Infant early education enlightenment system and method

Abstract

The invention relates to the technical field of infant early education, and discloses an infant early education enlightenment system and method, and the system comprises a learning data collection module which is used for collecting behavior data generated by an infant in a learning process, and an individual memory modeling module which is used for constructing an infant individual memory maintenance model based on the behavior data, and outputting memory decline parameters, the system comprises a memory maintenance model, a master degree evaluation module, a review scheduling generation module, a multi-sensory information optimization module, a learning path optimization module, a learning feedback closed-loop module, a learning feedback closed-loop module and a review scheduling generation module, wherein the review scheduling generation module is used for combining memory state parameters output by the memory maintenance model and a master degree scoring result; the method is used for performing statistical analysis and feedback judgment on learning results. By combining the learning behavior data and the memory retention model of the children, the optimal learning path and review scheduling strategy are realized, the review plan customized according to the actual performance of each child is obtained, and the learning efficiency and the memory retention rate are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of early childhood education, and in particular to a system and method for early childhood education enlightenment. Background Art

[0002] In the field of early childhood education, designing personalized learning pathways and review strategies has always been a key challenge in improving learning outcomes and memory retention. Traditional educational methods often adopt a one-size-fits-all approach, ignoring individual differences in learning progress, learning styles, and memory retention. This approach often fails to tailor learning content and review cadence to children's specific needs, resulting in low learning efficiency and even potential for forgetting or a lack of firm grasp of learning content.

[0003] To address this challenge, with the recent development of intelligent educational technology, a growing number of studies have begun exploring the design of data-driven personalized learning pathways. By collecting and analyzing children's learning behavior data, such as learning time, recognition accuracy, and reaction time, educational systems can assess each child's learning progress in real time and adjust learning tasks and review content in a timely manner. However, existing personalized learning methods mostly rely on static learning schedules and review plans, lacking the ability to accurately monitor and adjust dynamic changes in memory. Summary of the Invention

[0004] In response to the shortcomings of the existing technology, the present invention provides a system and method for early childhood education enlightenment, which solves the problems that the existing technology usually does not introduce a systematic memory retention model in review scheduling, lacks quantitative modeling of the memory decline process of young children, and cannot accurately predict the forgetting trend. This makes it difficult to scientifically plan the review strategy, causes waste of resources due to excessive review frequency, and affects knowledge retention due to insufficient review.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: a child early education enlightenment system, comprising: Learning data collection module, used to collect behavioral data generated by children during the learning process; An individual memory modeling module is used to construct an individual memory retention model for children based on the behavioral data and output memory decay parameters for subsequent learning scheduling; A mastery assessment module, a deep neural network model for processing learning history; A review schedule generation module, configured to automatically generate a personalized review schedule and review content set by combining the memory state parameters and the mastery level score results output by the memory retention model; The multi-sensory information optimization module is used to evaluate the learning effect of each sensory input channel during the learning process. By building a sensory input weight model, it adjusts the display method and input combination of subsequent learning materials; The learning path optimization module is used to construct a path optimization function using learning time and recognition errors as cost factors, and based on this function, it generates the optimal learning content presentation order and plans the learning path; The learning feedback closed-loop module is used to perform statistical analysis and feedback judgment on the learning results, and dynamically update the memory model, evaluation model and scheduling strategy according to the feedback results to form a closed loop.

[0006] Preferably, the behavioral data includes learning time, learning content identification, recognition accuracy, error frequency, recognition reaction time and the type of sensory input channel used. The individual memory modeling module constructs a memory model based on the exponential decay law, and fits the individual memory decay coefficient by minimizing the error between the recognition accuracy and the predicted memory amount, thereby establishing an individual memory retention curve that reflects the learning and forgetting speed.

[0007] Preferably, the retention model models the learning memory state by identifying the forgetting trend, and the mastery level assessment module uses a recurrent neural network or a long short-term memory network to construct a deep neural network structure, and uses the learning behavior time series data as input to train the model to predict the current mastery probability of the learning content.

[0008] Preferably, the mastery probability output by the mastery level assessment module is used to drive the review scheduling generation module to sort the learning content and calculate the review time interval. The system determines the review timing with the goal of minimizing the predicted forgetting loss. The network model uses learning behavior data as input and outputs the current child's mastery level score for each learning content.

[0009] Preferably, the review schedule generation module adopts a Bayesian optimization strategy to select the review content and time points with the greatest learning value by constructing a joint expected utility function of the mastery level prediction value and model uncertainty.

[0010] Preferably, the multisensory information optimization module calculates the information entropy value of each channel based on the sensory channel recognition results and reaction time, and dynamically adjusts the weight parameters of the multisensory channels according to the entropy value to improve the information efficiency of sensory input.

[0011] Preferably, the learning path optimization module constructs a learning path search space based on the review task set output by the review scheduling and the current mastery level score, and uses a weighted combination of time cost and error cost as the objective function to generate the optimal path solution through a dynamic programming algorithm.

[0012] Preferably, the learning path optimization module adaptively adjusts the ratio of time weight to error weight according to historical learning performance, and updates the learning path generation strategy in real time.

[0013] Preferably, the learning feedback closed-loop module uses hypothesis testing and variance analysis methods to evaluate the significance of learning results, and feeds back the statistical analysis results to the memory modeling module and review scheduling module for correcting individual model parameters and review rhythm.

[0014] A method for early childhood education enlightenment, comprising the following steps: First, collect behavioral data of children during learning activities, including learning time, recognition results, reaction time, and the type of sensory channels used; Then, an individual memory retention model is constructed based on the behavioral data, and memory decay parameters are extracted; Use a time series-based deep neural network model to predict and judge children’s mastery of various learning contents; Combine memory models with mastery level prediction results to generate personalized review schedules and review content sets; Then, based on the recognition performance and response efficiency of each sensory channel, the input value is evaluated to generate an optimized sensory input combination strategy; Construct a cost function with learning time and error rate as target factors, and use optimization algorithms to generate the optimal learning path; Finally, statistical feedback is given to the learning results, and the memory model parameters and review scheduling strategy are updated based on the feedback results to form a closed-loop optimization.

[0015] The present invention provides a system and method for early childhood education and enlightenment. It has the following beneficial effects: 1. The present invention combines children's learning behavior data with memory retention models to evaluate their mastery level in real time and generate personalized review schedules, thereby achieving optimized learning paths and review scheduling strategies, and obtaining review plans tailored to each child's actual performance, thereby improving learning efficiency and memory retention.

[0016] 2. The present invention generates an optimized sensory input combination strategy by evaluating the recognition performance and response efficiency of each sensory channel, realizes dynamic adjustment of the sensory input channel, improves the efficiency of information transmission, and enhances the effectiveness of sensory input on the absorption of learning content and optimizes the learning experience.

[0017] 3. The present invention uses a deep neural network model based on time series to predict the degree of mastery of various learning contents by young children, realizes accurate learning progress assessment and mastery prediction, and obtains a reasonable arrangement of the review priority of learning tasks, thereby maximizing the effect of learning effect.

[0018] 4. The present invention evaluates the significance of learning results through hypothesis testing and variance analysis, provides real-time feedback and updates memory models and review strategies, and adaptively adjusts individual models and review rhythms, thereby continuously optimizing the learning process and improving individual learning outcomes. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 This is a diagram showing the relationship between system modules of a preschool education enlightenment system according to the present invention; Figure 2 The present invention is a method flow chart of a method for early childhood education enlightenment. DETAILED DESCRIPTION

[0020] The following will clearly and completely describe the technical solution of the present invention in conjunction with the accompanying drawings. Obviously, the embodiments described are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0021] Please see the attached Figure 1 The embodiment of the present invention provides a system for early childhood education and enlightenment, comprising: Learning data collection module, used to collect behavioral data generated by children during the learning process; The individual memory modeling module is used to build an individual memory retention model for children based on behavioral data and output memory decay parameters for subsequent learning scheduling; A mastery assessment module, a deep neural network model for processing learning history; A review scheduling generation module is used to automatically generate a personalized review schedule and review content set by combining the memory state parameters and mastery level score results output by the memory retention model; The multi-sensory information optimization module is used to evaluate the learning effect of each sensory input channel during the learning process. By building a sensory input weight model, it adjusts the display method and input combination of subsequent learning materials; The learning path optimization module is used to construct a path optimization function using learning time and recognition errors as cost factors, and based on this function, it generates the optimal learning content presentation order and plans the learning path; The learning feedback closed-loop module is used to perform statistical analysis and feedback judgment on the learning results, and dynamically update the memory model, evaluation model and scheduling strategy based on the feedback results to form a closed loop.

[0022] Behavioral data include learning time, learning content identification, recognition accuracy, error frequency, recognition reaction time and the type of sensory input channel used. The individual memory modeling module constructs a memory model based on the exponential decay law. By minimizing the error between the recognition accuracy and the predicted memory amount, the individual memory decay coefficient is fitted, and then an individual memory retention curve reflecting the learning and forgetting speed is established.

[0023] Specifically, behavioral data includes: Learning time: The time spent on each learning task reflects the effort a child puts into the task. Longer learning times usually mean the task is more difficult, or the child's understanding of the content is lower, and more review and consolidation may be needed. Learning content identification: Each learning task has a unique identifier that helps the system distinguish different learning contents. These identifiers allow the system to track the specific content of each task, making it easier to analyze the impact of certain content on children's learning outcomes; Recognition accuracy rate: The percentage of correct answers given by children to a task during the learning process. Generally, a higher accuracy rate indicates a better grasp of the task, and the system can use this to assess learning outcomes. If the accuracy rate is low, the system will help children improve their mastery by adjusting the learning content or increasing the frequency of review. Error frequency: records the number of mistakes a child makes during the learning process. A higher error frequency usually indicates that the learning content is more difficult. The system can analyze error data to optimize review content or adjust learning plans to ensure that children better understand and master the content. Recognition reaction time: refers to the time it takes for a child to receive a learning task and then give feedback. The length of the reaction time reflects the child's mastery of the learning task. A longer reaction time may mean that the child needs more time to understand the task, or that the task itself is more difficult; Type of sensory input channel used: records the sensory input method used by the child to complete the task (such as vision, hearing, touch, etc.). Children may rely on different sensory channels in different learning tasks. For example, visual tasks may require more visual input, while language learning may require more auditory input. By analyzing the type and effect of sensory input, the system can adjust the presentation of learning tasks to optimize the learning experience; The individual memory modeling module uses the exponential decay law to construct a memory model for children. Based on behavioral data, this model predicts the child's memory retention for each learning task and adjusts the parameters of the memory model by minimizing the error between the recognition accuracy and the predicted memory amount. The memory decay model follows the following formula: ; in: Indicates time Afterwards, the degree to which children retain their memory of a certain learning content; is the initial memory retention value, which indicates the memory strength after learning a certain content; 𝜆 is the memory decay coefficient, which controls the speed of memory decay. The larger 𝜆 is, the faster the memory decay is. Indicates the time from the completion of the learning content to the present.

[0024] The individual memory modeling module fits the memory decay coefficient 𝜆 for each child through regression analysis of the children's learning data. The system estimates the memory decay rate by analyzing factors such as the children's recognition accuracy, error frequency, and reaction time during the learning process. The system adjusts the decay coefficient 𝜆 based on the error between the children's recognition accuracy and the predicted memory retention. If the system finds that the recognition accuracy of a certain learning content is low and the predicted memory retention value is far lower than the actual performance, then the decay coefficient 𝜆 will be increased, indicating that the memory decay rate of the content is faster and the review frequency needs to be increased. Conversely, if the accuracy is high and the reaction time is short, the system may reduce the decay coefficient and extend the review interval. The individual memory retention curve can clearly reflect each child's memory retention of the learning content. The system establishes this memory curve by continuously collecting children's behavioral data and dynamically adjusting the memory model. By continuously collecting children's learning performance, the system will gradually update the memory retention curve. For example, if a child performs poorly in a certain learning task, the system will shorten the review interval of the task and adjust the decay coefficient to adapt to the child's memory decay characteristics; through this dynamic update, the system can ensure that the review is carried out at the most appropriate time, avoiding review too early or too late, and ensuring the best memory retention of the learning content; By accurately collecting behavioral data and modeling individual memory decay patterns, this invention can provide each child with a personalized learning path and review plan. By minimizing the error between recognition accuracy and predicted memory retention, the system accurately fits the individual's memory decay coefficient and establishes a personalized memory retention curve, thereby optimizing learning outcomes and memory retention time.

[0025] The retention model models the learning memory status by identifying the forgetting trend. The mastery evaluation module uses a recurrent neural network or a long short-term memory network to construct a deep neural network structure, and uses the learning behavior time series data as input to train the model to predict the current mastery probability of the learning content.

[0026] Specifically, the mastery assessment module uses an RNN or LSTM structure to assess children's learning mastery. These networks are particularly suitable for processing time series data, such as learning behavior data (including learning time, error frequency, etc.); Using time series input, RNNs or LSTMs can analyze children's performance in different learning tasks and generate probabilities of mastery of the learning content. For example, by training children on their reaction times and number of errors during a task, the system can predict the degree of mastery of a particular task. LSTMs are particularly effective when processing long-span data, effectively capturing long-term dependencies in the learning process and avoiding information loss, thereby improving prediction accuracy. By combining forgetting tendency modeling with deep neural network assessment, the present invention can provide personalized learning and review strategies, ensuring that learning content is reviewed at the optimal time. This approach not only improves review efficiency, but also prolongs memory retention and optimizes learning paths.

[0027] The mastery probability output by the mastery level assessment module is used to drive the review scheduling generation module to sort learning content and calculate review time intervals. The system determines the review timing with the goal of minimizing predicted forgetting loss. The network model uses learning behavior data as input and outputs the current child's mastery level score for each learning content.

[0028] Specifically, the mastery level assessment module evaluates the child's mastery of each learning content by outputting the mastery probability. This assessment result directly drives the review schedule generation module. This module sorts the learning content according to the mastery probability and calculates the review time interval to determine the optimal review time. The system's goal is to minimize the predicted forgetting loss, that is, to avoid children forgetting the learning content before their memory declines as much as possible. The review schedule generation module sorts the learning content according to the mastery probability. Tasks with a high mastery probability can extend the review interval, while tasks with a low mastery probability will shorten the review interval to ensure that they are reviewed before memory declines. Input and output of the network model: Input: Children's learning behavior data, such as learning time, error frequency, reaction time, etc. Output: The mastery score of each learning content, reflecting the probability of the child's memory retention of the task; The review scheduling generation module calculates the review interval based on these scores and automatically adjusts the review priority of the task to ensure that the review is carried out at the appropriate time. The system determines the review timing by minimizing the predicted forgetting loss. Forgetting loss refers to the difference between the predicted memory retention value and the actual memory retention value. By optimizing this loss, the system can accurately determine the review time, avoid reviewing too early or too late, and ensure that each learning content is consolidated at the best time. This optimization method effectively improves learning efficiency, maximizes memory retention time, and ensures that children's learning content is not forgotten prematurely. By combining mastery assessment with review schedule generation, the system can precisely schedule review times and automatically adjust the order and time intervals of review content, thereby achieving personalized and intelligent learning review management. This approach not only improves the relevance and effectiveness of review, but also maximizes memory retention.

[0029] The review schedule generation module adopts the Bayesian optimization strategy to select the review content and time points with the most learning value by constructing a joint expected utility function of the mastery level prediction value and model uncertainty.

[0030] Specifically, the review schedule generation module uses a Bayesian optimization strategy to determine the optimal review content and timing. This strategy guides review scheduling by constructing a joint expected utility function that combines mastery predictions with model uncertainty, selecting the tasks that are most valuable for learning.

[0031] This function combines the predicted value of mastery (i.e., the probability of mastering the current learning content) with model uncertainty, weighing the learning value of the review task and the rationality of the review timing. Specifically, the expected utility function can be expressed as: ; in: μ(x) represents the predicted value of the mastery level of the current learning content; σ(x) is the uncertainty of the mastery prediction (i.e., the model’s confidence in the mastery of the task); β is the weight that balances mastery and uncertainty; By maximizing the expected utility function, the system selects the review content with the greatest learning value and reviews it at the appropriate time. The Bayesian optimization strategy takes into account the learning value of the review content and the uncertainty of the model. The system automatically prioritizes review tasks with lower predicted mastery and higher uncertainty, thereby improving review efficiency. By introducing Bayesian optimization, the review scheduling generation module can efficiently select review tasks and time points, enabling intelligent and personalized review management. This approach effectively balances the mastery of learning tasks with the uncertainty of review timing, maximizing the effectiveness and learning value of the review and ensuring that review tasks are performed when they are most needed, thereby improving learning efficiency and memory retention.

[0032] The multisensory information optimization module calculates the information entropy value of each channel based on the sensory channel recognition results and reaction time, and dynamically adjusts the weight parameters of the multisensory channels according to the entropy value to improve the information efficiency of sensory input.

[0033] Specifically, the multi-sensory information optimization module calculates the information entropy value of each sensory channel through the sensory channel recognition results and reaction time, and dynamically adjusts the weight parameters of each sensory channel based on these entropy values ​​to optimize the information efficiency of sensory input. Information entropy is used to measure the effective information volume of each sensory channel. The higher the entropy value, the greater the uncertainty of the information transmitted by the channel, which may require more time or resources to process; the lower the entropy value, the more certain the information is and can be transmitted more quickly. The module evaluates the contribution of each sensory channel to the learning task by calculating the information entropy value of the channel; The calculation formula of information entropy is: ; in: H(x) represents information entropy; is the probability of the i-th event in the sensory channel; By calculating the entropy value of each sensory channel, the system can determine the information efficiency of each channel in the current learning task. Based on the calculated information entropy value, the module will dynamically adjust the weight parameters of each sensory channel. For example, when the entropy value of a certain channel is high and the reaction time is long, the system may reduce the weight of the channel and increase the weight of the channel with lower information entropy and faster reaction. In this way, the system can improve the overall learning efficiency and reduce redundancy and delay in the information processing process. The goal of the module is to make information transmission more efficient, reduce redundancy and processing time, and thus improve the efficiency of information absorption in the learning process by reasonably adjusting the weight of each sensory channel. By dynamically adjusting weights based on information entropy, the Multisensory Information Optimization Module automatically optimizes the efficiency of sensory input across different learning tasks. This optimization method not only adjusts the priority of sensory channels based on actual learning needs, but also improves the overall efficiency of information processing, reduces redundancy, and enhances learning outcomes.

[0034] The learning path optimization module constructs a learning path search space based on the review task set output by the review schedule and the current mastery level score, and uses the weighted combination of time cost and error cost as the objective function to generate the optimal path solution through a dynamic programming algorithm.

[0035] Specifically, the learning path optimization module constructs a learning path search space based on the review task set and current mastery score output by the review schedule generation module. The module's goal is to generate the optimal learning path by minimizing the time cost and error cost during the review process. The objective function of learning path optimization is a weighted combination of time cost and error cost. Time cost takes into account the time required to review each task, while error cost reflects the risk of making mistakes during the review process. The objective function can be expressed as: C = αT + βE; in: C is the total cost; T is the time cost; E is the error cost; α and β are weight coefficients used to balance time cost and error cost; To find the optimal solution among multiple possible learning paths, the system uses a dynamic programming algorithm. Dynamic programming breaks down the problem into subproblems and gradually solves the optimal path. In this module, dynamic programming ultimately generates the optimal solution for the learning path by minimizing the total cost of the objective function. Each step takes into account the review order of tasks, time consumption, and potential error risks to ensure the optimization of the review path. Using a weighted objective function based on time and error costs, the Learning Path Optimization module generates a personalized learning path for each child, ensuring efficient and appropriate review schedules. The application of a dynamic programming algorithm makes path search more precise, effectively avoiding redundancy and inefficiency in the review process, and maximizing learning outcomes and time utilization.

[0036] The learning path optimization module adaptively adjusts the ratio of time weight to error weight based on historical learning performance and updates the learning path generation strategy in real time.

[0037] Specifically, the learning path optimization module dynamically adjusts the ratio of time weight to error weight based on historical learning performance, updating the learning path generation strategy in real time to achieve personalized optimization. The module evaluates the characteristics of the current learning task based on the child's historical learning performance data (such as learning time, error frequency, reaction time, etc.). This historical data reflects the child's strengths and weaknesses in past tasks, helping the system determine whether time cost or error cost is more important in the current learning path optimization. Based on historical learning performance, the module dynamically adjusts the weight ratio of time cost to error cost in the objective function. Specifically, if a child exhibits a high error frequency in certain tasks, the system increases the weight of error cost, forcing the system to prioritize review of tasks with high error risk. Conversely, if a child reacts quickly and has good mastery of certain tasks, the system decreases the weight of error cost and gives more consideration to time efficiency. By adaptively adjusting the weight ratio, the learning path optimization module can update the strategy in real time to ensure the optimality and personalization of the learning path. Each time a child completes a learning task, the system adjusts the path planning based on the latest learning data to ensure that the review schedule always matches the child's learning status. By adaptively adjusting the time-weighted and error-weighted ratios, the learning path optimization module can more accurately reflect children's actual learning needs. This dynamic adjustment allows the system to flexibly adapt to different learning situations, ensuring that review schedules for learning tasks are both efficient and targeted at addressing children's learning weaknesses, thereby improving overall learning outcomes and memory retention.

[0038] The learning feedback closed-loop module uses hypothesis testing and variance analysis methods to evaluate the significance of learning results, and feeds back the statistical analysis results to the memory modeling module and review scheduling module for correcting individual model parameters and review rhythm.

[0039] Specifically, the learning feedback closed-loop module uses hypothesis testing and variance analysis methods to evaluate the significance of learning outcomes, and feeds back the statistical analysis results to the memory modeling module and review scheduling module to modify individual model parameters and review rhythm to further optimize the learning process. The module first conducts hypothesis testing based on the child's learning performance (such as recognition accuracy, reaction time, etc.) to determine whether there is significant learning progress or decline. Through tests such as t-tests, the system can evaluate the effectiveness of the current learning task and confirm whether it meets the expected learning goals. The module uses analysis of variance (ANOVA) to evaluate the differences in learning outcomes across different tasks, different learning strategies, and sensory channels. Variance analysis can detect which factors (such as task type, learning method, sensory input type, etc.) have a significant impact on learning outcomes, helping the system identify key learning variables.

[0040] Through statistical results, the system can correct the parameters in the memory model, such as the memory decay coefficient, to ensure that the model more accurately reflects the actual memory retention status of the child. If the hypothesis test shows that the memory retention situation is significantly different from the model prediction, the memory modeling module will adjust its parameters to more accurately predict and adjust the review strategy. Based on the results of the variance analysis, the system can re-evaluate the review content and review time arrangements. If the learning effect of certain tasks is significantly lower than expected, the review scheduling module will adjust its review rhythm and increase the review frequency of the task to ensure that the learning effect is maximized. By introducing the significance evaluation and statistical analysis of learning results, the system can dynamically adjust the individual learning model to ensure that the learning path and review rhythm are highly matched with the actual learning effect, thereby optimizing the learning experience and results; The closed-loop learning feedback module enables the system to track learning outcomes in real time and make precise adjustments. By statistically analyzing the significance of learning outcomes, the system not only assesses various variables in the learning process but also effectively adjusts review cadence and memory modeling, improving the accuracy and efficiency of personalized learning and ensuring that each child reviews at the optimal time for optimal learning outcomes.

[0041] Please see the attached Figure 2, a method for early childhood education enlightenment, comprising the following steps: First, collect behavioral data of children during learning activities, including learning time, recognition results, reaction time, and the type of sensory channels used; Then, an individual memory retention model is constructed based on the behavioral data, and memory decay parameters are extracted; Use a time series-based deep neural network model to predict and judge children’s mastery of various learning contents; Combine memory models with mastery level prediction results to generate personalized review schedules and review content sets; Then, based on the recognition performance and response efficiency of each sensory channel, the input value is evaluated to generate an optimized sensory input combination strategy; Construct a cost function with learning time and error rate as target factors, and use optimization algorithms to generate the optimal learning path; Finally, statistical feedback is given to the learning results, and the memory model parameters and review scheduling strategy are updated based on the feedback results to form a closed-loop optimization.

[0042] Specifically, we optimize children's learning efficiency and memory retention through personalized learning paths and review scheduling: First, we collect children's behavioral data during learning activities. This data includes: learning time: the time required for each learning task, recognition results: the children's correct and incorrect judgments on the learning task, and reaction time: the time it takes for the children to respond to the task, reflecting their understanding of the learning content; Based on the collected behavioral data, the system constructs an individual memory retention model and extracts memory decay parameters from it. Using methods such as exponential decay models, the system can predict children's memory retention of learning content and assess the rate of memory decay. Using time series-based deep neural network models (such as RNN or LSTM) to train children's learning behavior data, the model can predict and judge the children's mastery of various learning contents. By learning from historical data, the model can accurately assess children's performance in specific tasks and their current mastery. Combining the memory model with the mastery level prediction results, the system generates a personalized review schedule and review content set. The system dynamically adjusts the review time and content to ensure that children review at the best time, maximizing learning outcomes. Based on the recognition performance and response efficiency of each sensory channel, the input value of each sensory channel is evaluated. By calculating the information entropy and response time of each channel, an optimized sensory input combination strategy is generated. This strategy will dynamically adjust the weight of each sensory channel to optimize the learning experience and improve learning efficiency. Construct a cost function with learning time and error rate as target factors. Utilize optimization algorithms (such as Bayesian optimization and dynamic programming) to generate the optimal learning path, ensuring that errors are minimized in the shortest possible time and improving learning outcomes. This optimization process considers the balance between time and error costs to determine the optimal review order for learning tasks. Finally, statistical feedback is provided on the learning results, and the significance of the learning results is evaluated using methods such as hypothesis testing and variance analysis. Based on the statistical analysis results, the system updates the parameters of the memory model and the review scheduling strategy in real time, forming a feedback loop. Through this closed-loop optimization process, the system can continuously adjust the learning path and review strategy to adapt to the child's actual learning progress and results. Based on the "multi-sensory teaching method", the "recognition-practice-reading-writing-expansion" literacy process is designed. Children perceive words in multiple dimensions through multiple senses such as "seeing", "hearing", "hands", "mouth reading" and "brain thinking", helping them learn Chinese characters in an interesting and efficient way. The explanation of individual Chinese characters is combined with learning in context, and the composition of Chinese characters is presented in an animated way, which is in line with the concrete thinking characteristics of children of this age group. Mechanical literacy is rejected. In combination with specific context, children can remember the shape of characters and understand their use in sentences and paragraphs, laying a foundation for future expression and writing. You can also conduct situational learning through the playback of interesting animations, allowing children to learn Chinese characters by watching animations. There are a variety of interesting courses, no boring test-oriented education, all of which are interesting animation courses. In combination with the forgetting law of the forgetting curve, the frequency of word repetition is designed, and difficult words that are easy to make mistakes and confuse are reviewed and practiced in time when the memory weakens, to ensure the formation of long-term learning and memory. It also includes the Chinese character writing link, allowing children to learn and practice Chinese character writing by hand. Sensory Channel Type: This system records the types of sensory input channels used by children during learning (such as vision, hearing, and touch) and analyzes the impact of different sensory channels on learning. Based on each child's behavioral data and learning progress, the system intelligently adjusts learning paths, review schedules, and sensory input strategies to provide a personalized learning experience. This approach not only optimizes learning efficiency but also maximizes memory retention, ensuring that children review and consolidate learning at the optimal time, ultimately achieving optimal learning outcomes.

[0043] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A child early education enlightenment system, characterized in that: include: Learning data collection module, used to collect behavioral data generated by children during the learning process; An individual memory modeling module is used to construct an individual memory retention model for children based on the behavioral data and output memory decay parameters for subsequent learning scheduling; A mastery assessment module, a deep neural network model for processing learning history; A review schedule generation module, configured to automatically generate a personalized review schedule and review content set by combining the memory state parameters and the mastery level score results output by the memory retention model; The multi-sensory information optimization module is used to evaluate the learning effect of each sensory input channel during the learning process. By building a sensory input weight model, it adjusts the display method and input combination of subsequent learning materials; The learning path optimization module is used to construct a path optimization function using learning time and recognition errors as cost factors, and based on this function, it generates the optimal learning content presentation order and plans the learning path; The learning feedback closed-loop module is used to perform statistical analysis and feedback judgment on the learning results, and dynamically update the memory model, evaluation model and scheduling strategy according to the feedback results to form a closed loop.

2. The early childhood education enlightenment system according to claim 1, characterized in that: The behavioral data includes learning time, learning content identification, recognition accuracy, error frequency, recognition reaction time and the type of sensory input channel used. The individual memory modeling module constructs a memory model based on the exponential decay law, and fits the individual memory decay coefficient by minimizing the error between the recognition accuracy and the predicted memory amount, thereby establishing an individual memory retention curve that reflects the learning and forgetting speed.

3. The early childhood education enlightenment system according to claim 1, characterized in that: The retention model models the learning memory state by identifying the forgetting trend, and the mastery level assessment module uses a recurrent neural network or a long short-term memory network to construct a deep neural network structure, and uses learning behavior time series data as input to train the model to predict the current mastery probability of the learning content.

4. The early childhood education enlightenment system according to claim 1, characterized in that: The mastery probability output by the mastery level assessment module is used to drive the review scheduling generation module to sort learning content and calculate review time intervals. The system determines the review timing with the goal of minimizing predicted forgetting loss. The network model uses learning behavior data as input and outputs the current child's mastery level score for each learning content.

5. The early childhood education enlightenment system according to claim 1, characterized in that: The review schedule generation module adopts a Bayesian optimization strategy to select the review content and time points with the most learning value by constructing a joint expected utility function of the mastery level prediction value and model uncertainty.

6. The early childhood education enlightenment system according to claim 1, characterized in that: The multisensory information optimization module calculates the information entropy value of each channel based on the sensory channel recognition results and reaction time, and dynamically adjusts the weight parameters of the multisensory channels according to the entropy value to improve the information efficiency of sensory input.

7. The early childhood education enlightenment system according to claim 1, characterized in that: The learning path optimization module constructs a learning path search space based on the review task set output by the review schedule and the current mastery level score, and generates an optimal path solution through a dynamic programming algorithm using a weighted combination of time cost and error cost as the objective function.

8. The early childhood education enlightenment system according to claim 1, characterized in that: The learning path optimization module adaptively adjusts the time weight and error weight ratio according to historical learning performance and updates the learning path generation strategy in real time.

9. The early childhood education enlightenment system according to claim 1, characterized in that: The learning feedback closed-loop module uses hypothesis testing and variance analysis methods to evaluate the significance of learning results, and feeds back the statistical analysis results to the memory modeling module and review scheduling module for correcting individual model parameters and review rhythm.

10. A method for early childhood education and enlightenment, applied to a system for early childhood education and enlightenment according to any one of claims 1 to 9, characterized in that: The following steps are involved: First, collect behavioral data of children during learning activities, including learning time, recognition results, reaction time, and the type of sensory channels used; Then, an individual memory retention model is constructed based on the behavioral data, and memory decay parameters are extracted; Use a time series-based deep neural network model to predict and judge children’s mastery of various learning contents; Combine memory models with mastery level prediction results to generate personalized review schedules and review content sets; Then, based on the recognition performance and response efficiency of each sensory channel, the input value is evaluated to generate an optimized sensory input combination strategy; Construct a cost function with learning time and error rate as target factors, and use optimization algorithms to generate the optimal learning path; Finally, statistical feedback is given to the learning results, and the memory model parameters and review scheduling strategy are updated based on the feedback results to form a closed-loop optimization.

Images