Education system data dynamic optimization query method based on deep reinforcement learning

Abstract

The invention discloses an education system data dynamic optimization query method based on deep reinforcement learning. The method comprises the following steps: S1, constructing a structured feature vectorization data set; s2, establishing a fuzzy cognitive map; s3, forming a dynamic state space of the education system query logic; s4, comprehensively measuring a query logic optimization effect; s5, generating an education system data dynamic optimization query strategy; and S6, generating a query plan, and performing dynamic optimization query on the target data in the education system according to the query plan. And S7, outputting an education system data dynamic optimization query result, recording a performance index of the query result, and feeding back the recorded performance index to the fuzzy cognitive map and the deep reinforcement learning model to update a causal relationship weight of the fuzzy cognitive map and a query strategy of the reinforcement learning model. According to the method, the efficiency, the accuracy and the adaptivity of dynamic data query of the education system are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the field of educational technology, and in particular to a dynamic optimization query method for educational system data based on deep reinforcement learning. Background Art

[0002] With the rapid development of artificial intelligence technology, educational informatization and intelligence have become important directions of modern education. Dynamic optimization query of data in the education system plays a vital role in supporting personalized learning, teaching assistance and educational resource management. However, traditional education data query methods are powerless in the face of dynamically changing educational scenarios and are difficult to adapt to the needs of modern education for efficiency, accuracy and flexibility.

[0003] At present, most education systems use pre-defined static query logic for data retrieval, which usually relies on fixed database query statements or rules and has obvious limitations when processing diverse data such as student learning records, course resources, teaching plans and test results. On the one hand, static query logic lacks dynamic adjustment capabilities and cannot adapt to the real-time changes in students' learning status, the continuous adjustment of teachers' teaching needs and the dynamic update of course resources in the education system; on the other hand, it cannot be optimized according to the characteristics of data types and user needs, resulting in low query efficiency and low relevance of results, making it difficult to meet the needs of personalized learning and resource recommendation.

[0004] In recent years, some improved query methods have begun to try to introduce artificial intelligence technology, such as query optimization tools based on traditional machine learning algorithms. Traditional machine learning algorithms improve query efficiency by analyzing data features. However, their optimization capabilities are still limited by manually defined feature sets and fixed optimization rules. Specifically, this type of technology lacks adaptability when processing complex and diverse data, and cannot adjust the query logic in real time to adapt to dynamic changes in educational scenarios. In addition, existing methods are difficult to optimize in a targeted manner based on the specific needs of educational scenarios, which makes them show obvious deficiencies in practical applications.

[0005] In summary, the existing technology has the following main problems when processing dynamic optimization query of education system data:

[0006] 1. Lack of dynamic adjustment capabilities: Static query logic cannot adapt to real-time changes in educational scenarios, and query rules need to be manually rewritten, which has poor flexibility;

[0007] 2. Low query efficiency: Existing methods are difficult to optimize query efficiency for the diversity and complexity of educational data, resulting in slow retrieval speed;

[0008] 3. Insufficient relevance of results: Due to the inability to dynamically adjust query strategies based on user needs and data characteristics, existing technologies often have problems with query results not matching needs;

[0009] 4. Lack of specificity for educational scenarios: Existing query methods usually aim at general optimization and lack customized support for specific goals in educational scenarios.

[0010] The above problems directly limit the practicality and intelligence level of dynamic optimization query of education system data. There is an urgent need for a new method that can combine dynamic state space, optimization reward mechanism and adaptive learning ability to improve the efficiency, accuracy and flexibility of education system data query and promote the further development of education informatization. Summary of the invention

[0011] One object of the present invention is to propose a dynamic optimization query method for education system data based on deep reinforcement learning. The present invention significantly improves the efficiency, accuracy and adaptability of dynamic query of education system data.

[0012] A method for dynamically optimizing querying education system data based on deep reinforcement learning according to an embodiment of the present invention comprises the following steps:

[0013] S1. Collect multi-source data in the education system, pre-process the collected multi-source data, and construct a structured feature vector quantization data set;

[0014] S2. Based on the structured feature vectorized data set, determine the key factor nodes that affect the query logic, define the causal relationship and association weight between each factor node, and establish a fuzzy cognitive map;

[0015] S3. Using the fuzzy cognitive map as a reasoning tool, inputting the node state values ​​in the education system data, calculating the dynamic state values ​​of the key factor nodes through fuzzy reasoning based on causal relationships and weights, and forming the dynamic state space of the education system query logic;

[0016] S4. Design a reward function based on the query objectives of the education system. The reward function takes the query efficiency of the education system data, the relevance of the query results and the user satisfaction as the core indicators to comprehensively measure the query logic optimization effect;

[0017] S5. Based on the dynamic state space and reward function, a deep reinforcement learning model is used to train the query logic of the education system. The dynamic state space is used as the input of the reinforcement learning model. Based on the strategy optimization mechanism of the reinforcement learning model, a dynamic optimization query strategy for education system data is generated through continuous iterative updates;

[0018] S6. Apply the dynamic optimization query strategy of education system data generated by the deep reinforcement learning model to the query process of the education system, dynamically adjust the query logic, generate a query plan, and dynamically optimize the query of the target data in the education system according to the query plan;

[0019] S7. Dynamically optimize the query result output of the education system data, record the performance indicators of the query results, and feed the recorded performance indicators back to the fuzzy cognitive map and the deep reinforcement learning model to update the causal weights of the fuzzy cognitive map and the query strategy of the reinforcement learning model.

[0020] Optionally, the S1 specifically includes the following steps:

[0021] S11. Collect multi-source data in the education system and represent the collected multi-source data as the original data set D raw :

[0022] D raw ={D record ,D course ,D plan ,D exam};

[0023] Among them, D record Record data for students’ learning, including the time of students’ learning activities, the progress of completing tasks, and test scores. course is the course resource data, including course content, course difficulty and course learning objectives, plan The teaching plan data includes teaching schedule, teaching modules and teacher resource allocation. exam Test result data, including test scores, knowledge point coverage, and error analysis data;

[0024] S12. For the original data set D raw Remove redundant items and noise data from the data to generate a denoised data set, interpolate and complete the missing data in the denoised data set, use linear interpolation or mean filling to generate a completed data set, normalize all data features in the completed data set, and the normalized data set D norm ;

[0025] S13. Based on the normalized dataset D norm , extract key features and construct feature vectorization dataset D vector :

[0026] D vector = {v 1 ,v 2 ,v 3 ,…,v m};

[0027] Each feature vector is represented as:

[0028] v i =[f 1 ,f 2 ,f 3 ,…,f n ],i∈{1,2...,m};

[0029] Among them, v i is the feature vector of a single data sample, f 1 ,f 2 ,…,f n They are the extracted learning progress characteristics, course difficulty characteristics, teacher resource allocation intensity characteristics and exam coverage characteristics.

[0030] Optionally, the S2 specifically includes the following steps:

[0031] S21. Based on feature vectorization dataset D vector , determine the key factor nodes that affect the query logic in the education system. The key factor nodes include:

[0032] Student learning efficiency node N efficiency Describe the efficiency of students in completing learning tasks, based on learning progress feature extraction;

[0033] Course resource difficulty node N difficulty Describe the complexity of course resources, extracted based on course difficulty characteristics;

[0034] Student learning interest node N interest Describe the students' interest in the course, based on the comprehensive extraction of learning progress characteristics and course difficulty characteristics;

[0035] Teacher query demand node N demand Describe the target needs of teachers for data query, based on the extraction of teacher resource allocation intensity characteristics;

[0036] Represent a node collection as:

[0037] N={N efficiency ,N difficulty ,N interest ,N demand};

[0038] S22. Perform causal relationship analysis on each node in the key factor node N, and determine the causal relationship direction and association weight between the nodes based on the actual needs of the education system query logic:

[0039] Define slave node N i To Node N j The causal relationship Eij , if node N i For node N j If there is a positive promotion effect, the association weight w ij >0; if there is an inhibitory effect, then w ij <0; if there is no direct relationship, then w ij =0;

[0040] The causal relationships between nodes and their associated weights are constructed as an adjacency matrix W, where the matrix elements w ij Represents slave node N i To Node N j The weight relationship is:

[0041]

[0042] Where n is the total number of nodes, w ij Represents slave node N i To Node N j The causal weight of

[0043] S23. Based on the node set N and the adjacency matrix W, construct the fuzzy cognitive graph FCM:

[0044] FCM = (N, E, W);

[0045] The state value A of each node in the fuzzy cognitive graph i Initialized as feature vectorized dataset D vector The normalized value of the corresponding feature in .

[0046] Optionally, S3 specifically includes the following steps:

[0047] S31. Vectorize the features into dataset D based on fuzzy cognitive map FCM vector The normalized characteristic value in is used as the initial state value input of the node, and the node initial state vector A (0) It is expressed as:

[0048]

[0049] in, For node N i The initial state value of , n is the total number of nodes in the fuzzy cognitive graph;

[0050] S32. Adjacency matrix W for each node N based on fuzzy cognitive graph i Update the state value of the dynamic node of the fuzzy cognitive graph:

[0051]

[0052] in, Indicates that after the t+1th round of reasoning, node N i The status value of Indicates that after the tth round of reasoning, node N j The state value of , f(x) is the activation function;

[0053] S33. Based on the node state values ​​calculated by fuzzy reasoning, the dynamic state space of the query logic of the education system is constructed:

[0054] S dynamic ={S query ,S data ,S user};

[0055] Among them, S query is the current query demand state, provided by the state value of the teacher query demand node, S data is the characteristic state of the education system data, which is provided by the state values ​​of the student learning efficiency node, the course resource difficulty node, and the student learning interest node. user is the user behavior pattern state, which is the state value set A of all nodes in the fuzzy cognitive graph (t+1) Provides overall dynamic characteristics representation.

[0056] Optionally, the S4 specifically includes the following steps:

[0057] S41. Based on the actual needs of the education system, determine the query target, which includes three core indicators: education system data query efficiency, query result relevance and user satisfaction:

[0058] Query efficiency E efficiency , used to measure the degree of optimization of the time required for data query;

[0059] Query result relevance E relevance , used to measure the matching degree between the returned data and the query target;

[0060] User satisfaction satisfaction , used to measure the user's subjective satisfaction with the query results;

[0061] S42. Construct a reward function R based on the query target. The reward function integrates core indicators to measure the query logic optimization effect:

[0062] R=α 1 ·E efficiency +α 2 ·E relevance +α 3 ·E satisfaction -β·T;

[0063] Among them, α 1 ,α2 ,α 3 is a weight factor used to balance the importance of different indicators, T represents the time required for the query, and β is the time penalty coefficient used to avoid the negative impact of high time overhead on query efficiency.

[0064] Optionally, the query efficiency E efficiency Calculation:

[0065]

[0066] Query result relevance E relevance Calculation:

[0067]

[0068] Where n is the number of data items returned by the query, R i is the relevance score between the ith data and the query target, S i is the weight factor of the i-th data;

[0069] User satisfaction satisfaction Calculation:

[0070]

[0071] Among them, m is the total number of user feedback, U j is the user's satisfaction score for the j-th query result, and its value range is [0,1].

[0072] Optionally, the S5 specifically includes the following steps:

[0073] S51. The dynamic state space S constructed dynamic As the input state space S of the reinforcement learning model t ;

[0074] S52. Based on the state space S t , the reinforcement learning model outputs a set of actions A t , Action A t It represents the dynamic adjustment of query logic, optimization of data retrieval path, and generation of query plan in the query logic of the education system;

[0075] S53. Calculate the current action A based on the reward function R, combined with the actual query efficiency, query result relevance and user satisfaction feedback value t The corresponding instant reward R t , instant reward R t Used to evaluate the current action A t The optimization effect:

[0076] R t =α1 ·E efficiency +α 2 ·E relevance +α 3 ·E satisfaction -β·T;

[0077] S54. Use deep reinforcement learning algorithm to optimize the model's strategy and use strategy update:

[0078]

[0079] Among them, Q(S t ,A t ) is the current state space S t and action A t The value function is η, the learning rate, and γ, the discount factor, which is used to balance the importance of immediate rewards and future rewards. t+1 is the state at the next moment, A t+1 For the next moment’s action;

[0080] S55. The reinforcement learning model generates a dynamic optimization query strategy for the education system data based on the optimized strategy, and uses the optimized query strategy to guide the execution of the actual query logic. At the same time, it iteratively updates the strategy of the reinforcement learning model by continuously collecting dynamic data of the education system query environment.

[0081] The beneficial effects of the present invention are:

[0082] (1) The present invention introduces fuzzy cognitive graphs to construct the dynamic state space of the education system. Through the dynamic adjustment of causal weights and the fuzzy reasoning calculation of the state values ​​of key factor nodes, the dynamic characteristics of students' learning status, course resource changes and teachers' query needs can be reflected in real time. Compared with the traditional fixed query logic, the dynamic state space of the present invention can flexibly adapt to the real-time changing multi-dimensional data characteristics in the education scenario, effectively improving the adaptability of the query logic and avoiding the complexity of frequent manual adjustment of query rules.

[0083] (2) The present invention adopts a deep reinforcement learning model to take the dynamic state space as input, and comprehensively measures the query efficiency, result relevance and user satisfaction through a reward function to generate an optimal query strategy. Compared with the traditional rule-based query optimization method, the deep reinforcement learning model can improve the execution efficiency of the query path and the accuracy of the results through continuous strategy optimization iteration. When faced with complex data types and dynamic requirements, the reinforcement learning strategy can automatically adjust the query logic to achieve efficient and accurate data retrieval.

[0084] (3) The present invention feeds back the actual data of query efficiency, result relevance and user satisfaction to the fuzzy cognitive map and reinforcement learning model through the feedback mechanism of query result performance indicators, so as to update the causal weights of the fuzzy cognitive map and the strategy of the reinforcement learning model, thereby ensuring that the query logic can be optimized and adaptively adjusted in real time as the educational scenario and user needs change. Compared with the existing static optimization method, the feedback mechanism of the present invention significantly improves the dynamic optimization capability of the query model. Experimental data show that in the scenario of changes in educational resources and changes in user behavior, the method of the present invention can achieve significant strategy optimization within 3 iterations, and the query efficiency and user satisfaction are both improved by more than 15%. BRIEF DESCRIPTION OF THE DRAWINGS

[0085] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0086] Figure 1 This is a flow chart of a method for dynamic optimization query of education system data based on deep reinforcement learning proposed by the present invention. DETAILED DESCRIPTION

[0087] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, which only illustrate the basic structure of the present invention in a schematic manner, and therefore only show the components related to the present invention.

[0088] refer to Figure 1 , a dynamic optimization query method for education system data based on deep reinforcement learning, comprising the following steps:

[0089] S1. Collect multi-source data in the education system, pre-process the collected multi-source data, and construct a structured feature vector quantization data set;

[0090] S2. Based on the structured feature vectorized data set, determine the key factor nodes that affect the query logic, define the causal relationship and association weight between each factor node, and establish a fuzzy cognitive map;

[0091] S3. Using the fuzzy cognitive map as a reasoning tool, inputting the node state values ​​in the education system data, calculating the dynamic state values ​​of the key factor nodes through fuzzy reasoning based on causal relationships and weights, and forming the dynamic state space of the education system query logic;

[0092] S4. Design a reward function based on the query objectives of the education system. The reward function takes the query efficiency of the education system data, the relevance of the query results and the user satisfaction as the core indicators to comprehensively measure the query logic optimization effect;

[0093] S5. Based on the dynamic state space and reward function, a deep reinforcement learning model is used to train the query logic of the education system. The dynamic state space is used as the input of the reinforcement learning model. Based on the strategy optimization mechanism of the reinforcement learning model, a dynamic optimization query strategy for education system data is generated through continuous iterative updates;

[0094] S6. Apply the dynamic optimization query strategy of education system data generated by the deep reinforcement learning model to the query process of the education system, dynamically adjust the query logic, generate a query plan, and dynamically optimize the query of the target data in the education system according to the query plan;

[0095] S7. Dynamically optimize the query result output of the education system data, record the performance indicators of the query results, and feed the recorded performance indicators back to the fuzzy cognitive map and the deep reinforcement learning model to update the causal weights of the fuzzy cognitive map and the query strategy of the reinforcement learning model.

[0096] In this implementation, S1 specifically includes the following steps:

[0097] S11. Collect multi-source data in the education system and represent the collected multi-source data as the original data set D raw :

[0098] D raw ={D record ,D course ,D plan ,D exam};

[0099] Among them, D record Record data for students’ learning, including the time of students’ learning activities, the progress of completing tasks, and test scores. course is the course resource data, including course content, course difficulty and course learning objectives, plan The teaching plan data includes teaching schedule, teaching modules and teacher resource allocation. exam Test result data, including test scores, knowledge point coverage, and error analysis data;

[0100] S12. For the original data set D raw Remove redundant items and noise data from the data to generate a denoised data set, interpolate and complete the missing data in the denoised data set, use linear interpolation or mean filling to generate a completed data set, normalize all data features in the completed data set, and the normalized data set D norm ;

[0101] S13. Based on the normalized dataset D norm , extract key features and construct feature vectorization dataset Dvector :

[0102] D vector = {v 1 ,v 2 ,v 3 ,…,v m};

[0103] Each feature vector is represented as:

[0104] v i =[f 1 ,f 2 ,f 3 ,…,f n ],i∈{1,2...,m};

[0105] Among them, v i is the feature vector of a single data sample, f 1 ,f 2 ,…,f n They are the extracted learning progress characteristics, course difficulty characteristics, teacher resource allocation intensity characteristics and exam coverage characteristics.

[0106] In this implementation, S2 specifically includes the following steps:

[0107] S21. Based on feature vectorization dataset D vector , determine the key factor nodes that affect the query logic in the education system. The key factor nodes include:

[0108] Student learning efficiency node N efficiency Describe the efficiency of students in completing learning tasks, based on learning progress feature extraction;

[0109] Course resource difficulty node N difficulty Describe the complexity of course resources, extracted based on course difficulty characteristics;

[0110] Student learning interest node N interest Describe the students' interest in the course, based on the comprehensive extraction of learning progress characteristics and course difficulty characteristics;

[0111] Teacher query demand node N demand Describe the target needs of teachers for data query, based on the extraction of teacher resource allocation intensity characteristics;

[0112] Represent a node collection as:

[0113] N={N efficiency ,N difficulty ,N interest ,N demand};

[0114] S22. Perform causal relationship analysis on each node in the key factor node N, and determine the causal relationship direction and association weight between the nodes based on the actual needs of the education system query logic:

[0115] Define slave node N i To Node N j The causal relationship E ij , if node N i For node N j If there is a positive promotion effect, the association weight w ij >0; if there is an inhibitory effect, then w ij <0; if there is no direct relationship, then w ij =0;

[0116] The causal relationships between nodes and their associated weights are constructed as an adjacency matrix W, where the matrix elements w ij Represents slave node N i To Node N j The weight relationship is:

[0117]

[0118] Where n is the total number of nodes, w ij Represents slave node N i To Node N j The causal weight of

[0119] S23. Based on the node set N and the adjacency matrix W, construct the fuzzy cognitive graph FCM:

[0120] FCM = (N, E, W);

[0121] The state value A of each node in the fuzzy cognitive graph i Initialized as feature vectorized dataset D vector The normalized value of the corresponding feature in .

[0122] In this implementation, S3 specifically includes the following steps:

[0123] S31. Vectorize the features into dataset D based on fuzzy cognitive map FCM vector The normalized characteristic value in is used as the initial state value input of the node, and the node initial state vector A (0) It is expressed as:

[0124]

[0125] in, For node N i The initial state value of , n is the total number of nodes in the fuzzy cognitive graph;

[0126] S32. Adjacency matrix W for each node N based on fuzzy cognitive graph i Update the state value of the dynamic node of the fuzzy cognitive graph:

[0127]

[0128] in, Indicates that after the t+1th round of reasoning, node N i The status value of Indicates that after the tth round of reasoning, node N j The state value of , f(x) is the activation function;

[0129] S33. Based on the node state values ​​calculated by fuzzy reasoning, the dynamic state space of the query logic of the education system is constructed:

[0130] S dynamic ={S query ,S data ,S user};

[0131] Among them, S query is the current query demand state, provided by the state value of the teacher query demand node, S data is the characteristic state of the education system data, which is provided by the state values ​​of the student learning efficiency node, the course resource difficulty node, and the student learning interest node. user is the user behavior pattern state, which is the state value set A of all nodes in the fuzzy cognitive graph (t+1) Provides overall dynamic characteristics representation.

[0132] In this implementation, S4 specifically includes the following steps:

[0133] S41. Based on the actual needs of the education system, determine the query target, which includes three core indicators: education system data query efficiency, query result relevance and user satisfaction:

[0134] Query efficiency E efficiency , used to measure the degree of optimization of the time required for data query;

[0135] Query result relevance E relevance , used to measure the matching degree between the returned data and the query target;

[0136] User satisfaction satisfaction , used to measure the user's subjective satisfaction with the query results;

[0137] S42. Construct a reward function R based on the query target. The reward function integrates core indicators to measure the query logic optimization effect:

[0138] R=α1 ·E efficiency +α 2 ·E relevance +α 3 ·E satisfaction -β·T;

[0139] Among them, α 1 ,α 2 ,α 3 is a weight factor used to balance the importance of different indicators, T represents the time required for the query, and β is the time penalty coefficient used to avoid the negative impact of high time overhead on query efficiency.

[0140] In this implementation, the query efficiency E efficiency Calculation:

[0141]

[0142] Query result relevance E relevance Calculation:

[0143]

[0144] Where n is the number of data items returned by the query, R i is the relevance score between the ith data and the query target, S i is the weight factor of the i-th data;

[0145] User satisfaction satisfaction Calculation:

[0146]

[0147] Among them, m is the total number of user feedback, U j is the user's satisfaction score for the j-th query result, and its value range is [0,1].

[0148] In this implementation, S5 specifically includes the following steps:

[0149] S51. The dynamic state space S constructed dynamic As the input state space S of the reinforcement learning model t ;

[0150] S52. Based on the state space S t , the reinforcement learning model outputs a set of actions A t , Action A t It represents the dynamic adjustment of query logic, optimization of data retrieval path, and generation of query plan in the query logic of the education system;

[0151] S53. Calculate the current action A based on the reward function R, combined with the actual query efficiency, query result relevance and user satisfaction feedback value t The corresponding instant reward R t , instant reward R t Used to evaluate the current action A t The optimization effect:

[0152] R t =α 1 ·E efficiency +α 2 ·E relevance +α 3 ·E satisfaction -β·T;

[0153] S54. Use deep reinforcement learning algorithm to optimize the model's strategy and use strategy update:

[0154]

[0155] Among them, Q(S t ,A t ) is the current state space S t and action A t The value function is η, the learning rate, and γ, the discount factor, which is used to balance the importance of immediate rewards and future rewards. t+1 is the state at the next moment, A t+1 For the next moment’s action;

[0156] S55. The reinforcement learning model generates a dynamic optimization query strategy for the education system data based on the optimized strategy, and uses the optimized query strategy to guide the execution of the actual query logic. At the same time, it iteratively updates the strategy of the reinforcement learning model by continuously collecting dynamic data of the education system query environment.

[0157] Embodiment 1:

[0158] In the application of a city education management platform, the platform is responsible for serving the teaching and learning data management of more than 100 primary and secondary schools in the city. One day at noon, a teacher of a school needs to quickly query the learning efficiency of a class of students and recommend learning resources for low-efficiency students. At the same time, analyze the matching degree between the difficulty of a newly launched course resource and the learning interest of students. The following is a specific application scenario:

[0159] At 11:30 am on December 15, 2024, a class teacher at the Seventh Middle School in a certain city wanted to understand the learning efficiency of the students in Class 3 of Grade 8 and make personalized resource recommendations for students with persistently low task completion rates. At the same time, the director of the Academic Affairs Office planned to analyze the matching between the resource difficulty of an AI course that the school had just introduced and the students' learning interests. The platform used the method of the present invention to complete multiple queries within 30 minutes and provided an accurate analysis report.

[0160] During the operation of the platform, the system dynamically collects students' learning record data, course resource data, teaching plan data and examination result data, which involves the following detailed data:

[0161] The learning task completion records of 50 students in Class 3 of Grade 8 (the time span is the past two weeks), with a data volume of about 5GB;

[0162] The newly introduced AI course resource data includes course chapter division, resource difficulty level, and course goal description, with a data volume of about 2GB;

[0163] The student learning interest data comes from students’ recent course selection, exam performance, and related survey feedback, with a total amount of approximately 1GB.

[0164] Phase 1: Identifying Ineffective Students

[0165] The platform first uses fuzzy cognitive graphs to infer student learning efficiency nodes. In the past 14 days, the learning efficiency score of student A (ID: S20231215-01) was lower than the threshold of 0.6 (the score range is 0-1), and his task completion time was significantly higher than the average (about 2 hours, while the class average is 45 minutes), and the learning efficiency status value was judged to be abnormal.

[0166] Further analysis shows that the learning interest node status value of Student A is 0.4, far lower than the class average of 0.8, and his usage rate of intermediate and advanced course resources is only 15%. Based on dynamic query logic, the system automatically generates personalized learning resource recommendations, including 3 basic-level resources (in the embodiment: "AI Basic Tasks for Beginners", "Image Processing Introduction", "Computer Processing Introduction") and 2 intermediate-level resources related to his interests (in the embodiment: "Fun Robot Programming", "Fun Computing"). The result generation time is 12 seconds.

[0167] Phase 2: Analyze the difficulty of course resources and the matching degree of interest

[0168] For the newly launched AI course, the director of the Academic Affairs Office hopes to analyze the match between students’ interest in the course and the difficulty of the course resources. The platform dynamically queries the data of eighth-grade students in the school and finds that:

[0169] The course resource difficulty score is 0.85, which is a relatively high difficulty level;

[0170] Among the 1,200 students in grade eight, 400 had an interest status value higher than 0.7, accounting for about 33%.

[0171] Further matching analysis shows that students with an interest status value higher than 0.7 have a task completion rate of 95% in the learning course, while students with an interest status value lower than 0.5 have a completion rate of only 48%. The system automatically generates a matching analysis report, and the report generation takes about 18 seconds.

[0172] Phase 3: Generating performance indicators and optimizing feedback

[0173] After completing the query task, the system automatically records the query performance indicators, including:

[0174] Query efficiency (average response time): 15 seconds;

[0175] Query result relevance (system evaluation relevance score): 92%;

[0176] User satisfaction rating: 89%.

[0177] The platform feeds the above performance indicators into the fuzzy cognitive graph and reinforcement learning model:

[0178] 1. The system dynamically adjusted the causal relationship weights of the "learning efficiency node" and the "learning interest node" in the fuzzy cognitive graph from the original weight of 0.7 to 0.8 to adapt to the actual situation that students' learning efficiency and interest are strongly correlated;

[0179] 2. The reinforcement learning model updates the query strategy based on immediate rewards. In subsequent similar queries, the query logic is dynamically adjusted to improve the execution efficiency of the query path.

[0180] To verify the effectiveness of the present invention, the platform used both the traditional method and the method of the present invention to conduct comparative tests on the above scenarios, and the results are shown in Table 1 below:

[0181] Table 1 Comparative test data of the traditional method and the method of the present invention for the above scenarios

[0182]

[0183] This embodiment demonstrates the application effect of the method of the present invention in actual educational scenarios. By dynamically adjusting the query logic and optimizing the query strategy, efficient and accurate query results are achieved, which significantly improves the query efficiency and user satisfaction, and provides an efficient solution for dynamic optimization query of education system data.

[0184] The present invention introduces fuzzy cognitive maps to construct the dynamic state space of the education system. Through the dynamic adjustment of causal weights and the fuzzy reasoning calculation of the state values ​​of key factor nodes, it can reflect the dynamic characteristics of students' learning status, course resource changes and teachers' query needs in real time. Compared with traditional fixed query logic, the dynamic state space of the present invention can flexibly adapt to the real-time changing multi-dimensional data characteristics in educational scenarios, effectively improves the adaptability of query logic, and avoids the complexity of frequent manual adjustment of query rules.

[0185] The present invention adopts a deep reinforcement learning model to take the dynamic state space as input, and comprehensively measures the query efficiency, result relevance and user satisfaction through a reward function to generate the optimal query strategy. Compared with the traditional rule-based query optimization method, the deep reinforcement learning model can improve the execution efficiency of the query path and the accuracy of the results through continuous strategy optimization iteration. When faced with complex data types and dynamic requirements, the reinforcement learning strategy can automatically adjust the query logic to achieve efficient and accurate data retrieval.

[0186] The present invention feeds back actual data of query efficiency, result relevance and user satisfaction to the fuzzy cognitive map and reinforcement learning model through the feedback mechanism of query result performance indicators, so as to update the causal weights of the fuzzy cognitive map and the strategy of the reinforcement learning model, thereby ensuring that the query logic can be optimized and adaptively adjusted in real time as the educational scenario and user needs change. Compared with the existing static optimization method, the feedback mechanism of the present invention significantly improves the dynamic optimization capability of the query model. Experimental data show that in the scenario of changes in educational resources and changes in user behavior, the method of the present invention can achieve significant strategy optimization within 3 iterations, and the query efficiency and user satisfaction are both improved by more than 15%.

[0187] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A dynamic optimization query method for education system data based on deep reinforcement learning, characterized in that: The steps include: S1. Collect multi-source data in the education system, pre-process the collected multi-source data, and construct a structured feature vector quantization data set; S2. Based on the structured feature vectorized data set, determine the key factor nodes that affect the query logic, define the causal relationship and association weight between each factor node, and establish a fuzzy cognitive map; S3. Using the fuzzy cognitive map as a reasoning tool, inputting the node state values ​​in the education system data, calculating the dynamic state values ​​of the key factor nodes through fuzzy reasoning based on causal relationships and weights, and forming the dynamic state space of the education system query logic; S4. Design a reward function based on the query objectives of the education system. The reward function takes the query efficiency of the education system data, the relevance of the query results and the user satisfaction as the core indicators to comprehensively measure the query logic optimization effect; S5. Based on the dynamic state space and reward function, a deep reinforcement learning model is used to train the query logic of the education system. The dynamic state space is used as the input of the reinforcement learning model. Based on the strategy optimization mechanism of the reinforcement learning model, a dynamic optimization query strategy for education system data is generated through continuous iterative updates; S6. Apply the dynamic optimization query strategy of education system data generated by the deep reinforcement learning model to the query process of the education system, dynamically adjust the query logic, generate a query plan, and dynamically optimize the query of the target data in the education system according to the query plan; S7. Dynamically optimize the query result output of the education system data, record the performance indicators of the query results, and feed the recorded performance indicators back to the fuzzy cognitive map and the deep reinforcement learning model to update the causal weights of the fuzzy cognitive map and the query strategy of the reinforcement learning model.

2. According to claim 1, a method for dynamic optimization query of education system data based on deep reinforcement learning is characterized in that: The S1 specifically includes the following steps: S11. Collect multi-source data in the education system and represent the collected multi-source data as the original data set D raw : D raw ={D record ,D course ,D plan ,D exam }; Among them, D record Record data for students’ learning, including the time of students’ learning activities, the progress of completing tasks, and test scores. course is the course resource data, including course content, course difficulty and course learning objectives, plan The teaching plan data includes teaching schedule, teaching modules and teacher resource allocation. exam Test result data, including test scores, knowledge point coverage, and error analysis data; S12. For the original data set D raw Remove redundant items and noise data from the data to generate a denoised data set, interpolate and complete the missing data in the denoised data set, use linear interpolation or mean filling to generate a completed data set, normalize all data features in the completed data set, and the normalized data set D norm ; S13. Based on the normalized dataset D norm , extract key features and construct feature vectorization dataset D vector : D vector ={v1,v2,v3,…,v m }; Each feature vector is represented as: <h2 style=";text-align:left;direction:ltr">v<h2 style=";text-align:left;direction:ltr"> i <h2 style=";text-align:left;direction:ltr"> (f1,f2,f3,…,f)<h2 style=";text-align:left;direction:ltr"> n <h2 style=";text-align:left;direction:ltr"> ],i∈{1,2...,m}; Among them, v i is the feature vector of a single data sample, f1,f2,…,f n They are the extracted learning progress characteristics, course difficulty characteristics, teacher resource allocation intensity characteristics and exam coverage characteristics.

3. According to claim 1, a method for dynamic optimization query of education system data based on deep reinforcement learning is characterized in that: The S2 specifically includes the following steps: S21. Based on feature vectorization dataset D vector , determine the key factor nodes that affect the query logic in the education system. The key factor nodes include: Student learning efficiency node N efficiency Describe the efficiency of students in completing learning tasks, based on learning progress feature extraction; Course resource difficulty node N difficulty Describe the complexity of course resources, extracted based on course difficulty characteristics; Student learning interest node N interest Describe the students' interest in the course, based on the comprehensive extraction of learning progress characteristics and course difficulty characteristics; Teacher query demand node N demand Describe the target needs of teachers for data query, based on the extraction of teacher resource allocation intensity characteristics; Represent a node collection as: N={N efficiency ,N difficulty ,N interest ,N demand }; S22. Perform causal relationship analysis on each node in the key factor node N, and determine the causal relationship direction and association weight between the nodes based on the actual needs of the education system query logic: Define slave node N i To Node N j The causal relationship E ij , if node N i For node N j If there is a positive promotion effect, the association weight w ij >0; if there is an inhibitory effect, then w ij <0; if there is no direct relationship, then w ij =0; The causal relationships between nodes and their associated weights are constructed as an adjacency matrix W, where the matrix elements w ij Represents slave node N i To Node N j The weight relationship is: Where n is the total number of nodes, w ij Represents slave node N i To Node N j The causal weight of S23. Based on the node set N and the adjacency matrix W, construct the fuzzy cognitive graph FCM: FCM = (N, E, W); The state value A of each node in the fuzzy cognitive graph i Initialize the feature vectorized dataset D vector The normalized value of the corresponding feature in .

4. According to claim 1, a method for dynamic optimization query of education system data based on deep reinforcement learning is characterized in that: The S3 specifically includes the following steps: S31. Vectorize the features into dataset D based on fuzzy cognitive map FCM vector The normalized characteristic value in is used as the initial state value input of the node, and the node initial state vector A (0) It is expressed as: in, For node N i The initial state value of , n is the total number of nodes in the fuzzy cognitive graph; S32. Adjacency matrix W for each node N based on fuzzy cognitive graph i Update the state value of the dynamic node of the fuzzy cognitive graph: in, Indicates that after the t+1th round of reasoning, node N i The status value of Indicates that after the tth round of reasoning, node N j The state value of , f(x) is the activation function; S33. Based on the node state values ​​calculated by fuzzy reasoning, the dynamic state space of the query logic of the education system is constructed: S dynamic ={S query ,S data ,S user }; Among them, S query is the current query demand state, provided by the state value of the teacher query demand node, S data is the characteristic state of the education system data, which is provided by the state values ​​of the student learning efficiency node, the course resource difficulty node, and the student learning interest node. user is the user behavior pattern state, which is the state value set A of all nodes in the fuzzy cognitive graph (t+1) Provides overall dynamic characteristics representation.

5. According to claim 1, a method for dynamic optimization query of education system data based on deep reinforcement learning is characterized in that: The S4 specifically comprises the following steps: S41. Based on the actual needs of the education system, determine the query target, which includes three core indicators: education system data query efficiency, query result relevance and user satisfaction: Query efficiency E efficiency , used to measure the degree of optimization of the time required for data query; Query result relevance E relevance , used to measure the matching degree between the returned data and the query target; User satisfaction satisfaction , used to measure the user's subjective satisfaction with the query results; S42. Construct a reward function R based on the query target. The reward function integrates core indicators to measure the query logic optimization effect: R=α1·E efficiency +α2·E relevance +α3·E satisfaction -β·T; Among them, α1, α2, and α3 are weight factors used to balance the importance of different indicators, T represents the time required for the query, and β is the time penalty coefficient used to avoid the negative impact of high time overhead on query efficiency.

6. According to claim 5, a method for dynamic optimization query of education system data based on deep reinforcement learning is characterized in that: The query efficiency E efficiency Calculation: Query result relevance E relevance Calculation: Where n is the number of data items returned by the query, R i is the relevance score between the ith data and the query target, S i is the weight factor of the i-th data; User satisfaction satisfaction Calculation: Among them, m is the total number of user feedback, U j is the user's satisfaction score for the j-th query result, and its value range is [0,1].

7. According to claim 1, a method for dynamic optimization query of education system data based on deep reinforcement learning is characterized in that: The S5 specifically includes the following steps: S51. The dynamic state space S constructed dynamic As the input state space S of the reinforcement learning model t ; S52. Based on the state space S t , the reinforcement learning model outputs a set of actions A t , Action A t It represents the dynamic adjustment of query logic, optimization of data retrieval path, and generation of query plan in the query logic of the education system; S53. Calculate the current action A based on the reward function R, combined with the actual query efficiency, query result relevance and user satisfaction feedback value t The corresponding instant reward R t , instant reward R t Used to evaluate the current action A t The optimization effect: R t =α1·E efficiency +α2·E relevance +α3·E satisfaction -β·T; S54. Use deep reinforcement learning algorithm to optimize the model's strategy and use strategy update: Among them, Q(S t ,A t ) is the current state space S t and action A t The value function is η, the learning rate, and γ, the discount factor, which is used to balance the importance of immediate rewards and future rewards. t+1 is the state at the next moment, A t+1 For the next moment’s action; S55. The reinforcement learning model generates a dynamic optimization query strategy for the education system data based on the optimized strategy, and uses the optimized query strategy to guide the execution of the actual query logic. At the same time, it iteratively updates the strategy of the reinforcement learning model by continuously collecting dynamic data of the education system query environment.

Images