship between the learning activity behaviors of each user's cognitive learning activity themes; Based on the learning activity behavior association relationship between each user's cognitive learning activity theme, the cognitive learning behavior association map of the user's historical cognitive learning content is analyzed to generate a user's historical cognitive learning behavior association map; Based on the user's historical cognitive learning timestamp, the cognitive learning trajectory analysis is performed on the user's historical cognitive learning behavior association map to obtain the user's historical cognitive learning activity trajectory.

4. The adaptive cognitive learning network simulation system based on a large language model according to claim 2 is characterized in that: The cognitive feature vector fusion conversion based on the user's historical cognitive learning data, the user's cognitive interaction record data, and the user's cognitive learning feedback information data includes: Perform cognitive learning content feature analysis on the user's historical cognitive learning data to obtain the user's cognitive learning content features, wherein the user's cognitive learning content features include the user's cognitive learning content theme, the user's cognitive learning difficulty level, and the user's cognitive learning time; Performing user interaction frequency feature analysis on user cognitive interaction record data to obtain user cognitive learning interaction frequency features, wherein the user cognitive learning interaction frequency features include user cognitive learning click-through rate, user cognitive learning viewing time, and user cognitive learning discussion participation; Perform cognitive learning feedback feature analysis on user cognitive learning feedback information data to obtain user cognitive learning feedback sentiment features, wherein the user cognitive learning feedback sentiment features include a user cognitive learning positive feedback sentiment index, a user cognitive learning neutral feedback sentiment index, and a user cognitive learning negative feedback sentiment index; The user cognitive learning content features, user cognitive learning interaction frequency features and user cognitive learning feedback sentiment features are transformed into cognitive feature vectors to obtain the user cognitive learning feature vector.

5. The adaptive cognitive learning network simulation system based on a large language model according to claim 1 is characterized in that: The hierarchical cognitive reasoning and learning network building module includes the following functions: Based on the low-level cognitive modules in the large language model, low-level task recognition and analysis are performed on the user's cognitive learning feature vector to obtain the user's cognitive knowledge low-level learning tasks; Perform task content classification on the user cognitive knowledge low-level learning tasks to obtain the user cognitive low-level learning task content classification matrix; A high-level cognitive reasoning framework that can handle low-level learning tasks is constructed through high-level cognitive modules in the large language model. Based on the high-level cognitive reasoning framework, a high-level cognitive reasoning path analysis is performed on the corresponding low-level learning tasks in the user's cognitive low-level learning task content classification matrix to obtain the high-level cognitive reasoning path between each low-level learning task. Based on the high-level cognitive reasoning path between each low-level learning task, the corresponding low-level learning task of user cognitive knowledge is adjusted and analyzed at a high level to obtain the high-level learning reasoning relationship of user cognitive knowledge; An adaptive cognitive learning network is constructed according to low-level learning tasks of user cognitive knowledge and high-level learning reasoning relationships of user cognitive knowledge to generate an initial architecture of the user adaptive cognitive learning network.

6. The adaptive cognitive learning network simulation system based on a large language model according to claim 5 is characterized in that: The high-level learning reasoning adjustment analysis of the corresponding user cognitive knowledge low-level learning tasks based on the high-level cognitive reasoning path between the low-level learning tasks includes: Based on the high-level cognitive reasoning paths between the low-level learning tasks, the corresponding user cognitive knowledge low-level learning tasks are connected with the cognitive reasoning paths to generate a high-level cognitive reasoning path relationship map between the low-level learning tasks; The learning task reasoning relationship strength calculation formula is used to evaluate and calculate the reasoning relationship strength of the high-level cognitive reasoning path relationship map between each low-level learning task, and the cognitive reasoning relationship strength between each low-level learning task is obtained; Based on the strength of cognitive reasoning relationships between low-level learning tasks, a high-level learning reasoning adjustment analysis is performed on the high-level cognitive reasoning path relationship map between low-level learning tasks to obtain the high-level learning reasoning relationship of user cognitive knowledge.

7. The adaptive cognitive learning network simulation system based on a large language model according to claim 6 is characterized in that: The specific calculation formula for the learning task reasoning relationship strength is: In the formula, R ij is the cognitive reasoning relationship strength between the i-th low-level learning task and the j-th low-level learning task, i and j are both item measurement parameters of low-level learning tasks, T is the upper limit of the time range for calculating the reasoning relationship strength, t is the time variable parameter, and w ij is the weight of the inference relationship between the i-th low-level learning task and the j-th low-level learning task, s ij (t) is the number of related reasoning paths between the i-th low-level learning task and the j-th low-level learning task at time t, α is the attenuation coefficient of the number of reasoning paths, c ij (t) is the task content similarity between the i-th low-level learning task and the j-th low-level learning task at time t, β is the attenuation coefficient of the task content similarity, p ij (t) is the context correlation parameter between the ith low-level learning task and the jth low-level learning task at time t, γ is the attenuation coefficient of the context correlation effect, and η is the correction coefficient of the cognitive reasoning relationship strength.

8. The adaptive cognitive learning network simulation system based on a large language model according to claim 5 is characterized in that: The construction of the adaptive cognitive learning network based on the low-level learning tasks of user cognitive knowledge and the high-level learning reasoning relationship of user cognitive knowledge includes: Performing base node configuration on the user cognitive knowledge low-level learning task to obtain the user cognitive low-level learning task base node; Performing adaptive learning node transformation on the user cognitive low-level learning task base node to obtain the user cognitive low-level adaptive learning node; Based on the high-level learning and reasoning relationship of user cognitive knowledge, an adaptive cognitive learning network is constructed for the corresponding user cognitive low-level adaptive learning nodes to generate the initial architecture of the user adaptive cognitive learning network.

9. The adaptive cognitive learning network simulation system based on a large language model according to claim 1, characterized in that: The cognitive learning interaction bottleneck simulation analysis module includes the following functions: Obtain users’ real-time cognitive learning behavior data; Perform learning behavior feature extraction and processing on the user's real-time cognitive learning behavior data to obtain the user's cognitive learning behavior features; Based on the user's cognitive learning behavior characteristics, cognitive learning interaction pattern recognition and analysis are performed on the user's real-time cognitive learning behavior data to obtain the user's cognitive learning interaction behavior pattern; Based on the user cognitive learning interactive behavior mode, the user cognitive learning behavior features are divided and integrated to obtain the learning feature fusion vector corresponding to each user cognitive learning interactive behavior; By inputting the learning feature fusion vector corresponding to each user's cognitive learning interaction behavior into the initial architecture of the user adaptive cognitive learning network to perform cognitive learning interaction simulation analysis, the cognitive learning interaction simulation process corresponding to each user's cognitive learning interaction behavior is obtained; The interactive simulation process bottleneck identification analysis is performed on the cognitive learning interactive simulation process corresponding to each user's cognitive learning interactive behavior to obtain the bottleneck point of the user's cognitive learning interactive simulation process.

10. The adaptive cognitive learning network simulation system based on a large language model according to claim 1, characterized in that: The cognitive learning network feedback optimization module includes the following functions: Conduct cognitive learning interaction bottleneck feature analysis on the bottleneck points of the user cognitive learning interaction simulation process to obtain bottleneck feature description data of the user cognitive learning interaction process; Based on the bottleneck feature description data of the user cognitive learning interaction process, the initial architecture of the user adaptive cognitive learning network is analyzed for learning network optimization objectives, and a list of user cognitive learning network optimization objectives is obtained; Conduct learning network optimization strategy analysis on the user cognitive learning network optimization target list to obtain the user cognitive learning network optimization adjustment strategy plan; According to the user cognitive learning network optimization and adjustment strategy scheme, the initial architecture of the user adaptive cognitive learning network is optimized and adjusted by cognitive learning network feedback to generate a user adaptive cognitive learning optimization and adjustment network.

Images